This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
& 297 & you know or something an’-298 so they leave, (131A, NS)
. Discourse markers and second language acquisition During the last two decades, analyses of discourse markers (a term which, in its widest definition, includes what has also been termed ‘discourse connective’, ‘discourse particle’, etc.) have occupied a large space in the literature on pragmatics. Discourse markers have been considered from a variety of perspectives and approaches, e.g. as signaling “a sequential relationship” between utterances (Fraser 1990; Fraser 1999), as marking discourse coherence (Schiffrin 1987; Lenk 1998), and from a relevance-theoretic point of view (Andersen 2001; Blakemore 2002; Blass 1990; Jucker 1993), to name just a few; they have been analyzed with regard to gender (Erman 1992; Holmes 1986) and age (Kyratzis and Ervin-Tripp 1999; Andersen 2001; Erman 2001), and in bilingual contexts (Goss and Salmons 2000; Maschler 2000; Matras 2000); they have been analyzed as a group and have been treated individually. There is a general agreement that discourse markers contribute to the pragmatic meaning of utterances and thus play an important role in the pragmatic competence of the speaker. Or, as Crystal comments in more everyday language, “I tend to think of [pragmatic expressions such as you know] as the oil which helps us perform the complex task of spontaneous speech production and interaction smoothly and efficiently” (Crystal 1988: 48). Most of these studies focus on native (or bilingual) speakers of English, who acquire this pragmatic competence in their childhood. How, then, do non-native speakers learn how to use so? Or well? Or like? In the literature on second language acquisition (SLA), the field of discourse markers has been largely omitted so far (cf. Ellis 1987; Ellis 1994; Bardovi-Harlig 2002; etc.), even though pragmatic competence in terms of knowing the cultural values of the second language, for example, is recognized as being essential for successful communication (House and Kasper 1981; Bialystok and Hakuta 1994: 160ff.; Kasper and Blum-Kulka 1993b; Bardovi-Harlig 1996; research on interlanguage pragmatics in general). Svartvik (1980: 171) illustrates this essentiality thus:
Discourse markers in English discourse
if a foreign language learner says five sheeps or he goed, he can be corrected by practically every native speaker. If, on the other hand, he omits a well, the likely reaction will be that he is dogmatic, impolite, boring, awkward to talk to etc, but a native speaker cannot pinpoint an ‘error’.
My assumption here is that, even if they do not attain a native-like competence (cf. Wilkins 1999: 14), most learners of English will aim at avoiding such judgments of their linguistic behavior. The general research question of this study then is how non-native speakers of English master the use of discourse markers. The structure of this book is as follows: The next two sections situate the present study in the existing literature, both on discourse markers and on second language acquisition / applied linguistics. Section 1.4 specifies the goals and research questions guiding the research, followed by a basic description of my approach. It also deals with my use of the term ‘discourse markers’ and the selection of markers to be analyzed in this study. Section 1.5 will introduce the reader to the corpus on which this research was carried out, the Giessen-Long Beach Chaplin Corpus. A descriptions of the material used follows. Section 1.7 discusses factors which have been identified as potential influences on the use and frequency of discourse markers. The last section finally presents the speakers and their distribution according to several of the factors introduced in Section 1.7. Chapters 2–5 are each dedicated to one discourse marker and form the main part of the book. Each chapter consists of two parts: First, I describe and illustrate the functions of the respective marker as found in the corpus. The second part then presents frequencies and statistical evidence in a variety of ways. This study focuses on the categorization and description of discourse marker use in native and non-native English. In addition, the last chapter tackles the question of which influential factors for the linguistic behavior of the German speakers can be identified. Information on some of these was elicited in a questionnaire, while other potential factors came up in the course of my investigations. In this last chapter I present and interpret statistical relationships between my results and those factors covered by the questionnaire and make reasonable guesses at the influence of other potential factors. Since we have little hard evidence for the actual influence of such factors yet, the last chapter should be taken as an incentive for further investigations on this matter.
. Discourse markers in the literature The beginning of discourse markers research can probably be seen in Robin Lakoff ’s early article entitled Questionable answers and answerable questions, in which she systematizes her observation that why and well at the beginning of answers can be used only under certain conditions (Lakoff 1973). Since then, a myriad of papers and
Chapter 1. Introduction
books have been written on individual markers and on the question of what discourse markers are.
.. Terminology When I use the term ‘discourse markers’ (DM), I am following Schourup (1999: 228), who states that “[t]he term DM used in this review is merely the most popular of a host of competing terms used with partially overlapping reference”. Schourup himself in an earlier work (1985) used the term ‘discourse particles’, which has recently also been used by Aijmer (2002). It is more commonly employed by researchers who (also) work on other languages such as German (e.g. Weydt 1969; Weydt and Ehlers 1987; Abraham 1991; Kryk 1992; Fischer and Drescher 1996), French (e.g. Mosegaard Hansen 1998), or Chinese (e.g. Lee-Wong 2001). One of Schourup’s own objections to the use of this term, however, lies in the fact that “ ‘particle’ has traditionally been a syntactic term” (1999: 229) which as such cannot cover the variety of syntactic classes of the linguistic items commonly considered to be discourse markers. Other authors prefer the term ‘connective’ (e.g. Fraser 1988; Bazzanella 1990; Lamiroy 1994; Unger 1996; Degand 2000). Another fairly frequent term is ‘pragmatic markers’ (Watts 1988; Redeker 1990; Caron-Prague and Caron 1991; Brinton 1996; Andersen 1998; Brinton 1998; Erman 2001). Andersen justifies his choice by saying that “the label ‘pragmatic’ is meant to suggest a relatively low degree of lexical specificity and a high degree of context-sensitivity” (Andersen 2001); he discards the term ‘discourse marker’ to avoid a confusion with Fraser’s account, in which discourse markers are seen as a subtype of pragmatic markers signaling “a sequential relationship between the current basic message and the previous discourse” (Fraser 1990: 383; Fraser 1996). This is a function Andersen refers to as the textual function of pragmatic markers. Lenk distinguishes between the two terms in a similar fashion: Studies that investigate pragmatic markers often focus more on the interactional aspects between the participants that are expressed through the use of particles. One of the most prominent functions of discourse markers, however, is to signal the kinds of relations a speaker perceives between different part of the discourse. (Lenk 1997: 2)
We will see that the markers analyzed in the present research show both textual and interactional functions, so that both terms would be justified. Following Schiffrin (1987), whose work has been most influential for the present study, as well as Schourup (1999) and Jucker and Ziv (1998), I will therefore use the term ‘discourse marker’. In spite of the various justifications given in the literature for the choice of a particular term, there is no agreement as to which linguistic items are to be considered discourse markers or pragmatic particles or whichever term is used. Suggestions for English range from multi-word expressions such as to return to my original point (Fraser 1988) to well and like (e.g. Watts 1989; Jucker 1993; Miller and Weinert 1995; Hasund 2002), or and but (Schiffrin 1987; Abraham 1979; Lauerbach 1993), oh and
Discourse markers in English discourse
mhm (e.g. Fox Tree and Schrock 1999; Jucker and Smith 1998). Commonly assumed features of discourse markers will be dealt with later in this chapter. Descriptions of other scholars’ results for individual markers will follow in the relevant chapters.
.. Properties of discourse markers For some authors of early publications on discourse or pragmatic markers or particles, it did not seem important yet to worry about a delimitation of what kind of linguistic elements should be subsumed under one or the other of these terms. Schourup (1985: 1), for example, commences his dissertation by declaring, “This is a study of several common items in English conversation known variously as ‘discourse particles’, ‘interjections’, ‘discourse markers’, and less respectfully as ‘hesitations’ or ‘fillers’.” No further details are given on any characteristics of these particles. Likewise, such a description is lacking in Östman (1981), who simply lists pragmatic markers as one type of “pragmatic devices” (p. 5) and gives some examples. Redeker also chooses not to define the class of discourse markers by means of syntactic features; instead, she insists on a functional definition (1990: 371f., 1991). Svartvik (1980: 168) in his treatment of well at least observes that there is “little agreement as to the function or word-class status of well” and that “lexicographers find it peculiarly awkward to define the particle well”. Several features of discourse markers are addressed by Schiffrin. Regarding syntactic position and grammaticality, she claims: Although markers often precede sentences, [...] they are independent of sentential structure. Removal of a marker from its sentence initial position, in other words, leaves the sentence structure intact. Furthermore, several markers – y’know, I mean, oh, like – can occur quite freely within a sentence at locations which are very difficult to define syntactically. (1987: 31f.)
The quotation actually contains two features: syntactic independence and place of occurrence in relation to the sentence structure. While she holds up syntactic independence as a feature of discourse markers in an attempt to delimitate them, she settles on an utterance-initial position as the common place of occurrence for discourse markers in her last chapter (1987: 328). Other characteristics she mentions are phonological reduction and no or vague meaning (1987: 328). In searching the literature for definitions of discourse markers, we will find that few of the features listed are defining criteria: Non-compliance with one of them will rarely lead to an exclusion of the linguistic item from the group of discourse markers. Rather, these features are descriptive of the group of linguistic items the respective author has in mind; in addition, they are supplied with a range of hedges, tentative expressions and modifying adverbs such as “tends to”, “typically”, “mostly”, etc. Let us take a look at the individual features:
Chapter 1. Introduction
... No single word class The difficulty of placing discourse markers within a traditional word class mentioned by Svartvik (1980: 168) is a feature which Fraser supports in several publications (e.g. 1988, 1990, 1999). It is not often discussed, but not questioned either (cf. Jucker and Ziv 1998b). ... Phonological features Schiffrin suggests that discourse markers “have a range of prosodic contours, e.g. tonic stress and followed by a pause, phonological reduction” (1987: 328). However, even though phonological reduction is cited by both Brinton (1996: 33) and by Jucker and Ziv (1998b: 3), it does not seem to have found many supporters. Sankoff et al. (1997: 197) put it tentatively, claiming that discourse markers “undergo greater phonological reduction than their source forms”. Erman (1992: 219) only mentions that “you know tends to form a separate tone unit”. The majority of researchers in this area do not dwell on phonological features for a definition of discourse markers. ... Syntactic position Regarding their positioning within the discourse, we find that often markers are supposed to be at the beginning of a discourse unit. However, there is not as much agreement as to what this discourse unit might be. Brinton claims that “[i]t is often said that pragmatic markers are restricted to sentence-initial position, or may always occur sentence initially” (1996: 33) with reference to Keller (1979) and to Quirk et al. (1985) (in Brinton 1990: 46). This claim is also made by Fraser (1988: 24) and Bell (1998: 515) and listed in Jucker and Ziv (1998b). For Schiffrin (1987), Fraser (1990: 389), and Biber et al., in contrast, the relevant discourse unit in this respect is not the sentence but the utterance or turn: “Discourse markers [...] are inserts which tend to occur at the beginning of a turn or utterance” (Biber et al. 1999: 1086). However, Erman found that her pragmatic expressions, including I mean and you know, which are treated as discourse markers by Schiffrin, were most frequent in the middle of an utterance or turn (1986: 132, 1987: 47). Moreover, they also occurred in the middle of sentences: “the typical syntactic position of PEs [i.e. pragmatic expressions, SM] is between clauses, although they frequently also appear within the boundaries of the clause” (Erman 1992: 219). Brinton (1996) also admits that markers have been frequently found in sentence medial and final positions in various studies. In addition, DuBois et al. (1992) report that they also occur in separate tone units (cited in Lenk 1998: 51). ... Syntactic independence and grammatical optionality There have been different formulations to describe syntactic independence. Fraser (1988: 22) says that “the absence of the discourse marker does not render a sentence ungrammatical and/or unintelligible” (cf. also the quote from Schiffrin 1987 above). That is, even if the marker occurs within a sentence, it is not tied to the sentence structure but remains “ ‘outside’ the syntactic structure” (Erman 2001: 1339, cf. also Bell 1998: 515; Brinton 1996: 34; Jucker and Ziv 1998b: 3). As Sankoff et al. put it, it
Discourse markers in English discourse
does not “enter into construction syntactically with other elements of the sentence” (1997: 195). What Fraser’s (and also Schiffrin’s) formulation also implies is the grammatical optionality of the marker, listed by Brinton and Jucker and Ziv as separate characteristics. It is important here, though, to emphasize that this optionality only concerns grammatical well-formedness of the relevant sentence, and not its pragmatic impact. Optionality seems to be a widely accepted feature of English discourse markers. It is the only one, in my opinion, which can be used to distinguish, for example, between discourse markers and their non-discourse marker homonyms (e.g. well as an adverb, you know in questions).
... Lack of semantic content Jucker and Ziv (1998: 4) consider the lack of semantic content to be a diagnostic characteristic of discourse markers, along with syntactic and phonological features. “Lack” in this case is not meant to imply a complete absence of meaning. Repeatedly, we find expressions such as markers having “no meaning [or] a vague meaning” (Schiffrin 1987: 328), “little or no meaning in themselves” (Erman 2001: 1339), “no apparent meaning” (Romero Trillo 2002: 774), “relatively little semantic content” (Simon-Vandenbergen 2001: 82), or containing “a residue of semantic meaning” (Ariel 1994: 3254, cf. also Blackwell 2000: 6).1 In addition to or instead of talking about the meaning of the marker itself, researchers also claim that markers do not add to the informational or propositional content of the utterance which contains the marker (Erman 1986: 131; Hölker 1991: 78f.; Brinton 1996: 34 citing Östman 1982; Sankoff et al. 1997: 197; cf. also Mosegaard Hansen 1998: 73). However, Andersen warns that non-propositionality is only partly a valid criterion, because some pragmatic markers can be seen to have truth-conditional implications. [...] I argue that some pragmatic markers affect the propositional meaning of utterances, though not necessarily as conceptual constituents of propositions but as constraints on the interpretational procedure. (2001: 40f.)
Thus, using non-propositionality as a defining criterion for discourse markers presupposes a clear definition and delimitation of propositional meaning of sentences or utterances.
... Orality The distinction between the oral and the written medium has been rejected as a defining criterion by Jucker and Ziv (1998b) as well as by Schourup. According to Schourup (1999: 234), most discourse marker studies are still based on speech data, and “[m]ost forms claimed to be DMs occur primarily in speech”. Thus, it seems that few researchers (with the marked exception of Fraser, cf. Fraser 1988, 1990, 1996, 1998, 1999) are interested in linguistic items which could be called discourse markers on other grounds, but which occur mainly in written language. For example, although Schiffrin does not list orality as one of the features of discourse markers, it is obvious from her definition of discourse markers as “contextual coordinates of talk” that she does not
Chapter 1. Introduction
have in mind elements which occur (exclusively or mainly) in written language. Erman (1986: 131, 2001: 1339) even claims that pragmatic markers are not only “abundant in spoken language” but “all restricted to spoken language”. This view would not necessarily be shared by Schiffrin, since she includes items such as and, but and or, which all occur in written texts as well. However, few researchers try to find “principled grounds [...] on which to deny DM status to similar items that are largely found in written discourse” (Schourup 1999: 234). The reason for this fact might simply be that most scholars work the other way round: There are certain linguistic elements which they find intriguing and which seem to fit with one of the definitions of discourse markers or have been called so by other scholars; they are not primarily interested in delimiting these elements against other potential discourse markers. In addition, Watts considers discourse marker use “one of the most perceptually salient features of oral style” (1989: 208). McCarthy, who looked at markers in written texts, explicitly speaks of “spoken discourse markers in written texts” and argues that “they play a major role in our judgement of the degree of spokenness present in the text” (1993: 180). Both Watts and McCarthy imply that discourse markers are frequent in spoken language and are a salient feature even to the lay person’s ear. This might contribute to the lack of interest in primarily written markers among many researchers. Moreover, items such as those suggested by Schourup (moreover, consequently, contrariwise, 1999: 234) arguably have more semantic content than, for example, well and are more restricted in terms of the contexts in which they can be used; therefore, they might also be less interesting. To summarize, we can say that even though orality may not be a defining feature, it describes a frequent condition of the linguistic items many academics treat as discourse or pragmatic markers.
... Multifunctionality About multifunctionality, Jucker and Ziv write: Whether a specific linguistic element is monofunctional or polyfunctional is not a useful criterion in deciding whether it is a discourse marker or not because of the obvious analytical vicious circularity it entails. Many studies actually set out to argue explicitly for the monofunctionality or polyfunctionality of specific markers, thus nullifying this as a valid criterion. (Jucker and Ziv 1998b: 4)
In contrast to their opinion, Lenk (1998: 50) distinguishes phrases like “summing up” from discourse markers because such phrases always have the same function, implying that discourse markers do not. She is not the only one who takes it for granted that markers have multiple functions; Erman (2001: 1338) claims that “[t]he importance and multifunctionality of pragmatic markers in everyday conversation is not a controversial issue”. Likewise, Aijmer talks about multiple functions quite naturally and states: “Discourse particles are different from ordinary words in the language because of the large number of pragmatic values that they can be associated with. Nevertheless speakers are not troubled by this multifunctionality” (Aijmer 2002: 3). Even though
Discourse markers in English discourse
Jucker and Ziv may be correct in rejecting it as a defining criterion, research has shown that generally the discourse markers under scrutiny fulfill more than one function or at least have sub-functions (cf. Fischer 2000; see also the discussion in Section 1.2.5). It is therefore acceptable if multifunctionality is assumed at least for those markers for which this feature has been convincingly shown.
.. Functions of discourse markers In Chapters 2–5, I will describe the functions of so, well, you know and like in detail and compare them to the functions other researchers have identified for these four markers. Therefore, I will here present only a short summary of how the functions of (groups of) discourse markers have been described in the literature. Most researchers agree that the use of discourse markers facilitates the hearer’s task of understanding the speaker’s utterances. As Aijmer (1996: 210) puts it, they “function as cues or guides to the hearer’s interpretation” (cf. also Ariel 1998: 223). There are two basic frameworks which look at discourse markers from different perspectives but eventually come to very similar conclusions: Coherence and Relevance Theory. Risselada and Spooren (1998: 131) still claim that most approaches to discourse markers are based on coherence (cf. for example Schiffrin 1985; Holmes 1986; Redeker 1990; Lenk 1995; Mosegaard Hansen 1998; Takahara 1999; Lee-Wong 2001; Degand and Sanders 2002; Fuller 2003); there are also a number of studies using Relevance Theory (cf. for example Blakemore 1988; Jucker 1988; Blass 1990; Helt and Foster-Cohen 1996; Unger 1996; Ariel 1998; Andersen 2001). While the coherence framework is said to focus more on the textual functions (e.g. Andersen et al. 1999: 1339), Relevance Theory focuses on cognitive processes (see Sperber and Wilson 1986). Thus, Schiffrin describes the contribution of discourse to coherence as follows: “discourse markers provide contextual coordinates for utterances: they index an utterance to the local contexts in which utterances are produced and in which they are to be interpreted” (1987: 326f.). Lenk (1995: 341) uses a more “everyday notion of coherence” as defined in the Collins Cobuild Dictionary: “If something is coherent, its parts fit together well so that it is clear and easy to understand”. According to Lenk, discourse markers tend to be used when “the speaker feels a need to verbally express how it ‘fits together well”’ (cf. also Aijmer 2002: 3), although, as Mosegaard-Hansen observed, markers such as by the way “may indicate rather that the host utterance is not intended to cohere, or is at best intended to cohere in a rather loose fashion with what preceded it” (1998: 197). As Schiffrin (1987: 318) also suggests, the interpretation process of the hearer is guided by the use of markers because “markers select a meaning relation from whatever potential meanings are provided through the content of talk, and display that relation”. A similar notion is found in Relevance Theory. To put it in very simplified words, the general idea is that the linguistic form of a sentence or an utterance (i.e. propositional representations) potentially gives rise to a number of possible interpretations. The hearer’s task then is to find the most relevant interpretation in the
Chapter 1. Introduction
given context (cf. Sperber and Wilson 1986; Blakemore 1992). According to Blakemore, discourse markers (or connectives) guide the hearer in this task by constraining the number of possible interpretations. Thus, they “encode instructions for processing propositional representations” (1992: 150f.), which Blakemore also terms “encoding procedural meaning”. Apart from marking coherence relations within the text and providing the hearer with processing instructions regarding possible interpretations, discourse markers are also said to mark “propositional attitude or illocutionary force” as well as “interpersonal relations” (Andersen 1998: 147). Because the latter has not received sufficient attention in the literature, Andersen et al. (1999) explicitly dedicate their paper to the role of discourse markers for the “social relationships between interlocutors”. For Biber et al. (1999: 1086), signaling an “interactive relationship between speaker, hearer, and message” is one of the two roles of discourse markers, and Jucker and Smith (1998) consider markers to be negotiating strategies of the common ground between speaker and hearer. So far, I have described very general formulations of the functions discourse markers have. Therefore, I would like to conclude this sub-section by presenting a slightly more detailed list of functions based on Brinton (1990: 47f.). Although somewhat dated, it is still relevant to current studies of the functions of discourse markers. According to this list, discourse markers are used – – – – – – – –
to initiate discourse, to mark a boundary in discourse (shift/partial shift in topic), to preface a response or a reaction, to serve as a filler or delaying tactic, to aid the speaker in holding the floor, to effect an interaction or sharing between speaker and hearer, to bracket the discourse either cataphorically or anaphorically, to mark either foregrounded or backgrounded information.
In the course of Chapters 2–5, I will come back to several of these functions.
.. Material for analysis As it was customary before the advent of computers, Lakoff ’s examples for the use of well and why were either invented or anecdotal. Invented and anecdotal evidence is also found in other works which often concentrate on classifying markers or on describing the meaning of selected markers, particularly in relation to a specific theory. Invented sentences are then frequently used to contrast markers. Examples are the works by Blakemore (1988, 1996, 2002), Fraser (1988, 1990, 1998), and Rouchota (1998). However, within the limits of descriptive linguistics, it can only be desirable to find out what (native and non-native) speakers actually do with language and how they use it. Accordingly, analyses are now more common which make use of real data – real in the
Discourse markers in English discourse
sense that people have been recorded while using language. There are several issues that are important.
... Recorded data – small and large corpora Since the eighties, investigations of small scale corpora abound. The studies listed below all concentrate on spoken language, and the database is typically collected by the researchers themselves. Östman (1981) taped dinner table conversations between adults and used Ervin-Tripp’s recordings of children’s conversations for his analysis of you know. Similarly, Watts (1989) recorded a family gathering to investigate speakers’ attitudes to discourse markers. Other settings for recording naturally occurring data are classroom interactions (e.g. De Fina 1997) or academic counseling encounters (e.g. He and Lindsey 1998). Schourup’s (1985) database consisted of radio tape shows and three conversations of free topic choice between friends in a lab situation. Lab conversations, albeit with prescribed topics, also form the basis for Jucker and Smith’s (1998) analysis of several discourse markers. Schiffrin (1987) conducted a series of sociolinguistic interviews with a small group of people and used them to describe the functions which eleven markers have in discourse. For her analysis of ideational and pragmatic markers, Redeker (1990) showed the participants in her lab experiment a movie and asked them to talk about it. Other lab experiments are conducted more in the style of psycholinguistics. Flowerdew and Tauroza (1995), for example, tested the difference which the presence or absence of discourse markers makes in understanding texts. They presented a videotape of an authentic lecture to one group of students and the same videotape with the discourse markers edited out to another group of students. The students were later asked to fill out a questionnaire designed to assess their understanding of the lecture. A similar experiment was carried out later by Fox Tree and Schrock (1999; cf. also Degand and Sanders 2002 for the comprehension of written texts). Miller and Weinert’s (1995) database partly came from recordings of a map task, in which one participant had to describe to his or her partner a route marked on a map. The partner had a similar but non-identical map available, but could not see the route. With the availability of larger corpora of written and spoken language, often collected through the combined efforts of several researchers or even institutions, scholars have increasingly turned to analyzing these corpora. Svartvik (1980), for example, based his analysis of well on a selection of texts from the London Lund Corpus of Spoken English (LLC). This is also the main source for various publications by Erman (1987, 1992, 2001) on you know and I mean. Tottie (1997) makes use of the spoken component of the British National Corpus, comparing her results with those from the London Lund Corpus. A number of publications have recently been brought out with data from the Bergen Corpus of London Teenage Language (COLT, conversations), including an edited volume entitled From the COLT’s Mouth (Breivik and Hasselgren 2002) and several publications by Andersen (1997, 1998, 2000, 2001). The International Corpus of English (ICE) provided the material for other research projects, e.g. Vine (2000, 2002) and Wichmann (2002). Probably the most recent large corpus made
Chapter 1. Introduction
available is the Santa Barbara Corpus of Spoken American English (cf. the book review by Kaufmann 2002). This corpus was designed “to make the American Corpus comparable with the spoken, conversational LLC corpus” (Lenk 1998: 7) and thus invites comparative investigations such as her own. However, large corpora are not necessarily a must for modern analyses of spoken language. Sinclair in his preface to an edited volume on Small Corpus Studies and ELT justifies the use of small corpora by using the following definition: A small corpus is seen as a body of relevant and reliable evidence, and is either small enough to be analysed manually, or is processed by the computer in a preliminary fashion, [...]; thereafter the evidence is interpreted by the scholar directly. (Sinclair 2001: xi)
Manual analysis is an important feature of all the studies mentioned above; if we want to discover the functions of discourse markers, we need to consider not only their lexical context, but also the pragmatic context. This is a task a computer is hardly able to do; instead, it requires human interpretation. A page later Sinclair states that “[t]he main investigative technique that is used here [...] is comparison; comparison uncovers differences almost regardless of size” (Sinclair 2001: xii). The “almost” in his statement is certainly significant if we want to gain reliable evidence; however, there is a point in claiming that technique and purpose of the investigation are more important for the type of corpus than its size. Thus, for example, if a researcher wants to compare how a certain linguistic element or structure is used in two different groups of speakers, it is desirable that all features the linguist is not interested in should be the same across the two groups. An advantage of small corpora which Sinclair does not mention is related to the fact that a small corpus is often collected by the researcher (and his project team) him/herself. Accordingly, s/he can design it to fit his/her purpose, whereas large corpora have to be taken as they are.
... Restricted vs. unrestricted context of recording Another relevant issue in this area is the context in which spoken material has been recorded. We have seen already that there is a wide range of possibilities. Lab recordings with their restricted context may stem from collecting material under rather artificial conditions, such as map tasks (Miller and Weinert 1995) or questionnaires testing the comprehension of lectures (Fox Tree and Schrock 1999); they may consist of guided interviews (Hays 1992), chatting about prescribed topics (Jucker and Smith 1998), or retelling a movie that was shown (Redeker 1990). On the other end, there are recordings of free and unrestricted conversations. For the Bergen Corpus of London Teenage Language (COLT), for example, teenagers were given recording sets and asked to tape conversations with friends and family members during a three-day period (Andersen 2001: 84). Participants of the Santa Barbara Corpus of Spoken American English switched on the recording sets in the morning and basically taped everything during the whole day (DuBois, lecture given at the ICAME 2002). Accordingly, not only topics and genres vary, but also the number of participants in the conversation.
Discourse markers in English discourse
These variables have to be taken into account particularly for comparative analyses between and within corpora.
... Using corpus material So far I have described the kind of material that has been used for research on discourse markers. Whatever the material, however, another important question is how the material is used, that is, which role it plays in the analysis. Tognini-Bonelli (2001: 84) claims that in the ‘corpus-based approach’ the linguist typically uses corpus data selectively to confirm pre-existing theories. This implies that instances of the linguistic phenomenon in question which do not fit pre-conceived notions or the theory outlined are discarded or at least remain unconsidered. In a corpus-driven approach, in contrast, “[e]xamples are normally taken verbatim, in other words they are not adjusted in any way to fit the predefined categories of the analyst; recurrent patterns and frequency distributions are expected to form the basic evidence for linguistic categories” (Tognini-Bonelli 2001: 84). This includes, in my opinion, dealing with all tokens of the linguistic phenomenon in question. If we want to describe language as it is used, this is paramount for our research, even if the results are not convenient. Unfortunately, few scholars explicitly address the issue as openly as Andersen (2001: 265f.) does for his study of discourse marker like in teenage language. He admits that only 39% of all tokens of like entered his analysis; any token which was not obviously a discourse marker or which was tagged “unclear” was discarded. Redeker (1990: 371f.) goes even further: She excludes all intraclausal relations, reduced clauses and fragments, false starts and other repaired utterances from her analysis and thus reduces the number and possible functions of discourse markers to account for. Similarly, Blakemore, who does not draw on recorded material, limits her study of so to “those uses of so in which it cannot be substituted by therefore” (1988: 194). The present study, in contrast, is dedicated to dealing with all tokens of the relevant linguistic items. .. Core meaning vs. multiple functions Of course such a corpus-driven approach poses the threat of toppling existing theories and descriptions of discourse markers (DMs). An excellent example is the issue of core meaning, sometimes also called ‘basic meaning’ (Fox Tree and Schrock 2002). As Schourup in his Tutorial overview on discourse markers points out, [s]tudies of individual DMs have in most cases attempted to isolate an invariant semantic content for each marker, usually referred to as its ‘core meaning’ [...]. The working assumption that there is a single core for each DM is ordinarily abandoned only if the data will not support a unified treatment, but in practice it is rare to resort to the option of multiple cores. (Schourup 1999: 249)
Such studies include Schourup (1985), Schiffrin (1987), Fraser (1990), Redeker (1991), and Jucker (1993). Naturally, it will be more often the case that the data does not support a unified treatment if the research is corpus-driven and therefore does not exclude
Chapter 1. Introduction
any instances of the discourse marker in the corpus, supposing that this corpus is large enough to yield a substantial number of tokens. According to Schourup, core meanings “are isolated by comparing the various discourse uses of a DM and attempting to determine what these uses have in common” (1999: 250). The more uses are found in a corpus and the more disparate they are, the more difficult it will be to find something they have in common. (Schourup’s use of ‘semantic content’ here should not be interpreted too narrowly; referring to Jespersen (1940) and Bolinger (1989), he lists ‘indicating surprise’ and ‘appeal to a norm’ as core meanings of why and well.) Nevertheless, scholars go to amazing lengths to cling to what seems to be the core meaning imperative. Schourup himself analyzed well based on the ideas of Bolinger (1989) and presents as the core meaning of well that “well evokes a norm with the more broadly epistemic notion that the speaker, in saying well, wishes to be seen as engaged in a prefatory act of epistemic consideration” (Schourup 2001: 1057). In my opinion, this “definition” is so general that it is rendered of little use. Even so, it does not cover all uses of well occurring in real spoken data, as will be seen in Chapter 3. Many researchers seem to feel the obligation to find a core meaning. Östman initially did not yield to this obligation: “At present I do not want to take a stand in the discussion of what general category you know should be assigned. And this is for a reason: I do not think it is a relevant question, or, more specifically, I think you know (and a number of other pragmatic expressions) can perform different kinds of functions in different situations and contexts” (Östman 1981: 16). In his historical overview of particle (and marker) research, more than a decade later, he still does not explicitly favor the core meaning approach, but he now strongly expresses himself against a multifunctionality approach to markers (Östman 1995: 101). Nevertheless, none of the researchers mentioned in this section claims that the discourse marker they analyzed has only one single function. All of them identify a range of clearly different uses, functions or sub-functions. This is what Jucker calls the “middle solution”, to “list a number of different uses while pointing out the common core of all of them” (Jucker 1993: 437). Other researchers, such as Sandra Thompson (personal communication, 2003), hold that it is of minor importance to define a core meaning. If there are indeed, as Jucker puts it, “a large number of – seemingly – disparate uses” (Jucker 1993: 440), then a comparison of discourse marker use between separate groups of speakers, such as the present study, should want to make this comparison at the level of these “disparate uses” as much as on the level of a core meaning.
.. Native vs. non-native English Among researchers who investigate discourse markers, it is no question that these elements fulfill crucial functions in communication. More generally, authors like Crystal and Svartvik find that discourse markers smooth spontaneous speech production and interaction (Crystal 1988: 48) and prevent the speaker from being perceived as impolite or “awkward to talk to” (Svartvik 1980: 171; see also the quotation in Section 1.1). Fox Tree and Schrock (1999) have been able to show that, for example, oh also facilitates
Discourse markers in English discourse
the comprehension of a spoken lecture. More specifically, Ferrara (1997: 373) in her study of anyway claims that it “is used to manage discourse [and] give clues to the listener about how to organize the talk”, a function which Östman (1995: 104) accords to discourse markers or pragmatic particles, as he calls them, in general. How do markers achieve this discourse-organizing task? Despite the fact that they use different frameworks, Schiffrin and Blakemore both take discourse markers to help the hearer interpret the following utterance. Schiffrin suggests that “markers select a meaning relation from whatever potential meanings are provided through the content of talk, and then display that relation” (Schiffrin 1987: 318, without emphasis). Since coherence relations depend “to a large extent on (the addressee’s interpretation of) the content of the units involved” (Risselada and Spooren 1998: 132), discourse markers thus contribute to coherence. Blakemore, who works within a relevance-theoretic framework, distinguishes conceptual and procedural meaning. In her account, discourse markers have procedural meaning, that is, they “encode a constraint on pragmatic inferences” (Blakemore 2002: 4; cf. also Blakemore 1988: 185). Through discourse markers, the number of potential interpretations which the speaker can draw from the utterance following the marker is narrowed down. If we take it for granted that discourse markers have such a decisive role to play in native speaker communication as the authors of discourse marker analyses claim, then we must assume that they are important elements to be learned by non-native speakers as well (see the quotation from Svartvik 1980 on page 1). In addition, as Cots (1992: 169) states, “[s]uccess in foreign language learning is graded in terms of how similar the linguistic behaviour of the learner is to that of the native speakers of the language”. This definitely includes the use of discourse markers as well, which poses the question of how successful learners of English are in this area. However, all of the studies mentioned in Section 1.2 use data which come from native speakers. Since the nineties, there have also been a number of publications on bilingual speakers. Maschler, for example, concentrates on discourse markers in bilingual Hebrew-English conversations (Maschler 1994, 2000). Salmons, Goss, and Fuller are interested in how discourse markers are used in German speech communities within America, for instance by Pennsylvania Dutch speakers (Salmons 1990; Goss and Salmons 2000; Fuller 2001). The examples given by Matras (2000) come from speakers who are fluent in German and one of various languages, while Olynak et al. (1990) deal with discourse markers in the speech of English-French bilinguals in Canada. Even though there is little agreement as to what counts as bilingualism (Romaine 1995: 11ff.), these studies on bilingual speech concentrate on speakers who live in a language contact situation. They either speak one language as their native language and live in an environment where the second language is spoken, or they grow up learning both languages simultaneously (cf. Bussmann 1996 for definitions). But what about speakers who learn another language, for example English, as a foreign language, i.e. without (permanently) living in an English-language context? In the beginning of this chapter, I quoted an illustration by Svartvik on the negative effect of lacking discourse markers in the speech of a foreign language learner. Svartvik
Chapter 1. Introduction
himself, however, investigated neither the performance of learners nor how they might be taught the use of markers. One of the first papers which dealt with how a foreign language learner might acquire competence in this area came from Germany in 1989, with Russian as a foreign language. Rathmayr’s aim was to identify means by which students would comprehend how to use particles. She came up with a number of ways: Analyzing spoken texts as to the frequency of particles and comparing texts with and without particles in order to sensitize students for their functions; discussing the use of particles as mitigators on various face-threatening speech acts to help the student understand particle functions; role play and discussions of real situations in which a particle was or could have been employed as a means of developing active particle use (Rathmayr 1989). Referring to German modal particles, Rösler (1982: 35ff.) suggests a “small project to teach modal particle”. It consists of showing short interactions with particles, having them re-enacted by the students, and asking the students to analyze and compare the different versions of the interaction. McCarthy and Carter propose “language-observation activities, problem-solving, perhaps cross-linguistic comparisons” (1994: 68). Observations tasks as well as role play, simulation, and drama are recommended by Kasper (1997). Fischer and Drescher (1996), who examined English and French plays and their translations into German, focus on explaining discourse marker functions rather than explicitly teaching them. They suggest that presenting all translational equivalents can indicate the range of functions a marker or particle has, well beyond the descriptions given for this marker/particle in mono- or bilingual dictionaries. Other authors concentrate more on the description of native and non-native speaker performance. Hays (1992) wanted to know whether Japanese students mastered the use of seven English discourse markers. To find out, he set up and recorded interviews of an instructor with a number of Japanese students, which then formed the data base of his analyses. Texts written by Chinese students are the learner corpus which Anping (2002) compared to native spoken English and spoken Chinese with regard to the discourse marker so. Written texts are also the material analyzed by Lamiroy (1994), albeit for French and Dutch as foreign languages. Nikula (1996) presents a rather comprehensive and large scale study. She recorded unmonitored, free conversations between native speakers of English and native speakers of Finnish, both in English (native and non-native) and in Finnish (native only). Within a politeness framework, she compares the use of a long list of what she calls “pragmatic force modifiers”, including items which are commonly considered discourse markers, in these conversations. Accordingly, her focus is not as much on individual markers but on the performance with regard to politeness as expressed through these “pragmatic force modifiers”. Finally, Romero Trillo (2002) explores the use of English discourse markers in the non-native discourse by Spanish speakers, as compared to their use in native English discourse. His approach is corpus-driven; based on several smaller corpora of adults and of children, he calculated the frequencies of markers such as listen, I mean, and well. None of these five studies, however, distinguishes individual functions of the markers investigated for the quantitative analyses and comparison.
Discourse markers in English discourse
. Second language acquisition, applied linguistics, and discourse markers So far, I have summarized discourse marker research and publications which also deal with foreign language learners. In general, the interest in the process of acquiring a language which is not the native one has generated a vast amount of research and subsequent publications, usually subsumed under the terms “second language acquisition” (SLA) and “applied linguistics”, “second/foreign language teaching” or “TESOL” (teaching English to speakers of other languages). While SLA research tends to concentrate more on developing theories, models and methods for analysis (cf. Phillipson 1991), applied linguistics research focuses more on teaching and practical aspects of learning a second/foreign language.
.. Second versus foreign language The distinction between second and foreign language is disputed; Bussmann (1996: 168) gives two definitions which operate at different levels: On the one hand, she defines a foreign language from a political perspective as “any language that is not officially recognized in a given country or state” whereas a second language (abbreviated: L2) is “an officially sanctioned language spoken by an identifiable population in a given country or state, such as French in Canada” (cf. also McDonough 1998). On the other hand, she presents a definition from the learners’ perspective: a second language is learned “with the goal of making the new language one’s own and of becoming a productive, functioning member in the L2 society”, while ‘foreign languages’ are usually learned with more specific goals in mind” (see also Kramsch 2002: 59f.). In practice, this distinction is of minor importance if made at all (cf. Crystal 1997). In the United States, “foreign language” has become a dispreferred term (Bussmann 1996: 168). Bialystok simply calls it “a new language” (1991: 63f.), while Phillipson et al. (1991) and Spolsky (1999) use both terms (“second and foreign language”) to include all languages which a person does not learn as his/her first language. In this study, I only use the data of German learners of English. For this reason, I will mostly refer to the non-native speakers as ‘the Germans’; only the extracts from the corpus will be marked ‘NS’ (native speaker) and ‘NNS’ (non-native, i.e. German speaker). Most of the Germans were recorded in Giessen, Germany. Of those who participated in California, all but one had arrived there for a semester abroad less than four weeks before the recording. Thus, at that stage they do not differ from German students recorded in Giessen who spent a semester abroad previous to the experiment in terms of foreign versus second language; all of them are speakers of English as a foreign language.
Chapter 1. Introduction
.. Applied linguistics and SLA According to Catford (1998), the beginnings of modern applied linguistics can be dated back to the late 19th century. Although the term is more general and could in principle also refer to other areas where the insights of linguistics are applied to practical purposes, most of its research is devoted to language learning and teaching. The title of one of the influential journals in the field makes this quite clear: It is called “Language Learning: A Quarterly Journal of Applied Linguistics”. Thus, applied linguistics also interrelates with second language acquisition. Gregg (1999) claims that modern SLA research as a scientific discipline developed since World War II. Much of this discipline is devoted to theories about how a learner is able to acquire a second language. This question can be divided into two sub-questions: What internal and innate (neural) features enable the human being to learn a first and a second language (cf. Bialystok 1998); and which external conditions (type of input, teaching style, etc.) are most favorable for the acquisition of a second language? Concerning the first of these sub-questions, one of the most well-known theories in first language acquisition holds that there is innate language knowledge, referred to as Universal Grammar (UG). The question for UG-based SLA research then is whether this innate knowledge is also available for second language acquisition, and if so, to what extent (White 1996). Naturally, researchers working with the notion of Universal Grammar focus on grammar, i.e. mostly on syntax. Three areas are of particular interest, according to Ellis (1997: 15): “the kind of errors learners make and how these errors change over time,” developmental patterns, i.e. “stages in the acquisition of particular grammatical features such as past tense”, and the systematic variability in learner language.
.. Communicative competence Bialystok and Hakuta (1994: 160ff.) acknowledge the importance of knowledge in a foreign language beyond grammar and vocabulary: “It is thus not sufficient to learn just the grammatical forms of the language. There is a relationship between the forms of language and how they are used to express meanings and intentions in appropriate ways”. What they have in mind, though, are culturally acknowledged rituals, etiquette, attitudes and belief systems, “both linguistic and beyond” (1994: 164) rather than specific linguistic items such as discourse markers. Of the relationship between SLA and pragmatics (i.e. language used in communication, as he defines it), Ellis writes: Early work in SLA research did not entirely neglect pragmatic aspects of learner language. However, they were attended to only in so far as they helped to explain how learners acquired grammatical competence. [...] Only recently, however, have researchers considered the acquisition of pragmatic competence in its own right. (Ellis 1994: 23)
Discourse markers in English discourse
The term ‘pragmatic competence’ is frequently mentioned in works on second language acquisition and foreign language teaching. The ‘Common European Framework for language learning, teaching and assessment’ acknowledges its status as part of what a student should learn about a language by providing descriptive categories for it (Trim 2000: 122f.). Pragmatic competence is defined as “an aspect of communicative competence [which] refers to the ability to communicate appropriately in particular contexts of use” (Jaworski 1998: 249). The term ‘communicative competence’, in turn, goes back to Hymes (1972), who proposed a model which combines four types of knowledge: knowledge of whether (and to what degree) something is possible, feasible, appropriate and actually done (cf. Cook 1998a). Although Hymes originally did not propose it as a model for language learning, ‘communicative competence’ was soon identified as “the overarching goal of English language teaching” (Piepho 1974). According to Cook (1998a), Canale and Swain (1980) later took up Hymes’ model and modified it for second language teaching; their version splits up communicative competence into three fields: grammatical competence, strategic competence, and sociolinguistic competence. The latter in turn is comprised of sociocultural competence and discourse competence. Cook summarizes sociocultural competence as “knowledge of the relation of language use to its non-linguistic context”, while discourse competence is “knowledge of rules for ‘the combination of utterances and communicative functions’ which may be conceived as knowledge of factors governing the creation of cohesion and coherence” (Cook 1998a: 66). In a later version (Canale 1983), sociolinguistic competence and discourse competence are separated to form autonomous sub-competences (Cook 1998a: 67). Strategic competence finally refers to strategies which a learner can use if communication threatens to break down, for example due to misunderstanding or because the learner lacks the vocabulary to express him/herself as initially intended (cf. Poulisse 1999). Concerning the term ‘communicative competence’, Gregg complains that “the term as used in the L2 acquisition literature is largely devoid of theoretical content” (Gregg 1996: 53). Nevertheless, it is still widely used (e.g. by Nunan 1999; Savignon 2000; Brumfit 2001; Kachru and Nelson 2001; Kramsch 2002; Magnan and Walz 2002; Shumin 2002). Communicative competence also relates to discourse markers. Besides grammatical competence, which is the necessary prerequisite for any communication at least to some degree, sociolinguistic, discourse and strategic competence may all manifest themselves in the use of discourse markers, as we will also see in the subsequent chapters of this study: Sociolinguistic knowledge is necessary for the negotiation of the relationship between speaker and hearer during a conversation, which can be done through, for example, the use of well or you know; a range of discourse markers are said to create coherence in discourse (cf. e.g. Schiffrin 1985; Lenk 1995; Risselada and Spooren 1998), the knowledge of which is part of discourse competence; and strategic competence manifests itself when non-native speakers use discourse markers to express or to introduce the expression of lexical difficulties (finding the appropriate/intended word or phrase) or to appeal for the hearer’s understanding. Nevertheless,
Chapter 1. Introduction
the term ‘discourse marker’ – or the terms ‘discourse particle’ / ‘pragmatic marker’ / ‘pragmatic particle’, for that matter – do not appear in any of the comprehensive encyclopedia and handbooks I checked (Ahrens et al. 1995; Ritchie and Bhatia 1996; Johnson and Johnson 1998; Spolsky 1999a; Byram 2000). Why should that be so?
.. Research in interlanguage pragmatics Referring to pragmatic competence, Ellis claims that “[t]he bulk of the work to date has focused on specific illocutionary acts”, and that “the performance and acquisition of speech acts by L2 learners [...] is this aspect of pragmatics which has received the greatest attention in SLA research”. Therefore, he argues, he will use the term ‘interlanguage pragmatics’ limited to speech acts (Ellis 1994: 23, 159f.). Likewise, Cenoz and Valencia (1996: 41), referring to Kasper and Blum-Kulka (1993b), equate interlanguage pragmatics with “identifying the strategies used by learners in the production of speech acts”. So does Barron (2003: 26), quoting Kasper and Dahl (1991). Before I proceed to give examples of research in this field, a definition of the term is in place here. Tarone explains: The term interlanguage was defined by Selinker (1972) as the separate linguistic system evidenced when adult second-language learners attempt to express meaning in a language they are in the process of learning. This linguistic system encompasses not just phonology, morphology, and syntax, but also the lexical, pragmatic, and discourse levels of the interlanguage. (Tarone 1999: 508; 2001: 476f.)
Selinker associated several features and processes with the concept of interlanguage which may also be relevant for the results concerning discourse markers in non-native speech. One of these processes is language transfer from the first language (L1), where features of the L1 are projected onto the foreign or second language (L2). Another process consists in the overgeneralization of L2 rules to situations in which these rules do not apply. Equally important is the idea that a learner’s interlanguage fossilizes, i.e. its development at some stage stops, short of the full attainment of native-like competence (Cook 1998b; Cherrington 2000). Interlanguage research is generally divided into interlanguage grammar and interlanguage pragmatics, with the latter comprising the pragmatic and discourse levels of the interlanguage. Since interlanguage grammar is less relevant for the research on discourse markers, I will deal only with interlanguage pragmatics here.
.. Speech acts and parallel corpora Various authors equate interlanguage pragmatics with speech act research. However, even though Kasper and Blum-Kulka (1993b: 3) do use expressions such as “actiontheoretical”, “linguistic action in context”, and “linguistic action patterns” to describe the scope of interlanguage pragmatics, the term “speech act” itself is not mentioned.
Discourse markers in English discourse
Judging from the contents of their edited volume (Kasper and Blum-Kulka 1993a) entitled Interlanguage Pragmatics, topics in discourse analysis are as much part of this area as speech act realizations. Nevertheless, the literature seems to be dominated by the latter. Blum-Kulka et al. (1989) themselves first focused on requests and apologies, based on the “Cross-Cultural Speech Act Realization Project (CCSARP)”. This project elicited native and non-native formulations of requests and apologies via a discourse completion test in seven different languages and language varieties (Edmondson and House 1991: 286). When Edmondson and House (1991) investigated the so-called “waffle phenomenon” (i.e. learners being more verbose than native speakers), they also referred to this project. Olshtain and Cohen (1991), whose paper deals with teaching speech acts to non-native speakers, list two more speech acts in addition to requests and apologies: complaints and compliments. These four plus the speech acts of offering (Barron 2003) and refusal (Gass 1999a) tend to re-occur frequently in interlanguage pragmatics research (e.g. Trosborg 1995; Hartford and Bardovi-Harlig 1996; Baba 1999; Golato 2002). Discourse completion tests, as used in the Cross-Cultural Speech Act Realization Project (CCSARP), and questionnaires have the advantage of providing perfectly comparable data for different groups of speakers. However, according to Tarone (1999: 510), Selinker held the opinion that the data to be used in interlanguage research should be “utterances produced by second-language learners when they were trying to communicate meaning in the target language”. I understand this to include the voluntary intention to communicate, which requires that the learner has to or wants to fulfill a communicative task in the situation s/he is in at that moment. In contrast, discourse completion tests and role plays ask the learner to imagine a communicative situation rather than putting him/her into such a situation. Sinclair expresses a similar idea using the term ‘authentic’, but he is even more stringent on his requirements. He defines ‘authentic data’ as material which “is gathered from the genuine communications of people going about their normal business” (Sinclair 1996) and demands that all data collected under artificial or experimental conditions be marked as such. A very recently completed corpus that, at the same time, is highly comparable and fulfills Selinker’s requirement better than the CCSARP is the International Corpus of Learner English (ICLE). Initiated in the late 1990s by Granger at the University of Louvain in Belgium, it comprises fourteen sub-corpora representing fourteen different first languages. The data consist of argumentative essays of about 500 words written by advanced learners of English, i.e. students of English language and literature in their third or fourth year. All of them are learners of English as a foreign language: they do not live in an English-speaking country and have learned English primarily in a classroom context (Granger 1996b; De Cock et al. 2003). This corpus fulfills Selinker’s demand better than discourse completion tasks because the task for the student (to write an argumentative essay about a topic of the teacher’s or their own choice) is a genuine one within the classroom setting. As Granger argues, referring to Sinclair’s term ‘authentic data’, “[i]n as far as essay writing is an authentic classroom
Chapter 1. Introduction
activity, learner corpora of essay writing can be considered to be authentic written data” (2002: 8). Another parallel corpus initiated at the University of Louvain in 1995 and in the process of compilation, the Louvain International Database of Spoken English Interlanguage (LINDSEI), will complement the ICLE. It is supposed to contain transcripts of fifty interviews with learners of English as a foreign language with a variety of first language backgrounds (De Cock et al. 2003). This corpus, once it is completed, can serve as a comparison for the Giessen-Long Beach Chaplin Corpus (GLBCC). The latter was not constructed to be a part of the LINDSEI; nevertheless, it perfectly fulfills its requirements for comparability. More details about the GLBCC will be given in Section 1.5.
.. Gambits Interesting as these corpora and the accompanying research projects are, they have not included much research on discourse markers in learner English so far. The closest to discourse markers we can find in the literature are linguistic elements which have been termed ‘gambits’. Bublitz describes ‘gambits’ as follows: Tatsächlich sind Gespräche (und andere Textsorten) durchzogen mit zahlreichen metakommunikativen Ausdrücken, die wie Verkehrsschilder fungieren. Ihnen kommen vielfältige Aufgaben zu, darunter diese: – Sie sorgen für den (möglichst reibungslosen) Ablauf eines Gesprächs oder strukturieren einen monologischen Text (textuelle Funktion), – sie geben die vom Sprecher gewünschte Interpretationsrichtung an (propositionale/ideational Funktion), – sie ”regeln” das soziale Miteinander von Sprecher und Hörer (interpersonale Funktion). Der Vielfalt ihrer Funktionen entspricht in der Forschung auch eine Vielfalt der Definitionen und Bezeichnungen solcher sprachlicher “Verkehrsschilder”; verbreitet ist der Terminus gambits. Gambits sind feste Syntagmen (selten Simplizia), [...] die vornehmlich, aber nicht ausschließlich, in dialogischen Textsorten vorkommen und dort im Sinne der oben spezifizierten drei Aufgaben als Struktur- und Interpretationshilfen fungieren. (Bublitz 1995: 187f.)
When Bublitz says that the term ‘gambits’ is wide-spread, he is right and wrong at the same time. A Google internet search shows that there are more than three thousand web pages in which both language and gambits occur, but it seems that the term is used mainly in resource (do-it-yourself) books for English learners. In second language research, the term ‘gambits’ (or gambits as a topic) apparently did not find many supporters; it used mainly by Eric Keller (e.g. Keller 1981), Juliane House and Willis Edmondson (e.g. House 1982; Edmondson and House 1991; House 1996). To my knowledge, there is only one other (more recent) publication, which might have
Discourse markers in English discourse
been based on a master’s thesis (Trommer 1990). Furthermore, several major works on SLA and teaching and encyclopedia of the field do not even mention the term ‘gambits’ (e.g. Ellis 1994; Spolsky 1999a; Ritchie and Bhatia 1996; Johnson and Johnson 1998; Byram 2000; Johnson 2001). But let us have a closer look at how the term ‘gambits’ is used. As we will see in the following chapter, the discourse markers analyzed in this study do, like gambits (as described by Bublitz), structure text, direct the interpretation of an utterance and serve interpersonal functions. However, while the linguistic elements generally treated as discourse markers in the literature typically consist of one or maximally two words, such as well, so, actually, you know, I mean, and so forth (with the exception of Fraser’s 1988; 1999 usage), the elements Bublitz lists are multi-word expressions of the kind in my opinion, by the way, to cut a long story short, let’s be fair, I’d rather not, etc. (1995: 188f.). Thus, even though the general functions of gambits and discourse markers may be described in the same way, we are apparently not dealing with the same kind of elements. However, despite Bublitz’ statement that gambits are rarely “Simplizia”, House also includes well, oh, yeah, and okay (1982). In her 1996 publication on pragmatic fluency, she treats gambits “or discourse lubricants” as elements which also enhance fluency. According to her, they “are used to establish, maintain, and end contact. Gambits are typical of oral interaction, helping to cement segments of talk into a discourse” (House 1996: 232). There is no attempt to describe the syntactic properties of gambits in the way Bublitz does. This would indeed be a futile undertaking if the list of relevant expressions contained “some 500 such gambits” and if it is “impossible to have a complete list of gambits” as Keller claims (1981: 95f.). However, we are dealing here with two approaches to gambits. For Keller, they are used “to introduce a semantic frame, signal social context or a person’s state of consciousness, serve an explicit communication control function, or assume any combination of these functions” (ibid.: 96). Accordingly, he arrives at four categories, each of which is divided into two or more sub-categories. Semantic framing is the most frequent of the four main functions; it includes indicating a digression, marking a main aspect of a topic, marking the expression of an opinion, subject evaluation, etc. Social context signals are mostly used in turn-taking (“And what about you?” p. 101). “Gambits “indicating a person’s readiness to receive or provide new information, opinions, or emotions” (p. 102) fall under the category of state-of-consciousness signals. Communication control gambits, finally, signal and negotiate (mis-)understanding mostly based on (assumed) acoustic problems (“Pardon me?”). Although Keller mentions you know and you see as hesitations which fulfill a communication control function (p. 94), the examples he gives for the different subcategories are generally longer. They range in complexity from “In general” to “May I interrupt you for a moment” and may contain what is elsewhere termed a discourse marker, for example “Well, let’s see now” (pp. 99, 101). If they are not a proposition in themselves, they at least have, with very few exceptions, a clear conceptual content. House’ approach is based on the categorization system developed by Edmondson in 1977 (House 1982: 111):
Chapter 1. Introduction
Three classes of gambits are differentiated through their pragmatic focus on either the previous turn (Uptakers), the following turn (Appealers), or the current turn (Clarifiers). Two of these classes (Uptakers and Clarifiers) are further subdivided. Over and above these major directional classes the gambit type “Starter” immediately follows turn change. (House 1996: 232)
Uptakers are typically represented by elements such as yeah and oh. The two types of Clarifiers are glossed “Please be agreeable to my message” (Cajoler) and “Listen carefully now” (Underscorer), respectively. Examples of such gambits are I mean, you know, look, and listen. Question tags, right, and okay are said to function as Appealers, soliciting the hearer’s agreement. Finally, a Starter gambit such as well is “used to signal that the speaker is now assuming his or her turn” (ibid.: 233). Even though House presents expressions such as “to tell the truth” as exemplifications of gambits (a Cajoler, in this case) in her 1982 publication, the typical tokens listed in House (1996) are almost without exception one- or two-word expressions. Thus, not only the categorization systems of Keller and Edmondson/House differ but also the kind of linguistic items they consider to be gambits. Keller’s range of items does not overlap with the items covered in this study, while Edmondson/House’ gambits include several of the expressions that are elsewhere considered discourse/pragmatic markers (e.g. Erman 2001 for you know; Schiffrin 1987 for I mean; Jucker and Smith 1998 for yeah; Romero Trillo 2002 for look and listen). Two of their gambits will also form part of my analysis. Besides using the same term, however, the two approaches have one thing in common, which decidedly distinguishes them from my approach. Both emphasize the functions of the gambits and then find linguistic examples of elements which fulfill these functions. In contrast, my approach will start with selected linguistic items and then look at the functions these items have. We have seen that there is little in the area of second language acquisition and applied linguistics which deals explicitly with discourse markers. The focus in this area is either on grammatical features or, as far as pragmatic competence goes, on speech acts. The approaches which deal with particular linguistic items that are part of pragmatic competence (gambits) start with functions rather than with the linguistic items. They do not necessarily limit the range of gambits which fulfill these functions and include items which go far beyond the range of what is generally called a discourse or pragmatic marker.
. The present study In the beginning of this chapter, I asked how non-native speakers of English master the use of discourse markers. As I have shown, research on discourse markers and research on language of learners of English overlap only to a very small degree. My aim is to contribute to this overlap by systematically comparing native and non-native speaker data with regard to a selected number of discourse markers, so, well, you know, and like.
Discourse markers in English discourse
To do so, I will use a section of the Giessen-Long Beach Chaplin Corpus (GLBCC). The GLBCC is a 350,000 word corpus of native and non-native spoken data which is large enough to allow for statistical analyses (cf. Section 1.5 on the corpus details). The overwhelming majority of the native speakers are American; the largest group of nonnative speakers in this corpus are German. Since the remaining non-native speakers have a variety of different first languages, my study will be restricted to a comparison between American and German speakers. All of the speakers in the corpus are university students, and most of the Germans are enrolled in English philology or linguistics. Therefore, the question above might be re-formulated as “How do German university students who are reasonably fluent in English master the use of four selected discourse markers?” This question is only answerable if we have a standard for comparison; my standard will be the native speakers’ usage of these discourse markers (see Section 1.7.2.1 for a discussion of the native speaker definition). Furthermore, I aim at quantitatively and qualitatively relevant results. The qualitative side will consist in developing a categorization system for the functions and sub-functions of each discourse marker, classifying their tokens accordingly and comparing the German speakers’ usage to that of the native speakers. The quantitative side will consist in frequency counts and statistical analyses to show whether these frequencies are significantly different across speaker groups. Thus, the re-formulated question breaks down into several sub-questions guiding the analysis: a.
How does the frequency of the selected discourse markers in German non-native speakers’ (NNS) discourse compare to the frequency in native discourse? b. Which functions do the selected discourse markers have in the native discourse? c. Which of these functions also occur in the non-native discourse? d. How do the frequencies of the functions of discourse markers compare in native and non-native discourse? Once we have established to what extent the German speakers’ use of discourse markers conforms to the native speakers’ use, a fifth question will impose itself: What are the factors which influenced the results? This question was anticipated to some degree when the experiment which provides the data for the GLBCC was designed (see Section 1.5.2 on the experiment design). After participating in the experiment, the students filled out a questionnaire which, for the non-native speakers, elicited information about their use of English in formal and informal contexts, their acquisition of English, and about time spent abroad. Thus, the final question for this study is: e.
Which factors related to learning and using English in school and in informal contexts influence the NNS’s use of the discourse markers?
Allowing some informed speculation, this question was expanded into asking whether there are any other factors which might have a potential influence. Answers to this are sketched out in the last chapter.
Chapter 1. Introduction
.. Basic assumption of the present study: Multifunctionality An assumption which plays an important role for this research is that discourse markers are multifunctional. In Section 1.2.5, I have reported on the issue of core meaning versus multifunctionality, showing that there is (a) no agreement on whether it is obligatory, useful, or rather counterproductive to try to identify a core meaning, and that (b) corpus linguistics may present an obstacle to giving a unified description of the discourse marker function because it may return instances which do not fit this description. In Section 1.2.2.7, furthermore, I have discussed attitudes in the literature to multifunctionality as a feature of discourse markers and come to the conclusion that it is acceptable to assume it at least for those markers for which it has been shown convincingly. This is definitely the case for so, well, you know (e.g. Schiffrin 1987) and like (e.g. Andersen 2001). In assuming that discourse markers are multifunctional, I do not mean to say that they cannot have a core or basic meaning. Even when they do, however, they serve a number of sub-functions (cf. for example Jucker 1993, Östman 1981; Erman 1987; Erman 2001; Fox Tree and Schrock 2002). In the context of comparing the performance of native and non-native speakers, it may even be misleading to think of a linguistic element like so or well as having only one discourse marker function. Let me illustrate this point with an example. Hays (1992) did a study with Japanese students speaking English (in Japan), trying to find out whether they mastered a range of markers. In his paper, Hays describes Schiffrin’s model of planes, in which so has its main function on the ideational plane. His analysis of the interviews revealed that, while only three students mastered well, an overwhelming number of students were able to use the markers and, but, and so, all of which he finds to be in the ideational plane (1992: 29). Based on Schiffrin’s model and on these results, he concludes that “semantic words, and therefore ideational discourse markers are to be expected earlier, because they are overtly taught. [...] [Markers on other planes] would not be expected unless there has been exposure to that discourse community” (Hays 1992: 29). This argument is plausible. However, Hays simply presents as his result the number of students who “demonstrated the ability to use” and, but, so, well, etc., without any further sub-categorization of these markers (1992: 28). Schiffrin, in contrast, distinguishes various functions of both so and well, on the ideational plane as well as on other planes. If Hays had considered this distinction in his study, he might have found that his students do master some functions of the markers he tested but not others. Stubbe and Holmes (1995: 66), like Schiffrin, found it important to look at individual functions: “it has been demonstrated in a number of New Zealand studies [...] that it is often essential to take specific functions into account in order to interpret distributional information sensibly; this has been demonstrated clearly with regard to gender differences”. In my own research, I also distinguish a number of individual functions for each discourse marker, since I expect to find differences between native and non-native speakers both at the level of individual discourse markers as well as at the level of
Discourse markers in English discourse
individual functions. The general outline of my classification system will be described in Section 1.4.5, while details are given in Chapters 2–5.
.. Basic approach to data material: Corpus-driven or bottom-up The approach followed in this research project is not based on any existing framework or theory; rather, it is corpus-driven in the sense described by Tognini-Bonelli (2001: 84): “Examples are normally taken verbatim, in other words they are not adjusted in any way to fit the predefined categories of the analyst; recurrent patterns and frequency distributions are expected to form the basic evidence for linguistic categories”. It is virtually impossible and in most cases not even desirable to approach data without having any theory or category in mind, since such categories are typically necessary to select and look for certain phenomena in the data. However, in a corpus-driven approach, evidence from the data takes precedence over theoretical constructions, i.e., as Tognini-Bonelli says, the categories to describe the data are adjusted to fit the data. To avoid the laden term ‘corpus-driven’, I will refer to this procedure as a ‘bottom-up approach to corpus data’. In accordance with this approach, for each of the discourse markers treated here a number of categories were set up based on what I found in the data. And while it used to be common in the literature to be selective about the pieces of data that are used for analysis or which instances of the discourse markers are not to be considered (e.g. Redeker 1990: 371), in a corpus-driven or bottom-up approach, all occurrences of the linguistic items under scrutiny must be analyzed, which has been done in the present study. The result is a categorization system with two levels and a range of subfunctions, described in Section 1.4.5 and Chapters 2–5. These chapters will also show in how far my categories coincide with those found by other researchers.
.. Selection and definition of discourse markers For this research, four discourse markers were selected on both practical and theoretical grounds: so, well, you know, and like. In practical terms, I selected four linguistic items that occurred in sufficient numbers in the corpus for meaningful quantitative analysis. In theoretical terms, I was interested in discourse markers with a range of different characteristics which might potentially be relevant for the usage by EFL speakers. Thus, so is a conjunction and used not only in spoken but also in written language; well, in contrast, is said to pertain exclusively to the oral medium (Svartvik 1980: 169). You know apparently has a translational equivalent in German, while like does not. So and you know as discourse markers may seem transparent to the EFL speakers; this might less be the case with well. Like is claimed to be a fairly recent marker (at least in the present frequency of use it is often attributed to young people, cf. Dailey-O’Cain 2000; Andersen 2001), while the other three are well-established and accepted as long as their frequency in a person’s speech is not experienced as irritating. Finally, I wanted to investigate items which had received some coverage in the
Chapter 1. Introduction
discourse marker literature but were acknowledged as discourse markers by various analysts to variable degrees. Thus, Schiffrin (1987) and Schourup (1982) agree on including well and you know, and Schiffrin and Fraser (1990, 1998, 1999) agree on so. Of the three, however, only Schourup includes like. In addition, coverage in the literature such as Hays’ (1992) study of so, well, and you know, among other markers, in Japanese students’ speech provides me with results of other researchers for comparison with my own results. Since the conversations for this research were collected as part of another project focusing on reference assignment, the setup of the experiment was not specifically meant to elicit the whole range or all possible functions of discourse markers. Nevertheless, as mentioned before, it provided the same setting and task for all participants, natives and non-natives alike, so that the resulting transcripts were perfectly comparable and ideally suited for the purpose of my research. In addition, the selected markers occurred with sufficient frequencies to allow for statistical analyses. For the purpose of this study, I use the term discourse marker to include at least the four selected linguistic items mentioned above, so, well, you know, and like. Following the list of features discussed in Section 1.2.2, I take (these) discourse markers to be linguistic elements – – – – – –
which are, as a group, difficult to place within a traditional word class, which are syntactically optional, which may occur at the beginning, middle, or end of a discourse unit or form a discourse unit of their own, which have little or no semantic meaning in themselves, which are multifunctional, which occur in oral rather than written discourse. If they are found in both, they often if not always assume functions in the oral medium that go beyond those they have in the written medium (cf. Stenström and Andersen 1996; Svartvik 1980: 169).
.. Methodology As explained in previous sections, my assumption concerning the use of discourse markers is, firstly, that it is possible to identify a range of functions for each of them, and secondly, that we will find differences in frequency between native and non-native speakers of English at the level of these individual functions. For this reason, it is not useful, in the context of this study, to focus on a core or basic meaning, as many other authors have done (e.g. Jucker 1993; Schourup 2001 for well; Fox Tree and Schrock 2002 for you know and I mean; Fraser 1990). I will therefore begin each of the following chapters by first describing the individual functions of each marker identified in the data. As mentioned before, the approach to corpus data followed in this research project is bottom-up. Accordingly, the data was not analyzed based on a given framework;
Discourse markers in English discourse
rather, the categorization system described in the following chapters was developed on the basis of the corpus. The method applied was basically the same for all four markers. First, all transcript lines containing the linguistic item in question (i.e. so, well, you know, and like), whether discourse marker or not, were extracted with the help of a PERL program2 and sorted according to speakers. Of these, I randomly selected a test corpus of about one hundred instances per item, supplied them with context, and identified the function of each item. In this process, I had the functions mentioned by other researchers as a mental background, but did not necessarily try to apply them to the data; rather, the data themselves were the source of whatever functions and recurrent patterns I would find. These recurrent patterns formed the basis for setting up categories (see Chapters 2–5 for details on the categories) for each linguistic item. Within several days, I repeatedly categorized the test corpus and refined the categories until I reached a certain level of coincidence in each categorization process. After an interval of one to three weeks away from the data, I recoded the test corpus and achieved a reliability of minimally 71% even when excluding those instances which were obviously not discourse markers (such as so in so much, well in he did well, you know in did you know, or verbal like). In psycholinguistics, with its rigorous methodological standards, a recoding reliability of 70% is generally considered sufficient for rendering reliable results; therefore, my category descriptions could be deemed stable enough to be used for the whole corpus. Based on these descriptions, I then categorized all instances of the linguistic item in question, with constant reference to the full transcripts for the context of the item (cf. Biber 2000: 287). According to Erman (1987: 121; cf. also Erman 2001: 1342), discourse markers may fulfill more than one function at once. However, Stubbe and Holmes (1995: 64) cite Holmes (1984) saying that “it is usually possible, in context, to identify one primary function”. This coincides with my own experience. Where more than one function was present in a single instance of a discourse marker, I always chose the function that was most salient in that particular instance. In general, all categories had to comprise at least about ten instances. When there were fewer instances of a function, or when it was difficult to draw a clear line between adjacent functions or to repeatedly classify such instances in the same way, they were subsumed (with similar functions) under one category. Many discourse markers also have homonyms which do not manifest the features listed above, such as well as a noun, an adverb, an adjective, or in an idiomatic or phrasal use (cf. for example COED 1976; Cobuild 1987; OALD 1989). These homonyms were also categorized in the data, but they will not be counted as discourse markers. Another class of elements which does not fit the description of discourse markers given above is like functioning as quotative device. It is generally not optional, and it contributes to the proposition which contains it. Nevertheless, I will follow common practice in literature and consider it to some extent along with other functions of like. Another PERL program transformed the results of my categorization into quantitative data, which were then submitted to a multitude of statistical analyses, using
Chapter 1. Introduction
the software package SPSS, to compare the frequencies of native and German speakers (and subgroups thereof) per discourse marker function or function groups. Due to the lack of normal distribution, which is required for a number of statistical tests, the Median test (which does not require normal distribution) proved to be the most useful test to compare speaker groups. Detailed results are given in the respective sub-sections of the following chapters.
.. Discourse marker functions and levels of discourse In her seminal book on discourse markers, Schiffrin (1987) establishes a discourse model with five planes of talk on which markers function: exchange structure, action structure, ideational structure, participation framework, and information state. These planes differ from each other in the type of their basic units. Of the discourse markers Schiffrin analyzes, each functions on at least two and up to all five of the planes but has its primary use on one specific plane. The basic unit of the exchange structure is the turn, and the task of discourse markers consists of organizing and smoothing turn-taking. Six markers (well, and, but, or, so, y’know) function in the exchange structure; however, none of them has its primary use on this plane. Speech acts are the basic unit of the action structure; at its center is the sequencing of acts. Again, there are seven discourse markers which, as one of their secondary functions, mark speech acts (oh, well, and, but, so, because, then), but none have this function as their primary use. The ideational structure has as its basic units propositions, “or what I‘ll just call ideas” (Schiffrin 1987: 25f.). Seven markers have their primary use on this plane: and, but, or, so, because, now, and then. In addition, well, I mean, and y’know are represented as markers with secondary uses. They provide cohesive ties between clauses and utterances, and they organize the transition between different topics and between topics and subtopics. A third function lies in indicating relations between ideas such as marking background information, explanations, illustrations, or reasons. Within the participation framework, relations between speaker and hearer and also relations between the speaker and his/her utterance, for example commitment or evaluation, are expressed and negotiated. The primary markers to do this job are well and I mean, supported by the secondary uses of oh, so, now, and y’know. Finally, oh and y’know (plus well, so, because, then, and I mean) help to manage and organize knowledge and meta-knowledge, what the speaker knows and what s/he assumes about the knowledge of the hearer. While the focus of the participation framework is on speaker and hearer in their social interactional capacities, here it is on their cognitive capacities. Redeker, in a review article, criticizes Schiffrin’s model for several reasons. Two of the main points are (a) that there are inconsistencies between the descriptions of the individual discourse markers and their place in her model (resp. a table assigning markers to planes of the model), and (b) that the planes in her model are not all
Discourse markers in English discourse
comparable nor well-defined or consistently treated. Regarding the first point of criticism, Redeker writes: “Examples of marker functions not included in DS’s [Schiffrin’s] table can be found in DM [the book Discourse markers], in the narratives described in Redeker (1986), and elsewhere in the literature on spontaneous talk” (Redeker 1991: 1151). Following this statement, Redeker presents examples of markers which Schiffrin describes as functioning at a particular plane, and analogous examples with other markers which have not been accorded the same place in the her model. Furthermore, Redeker complains about the lack of examples for marker-plane relationships; some markers’ function on particular planes (e.g. so in action structure) are only scantily illustrated (Redeker 1991: 1149). Regarding the second point of criticism, Redeker argues that “the components information structure and participation framework are clearly not on par with the other three planes. The cognitions and attitudes composing those two components concern individual utterances, while the building blocks at the other three planes are relational concepts” (Redeker 1991: 1162). In the course of the book, moreover, the notion of ideational structure seems to be defined in various ways, surrounding it with “unnecessary haziness” (ibid.). Also, she considers it a shortcoming that Schiffrin does not include monological discourse in her model. Redeker then (p. 1167f.) suggests a model with only three levels instead of five to accommodate the discourse marker functions described by Schiffrin and found in Redeker’s own data. Her ideational structure is roughly identical with Schiffrin’s ideational structure, comprising temporal sequence, elaboration, cause, etc.; her rhetorical structure is roughly equivalent to Schiffrin’s action structure. The third level is called “sequential structure” and supposedly is “an extended variant” of Schiffrin’s exchange structure. However, the basic units in the exchange structure are turns, while Redeker’s definition of sequential transitions rather reminds of ideational relations: “A paratactic sequential relation is a transition between issues or topics [...]. Hypotactic sequential relations are those leading into or out of a commentary, correction, paraphrase, aside, digression or interruption segment” (p. 1168). While I endorse Redeker’s criticism about inconsistencies in Schiffrin’s work, I do not find her model more convincing. It does not accommodate all functions of discourse markers either (for example in turn-transitions or the negotiation of interpersonal relationships) and does not seem to be any more precise in its definitions than Schiffrin’s model. My own research was not intended to result in a comprehensive model of discourse. Nevertheless, when I developed the functional categories for each of the four markers based on repetitive patterns in my data, I found that three of the four markers have both textual and interactional functions, with several functions at each of the two levels. One marker, like, functions only at the textual level, as we will see in Chapter 5. My categorization of textual functions comes close to Schiffrin’s functions in the ideational structure, but it goes slightly beyond. Most of these occurred in the narrative parts of the GLBCC, although they were not entirely restricted to the narratives. These functions have in common that they do not directly address the hearer; instead, they remain focused on lexical expressions and propositional content expressed in units of
Chapter 1. Introduction
various length, from single words or phrases to a sequence of utterances describing a particular scene. Thus, focus on lexical expressions is found in instances where the discourse marker is used to indicate a search for a phrase which expresses what the speaker has in mind, and where it is used to mark false starts and repair. Also, the marker can indicate that the following term is not necessarily an exact one. Other functions at the textual level are to structure propositional content: for example, they mark transitions from one scene to another, from main ideas or the main story line to explanations and exemplifications and back to the main idea again. Markers may also serve to distinguish between the speaker’s own voice and the voice of somebody else by introducing quotations. In the course of my analysis, it became obvious that several of the textual functions are served by more than one marker, though in different ways. Instead of focusing on the text (i.e. the exactness and appropriateness of verbal expressions or relationships between propositional contents), discourse markers used interactionally focus on the relationship between speaker and hearer. They are found more often in the discussion part of the recording sessions. In several functions, the respective discourse markers occur at transition relevance places and mark a speech act, a response, an opinion, or an evaluation. In other functions, they contain an appeal to the hearer; this is particularly the case for you know. Like can also appeal to the hearer by directing his/her focus to an individual phrase or word. In general, it can be said that the interactional functions are more particular to the individual markers than are the textual functions. Again, this list is not complete, and the details will be presented in Chapters 2–5.
. The Giessen-Long Beach Chaplin Corpus (GLBCC) The Giessen-Long Beach Chaplin Corpus (GLBCC) was collected during a three-year project funded by the Deutsche Forschungsgemeinschaft (DFG) for research on reference assignment. This project was conducted by Andreas H. Jucker (Justus Liebig University Giessen) in cooperation with Sara W. Smith (California State University at Long Beach); the recordings were made both at Giessen and at Long Beach, mostly by the author of the present study.
.. Purpose of the corpus To obtain spoken data that would suit the purposes of the project, an experiment was designed in which pairs of students told each other about a film they had been shown. Several considerations played a role in the experiment design. The foremost consideration was to create suitable data for the topic of the research project, reference assignment strategies. How do speakers introduce characters they want to talk about? How do they introduce main characters versus minor characters? Those who are known to the interlocutor versus those whom the interlocutor does not know?
Discourse markers in English discourse
These were some of the questions at the center of the research project. Previously, a set of sessions had been recorded by Jucker and Smith in which pairs of university students talked about topics which the researchers had assigned. However, these recordings turned out to be of little use for the project, which led to the initiation of the GLBCC, because the researchers often did not know the characters referred to in the interaction nor to what extent they were known to the interlocutors. What was needed was a topic for the speakers to talk about which would contain references to a limited number of characters, known to the researchers, but both known and unknown to at least one of the participants. The ideal solution consisted in showing a short movie to them, with one partner watching all of the movie and the other being called out in the middle of it. The person being called out first retold the first half of the movie in a monolog, while the film continued in another room. The participant who had seen the whole movie was then asked to tell his/her partner what had happened after s/he had been called out and discuss it with him/her. Thus, the researchers had control over the potential characters referred to, and there were unknown and known characters mentioned in the narrative and ensuing discussion. At the same time, the experimental setup was chosen as a compromise between controlled laboratory experiments, such as are often found in psycholinguistics, and free conversations. Admittedly, the surroundings (offices and other rooms at the university) were a rather artificial environment. However, the task of telling another student about a movie and then talking about it seems to be authentic enough. Erman (2001: 1343) reports: “Young people seem to be more preoccupied with telling stories and reporting events which they have heard of, seen in films, or experienced themselves, than engage in argumentative discourse”. The participating students mostly signed up in pairs, sometimes with friends, sometimes with strangers, so that they usually were not talking to the researcher or someone whom they might have expected to already know the movie. Another focus of the research project lay in comparing native with non-native performance. For this purpose, recordings were made in two places, Giessen (Germany) and Long Beach (California, USA), and with a mind to involving native speakers of English and non-native speakers of various linguistic backgrounds. Due to practical reasons, most of these non-native speakers were Germans; however, considerable numbers of speakers also came from Spanish-speaking, Japanese, Chinese, and Khmer (Cambodian) language backgrounds. In total, speakers of twenty-three first languages are represented in the corpus. For my own research, though, I concentrated on American native speakers of English and German speakers of English as a foreign language. All of the speakers were students and had mastered English well enough to take classes taught in English. In addition, participation was voluntary and the students knew that they would have to talk in English, which means they were confident enough in their English language skills to venture participating in the project. Nevertheless, we wanted to keep the influence of language in the movie on the participants’ verbal production to a minimum and therefore chose a silent Chaplin film, The Immigrant, as the movie to be shown. Admittedly, it is not completely without language; occa-
Chapter 1. Introduction
sionally a black screen with some words is shown (inter-titles), such as I’m an artist. The participants regularly included the content of several of the fourteen inter-titles in their narrative, but they did not necessarily use the words shown on the screen. Likewise, the inter-titles did not affect the understanding of the non-natives negatively; it was rather some of the native speakers who had difficulties interpreting two of the inter-titles correctly. In terms of cultural particularities (cf. Du Bois 1980), we found that, even though the movie related to speakers from different cultural backgrounds in different ways, the native speakers did not have any advantage over the non-native speakers. Quite on the contrary, several of the non-native speakers identified to some extent with the situation of an immigrant. In total, 136 recordings in English were transcribed; later, another set of twentyfive recordings with the same experiment setup was added, but with the narratives and discussions conducted in German language. How does this corpus compare to other narrative corpora? When designing this corpus, we had two other, similar corpora in mind, the Pear Stories and Redeker’s movie experiment. There are a number of obvious differences between the GLBCC and the Pear Stories (see Chafe 1980a, 1980b; and Du Bois 1980). First of all, the latter corpus was meant from the beginning to include narratives in several very different languages. For Chafe and his team, a film without any words was therefore imperative. In contrast, for this study as well as for the research on reference assignment strategies, some language in the movie was considered acceptable, as long as it did not present an obstacle for understanding, since (in the beginning) all recording sessions were supposed to be conducted in English. While Chafe et al. wanted to record native speakers in their home countries, we were intent on native and non-native English and therefore could restrict ourselves to recording in Germany and the USA. Our participants were exclusively university students; while most of the narrators of the Pear Stories were also university students, several participants were also included who came from less educated backgrounds or were children in the first and fourth grade. In addition, unlike the Pear Story narratives, the GLBCC data do not contain written narratives nor repeated narratives from the same speaker. Redeker (1990) also used silent movies, although hers were not complete films but extracts, which made the experiment setup somewhat less natural. Her narrators also were exclusively university students, while half of the listeners were non-student Berkeley residents, but all were native speakers of English. Another major difference from the GLBCC consists in the fact that Redeker’s participants communicated only via headphones and microphones; they did not see each other. This further contributed to the setting being more artificial. Finally, the two corpora immensely differ in size: Redeker recorded only sixteen sessions, as compared to 136 in the GLBCC. Table 1.1 summarizes the similarities and differences of the three corpora/projects. As mentioned above, only those of the well over two hundred participants who were either (American) native or German EFL speakers were selected for this study. This resulted in seventy recordings being analyzed; thirty-four of the participants being American NS and seventy-seven German EFL speakers.3
Discourse markers in English discourse
Table 1.1 Differences between the GLBCC and two other corpora based on silent movie narratives Giessen-Long Beach Chaplin Corpus (Jucker et al.) language(s) of the recordings language in the movie type of movie
Pear Stories (Chafe et al.)
English 10 different languages (plus some recordings in German) silent with (English) silent inter-titles Chaplin self-produced
recording situation face-to-face
face-to-face
type of material
oral and written narratives students, children, participants from less educated background interviewers (research team) more than 230?*
speakers
oral narratives discussions students
hearers
students
corpus size
136 recordings (plus 25 recordings in German)
Redeker’s silent movie experiment English
silent extracts from two silent movies via headphones and microphones oral narratives students
students/other Berkeley residents 16 recordings
* Chafe (1980b) says that the film was shown to fifty speakers of English and to twenty speakers of each of the other nine languages. He also mentions that the research teams obtained repeated narratives from some of the participants, without giving the total number of recordings.
Due to the experiment design, it could not be expected that all discourse markers would be used, nor that those which occurred would be used in all their functions. For the present study, however, this potential disadvantage was more than offset by the particular advantage of the GLBCC: All participants, whether native or non-native, saw the same movie, had the same task, and participated under the same circumstances, which makes the data of native and non-native speakers perfectly comparable. This is an advantage that none of the multinational corpora published so far has been able to offer.
.. Experiment design The exact design was as follows: In the experiment, pairs of university students (A and B) watched a 24-minute silent Chaplin movie, The Immigrant, which consists of two parts. The first part is set on an immigrant ship, where Chaplin (among other things) wins money in card and dice games and falls in love with a young lady. Partner A was called out in the middle of the movie, when the ship arrives in New York, and asked to orally retell the first part of the story. Partner B saw the whole movie and
Chapter 1. Introduction
was afterwards asked to tell his/her partner what had happened in the second part. This part is set mostly at a restaurant, presumably in New York, where Chaplin finds and loses money, orders food, meets the lady again, and is rescued from his financial problems by an artist who wants to paint the two of them. (For more details of the movie see the summary in Appendix 1.) Both partners then discussed the movie, often making use of some questions that they had been given as suggestions or a guideline for their discussion. Thus, some parts of the discussion and also the function of some discourse markers become clear only if the existence of these questions is considered and their content known. The instructions after the movie read: Only one of you, Partner B, has seen the second half of the video. We would like for Partner B to describe (orally) what happened during that section. Then we would like for both of you to discuss how you liked the film. Please discuss questions such as: – Did you like it? Why? – What did you think was the funniest part? – What did you think of the acting by major characters? – What did you think of the acting by minor characters? – What do you think were the major challenges when making a silent movie? For pairs, the role of the participants (which one would watch only half of the movie, i.e. role A, and which one would watch all of it, i.e. role B) was determined by drawing straws. Sometimes, only one person signed up for a recording session; in other cases, only one of two students showed up at the time assigned for the recording. When this was the case, the single participant watched and retold the whole movie and expressed his/her opinion about it, also along the lines of the questions presented above. These single speakers in the corpus are referred to as speakers in role C. In addition, the participants answered questions about their relation to the movie and filled out a demographic questionnaire. Apart from questions about age, gender, and their relationship to the partner, we also asked the participants, “What is your first language?” If they gave English for their first language, we considered them to be native speakers (see Section 1.7.2.1 for a discussion of the native speaker definition). In this part of the experiment, I was able to add questions to the experiment design to meet the needs of my own research. The questionnaire was supposed to give us information not only about the age and first language of the participant, but also about where and how the non-native speakers had acquired their English skills and how much they used them in which contexts. In short, the additional questions should give us information about potential factors influencing the linguistic performance of the non-natives with regard to discourse markers. For the non-native speakers, we therefore included questions about where (i.e. in which country), how long and how they had learned English: in school, at the university, in a family setting or with friends? We also inquired about the language(s) they used in formal and informal contexts, with same age conversational partner and with older family members. A third ques-
Discourse markers in English discourse
tion asked how long they had been to the States or another English-speaking country, and at which age (see Appendix 2). The participants knew they would be recorded on audio and video tape; however, during the experiment they frequently forgot at least about the video tape. They had also been informed that we were interested in how they used language, but did not receive any details on the exact interest of our research until after the experiment.
.. Transcription and intonation unit (2)
14 and yo- you imagine that he’s vomiting as well, 15 but (H) @ then (H) he turns around and he & 15a & he was16 he just got a fish out of the .. out of the sea, 17 @@ 18 so. 19 ...(1.4) (H=) um then they all-20 all the people on the boat go and have & 20a & their mea=l, 21 ...(1.6) a=nd .. well in the between time & 22 & you you sometimes you see uh a woman .. & 22a & together with her mother, (210C, non-native speaker (NNS))
The conversations/narratives took about fifteen minutes; approximately five minutes for A’s narrative and ten minutes for B’s narrative and the discussion.4 They were recorded on audio and video tape; the transcriptions were done at the Justus Liebig University, Giessen, Germany, basically following Du Bois’ (1991) transcription design principles (for a complete list of the transcription symbols see Appendix 3). This means that the result is an orthographic transcription, in which non-standard pronunciation is only marked for forms which are acknowledged in dictionaries, such as gonna, wanna, etc., and common contractions. Like the Pear Stories, “[r]ecording of intonation in the transcriptions was limited to the use of a period for sentence-final falling pitch, a question mark for sentence-final non-falling pitch, and a comma for any of several clause-final but not sentence-final contours” (Chafe 1980b: xv). This description implies that the transcribers place as much value on syntactic boundaries (i.e. clause and sentence boundaries) as on intonation contours to determine intonation units. At first, this procedure might seem an unwelcome mixing of levels, but in any case it is a necessary one. Erman (1987: 38) says that already “Brown & Yule [...] found that people, even trained phoneticians, had difficulties in identifying tone groups by phonological criteria alone (1983: 161)”. In addition, she cites Laver (1970: 69) claiming that “the boundaries of the tone group often, though not always, coincide with those of the syntactic clause” (Erman 1987: 37). This statement fits with the experience of the transcribers of the GLBCC. However, terminology differs between Erman and Du Bois: what Erman dubs “tone group” is called “intonation unit” by Du Bois and also by the transcribers of the GLBCC. Its
Chapter 1. Introduction
definition, in principle, coincides with the one which the researchers He and Lindsey (1998: 137) give: “intonation units (IUs), defined as a stretch of discourse falling under a single coherent intonation contour (Ford and Thompson 1996, Tao 1996)”. However, in accordance with the authors mentioned above, we have to add that the coherent intonation contour can only be judged by drawing – at least implicitly – on syntactic considerations.5 As Erman observed, this coherent intonation contour may also contain short and longer pauses, for example after a conjunction in order to hold the turn (Erman 1987: 38; cf. line 21 in extract (2)), or within the intonation unit when the speaker is searching for the appropriate words, as for example in line 16 of the extract. When the intonation contour seemed to be incomplete, it was marked as a truncated intonation unit (lines 15 and 19). In general, one intonation unit occupies one line in the transcript; the following IU is presented in a new line. When an intonation unit is too long for a single line, it is marked with an ampersand (&) at the end of this line and continues with another ampersand on the following line (lines 21–22a in the extract). Great care has been taken in the GLBCC not only to identify intonation units, but also to correctly and consistently represent pauses, repetitions and truncated words (cf. line 14: yo- you) as well as overlaps. Overlaps are marked by square brackets and vertical alignment; where several overlaps occur in subsequent intonation units, the second and third overlap are marked by two or three square brackets, as can be seen in extract (3). (3)
98 A: [it was] not so [[funny or so whatever]]. 99 B: [I] [[I= yea=h]], 100 I got sick by watching them= .. going & 100a & like [[[this .. just]]]-101 A: [[[up and down]]], (46A+B, NNSs)
The initial transcriptions were done by student assistants; however, the author of the present study diligently corrected all of the transcripts used in this study herself to obtain the highest possible level of consistency.6 In addition to intonation units, two more features have been marked that fall into the realm of intonation in discourse (cf. Couper-Kuhlen 2001). Thus, words spoken with particular emphatic stress are rendered in capital letters (cf. extract (4)), and stretches of speech in a particular speech quality are bracketed to indicate that speech quality. We differentiated four qualities: – – – –
whisper (extract (5), line 184), laughter quality (extract (5), lines 185f), low voice (extract (6)), and quotation quality (extract (7)).
The last one was attributed when the speaker changed his/her voice to indicate that the words spoken in this different voice quality were not his/her own.
Discourse markers in English discourse
(4)
148 and he started checking his pockets to see & 149 & if HE had enough money to pay HIS bill. (69B, native speaker (NS))
(5)
184 185 185a (57B,
(6)
61 <SV that’s all SV>, (27A, NNS)
(7)
295 he said & 295a &, (28B, NNS)
<WH what do you WH>-...(2.2) <@ what did you think of the & & acting by major characters @>. NS)
One further issue might be mentioned here regarding anonymity (cf. Hasund 1998). Since participants in our project talked about a movie, references to their private life did not result in any recognizable details. No full names of private persons were mentioned in the recording. First names were mentioned a few times and were substituted in the transcripts arbitrarily by very common names, such as “Jim” or “Sue”. So far, the recordings of the GLBCC have not been published, but in any case it seems unlikely that these occasional first names could be sufficient to identify a speaker. The speakers themselves are always referred to by the number of their recording and their role in the experiment (e.g. “speaker 200B”); there are no records correlating these references and the names of the speakers. Thus, full anonymity is guaranteed.
. The data For many corpora and smaller projects, the size of the corpus is given in recorded time and/or tokens of words (e.g. Watts 1989; Johansson 1998; Svennevig 1999; Smith and Jucker 2000; Andersen 2001; Kaufmann 2002; Rieger 2003). In the GLBCC, speakers in role A needed approximately five minutes for their narrative of the first part of the movie. For the retelling of the second half and the subsequent discussion of the movie, both speakers, A and B, were allotted ten minutes. Single participants usually fulfilled their task (retelling the whole movie and giving their opinion of it) in about ten minutes. The subsection of the GLBCC which I have been using therefore consists of approximately 6.5 hours of recording and a total of 155,508 words. While the recording time cannot be assigned to individual speakers, since speakers interact in the narrative of speaker B and in the discussion section, total words per speaker and per speaker groups could easily be calculated (see Figures 1.1 and 1.2). Thus, the Americans spoke 53,028 words, with an average of 1,560 words per speaker, the four British were slightly more talkative with an average of 1,731 words (total: 6925), while the German speakers’ speed of talk was apparently impeded by imperfect language skills, especially in terms of vocabulary. They only managed an average of 1,241 words (total: 95,555 words), even though their range of words per speaker was
Chapter 1. Introduction US NS 34%
1560 words
1731 words 1241 words
GB NS 4%
Germans 62%
Figure 1.1 Distribution of words per speaker group and average words per speaker role A 36%
1133.45 words
1542.76 words
1374.27 words
role C 10%
role B 54%
Figure 1.2 Distribution of words per role and average words per speaker
not considerably different from those of the Americans: While the Germans covered from 531 to 2323 words per speaker, the Americans figured from 522 to 2765 words per speaker. Speakers in role B contributed more than half of the total sub-corpus, while speakers in role A contributed more than a third. In general, the highest average of words per speaker was produced by participants in role B, followed by those in role C. However, German single participants were more talkative than other German subjects and more
Discourse markers in English discourse
than the Americans in roles A and C, while American in those two roles produced approximately the same average of words. The (average) number of words per speaker plays an important role in my statistical analyses; frequencies of discourse markers or their individual functions are all calculated based on this value.
. Influential factors for the use and distribution of discourse markers So far I have hypothesized that I will find differences in how native and non-native speakers use discourse markers and in the frequencies of their individual functions in the two speaker groups. Being a native or non-native speaker, however, is not the only factor which might influence the frequency of discourse markers. There are also non-linguistic factors said to be relevant, for example demographic features. In Section 1.7.1, I will present such features, present the findings of other authors and describe the extent to which they have been considered in the GLBCC. Section 1.7.2 then deals with a number of linguistic factors. Table 1.2 gives an overview of both types; factors which are not considered in Chapter 2–5 are printed in italics.
.. Non-linguistic factors ... Gender Most studies which investigate social distribution include gender as a potentially influential demographic feature. Ever since Lakoff wrote about Language and Woman’s Place in the 1970s, characterizing women’s language as powerless and suggesting that women use more hedges and tag questions (e.g. 1975), certain discourse markers have also been among the linguistic elements which are supposedly more frequently found in women’s language. Two markers have probably been associated with females most Table 1.2 Factors which may have an influence on discourse marker use and frequency non-linguistic factors
linguistic factors*
– gender – age – social class – ethnicity – relationship between the partners – role – formal vs. informal context for the recording
– native versus non-native speakers – acquisition of English in formal and informal contexts – usage of English in formal and informal contexts – abroad versus not abroad – British versus American influence – native speaker contact
* Normally, the term ‘linguistic factors’ refers to the textual context, such as syntax and collocations. I am using the term here to refer to factors which are not demographic or social but directly related to language use, such as the acquisition and use of English in different environments.
Chapter 1. Introduction
often: you know and like. Östman (1981: 71f.), for example, concludes that “you know tends to be used more by women than by men” even though this was not the case in all conversations. In addition, he found that the scope of you know seemed to play a role; while women qualified larger units, such as speech acts or information units, with you know, men used it rather to modify phrases and lexemes (1981: 75). Holmes (1986: 12f.) arrives at slightly different results. In her (presumably New Zealand) data, you know was equally frequent in the speech of men and women. Like Östman, though, there are particularities in the purpose of you know, in that “women use you know significantly more frequently to convey speaker certainty, while men use it more often to convey uncertainty” (1986: 13). Men indicate linguistic imprecision with you know; women put it to attributive and emphatic use. Because such differences can be found, Holmes places particular emphasis on the distinction of functions of an individual marker. According to her, much previous work on women’s language has been done without distinguishing different uses of you know, or counting absolute numbers of hedges without taking into account the total number of words by each speaker, thus rendering itself useless (1986: 4). Erman (1992: 231f.) basically supports Holmes’ findings and argument, even though she describes her results differently. For instance, men’s you know tended to occur between incomplete propositions, while women’s you know occurred between complete propositions. Another interesting observation she makes is that pragmatic expressions (including you know) “are much more frequent in same-sex as opposed to mixed-sex interaction” (1992: 228). Ten years later, Macaulay (2002) investigated Scottish speech, only to find again that, based on the data, “[w]omen are more likely to use you know than men”. However, this statement is not followed by an evaluation of gender speech; in addition, Macaulay warns that his conclusions are tentative and “their significance may not extend beyond the boundaries of Scotland” (2002: 765). Concerning the discourse marker like, both Romaine and Lange (1991) in a study of American English and Andersen (2001) in his investigation of London teenage language found that females use it more. Tagliamonte and Hudson (1999) corroborate this tendency also for quotative be like in British data from university students. In Andersen’s teenage data, however, quotative be like shows no significant gender differences. Blyth et al.’s (1990) analysis even revealed a higher probability for men to use be like, which contradicted their hypothesis and might possibly be due to a low number of participants (10 male, 20 female). Ferrara and Bell (1995) were able to show that, in 1990, there was still a tendency of be like being used more by females than males; in 1992, however, frequencies were equal at least in the group of the 16- to 39-year-old people. This is also confirmed in a study of informal U.S. English by Dailey-O’Cain (2000) for both the discourse marker like and quotative be like. However, obviously the perception of people regarding like and gender still is biased towards a higher frequency on the female side (2000: 60). All four discourse markers treated in this study and their respective functions will be checked for possible differences not only for male versus female speakers but also for same-sex versus mixed-sex pairs. The results are presented in Chapters 2–5.
Discourse markers in English discourse
... Age Another frequently mentioned demographic factor is age. Again, relevant studies focus mainly (but not exclusively) on you know and like / be like. Thus, Blyth et al. (1990: 219) report that the “use of be like dropped off sharply after the age of 25 and disappeared altogether at the age of 38”. Likewise, Ferrara and Bell (1995) found that, in 1992, their participants of age 40 and older did not use be like at all. By the end of the 20th century, even older people seem to use both be like and like as a discourse marker. Older people in Dailey-O’Cain’s (2000) data means speakers aged 30 to 49, and a similar, though less restricted age range in Andersen (2001, referring to the spoken section of the BNC). In both cases, younger speakers employ both types of like much more often than the older speakers. Andersen, however, who divided the teenagers of the COLT into three age groups (10–13, 14–16, 17–19), found that among the London teenagers, discourse marker like (but not quotative be like) increases with age (2001: 288f.). For you know, Stubbe and Holmes (1995) did not find any age effect among New Zealand middle-class speakers in conversation, considering “young speakers (18–34 years old) and middle-aged speakers (40–55 years old)”. Macaulay (2002: 765), however, concluded from his data that “[a]dolescents have not yet developed the use of you know as a characteristic of their discourse style”, in a comparison of 13–14-year-old adolescents with adults aged 40 and older (2002: 750). From these examples, we can already see that “old” and “young” may be defined in a variety of ways. The question arises whether there is any reasonable ground on which to divide the age range into groups. Andersen (2001), for example, applies his age division to the COLT data without providing any reasoning for it. Dailey-O’Cain chose her informants according to age groups; like Andersen, she does not give any explanation for the boundaries of these age groups either. In the case of the GLBCC data, one division offers itself: The American speakers are aged 18 to 21 years and 25 years and older. There are no speakers aged 22 or 24, and only one aged 23. With this division, the younger group has twenty-one members, the older one nine. For the Germans, a third, intermediate group of speakers aged 22 to 24 was introduced. This group is the largest with 37 members; the youngest group and the oldest are of approximately the same size (17 and 23 members). ... Social class Several of the authors mentioned so far also considered the social class of their speakers. For like, excluding it’s like and quotative be like, Andersen (2001) maintains that it is the higher class adolescents who use it more than the middle and lower class adolescents from London. Dailey-O’Cain does not directly refer to social class, but she mentions that several of her informants associated like with southern Californian ‘Valley Speak’ and ‘Valley Girls’ (2000: 70, 76). This term usually refers to the language of middle-class white people or, viz., to middle-class white teenage girls in southern California. Of the New Zealand speakers represented in Stubbe and Holmes’ (1995) study, working class men use you know considerably more than middle-class men, even
Chapter 1. Introduction
though it is by no means infrequent in the middle class. This effect is particularly pronounced for young men. In contrast, Macaulay did not find a class difference for you know overall. Nevertheless, he observed that “middle-class speakers are more likely to use you know medially in an utterance for purposes of self-repair or elaboration, while working-class speakers mainly use you know at the end of an utterance” (2002: 765). In designing the GLBCC, we opted for including only students as participants. Thus, we made sure the non-native participants would have mastered a certain level of English sufficient to conduct a conversation. Beyond this, we have not been able nor did we consider it necessary to check on the social class of the students’ background.
... Ethnicity I mentioned above that speakers in Dailey-O’Cain’s (2000) associated like with ‘Valley Girl’, i.e. white middle-class girls from southern California. Even though ethnicity was not explicitly mentioned as a factor in her study, the association implies that other ethnic groups do not use like at all or as frequently. Andersen confirms this association at least for adolescent language in London, albeit on the basis of scant information and less than obvious results regarding ethnic distribution of like. Against this background, it would be interesting to see whether the ethnicity of a speaker (e.g. African American vs. white) makes a difference in discourse marker use and frequency. Unfortunately, we know little of the ethnic background of the American speakers, since we were not allowed to make distinctions based on ethnicity in the GLBCC. It is obvious, however, that the group included some African American speakers. We are not permitted to identify them (e.g. based on the video tapes); though several transcripts contain non-standard linguistic features such as regular past forms of verbs instead of the irregular standard, we cannot correlate the transcripts to ethnicity. Presumably, there are also some speakers with an originally Hispanic family background who participated in the experiment. Nevertheless, in the questionnaire, all of them answered the question “What is your first language” with “English” (see Section 1.7.2.1 below for a more detailed treatment of the issue). ... Relationship between the partners Now we are leaving demographic features of the speakers and turn to other factors which have been claimed to be influential. In Redeker’s (1990) experiment setup, half of the participants retold the movie extract they had seen to a friend. The other half told it to a complete stranger. In addition, while the speakers and their friends were all students, the stranger listeners were “non-student Berkeley residents (members of the P.E.O. Sisterhood)” (1990: 370). In her data, “speakers who were talking to a stranger used more ideational markers than those who were talking to a friend”. ‘Ideational markers’ in her definition include “semantically rich connectives” such as because and so, and cover a wide range of markers, from question words to temporal connectives and causal conjunctions (1990: 372).
Discourse markers in English discourse
Jucker and Smith (1998) also found an influence of the relationship between the speakers on discourse marker frequency, not only for particular markers (like, oh, well) but for two groups of markers. Jucker and Smith distinguish between reception markers and presentation markers. “Reception markers are used to signal a reaction to information provided by another speaker. Typical examples are yeah, oh and okay. Presentation markers, on the other hand, accompany and modify the speaker’s own information”, for example like, you know, and I mean (1998: 174). The speakers themselves rated their relationship, choosing from “strangers, acquaintances, friends, or good friends” (1998: 175). It turned out that strangers showed a higher rate of reception markers than friends, while friends used presentation markers more frequently (1998: 177). However, in comparing results from the GLBCC with the previous project by Jucker and Smith on which their 1998 publication is based, some care must be exercised, since the respective experiment setups are different. In the previous Jucker and Smith project, participants were given topics to talk about, including time to “get to know each other” or to “catch up with each other (1998: 175), while in the GLBCC, participants were given a more narrowly defined task, i.e. retell a movie and discuss it. Thus, personal matters did not play a dominant role, and at the same time, communication was facilitated through the task. In the GLBCC, no provisions were made to ensure a systematic distribution of strangers and friends. The participants were just asked in the questionnaire how they rated their relationship: friend, acquaintance, stranger, without any further explanations. However, from the interaction between partners, the degree of acquaintance between them was not always obvious. Rather, it seemed to depend on sympathy and/or discourse style. Thus, even pairs of complete strangers were sometimes found to discuss the movie with a degree of intimacy as if they had known each other for years. Nevertheless, the students’ self-assessment of their partner relation will be considered in the statistical analysis of each marker. Participants will be divided into two groups, friends forming one group, and acquaintances and strangers forming the second group under the label ‘strangers’. This is necessary for the sake of comparability, because among both Americans and Germans there were about twice as many friends as strangers and acquaintances taken together.
... Role A function which is also potentially relevant in my data is the role of the speaker. To my knowledge, there is no literature on discourse marker use in data with the same or a similar experiment setup. Fuller (2003) analyzed marker frequency in different contexts, interview and conversation, and for the roles of interviewer and interviewee. However, these roles do not compare to the specific conditions of the speakers in the GLBCC. In this corpus, all speakers’ data contains a narrative section and a discussion section. While speakers in role B always had a listener who was not only visible but gave backchannels, asked questions, and contributed equally to the discussion part, speakers in role A completed their narrative of the first part of the movie with maximally a
Chapter 1. Introduction
visible but silent listener. Speaker A was told to imagine that a friend asked for details about the movie, and had requested him/her to leave a message on the answering machine. Some participants vividly imagined the scene and talked as if they really talked to a friend. Others asked for permission to look at the researcher. This request was granted; none of them, however, got a verbal response to their narrative. For the speakers in role C, i.e. those who did not have a partner, the situation was similar. They also talked “on the phone” or were looking at the silent researcher, but they retold the whole movie and added their opinion of it, while speakers in role A had a partner at least for the discussion part. In contrast, speaker B retold the second half of the movie to another person who listened actively and could respond with backchannels, clarifying questions, etc. Therefore, it should not be surprising to find differences in the frequency of certain discourse markers, e.g. you know. More details will be given in the chapters on the respective markers.
... Formal versus informal context for the recording Even though Erman (1987) claims that it is not useful to analyze what she calls ‘pragmatic expressions’ in terms of informal language, many others seem to regard it as an important notion. The formality or informality of a speech situation has often been said to influence the use of discourse markers, or, to put it differently, discourse marker use has been associated with informality (cf. Jucker and Ziv 1998b; Andersen 1998). Svartvik (1980: 172) cites Crystal and Davy (1975: 101f.) saying that well may be used “to maintain the informality of a situation”. Östman (1981) tells about a professor who, after about an hour of teaching, suddenly starts using you know and thus indicates to the students that he has moved to a more informal style. Holmes (1986) explicitly uses data from formal context (TV and radio interviews) and compares them with data from informal contexts (conversations in private homes). She found you know to occur predominantly in informal conversations. For other researchers, it is rather a fact than a hypothesis that at least certain discourse markers such as like are more frequent in informal contexts (e.g. Hasund 2002). Many accordingly emphasize the informality of the recording contexts of their data (for example Sankoff et al. 1997; Dailey-O’Cain 2000; Rieger 2000). Informality typically means that the speakers do not act in public, as in a radio or TV interview; other than that, there does not seem to be any definition but rather a scale of informality. The initiators of corpora such as the COLT (Corpus of London Teenage Language) or the Santa Barbara Corpus of Spoken American Language aimed at recording situations that were as close to natural situations as possible; they provided their recruits with microphones and had them record everyday interactions with friends and family members. Since the fact that the speakers are being recorded may already inhibit total naturalness, Du Bois called them ‘naturalistic’ (2002) rather than natural. For smaller corpora, authors also may record themselves in such interactions (e.g. Watts 1989). Higher up on the formality scale are socio-linguistic interviews with the researcher, which Dailey-O’Cain (2000: 64) also calls informal. Certainly, a lot depends on the relationship between interviewer and interviewee, on the atmo-
Discourse markers in English discourse
sphere which the interviewer creates, and on how conscious the speakers are of being recorded. Hays, who conducted interviews, even claims that the latter plays the more important role: “In even a very free form discussion, when the conversation is being recorded, students are more likely to be monitoring their production than in a casual conversation among friends. [...] Therefore the fact that the interviewer is an instructor, rather than a peer, is not so important” (1992: 28). Romero Trillo, who investigated attention-getting devices such as look or listen, would probably not agree with Hays’ assessment. He found that when the informants are answering the interviewer’s questions, the rate of use of the attention-getting devices is much lower than when the informants are speaking freely, [...] the function of these elements is closely linked [...] to the degree of formality or equality among the speakers in the situation. (Romero Trillo 1997: 218)
In the GLBCC, several factors admittedly contributed to the formality of the situation: university rooms (offices, lab rooms, a class room, a student lounge), the existence of an instructor, consent sheets to sign, a given task and topic rather than free conversations, microphones. And yet, some participants did not seem to mind these factors at all, while others were fairly self-conscious at first, but in the course of the narrative and discussion with the partner obviously became so involved that they forgot about being recorded. Part of the reason certainly is the fact that the task, i.e. talking about a movie, obviously is a very natural activity for teenagers and students. In addition, pair interaction always was with a peer and not with an older researcher. Even the instructor (in most cases the author of the present study herself) often was not considered on par with a member of staff but rather taken for a student assistant. Both facts contributed to the informality of the situation, so that a fair balance between a controlled content of the recordings and a relaxed, informal atmosphere was achieved for most of the session. Since we recorded only in one type of context, the informality of this context has not been the object of or a factor in the analysis.
.. Linguistic factors So far, I have referred to native and non-native speakers without defining the terms. This is a quite common procedure even in papers which focus on these two terms or speaker groups, respectively (for example Varonis and Gass 1985; Watts 1989; Cots 1992; Greenbaum 1996; de Haan 1998; Gass 1999b; Bürki-Cohen et al. 2001; Spolsky 2002). However, the term ‘native speaker’ is not undisputed (cf. Davies 2003). Therefore, the first subsection below is dedicated to this dispute and an explanation of how I use the term. For the native speakers in my corpus, a number of non-linguistic factors (as listed in Section 1.7.2) are analyzed in terms of their potential influence on discourse marker use. Naturally, however, linguistic factors are a focal point in trying to find (tentative and partial) explanations for the marker use by the non-natives, i.e. the German
Chapter 1. Introduction
speakers in this case. As mentioned in the discussion of the experiment set-up, the participants also filled out a questionnaire on a variety of linguistic factors, which are explained in detail below. Here informality does play a role as a potential influence on speech behavior, informality not of the recording context but of the language learning context; and it does so for a simple reason: If discourse marker use is, as shown above, regarded as being related to informal contexts, it must be assumed that competence in this use accordingly can only be acquired in such contexts. Richards and Schmidt (1983) think that conversational competence in another language will not be achieved in second language teaching at school unless the learner is treated as conversational partner (as ‘talker’, as they put it) and real, serious, conversation – i.e. no role-playing – takes place. However, the required kind of real conversation is more likely to be found in (informal) interactions with native speakers than in the classroom. Sankoff et al. (1997: 193) explicitly apply this opinion to discourse markers: “Discourse markers are of particular interest because they constitute an aspect of the language not taught in school. [...] That is, only L2 speakers with a high degree of contact with native speakers will master the use of discourse markers.” Hays, in his study of Japanese students’ competence regarding discourse markers, expresses himself in a similar manner, at least concerning the importance of native speaker contact, even though he would not agree with Sankoff et al.’s view that no discourse markers are part of classroom instruction. Considering his results on so, well, and other markers he concludes that “semantic words, and therefore ideational discourse markers [e.g. so] are to be expected earlier, because they are overtly taught. (. . .) [Markers on other planes] would not be expected unless there has been exposure to that discourse community” (Hays 1992: 29). My claim goes even more into detail: It is not just the level of individual discourse markers at which we find an influence of formal vs. informal acquisition contexts but also at the level of individual functions of a discourse marker. In order to obtain relevant information, our questions were divided into three sections, the acquisition of English, the usage of English, and times spent in an English-speaking country. According to my hypothesis, non-native (German) speakers with a higher degree and longer time of either usage or acquisition of English in informal contexts should use discourse markers and/or some of their individual functions more than those students with less use or acquisition in informal contexts.
... Native versus non-native speakers There are at least two perspectives on the matter of the ‘native speaker’, a linguistic and a socio-political perspective. In discussions from both perspectives, the term is discarded on the grounds of the difficulties to determine the boundaries. Who may and may not count as a native speaker? Is the American who has been living in Brazil for a number of years still a native speaker (Souto Silva 2000)? Or does the Hispanic teenager who came to the United States before the age of ten now count as a native speaker of English (cf. Holmes 1996: 165)? Can only those speakers be considered native who
Discourse markers in English discourse
learned the Standard variety (if there is such a thing) from the very beginning of their life? Or is the native speaker dead, as Paikeday claims (1985)? In the literature about this issue from a socio-political perspective, such as Nayar (1998), we find the complaint about the dominance of American and British varieties in a range of fields: from the teaching of English, to attitudes in the population, to the economic situation. Consequently, Nayar demands the legal, linguistic, and attitudinal acknowledgement and acceptance of national standard varieties, which should also include the replacement of American/British English as a model in language teaching (1998: 288). Considering the issue from a linguistic perspective, Piller (2001: 118f.) claims that so-called native speakers do not necessarily speak their language without errors. In contrast to errors coming from non-native speakers, however, these errors are likely to be excused as poetic license, simplification for a foreigner, negligence, lack of concentration, typo, etc. In addition, there are always people who, for some reason, have not completely mastered their native language, such as speaking-impaired people or children. Even adults without speaking impairment may not speak their language well, according to Rampton (1990: 98). For him, speaking a language well apparently includes the ability to tell or write stories, a skill which obviously not all native speakers possess. Rampton suggests that the native speaker as a model in language teaching should be substituted by language expertise, which is much more easily to define linguistically. It is most probably true that not all native speakers conform to what prescriptive grammarians would see as the correct use of the language. Therefore, Rampton’s suggestion is valuable as far as grammar is concerned. However, where the matter of interest is not the correct usage but the pragmalinguistic behavior of real people, as in this research project, the idea of language expertise is of little value. In other words, when we want to find out about the linguistic behavior of the native speakers, we cannot select speakers on the basis of their linguistic competence, since this would lead us into circularity (Escudero and Sharwood Smith 2001: 277). Thus, the characteristics chosen for the selection of speakers must either be extra-linguistic, or, at least, not be based on our judgment of the speakers’ linguistic competence. It is surprising to see that few of the authors mentioned so far explicitly distinguish the purpose for which the term ‘native speaker’ is used: whether we use it to include or exclude people from a group of speakers, or whether we want to describe the prototype of this group (cf. Escudero and Sharwood Smith 2001: 281; Mukherjee 2002: 111); whether we use it to describe the (average) linguistic behavior of a group of speakers, or whether we want to set up a linguistic norm for language learners. For the GLBCC, we wanted to include prototypical native speakers of English, speakers for whom English was a foreign language (cf. Section 1.3.1), and bilingual speakers. A prototypical native speaker of English in our view would be a person who speaks only English, or a person who learned another language later in life through formal instruction, but still predominantly uses English. To gather the relevant in-
Chapter 1. Introduction
formation, the questionnaire accompanying the experiments for the GLBCC asked: “What is your first language?”, giving the options of “English” and “other (please specify)”. The participants were also told that they could give both English and another language as their first language. For those with a first language other than English (or with two first languages), the questionnaire contained further questions on language use (see Appendix 2). The question about the first language was unproblematic in the case of prototypical native speakers of English and speakers of English as a foreign language, but it contained the risk that participants in Long Beach would name (only) English as their first language who did not conform to our concept of a native speaker. However, an informal survey among California students showed that only 7 of 27 bilinguals considered English as their first language. Six of these 7 maintained that they were more fluent and felt more comfortable in English than in their second language; they also used predominantly English, at home and elsewhere. Such individuals were considered close enough to our concept to be included as native speakers. Therefore, the present study refers to participants in the GLBCC as native speakers if they gave English as their (only) first language.
... Acquisition of English in formal and informal contexts In our questionnaire, we asked the non-native participants in which context they had learned English. They had the following options, each provided with a blank to indicate for how many years this context was relevant: Family, friends, school, university, other (to be specified). The first two counted as informal contexts, the next two as formal context. As which the category “other” was counted depended on how the participant specified it. ... Usage of English in formal and informal contexts Certain elements of a language are not only learned consciously (e.g. through instruction) but also through interaction with native or at least very advanced or near-native speakers. Therefore, we wanted to know whether the non-natives in our data used English outside the language learning classroom. The question put to them in the questionnaire read: What is your primary means of communication? For the Germans, both German and English were presented as explicit options, with another blank for a further language, if applicable. Again, we distinguished several contexts to which the question applied: with family members of the older generation, with family members of the same generation, with friends, or at school/university. The distinction ‘older family members’ versus ‘family members of the same generation’ had been introduced to capture the language situation of non-native speakers in the USA and was retained in the questionnaire used in Giessen. Again, the usage of English with family or friends was considered informal usage, while school/university counted as formal context. ... Abroad vs. not abroad For language students, who made up the majority of our German participants, it is not uncommon to spend some time abroad; in one particular course of study, this is even
Discourse markers in English discourse
an obligatory part of the curriculum. Therefore, we asked the German participants whether they had spent time abroad, and if so, where and for how long. This type of information invites a number of comparisons and divisions. First of all, students who never spent time abroad could be compared to those who did. However, we hypothesized that the length of time spent abroad might also play a role for the acquisition of discourse marker use and other features. Therefore, we divided the time length into four sections: up to four weeks, between four weeks and one year, one to three years, and more than three years. These could then be combined into various groups. Since I suspected that four weeks might not be enough time yet to pick up discourse marker use, the various combinatory options included counting those who had spent less than four weeks in an English-speaking country with those who had never been abroad.
... British versus American influence In their study of quotative be like in British and Canadian English, Tagliamonte and Hudson (1999) found differences between the two varieties of English studied, and between these and American English. Biber et al. (1999: 1095ff.) report on the distribution of inserts in American and British English and find that well is slightly more and you know much more common in American English than in British English. Accordingly, if we assume that discourse markers are acquired mainly in informal interaction with native speakers, we also have to consider the variety of English which influences the non-native speakers. Unfortunately, we did not ask the German participants whether the native speakers with whom they used English were American or British (or of any other nationality). However, we do know in which country they spent some time. Of those Germans who had spent time abroad, most had been to the United States of America and/or Great Britain. Some lived only in one country for some time, others spent time in both Great Britain and the United States. Usually, though, one stay was much longer than the other. Therefore, the German speakers were divided into a group which never went abroad, a second which had spent more time in Great Britain, and a third group which had spent more time in the US, to see whether the country had any influence on the German speakers’ performance regarding discourse markers. ... Native speaker contact Finally, all of this information was combined to create a factor entitled “native speaker contact” in informal contexts. Again, this factor was defined in two variants. In the first variant, even short periods of time abroad or speaking English with some friends (i.e. both English and German as primary means of communication) are sufficient for an inclusion. In the second variant, the selection of the group with native speaker contact is more restricted and based on contacts which I deemed extensive enough to allow for the German speakers to pick up native competence features. That is, the speakers in that group had either spent more than four weeks in an English-speaking country and/or used English as primary means of communication more than just occasionally with friends and/or learned English in an informal environment.
Chapter 1. Introduction
. The speakers 77 Germans, 34 American native speakers, and 4 British citizens are represented in the GLBCC. However, since there are only these 4 British speakers, which is not enough for any meaningful statistical comparison, they will be excluded from any further statistics unless otherwise noted. I have said before that all of the participants were students, i.e. enrolled at a university program. The American students were recorded mainly at the California State University at Long Beach (CSULB), while most of the German students were recorded at the Justus Liebig University in Giessen (JLU). There are some exceptions, though: Two of the Americans were PhD students and belonged to the staff of CSULB; seven of the German students were recorded at Long Beach, one of them being a long-term student there, the others having arrived shortly before the recording to stay for one or two semesters. Five of the American and the four British participants were exchange students in Giessen and were recorded there. Students were recruited in class and through posters on a voluntary basis, either for credit or for a small financial reward. The Americans were mostly students of linguistics and/or psychology, while more than three quarters of the Germans were students of English literature or English linguistics. The following sections present the distribution of speakers according nonlinguistic and linguistic factors.
.. Native-nonnative pair distribution Most speakers in a pair had a speaker with the same first language as a partner, but there are a few exceptions. Three of the seventy-seven Germans participated with an American student in Giessen; another two had non-native partners with other first languages (also in Giessen). Of the twenty-nine American participants who were recorded in California, five had partners with Hispanic or Filipino background. In addition, two of the British speakers signed up with American partners in Giessen.
.. Gender distribution One quarter of the German participant group and slightly more of the American group was male, which corresponds to 19 German and 10 American participants. Again, these figures mirror the quantitative relationships of students at the universities, this time in the language and psychology departments. Of the British students, three were male and one female. Since we depended on volunteers and needed as many of them as possible, we were not able to guarantee an equal distribution of gender, even though this might have been desirable. Nevertheless, the distribution at least turned out to be more or less equal between the two main groups, allowing for a meaningful comparison.
Discourse markers in English discourse
In contrast, the gender pairings are not as equally distributed (see Figures 1.3a and 1.3b). While 20% of the American males participated without a partner and another 20% with a male partner (and, accordingly, 60% with a female one), only 5% of the German males participated alone, and 58% had a male partner, leaving only 37% with a female one. The figures for the females are more similar. 13% of the Amerifemale without partner 9%
male with female 18% male same sex 6% male without partner 6%
female with male 18%
female same sex 43%
Figure 1.3a Gender distribution of the American speakers female without partner 6%
male with female 9%
male same sex 15%
male without partner 1%
female same sex 59%
Figure 1.3b Gender distribution of the German speakers
female with male 10%
Chapter 1. Introduction
can females were single (German: 9%), a quarter had a male partner (Germans: 14%), and the final 62% participated with another female (Germans: 77%). The numbers for males and females do not quite match because in several cases Americans and Germans had as their partners students with other language backgrounds, whose data I did not use.
age 18–21 62% missing data 9%
age 25+ 26% age 22–24 3%
Figure 1.4a Age distribution of the American speakers
age 22–24 48%
age 18–21 22%
age 25+ 30%
Figure 1.4b Age distribution of the German speakers
Discourse markers in English discourse
.. Age distribution The age of the American participants ranged from 18 to 40 years, while the Germans were between 19 and 31 years old. For the reasons mentioned in Section 1.7.1.2, I have divided the age range into three age groups: 18 to 21 years, 22 to 24 years, and 25+ years. The distribution of the American and the German students according to these age groups is shown in Figures 1.4a and 1.4b. Almost two thirds of the Americans were between 18 and 21 years old, while this age group accounts only for 22% in the German group. The majority of the Germans was between 22 and 24 years old; these differences presumably mirror the general age difference at the respective universities. Only one American in this age group (22–24 years) participated. Seven Americans were found in the 25+ year-old group. Thirteen of the Germans (i.e. 17%) were in that latter age group, and only one person of 30 years or older. Three Americans did not fill out the blank asking for their age. Of the four British students, two were 20 and two 21 years old, but they are not represented in the charts nor in other statistics concerning age. Age turned out to be relevant in some cases; see Chapters 2–5 for details.
.. Relationship between speakers Most of the students (Germans: 69%, Americans: 61%) participated with a person whom the considered as a friend, according to their own judgment as indicated in the questionnaire. About a fifth (23% of the Germans, 21% of the Americans) had a stranger as partner. 8% of the Germans and 17% of the Americans thought of their partner as an acquaintance (see Figures 1.5a+b). However, since people define the term friend 62%
stranger 21% acquaintance 17%
Figure 1.5a Relationship between American partners
Chapter 1. Introduction friend 69%
stranger 23% acquaintance 8%
Figure 1.5b Relationship between German partners
‘friend’ in different ways, this may not be a very reliable factor. In addition, occasionally even complete strangers conversed in the study as if they were long-term friends (cf. Section 1.7.1.5).
.. Role distribution Figures 1.6a and 1.6b illustrate the distribution of speakers according to their roles. Of the American speakers, almost a third acted in role A, 9% male, 21% female. Almost twice as many retold the second half of the story as speaker B (15% male, 40% female). Role C remained for 15% of the speakers, most of them (9%) female. The role distribution among the Germans was more equal. 12% of the speakers were males acting role A, another 12% males in role B. In the same way, a about a third of the speakers were females in role A, another third female participants in role B. However, there was only one single male, as opposed to 5 single females (i.e. 6%). Of the British speakers, two of the males acted in role A, the other male and the female in role B. Even though the role distribution in the two groups, American vs. German, is not exactly the same, I deem it similar enough not to skew a comparison between these groups. Further details will be given in the chapters on the markers.
.. The German speakers according to their acquisition and use of English The next subsections give a more detailed description of the German speakers and their use and acquisition of English. These issues will also be dealt with in Chapters 2–5; in the sections on quantitative results, I will address the relationships between
Discourse markers in English discourse role A female 21%
role B male 15%
role A male 9%
role C female 9% role C male 6%
role B female 40%
Figure 1.6a Role distribution of the American speakers
role A female 37%
role A male 12%
role C female 6% role C male 1%
role B male 12%
role B female 32%
Figure 1.6b Role distribution of the German speakers
Chapter 1. Introduction 5 to 8 years 16% missing data 3%
> 12 years 24%
9 to 12 years 57%
Figure 1.7 Distribution of German speakers according to years of learning English in a formal environment
the use and acquisition of English and the frequency of the various discourse marker functions.
... Learning English in a formal environment The vast majority of the participants (72%) learned English for eight or nine years, which means they had English as subject up to their graduation from school. 17% had fewer years of English instruction in school, although nine of them (12%) later studied English at university. 8% of the participants claimed to have had more years of formal instruction at school. Almost a quarter of the students did not enroll in English at university level but studied some other subject. A third had already studied it for more than three years at the university, while 42% had spent between one and three years on English as their minor or major subject. In general, it can be said that more than 80% of all German participants had at least nine years of formal instruction, most of them in Germany. In addition, three students said they had attended a language institute for one year. Figure 1.7 summarizes the distribution of German speakers according to their years of learning English in a formal environment. ... Learning English in an informal environment Beyond their formal instruction, some participants also learned English in a more informal environment. One had been living in the United States as a very young child for a couple of years, another spent 5 years in Bangladesh as a teenager; both indicated these times as times of learning English in a family. A third participant had an English-speaking parent but claimed that she used almost exclusively German at home.
Discourse markers in English discourse
Eight students wrote that they had learned English from friends, for a time of between some weeks and ten years. One had lived as an Au Pair in an American household, one attended an (apparently English language) boarding school, and a third had an English speaking nanny for 25 years. Finally, three claimed they had learned English through singing and reading. In my investigations of the relationship between discourse marker use and the acquisition of English in an informal environment, not all of these students were included. It is highly unlikely that the (i.e. 1–3-year-old) child had already learned how to use markers, nor does this knowledge come from songs or through reading if markers occur mainly in spoken language. Also excluded were those students who said they had learned English from friends for half a year and less, since I did not consider these time frames sufficient for the acquisition of native-like discourse marker use if the learner is not immersed in the target language culture. Thus, ten participants remained who qualified for the statistical group which had learned English in informal contexts.
... Using English in informal contexts We asked the non-native participants of GLBCC whether they used English (and German) as primary means of communication with friends and with family members. It turned out that half of the Germans spoke only German in these informal settings. A third used both German and English with friends, but not with family members. Four percent used German and English both with friends and the family; another four only German 49%
missing data 8%
German and English with friends 35%
English + German with family and friend 4% mainly English with friends 4%
Figure 1.8 Distribution of German speakers using English in informal settings
Chapter 1. Introduction
percent communicated mainly in English with their friends, but in German with the family. The final 8% did not give complete information on this subject (see Figure 1.8).
... Spending time in an English-speaking country Some of the Germans were recorded at Long Beach. Students in Giessen were recruited by poster and in class, mainly in the English department. Some students are required to spend a semester abroad; others do it voluntarily to improve their language skills. For these reasons, it made sense to ask the German participants whether they had spent time in an English-speaking country, and if so, where and for how long. It turned out that less than a third (31%) had never been to an English-speaking country. 22% had been in Great Britain for some time, one of them for only 3 weeks, the others for more than a month. Many more (38%) spent time in the USA or in Canada. Again, seven of these could account for less than 4 weeks; the other twentytwo had lived there for up to four years. A fourth group (9%) of the Germans had lived in Great Britain and America, for more than a month in each country. In Figure 1.9, the participants who spent less than 4 weeks in an English-speaking country are counted as not having been abroad. Thus, the percentages given above are slightly modified. In my analysis of the relationship between time abroad and discourse marker use, I will consider both to some extent. Having presented the corpus on which this study is based, the speakers involved, and factors which may be influential, I will now turn to a detailed description of how the four selected discourse markers were used in my data. no time abroad 41% time in GB + US 9%
time in GB 21%
time in the US 29%
Figure 1.9 Distribution of German speakers according to times spent in an Englishspeaking country
Chapter 2
So
. Functions of so in the literature So as a discourse marker has not received much attention in the literature, at least not compared to well, you know and like. Schourup, for example, treats the latter three in detail and several others (such as I mean and now) shortly; so, however, is not mentioned (Schourup 1985). Likewise, Östman (1981) presents a list of particles which includes well, you know, like and some others, but not so. Aijmer (2002: 2) illustrates what she means by discourse particles with a list of twenty-nine items taken from Stenström (1994); this list also does not contain so. In Biber et al’s grammar, we find the following definition of discourse markers: “Discourse markers [...] are inserts which tend to occur at the beginning of a turn or utterance, and to combine two roles: (a) to signal a transition in the evolving progress of the conversation, and (b) to signal an interactive relationship between speaker, hearer, and message” (1999: 1086). Even though so can arguably be said to meet this definition, as we will see in the course of this chapter, it does not seem to fit Biber et al.’s idea of inserts, since it is not mentioned under that heading. Jucker and Ziv comment, with reference to characteristic features of discourse markers: “elements demonstrating more of the critical features may be taken to be more prototypical members of the class of discourse markers and those showing fewer characteristic properties may be considered more peripheral” (1998b: 2f.). Could it then be that so is considered to be more peripheral? Based on Brinton (1996), Jucker and Ziv present a list of features which they have re-ordered into five levels of linguistic description (1998: 3): – – – – –
phonological and lexical features, syntactic features, semantic feature, functional feature, sociolinguistic and stylistic features.
If we go through Jucker and Ziv’s list, we find that, even though so is short, it does not form a separate tone group nor is it difficult to place within a traditional word class (phonological and lexical features). Two main uses of the linguistic item are readily identified in probably all dictionaries: so as an adverb of degree or manner and so as a conjunction (cf. for example COED 1976; Cobuild 1987; OALD 1989; OED 1989; Webster’s 1994). Regarding the syntactic features, so does better: It is indeed (typically)
Discourse markers in English discourse
restricted to sentence-initial position, only loosely attached to the syntactic structure and optional, at least in the sense that the sentence containing it remains grammatically well-formed even without so. Whether it is multifunctional – a criterion which Jucker and Ziv reject as defining one – remains to be seen after the analysis. The same applies to the assumed high frequency. In any case, it has not been regarded as gender specific nor is it stylistically stigmatized; it is not exclusively a feature of oral discourse and not generally associated with informality. Thus, so does indeed not seem to be a very prototypical discourse marker. In addition, there remains the issue of propositional meaning. It is remarkable in this context that the dictionaries mentioned above do not seem to have any difficulties in describing the function of the conjunction so as to “introduce the result or consequence/reason” (e.g. Cobuild 1987), or “to show the reason/result for something” (e.g. OALD 2000) or paraphrasing its meaning as “accordingly, consequently, therefore”, for example in the COED (1976). However, in Section 1.2.2.5 I have shown that numerous authors are unwilling to accord discourse markers much semantic (or, in Relevance theoretic terms, conceptual) meaning of their own, and, in the opinion of many researchers, neither do markers contribute to the propositional content of the sentence which contains them. Rather, they are said to have pragmatic meaning, i.e. they function only in the situational context of language in use, between speaker and hearer. Thus, Schiffrin suggests that “markers select a meaning relation from whatever potential meanings are provided through the content of talk, and then display that relation” (Schiffrin 1987: 318, without emphasis). For Blakemore, discourse markers have procedural meaning, that is, they “encode a constraint on pragmatic inferences” (Blakemore 2002: 4; cf. also Blakemore 1988: 185) and guide the hearer in the interpretation of what has been said (see Section 1.2.6 for a more detailed description of the contribution of discourse markers in Schiffrin’s and Blakemore’s approaches). Similarly, Fraser states that “a discourse marker does not create meaning [...] but only orients the hearer”. Discourse markers “are extremely useful guides for clarifying a speaker’s communicative intention” (1990: 390). These statements are also made with a view to so. How then should the matter be decided? Ariel suggests that there are differences between discourse markers: Markers usually carry a non-truth-functional (non-semantic) meaning [...], but they often contain at least a residue of semantic meaning. They can thus be distinguished as relatively semantically transparent markers as opposed to relatively opaque ones. Transparent markers are those whose primary function is semantic, since they are assumed to carry a truth-functional content. Notable examples are and, but, or, so, and because. [...] Opaque markers are those whose discourse function is (virtually) not semantic at all. (Ariel 1994: 3251)
As an opaque marker she mentions well; intermediate markers are, for instance, you know or I mean. These, she suggests, have undergone semantic bleaching (cf. also
Chapter 2. So
Ariel 1994: 223f.). It is obvious then that there is no agreement among researchers on discourse markers regarding the question whether so has or has not semantic or conceptual meaning and if it does, how much. However, even if one does not want to subscribe to Ariel’s statement that transparent markers carry a truth-functional content, few researchers would deny that there are indeed differences in the transparency/opaqueness of marker meaning, and that so belongs to the category of (rather) transparent markers. This may be another reason why well and you know attracted more research than so. So far I have measured so against the discourse marker criteria presented in Jucker and Ziv (1998), rather than against my own definition. This is not without reason. Since I will consider all tokens of the linguistic item so, I consider it more useful to postpone the (non-) discourse marker assessment until Section 2.1. But let me now focus on those studies which give accounts of so as a discourse marker (DM). Fraser, for example, excludes a number of other expressions from the class of ‘discourse markers’ which other authors clearly include, such as because, you know, I mean, and interjections in general; so, however, is treated as a member of this class (e.g. Fraser 1988, 1990, 1996). He approaches DMs from a grammaticalpragmatic perspective and characterizes a DM as a linguistic expression which “signals the relationship that the speaker intends between the utterance the DM introduces and the foregoing utterance” (Fraser 1999: 936). Concerning so, Fraser refuses to see it solely as a marker of result or consequence. He presents a range of examples, unfortunately without any discussion, to make the point “that so as a discourse marker permits a range of interpretations” (1990: 393). His examples include the following (his numbering): (11) d. Son: My clothes are still wet. Mother: So put the drier on for 30 minutes more. e. Teenage son: The Celtics have an important game today. Disinterested parent: So? g. [Grandmother to granddaughter] So tell me about this wonderful young man you’re seeing. Admittedly, it would be difficult in these three examples to read so as marking a result or consequence on a propositional basis. What so in all three cases has in common, according to Fraser, is to indicate the fact “that the speaker takes the message following to have a consequential relationship to the prior material” (1990: 394). This is easy to see in example (11d). If the clothes that should be dry are still wet, the consequential action is to put them in the drier. Notice, however, that we are not talking about states of affairs here which logically follow one from the other, but actions viz. a directive speech act to carry out an action which is performed because of the state of affair expressed in the previous proposition. Fraser continues: “The specific consequential relationship in a given instance is the result of filling out this general signal based on the details of the particular discourse context” (1990: 394). That is, it depends on the situation and speech context exactly what relationship so is meant to convey. In Fraser’s
Discourse markers in English discourse
example (11e), for instance, the parent either does not see the significance of the son’s statement or questions the son’s implication that the Celtics having an important game should have consequences. To arrive at the latter interpretation, however, a lot of world knowledge (e.g. about a team called ‘Celtics’, the relationship the teenage son has to this team, and the fact that parents are not always willing to concede their teenage children’s wishes) and knowledge about how discourse works (that the son’s statement may not simply be meant as a statement but may function as a pre-request, cf. Yule 1996: 67) comes into play. In (11g), we even lack sufficient context to judge the relationship the grandmother wants to communicate by using so. Apart from giving examples and indicating the general function, Fraser does not describe any detailed functions of the discourse marker so. Blakemore analyzes so within a relevance-theoretic framework, where it functions as a constraint on relevance (1988: 185ff.) and specifies the relation between two utterances (Blakemore 1992: 136).7 While she avoids giving so a label in her 1988 paper, she later terms it a ‘discourse connective’ (Blakemore 1992: 136), as do some other authors working within Relevance Theory (e.g. Blass 1990; Unger 1996). In Fraser’s grandmother example (11g, his numbering) so cannot be interpreted with certainty, because we lack knowledge of the context. Blakemore gives a similar example in which so does not have a linguistic antecedent either, but she provides a context for it: Speaker A comes home laden with parcels, and speaker B comments “So, you’ve spent all your money” (Blakemore 1988: 188f.). Unlike therefore, which seems to be interchangeable with so in a number of contexts, so does not need a linguistic antecedent. Instead, it construes the following utterance as one which is relevant to the situational context. Apart from so without a linguistic antecedent, Blakemore distinguishes three other functions of so, or rather three other situations in which so may occur. So obviously can indicate that the state of affairs described by the proposition following it is a result of the state of affairs described by the proposition previous to it, as in Blakemore’s example (5a): “Tom ate the condemned meat. So he fell ill.” (1988: 184). This is one function of so. Moreover, it “can be used to indicate that the relevance of the proposition it introduces lies in the fact that it is a contextual implication of the first proposition” as in “There’s $5 in my wallet. So I didn’t spend all the money then” (1988: 188). So here does not introduce a logical consequence but an inferential conclusion which the speaker draws from the fact that there are $5 in the wallet. In both cases, so is not necessary for the hearer’s judgment of the relationship between two propositions, but it facilitates this judgment by guiding “the interpretation process by imposing a constraint on the inferential (or pragmatic) computations a proposition may enter into” (1988: 185). Finally, Blakemore also mentions, like Fraser, so or so what uttered as a question after a statement made by another person (i.e. Fraser’s example (11e), cf. Blakemore 1992: 139). Blakemore interprets this question as an indicator that the speaker of so “is unable to see the significance of what someone has said” (1990: 189). Schiffrin (1987) includes an extensive treatment of so in her book entitled Discourse Markers, in which she analyzes DMs as being multifunctional in a model of
Chapter 2. So
discourse consisting of several planes (see Section 1.4.5). According to her, DMs are contextual coordinates (1987: 327) which contribute to coherence in conversation by allowing “speakers to construct and integrate multiple planes and dimensions of an emergent reality” (1987: 330). Schiffrin provides us with a very detailed account of so, along with a discussion of because. Similar to Blakemore, but on the basis of her own levels of discourse, she distinguishes between so marking a fact-based resultative relationship which holds between states and events, and so marking a knowledgebased relationship. She explains: “A knowledge-based causal relation holds when a speaker uses some piece(s) of information as a warrant for an inference (a speakerinference), or when a speaker intends a hearer to do so (a hearer-inference)” (Schiffrin 1987: 201f.). The fact-based relationship, which functions in Schiffrin’s ideational structure, works like Blakemore’s “John ate the condemned meat, so he fell ill” example. The knowledge-based relation (at the level of the information state) is illustrated by Schiffrin with “John’s lights are burning. So, John is home” (1987: 205). Here, the speaker assumes or perhaps knows that the lights are burning only if John is home, and thus inferentially deduces from the fact that the lights are burning that John is / must be home. In addition, so in the action structure may mark an action as resulting from a state of affairs. Thus, a request for information may be initiated with so, indicating that the request was motivated by the preceding talk (1987: 208; cf. Fraser’s wet clothes example above). Unfortunately, Schiffrin gives only one example for action-based relationships expressed by so and otherwise concentrates on because. I have already mentioned that so indicates a resultative relationship in the ideational structure. However, in this structure, so has an additional function, in which it complements because. Schiffrin claims that these two markers organize the transitions from main to subordinate to main ideational units in talk (1987: 191ff.). Because is used to go from main units to subordinate units, i.e. to information which provides the background or explanations to the current argument or narrative line. Here the terms ‘main’ and ‘subordinate’ have to be seen relative to their context; explanations which are subordinate to the main story line may be further supported by additional background information, which in turn is subordinate to the explanation. Each may be prefaced by because. While the speaker “steps down” with because, s/he “steps up” with so. In a fashion complementary to because, it marks propositions as being main units relative to those expressed before. Finally, “[s]o functions in the organization of transitions in participation framework” (1987: 217). It may be used to indicate that a transition is being offered; in the examples Schiffrin gives, it frequently marks the completion of adjacency pairs or the completion of a story and thus gives other speakers the opportunity to take their turn. In other cases, it introduces an explicit selection of the next speaker, as when Schiffrin directs a question to one of her interviewees. So may also be employed to introduce the utterance after the transition, i.e. as a means of self-selection (1987: 219). Even on this plane (participation framework), so, if it is used elliptically in an utterance-final position, can convey a notion of result, even if no result follows: In this
Discourse markers in English discourse
function, “so instructs the hearer to recover a conclusion (an inference, a claim) which has already been presented, or which is otherwise mutually known” (1987: 223). Not all of the functions Schiffrin describes for so turned out to be useful for my own category system. In practice, for example, so marking a fact-based resultative relationship and so marking a knowledge-based resultative relationship were difficult to distinguish. Schiffrin herself admits that with so, multiple realizations of fact-based, knowledge-based, and action-based relations are possible, “due to our understandings of causality” (1987: 211). Nevertheless, much of Schiffrin’s approach will be found again in the categories developed in this paper. So also plays a role in Redeker’s approach (1990, 1991). She takes up the idea of a model with several planes within a coherence framework, like Schiffrin, but reduces their number to two levels, the ideational and the pragmatic structure. The pragmatic level is further subdivided into rhetorical and sequential relations. While ideational relationships exist between states of affairs rhetorical relations hold between beliefs and intentions motivating the relevant utterances. A sequential relation is assumed “when two adjacent discourse units do not have any obvious ideational or rhetorical relation – while still being understood as belonging to the same discourse” (1990: 369). In her 1991 review of Schiffrin’s book Discourse Markers, rhetorical and sequential relations are presented as on par with the ideational structure. So can be used in both the ideational and the pragmatic structure: as a “semantically rich connective” to mark ideational relations (no examples or further explanations are given in either publication) and as a “pragmatic connective” to mark pragmatic relations (1990: 372). In practice, this meant that it “was used between successive elements in a chain of events as in example (3), or to mark the speakers’ summing-up or conclusion” (1990: 372f.). Even though she does not explicitly say so, it can be assumed that at least the former function exemplifies sequential relations. Redeker’s idea of a sequential function is lacking in Schiffrin’s description, but it nevertheless proved relevant for dealing with my data. Also, like Redeker I frequently found so being employed to summarize. Three other studies look at so from a contrastive point of view, with a focus on how non-native speakers use it. Nikula (1996) works within a politeness framework and investigates a huge number of what she calls ‘pragmatic force modifiers’ in the English and Finnish language conversations of British and Finnish students. Her data consist of three corpora: one consisting of conversations between British and Finnish students in English, and two native speaker corpora, one each for Finnish and for English, which serve as a comparative basis for the native/non-native corpus. The modifiers which she looks at are exemplified by expressions such as I mean, you know, like, well, I think, really, sort of, I don’t know, etc. They may be used either to diminish or hedge the pragmatic force of the utterance, or to intensify it. So is just one modifier in her list, and a rather infrequent one in her data. It occurs only eight times in the conversations between native speakers of English, as compared to 277 instances of I mean. Unfortunately, it is not treated in more detail in her dissertation, so that we do not know how she defines it nor in which contexts so counts as a modifier.
Chapter 2. So
Hays (1992) carried out a study with Japanese students speaking English (in Japan) with the aim of finding out which type of discourse markers they controlled. He basically follows Schiffrin’s (1987) model of planes, but he exclusively refers to so as an ideational marker. No distinction whatsoever is made of individual functions of so, nor does he define any function besides referring to Schiffrin. From his results, he concludes that ideational markers are acquired earlier than markers on other planes. However, since he ignores the fact that, in Schiffrin’s model, so has functions on all five planes, his results have to be interpreted more carefully. A comparison between Asian students’ performance and native speakers’ performance in English writing and speaking is at the center of Anping’s (2002) work. She tries to find reasons for the overuse of so in Chinese EFL (English as a Foreign Language) learners’ written English. To that end, she examines a number of corpora, with spoken and written data, native and non-native English, and native Chinese. In the second section of her paper, she presents several functions of so from the Chinese version of the Longman Dictionary of Contemporary English (1997), including so as a resulting connector, question initiator, turn taking signal, and attitudinal signal. However, the study itself is restricted to “the connective use of so with meanings similar to so that (‘with the result that’), thus (‘in this way’) and therefore (‘for that reason’)” (2002: 43). Other functions of so are not treated. In the written EFL corpus collected at the university, the Chinese learners tended to use so as an intersentential connector. Anping comments: “This preferred position of so in EFL learners’ English indicates it is more like a discourse marker than a syntactic conjunction” (2002: 45). Their use of so in written EFL English compares with the use of so in spoken native English data, which might in part be due to the learners’ unawareness of the stylistic impact of so. In contrast to the two studies described above, we find a focus in Fischer and Drescher (1996) on the fact that discourse markers, or discourse particles, as they term them, have a range of functions. They claim that the variety of translations for a discourse particle that can be found in translations of plays, for example, “show[s] a very broad sense spectrum for each discourse particle” (1996: 857). Thus, in the German translations of two English plays, they found that “English so is translated as the German discourse marker also, as the interjection na und and as the modal particles denn, aber and dann” (1996: 856). Unfortunately, the authors only present these translations of so, without describing the functions that become apparent through these translations.
. The functional categories of so in this study Describing the functions of so in my data is the purpose of the following sections. With 1819 instances, so is a frequent lexical item. For this item, I identified a total of fourteen functions, non-discourse marker uses included. The discourse marker functions are divided into those working at the textual level and those working at the interactional level. Table 2.1 gives an overview of all functions.
Discourse markers in English discourse
Table 2.1 Functions of the lexical item so as used by German and American speakers in the GLBCC Non-Discourse Marker Functions: – adverb of degree or manner – expressing purpose – so in fixed expressions – direct translations of a German expression – so as a substitute Textual Level: – marking result or consequence – main idea unit marker – summarizing / rewording / giving an example – sequential so – boundary marker Interactional Level: – speech act marker – question or request – speech act marker – opinion – marking implied result – marker of a transition relevance place
Let me begin the descriptions with the non-discourse marker functions, which were classified mostly for the purpose of dealing with all instances of the lexical item. The distinction between non-discourse marker and marker functions is based on the list of features given in Section 1.4.3, i.e. so in the examples in the following section cannot be omitted without leaving the grammatical construction incomplete and/or without changing the semantic content of the utterance.
.. Non-discourse marker functions of so ... Adverb of degree or manner The first function I want to describe is often referred to as ‘adverb of degree’ and/or ‘adverb of manner’ (cf. for example COED 1976; Cobuild 1987; Webster’s 1994; CIDE 1995). Biber et al. (1999: 564f.), distinguish between amplifiers “and other modifying adverbs of degree”; so in this function is included in the group of amplifiers. Biber et al. find that it is more common in conversations than in academic prose and more common in American than in British English. For my categorization system, I prefer using the traditional term ‘adverb of degree or manner’ rather than amplifier. The use of so in this function is illustrated in the following extracts from my data, with no need for further explanation:8 (8)
360 A: but .. I I didn’t think he looked .. & 360a & like that quite so much, (22A, NS)
Chapter 2. So
(9)
(10)
193 and he did it so many times that the & 193a & artist finally said oh OK. (22B, NS) 471 but it’s so different to what we’re used to. (22B, NS)
The German speakers used so as an adverb of degree or manner on average about twice as much as their native counterparts, although this difference did not reach statistical significance in a median test. Frequency here and elsewhere was, as explained in Chapter 1, always calculated as instances per 100 words.
... Expressing purpose So as a conjunction can be used to express purpose, mostly but not only in combination with that. It is also paraphrased as in order that (e.g. Collins-PONS 1983; OALD 1989; Webster’s 1994). The first example (no. 11), with so only, describes a scene in which Chaplin does not have the money to pay his restaurant meal and is afraid of being beaten up. In order to avoid the beating, he tries to postpone paying, presumably until he has found a solution to his financial problem. Even though so stands alone in this extract (i.e. without that), the “purpose” reading is made clear through the use of the conditional (wouldn’t) and our understanding of paying, not having enough money and being beaten up in this story. (11)
(12)
287 287a 288 289 290 290a (47B,
B:
342 342a 343 344 345 (40A,
A:
A: B:
.. so then he .. keeps .. beating & & around the bush, trying to avoid .. paying. [<SV mhm SV>], [so he wouldn’t] get the snot & & knocked out of him,
NS) yeah the thing is you always have to & & exaggerate if you’re doing that. [because you can’t say anything]. B: [yes yes .. yeah yeah]. A: so that everyone understands it. NNS)
In the second example (no. 12), in which so is combined with that, the two participants discuss the challenges of making a silent movie. Speaker A thinks that the purpose of exaggerating in such a movie is having the audience understand what the actor wants to express. On average, the Germans used so for expressing purpose somewhat more frequently than the Americans in my data (0.0031 vs. 0.0026 times per 100 words).
Discourse markers in English discourse
... So in fixed expressions Dictionaries list them under various sub-entries, but most of those referred to above do mention fixed expressions containing so. Besides and so on and or so, they also list so what? In my data, so what? does not occur; with three exceptions all so in fixed expressions occurred as and so on or or so, and all but two were used by German speakers. Example (13) comes from an American speaker, example (14) from a German one: (13)
177 & a minister or so for marrying or something. (44B, NS)
(14)
459 & the the real sounds and surroundings & 459a & and so on, (28A, NNS)
The reason for the fact that these expressions are fairly frequent with the German speakers and virtually non-existent in the native speakers’ speech can probably be seen in the existence and frequency of their literal translations in German. The German translation of or so is ‘oder so’, which is presumably pragmatically equivalent and, I would assume, about as frequent as the English whatever. And so on is translated as ‘und so weiter’; its frequency in the data may be partially due to the Germans’ lacking linguistic skills or vocabulary to express in English all they would have said in German. Furthermore, the German expression seems to be fairly popular: In a smaller German corpus recorded with the exactly identical experiment design, und so weiter occurred in seven of 25 transcripts, in one of them even fifteen times. This is the only non-discourse marker function in which the frequency difference between the non-native speakers and the American native speakers achieved statistical significance, if we assume a five percent threshold for significance (p = 0.016). On average, the Germans used so in expressions 0.02 times in 100 words. The American participants, in contrast, used it hardly at all.
... Direct translations of a German expression This category is closely related to the previous one in that the German speakers directly transferred expressions or sentences from their first language. Some of these instances may not necessarily be wrong, but they are all clearly recognizable as literal translations from German. The examples below present extracts from the corpus, with a literal translation for (15) and (17) and in (17) also the German equivalent of the sentence. Example (15) presents an extract from the data which is a word by word translation of a grammatically well-formed German sentence. I do not mean to say that the speaker first thought of the German sentence and then translated it. Nevertheless, the English sentence which speaker 27A utters exactly corresponds lexically and syntactically to the German sentence and sounds rather unusual if not ungrammatical. As to the end of example (16), I would intuitively, as a native speaker of German, judge ‘und so’ as a very frequent expression in informal speech, while its literal translation and so seems rather odd in English. Similarly, the beginning of the sentence in (17) is completely natural in German (‘so weit wie’). A correct translation would read ‘as far
Chapter 2. So
as’; however, speaker 42A apparently does not translate it as an idiomatic expression but as separate words, arriving at so far as. The quotative construction in example (18) was caused by interference and is definitely not possible in English. In German, however, the informal quotative construction often has the form ‘und PRONOUN so’, equivalent to the English ‘BE+like’ (cf. Golato 2000 for a comparison). The result of the transfer is a mix of a German and an English quotative construction: and she is correct in both languages, the verb form ‘s comes from the English verb form of quotative BE+like, but the speaker fails to substitute German quotative so by the English quotative like. (15)
yeah they weren’t so advanced like today ‘ja sie waren nicht so fortgeschritten wie heute’ (27B, NNS)
(16)
.. and he always plays to be ashamed and so, und so, (75B, NNS)
(17)
so far as I saw it I I liked it, so weit wie ich sah es ich ich mochte es ‘so weit wie ich es gesehen habe, fand ich es gut’ (42A, NNS)
(18)
254 and she’s so, (72B, NNS)
237
200
557
... So as a substitute Finally, so also functions as what the Cobuild (1987) and Webster’s (1994) dictionary,9 as well as Biber et al. (1999), call a substitute, for something (typically a post-predicate clause) which the speaker does not want to repeat. In my data, substitute so occurred mainly within the expression I think so. Again, it was used most by the German speakers: more than three times as often per total words as by the American speakers, even though this difference did not achieve statistical significance (p = 0.071). The two examples below, however, come from natives: (19)
505 B: ...(2.9) I think so too even-(44B, NS)
(20)
39 and so did Charlie Chapman. (142A, NS)
.. Discourse marker functions of so ... Marking result or consequence Let us now turn to the discourse marker functions of so. As I have said before, all dictionaries describe the marking of result or consequence as one of the functions of so. We also find it in most of the papers and chapters dealing with so mentioned previously
Discourse markers in English discourse
(Blakemore 1988; Schiffrin 1987; Fraser 1990; Anping 2002, etc.). In these papers, this function is always considered a discourse marker function. I will argue with Schiffrin and Blakemore (see the beginning of this chapter) that so in this function is not only syntactically optional but also semantically optional. That is, the relationship expressed by so between the propositions before and after it is one of the interpretative options the hearer has anyway. The task of so consists in facilitating the hearer’s task by selecting one of these interpretative options. Examples (21), from a native speaker, and (22), from a German speaker, illustrate the result/consequence function of so: (21)
50 and then ((sniffs)) um ... there is like a & 51 & ruckus going on in front of them, 52 which is uh some guy wa- you know short & 52a & on his bill, 53 so they beat him up and kicked him out. (2B, NS)
Here, so marks they beat him up and kicked him out as a consequence of the customer’s not having enough money to pay his bill. Since our culturally determined world knowledge includes the knowledge about how waiters or restaurant owners may react when a customer cannot pay the bill, we would be able to infer a resultative relationship between “customer cannot pay the bill” and “customer is beaten and kicked out” even without so. The chronological sequence additionally supports the ‘result’ reading (cf. Biber et al. 1999: 883). However, by using so, the speaker on the one hand helps the hearer to arrive at this interpretation and on the other hand makes clear that she intends this interpretation. The next extract describes a scene on the immigrant boat. The same line of argumentation as above also works here. Our world knowledge also includes the fact that people on a boat get sick when it is thrown about by the waves. Thus, as a result of the sea being rocky, people get sick. (22)
13 13a 14 (27A,
uh the the sea were ... uh .. or was & & uh very rocky, so um all people get um .. seasick, NNS)
In both extracts, the consequence marked by so and the cause for this consequence are dealt with in one intonation unit. In many instances, however, the picture is more complex, as in example (23). The reason why you had to really watch (line 85) extends over a number of intonation units and is summarized again in line 88; the consequence itself is not only given in line 85 but further explained in the following two intonation units. Similarly, the proposition following so sometimes is already treated as a result or consequence, but the speaker, recognizing that perhaps s/he assumed too much knowledge on the part of the hearer, supplies the underlying reasoning after that, as in extract (24).
Chapter 2. So
(23)
75 of course they had some captions to explain & 76 & parts of the story that just cannot be & 76a & explained by .. well changes, 77 and the volume of the tone, 78 and the melody of the music, 79 what that cannot just be acted out, 80 so that everybody will get what was going on, 81 (H) but these .. captions were-82 (Hx) I don’t know maybe there were like ten & 82a & in the whole movie, 83 maybe fifteen, 84 but I don’t think more than that, 85 (H) .. so you had to really WATCH. 86 you cannot just look away and then tune in & 87 & to the movie again after ten minutes, 88 then of course then you definitely & 88a & missed the story line. (25C, NNS)
(24)
163 so he orders more and more coffee, 164 because he can’t pay, 165 and he wants to-166 A: ...(1.4) just stay there as long as possible. (48B, NNS)
In her treatment of so, Schiffrin distinguishes between so marking a fact-based result in the ideational structure and so marking a knowledge-based inference in information structures (1987: 202ff.). The first two examples above illustrate resultative relations between facts: passengers become seasick because the sea is rough, and a customer is beaten up because he cannot pay his bill. So in the following extract, in contrast, clearly marks an inference: (25)
250 B: and then the guy says oh .. I’ll ma- I’ll & 250a & engage you both today, 251 you get married tomorrow or something. 252 so he obviously wasn’t an artist. 253 he was probably a priest or something. (63B, NS)
In the movie, Chaplin and his lady love are sitting in a restaurant when another customer comes over and, in the course of a conversation, says (which is shown by an inter-title, i.e. white text on a black screen), “I’m an artist”. Later he says he will engage them. B does not think of ‘engage’ as a term referring to a work contract but only of its sense in human love relationships. Therefore, he concludes that this customer must be a priest rather than an artist, especially since the movie ends with Chaplin and the lady entering an office for marriage licenses and the wedding march being played. Example (23) above, however, where speaker 25C talks about the particularities of watching a silent movie, is not such a clear-cut case. It is a fact that several inter-titles or captions, as speaker 25C calls them, occur in the movie, and that at least some of
Discourse markers in English discourse
them are essential to understand the plot. But is it a fact or an inference that you have to watch the movie with particular attention? Also consider the following extract: (26)
99 ...(1.2) (H) so then the um the waiter takes it, 100 and puts it in his pocket, 101 but it falls out of the bottom of his pant leg. 102 so Charlie tries to get it, (4B, NS)
The background is the following: Chaplin had found a coin in the street but lost it immediately, which he later finds out in the restaurant after having ordered a meal. A beggar comes in and shines a coin, presumably the one Chaplin lost, and pays a cup of coffee with this coin. When the waiter loses it, it is not a natural (or, as Blakemore 1988 puts it, a logical) consequence that Chaplin tries to get it; rather, it is inferred from facts, common knowledge, and assumptions based on this common knowledge. We know that Chaplin needs money, we have been told that the waiter lost money, and we assume that he will take advantage of any situation which might provide him with money. Thus, we (or rather the speaker of this sequence) conclude that there is a resultative relationship between the waiter losing the money and Chaplin trying to get it. In Biber et al. (1999: 877), we find that they treat “linking adverbials” which indicate relationships of result and inference under one single heading, because the boundary between fact-based (result) and knowledge-based (inference) usages of so is not always as clear as it seems to be, judging from Schiffrin’s presentation (1987: 202). In my data, the instances making up the category ‘marking result or consequence’ are more set along a scale (of more or less world knowledge being included in the so) than a homogeneous category. Therefore, I grouped all instances, whether more fact-based or more knowledge-based, under the same category for the sake of consistency. In any case, however, so as a marker of result or consequence functions at the textual level, since it indicates particular relationships between propositional contents expressed in the narrative or discussion. Marking result or consequence accounts for more than forty percent of the tokens of the linguistic item so. In terms of frequency per total words, the difference between the American and the German speakers in the Median test was significant at p = 0.011, even though the American speakers used it on average only one and a half times as much as the German speakers (0.65 vs. 0.43 tokens per 100 words).
... Main idea unit marker Another category introduced by Schiffrin (1987: 191–203), which I found useful for my data, is so as a marker of main idea units. Schiffrin describes it as complementary to because: while the speaker leaves the main story line or the main argumentative line with because to introduce explanations and background information, s/he returns to the main level with so. Although Schiffrin does not explicitly situate the ‘main idea
Chapter 2. So
marker’ function within her five-plane model, it is clear from how she describes it that it works in the ideational structure. In my data, this category was applied whenever the speaker came back after a digression or explanation to the main thread of the narrative or to a topic or an opinion mentioned before and then repeated or alluded to this main idea. Like so as a marker of result or consequence, so as a main idea unit marker relates two propositional ideas, not as one being the result of the other but as one being part of the main story or argumentative line. I will first illustrate so where it refers to main idea units in the narrative part. In example (27), the narrative before line 27 included that Chaplin entered a restaurant and sat down. After the utterance he orders, the speaker remembers that there was another slapstick scene before Chaplin actually orders a meal, in which the waiter tries to make Chaplin take off his hat. This scene is described in lines 28–46. Then the speaker continues with so and a repetition of what she had said before, and thus comes back to what she apparently considers to be the main thread of the narrative. Something similar happens in example (28). Speaker 9C has narrated how Chaplin had been able to pay his bill by using another customer (the guy)’s tip. This gave rise to the scene which speaker 9C describes in the example (lines 139–142) as an aside to the main story line (lines 138, 143). The aside is additionally set off against the main story by the pauses at the beginning of lines 139 and 143. (27)
27 he orders, 47 (H) so basically he orders like some bea=ns, (8B, NS)
(28)
138 and then they got ready to leave, 139 ...(1.5) and I’m not sure. 140 well it seemed like .. the waiter was mad & 141 & at the guy cause he didn’t give him a tip, 142 and he just left without saying anything, 143 ... (H) so then they left, (9C, NS)
Examples like these make up the vast majority of the cases classified as “main idea unit marker”. However, this function also occasionally occurs in the discussion part, with so leading back to an opinion (example (28)) or a topic (example (29)) mentioned before: (29)
252 did you like the film? 253 A: yeah. 254 I did like it. 255 because .. it was very funny an’ entertaining, 256 and um, 257 .. in general I like Charlie Chaplin. 258 .. [so .. I enjoyed the movie], (51A, NNS)
Discourse markers in English discourse
(30)
295 295a 296 (...) 344 (45A,
because I think it’s .. it is ...(1.3) & & more of a challenge to= do .. & & acting without .. speaking. ...(4.0) so we .. we have the speech, NNS)
In (29), the speaker first answers the question did you like the film? with an affirmative statement, then explains why she liked it, and then repeats her opinion. In the discourse preceding extract (30), the speaker started off in line 295 with answering the question “What do you think were the major challenges when making a silent movie?” The partners then talk about facial expressions, about the inter-titles and about how the young lady played her part. Finally, speaker 45A comes back to the challenges of making a silent movie and to lack of speech as one of them. This function of so was more than twice as frequent (0.11 vs. 0.04 tokens per 100 words) in the American speakers’ data as in the Germans’ data, though without achieving statistical significance (p = 0.071) in the Median test.
... Summarizing/rewording/giving an example In the GLBCC, so was repeatedly used to summarize what the speaker had said before, to put it in different words, or to give an example for it. Neither of the three functions seems to have occurred in Schiffrin’s (1987) data, since there is no reference to any of them. Redeker (1990: 372f.), however, mentions that (and) so was used in her data “to mark the speaker’s summing-up or conclusion”. It may, at first, be difficult to perceive how functions such as summarizing, rewording, and giving an example can be combined to form one single category. However, when we look at the relevant instances, it becomes clear that they have a common feature. In all three, the utterance following so expresses the same propositional idea as a previous utterance. The function discussed in the last section, so as a main idea unit marker, also included a repetition of the propositional idea, but while the definition of the main idea unit marker necessarily required a digression of some sort, no digression occurs with the functions described in this section. In contrast, the idea summarized/reworded/exemplified immediately precedes the one prefaced with so, being expressed sometimes in only one intonation unit, sometimes in several. Let me illustrate the three functions, what they have in common, and what distinguishes them from so as a main idea unit marker. Extract (31) gives an example of so introducing a summary. Speaker 4B first describes in detail how Chaplin eats his bean dish in a restaurant. Before she moves on to what else happened in the restaurant, she summarizes her description by saying so he keeps u=m .. eating. (31)
50 50a 51 52
and the guy next to him is like & &. ... so then he continues eating <@ beans @>, <@ and @> um,
Chapter 2. So
53 so obviously he’s eating them with his knife. 54 <SV he just keeps them with his knife and puts & 55 & them in his mouth instead of using his & 55a & fork or his spoon SV>, 56 (H) so he keeps u=m .. eating. (4B, NS)
Another scene which was repeatedly summed up with so is the one in which Chaplin manages to pay his bill by using the other customer’s tip, for example as so he pays the bill, or as so he got out of that one. The next extracts illustrate rewording. In the scene described in (32), Chaplin is desperate because he does not have the money to pay his bill. Another customer (the man) comes up to him, talks to him and offers to pay, without knowing about Chaplin’s distress. Chaplin in turn is too proud to let this customer know about it and politely refuses his offer. This is where extract (32) sets in. Lines 199 and 200 both express Chaplin’s reaction, though in different words, when the other customer finally accepts Chaplin’s refusal. In extract (33), this other customer, who is an artist, offers Chaplin and his lady-love a job. This is expressed both in lines 208–209, where the artist’s words (appearing in the movie as an inter-title) are quoted, and in lines 210– 211, prefaced by so, as a description of what these quoted words meant. Incidentally, both examples involve an interpretation; however, this need not be the case. Speaker 8B, for example, elsewhere rewords they get ready to leave as and so like they get up whatever, both being descriptions of the same obvious action. (32)
196 and they go back and forth like two or & 196a & three times, 197 and then (H) finally the man’s likeyou know, 199 and Charlie is like. 200 and so he’s like I should’ve let him & 200a & pay whatever. (8B, NS)
(33)
208 209 210 211 211a (22B,
(34)
138 (H) and he starts ordering his food. 139 ... right? 140 so he orders some bea=ns, 141 .. and a bread, 142 (H) a piece um-143 .. a piece of long-144 piece-145 long piece of bread and .. some coffee. (140B, NS)
but the artist said, um I’ll engage both of you tomorrow. .. <SV and so he was gonna hire them .. to & & be... part of his art studio or something & & I don’t know SV>. NS)
Discourse markers in English discourse
While ‘summarizing’ means that an idea originally expressed in more than one intonation unit is summarized in one intonation unit, giving an example may (though does not necessarily have to) do the opposite. In extract (34), for instance, his food (line 138) is specified as beans and bread and coffee, stretched over several intonation units (lines 140–145). Another, though minor, reason why these three functions were combined into one category is their frequency. Rewording occurred most often; it makes up half of all the instances in this category. Summing up was found sixteen times, and giving an example only seven times. Compared to the overall frequency of the linguistic item so (1819 tokens), these seemed to be frequencies too small to make up categories of their own. This is the second single discourse marker category with statistical significance between the speaker groups. The American participants showed an average frequency of 0.041 tokens per 100 words, while the German speakers had a frequency of only 0.025.
... Sequential so How does a speaker achieve a coherent transition from one event within a narrative to the next? Redeker in her paper on ideational and pragmatic markers writes about relations between discourse segments, when two adjacent discourse units do not have any obvious ideational or rhetorical relation – while still being understood as belonging to the same discourse – their relation is called sequential. Sequential relations can be paratactic, that is, transitions to the next topic or to the next point, or hypotactic, that is, leading into or out of a commentary, correction, paraphrase, aside, digression, or interruption segment. (Redeker 1990: 369)
Again, as mentioned above, the conversations in the GLBCC are of a comparatively simple internal structure, since the narratives do not serve any argumentative or interpersonal function. They are motivated only – or at least mostly – by the task within the experiment set-up, which is to retell what happened in the movie they were shown. All instances classified as sequential occur in the narrative part. This means that sequential so in my data does not cover the wide range of sequential relations described by Redeker. Rather, it is restricted to what we find in the Cobuild English Dictionary (1995) as one of the functions of so, namely, that it introduces “the next event in a series of events”. The Oxford Advanced Learner’s Dictionary (OALD 1989; OALD 2000) puts it into slightly different words, saying that it is “used to introduce the next part of the story”. This is exactly what so apparently did in almost 140 instances of my data. The extract (35) sets in when Chaplin and the artist both have paid their bills in the restaurant. The two men and Chaplin’s lady-love leave the restaurant; outside, the artist tells them (shown as an inter-title) that he engages both of them starting the following day and turns to walk away. Chaplin goes after him to ask whether he could give them “a couple of dollars in advance”. Speaker 4B substitutes this whole scene with the standard leave-taking formula, it was nice meeting you (line 163). Then
Chapter 2. So
Chaplin and the lady turn around a corner, where the next scene takes place: Chaplin courts the lady and finally enters the marriage license office with her. This transition from the good-bye scene (in whichever form) to the courting scene is marked with so. (35)
161 all of them get up and leave, 162 like arm in arm, 163 an’ .. the artist says it was nice meeting you, 164 so-165 it’s raining outside. so they go around the corner, 166 167 and there’s a big s- door. 168 there is a big door that says & 168a & marriage licenses, (4B, NS)
The same kind of transition takes place in extract (36). The first scene described begins when Chaplin spots the lady which he fell in love with on the immigrant boat sitting right next to him in the restaurant. The next scene, in which several waiters beat up another customer, is introduced with so, as in the extract above. In (36), the sequential function of so is further enforced by adding then, a combination which is frequently found. (36)
49 50 51 52 53 54 A: 55 B: 56 A: 57 B: 58 58a 59 (6B, NS)
((clears his throat)) he invited her over, and then he felt her handkerchief, that she’d been crying. so apparently her mother died. because-[uh huh]. [that’s kind of the] impression he gave. right. and uh-.. so then the waiter .. uh goes over to & & another guy’s table, and he starts b=eating up this guy=,
I mentioned in Section 1.4.4 that I always categorized the instances of the selected discourse markers according to the function which was most salient. In twenty-six cases of so, however, both the resultative and the sequential notion were present to an extent. So in these instances was not purely sequential; it introduced a new scene, but this scene was not as unrelated to the previous one as otherwise the case with so classified as sequential. On the other hand, even though the movie content described before so is essential for the scene following so, this relationship is not strongly resultative. It is illustrated in extracts (37) and (38). After describing Chaplin’s attempts to postpone paying his bill (by ordering a cup of coffee for his lady-love, line 259), speaker 131B in extract (37) mentions that there is another customer in the restaurant. This mention in line 264 is a prerequisite for speaker B to continue describing the ensuing scene using the pronoun he for this cus-
Discourse markers in English discourse
tomer. The fact that the customer is present in the restaurant may not be the reason for him going over to Chaplin and the lady, but it is a necessary condition. (37)
259 B: and then he’s like well the .. lady & 259 & wants a coffee. 260 A: ...(1.5) [mhm], 261 B: [or some]thing. 262 and so she goes back there, 263 ... he goes back there an’-264 ... there’s this OTHer guy in the & 264a & restaurant he’s an artist. 265 so he comes over and starts talking to them. 266 ... a=n’ .. the waiter brings their check, 267 ...(1.6) and the artist .. offers to pay, (131B, NS)
A similar situation is given in extract (38). Starting with sequential so, speaker 6B describes the scene in which a beggar comes in and shines a coin, with an expression of joy and surprise on his face. This scene is a prerequisite for the following scene to take place, introduced with so, in which Chaplin tries to talk to the beggar. The notion of result or consequence may be stronger here than in extract (37); yet, in both cases speaker B ends the intonation unit before so with falling intonation and thus indicates the end of one section of the narrative. In extract (38), this even induces speaker A to give a backchannel. (38)
80 B: 80a 81 82 83 84 A: 85 B: 85a 86 87 88 (6B, NS)
so then this like other bummy type of & & a guy comes in, and he has this .. coin, he is like polishing it, apparently he found Charlie’s coin. mhm, so then uh Charlie Chaplin is sitting & & there with his girlfriend, and he like wants to .. say something to & & this guy who has found his coin, but he doesn’t.
The twenty-six instances of sequential so with a resultative notion have not been separated to form a category of their own in the statistics but have been counted with other sequential instances. Sequential so is the most frequent single function after result/consequence in the American speakers’ data. They used it 0.25 times in 100 words, while the German speakers used it only 0.03 times. Therefore, it is not surprising that this difference is significant (p < 0.001).
... Boundary marker This is another function which neither Schiffrin nor Blakemore or Fraser have described in their publications, to the best of my knowledge: so being used as a boundary
Chapter 2. So
marker between types of talk, or, more specifically in my data, as a boundary marker between instructions and the beginning of the narrative. Almost a quarter of all speakers started their narrative either with so as their very first word, as in extract (39), or, more often, with some kind of acknowledgement (usually OK) and so, as in extract (40). In all of the following examples, the first line presented is the first utterance of that speaker after the researcher had given them instructions on what they were supposed to do. (39)
27 27a 28 29 30 (46B,
(40)
4 A: OK. 5 (H) .. so. 6 ...(1.1) it’s the beginning of the movie, 7 and it opens up .. with .. showing a whole & 7a & bunch of immigrants, 8 (H) and Charlie Chaplin is one of the immigrants. (134A, NS)
so what happened when you went out was & & that they arrived in America, A: mhm. B: a=nd they went through the immigration, they got separated, NNS) B:
All speakers in roles A and C who used so as a boundary marker, and most of the speakers in role B, immediately began their narrative after so, as illustrated in the extracts above. However, some speakers in role B first checked with their partners to make sure they started their narrative at the correct point. They did so either by explicitly asking at which point the partner had left the room in which the movie was shown, or by stating their assumptions about this point and waiting for positive feedback from the partner. Unlike the functions discussed before, so in this function does not relate propositional ideas as resultative, main versus subordinate, summary/example or sequential. Rather, it structures the spoken material into types of speech, that is, into instruction versus narrative and thus can also be seen as functioning at the textual level. This function naturally was used only once by a single speaker. Nine Americans and eighteen Germans marked the boundary between text types with so, which corresponds to 26% of the former and 23% of the latter group and about the same frequency per total words in both groups.
... Speech act marker – question or request As evidence that the functions of so are not limited to indicating result or consequence, Fraser (1990: 393) gives a number of examples, three of which I have reproduced in the beginning of this chapter. Two of them preface directive speech acts: So put the drier on for 30 minutes more and So tell me about this wonderful young man you’re seeing. As a speech act marker, so may also preface other speech acts such as requests or questions. Even though Schiffrin does not use the term ‘speech act’ in the context of the analysis
Discourse markers in English discourse
of so, her concept seems to be similar. In her data, so occurred between account and request (1987: 208), to “mark an action which has just been motivated” (1987: 209). Thus, so realizes causal relationships in the action structure (1987: 207, 316). Schiffrin obviously does not use the term ‘account’ necessarily in the sense that the speaker first gives a reason or justification before uttering the request; it is sufficient that the request be motivated by what other participants said (1987: 208, her example (26)). With this category, we leave the textual level and turn to the interactional level, since speech acts are in principle only “felicitous”, to use Austin’s (1962) term, if there is interaction with a receiver of the speech act. So in this and the following functions prefaces utterances which are directed at the hearer, or is itself directed at the hearer, as we will see. I have applied the category ‘speech act marker’ whenever a request or a question was prefaced with so. As Schiffrin pointed out (1987: 208f.), so not only marks the speech act, but also indicates that there is a resultative relationship between the request or question and some piece of previous discourse (i.e. the request/question is “motivated”). In my data, the motivation for the speech act may or may not be found in the immediately preceding intonation units, occasionally not even in the discourse itself but in the context of the experiment. In any case, however, the speaker presents the speech act as being motivated. Extract (41) is a nice example of a question being presented as motivated on the one hand, and on the other as being motivated by the non-linguistic context. In the movie, Chaplin has learned from the young lady he had fallen in love with that her mother lost her money. Since he has just won a considerable amount of money gambling, he decides to slip it into her jacket pocket without her noticing. Having done so, he thinks about it again, decides to keep some dollars for himself, slips the rest back into her pocket. This procedure, which Chaplin repeats a couple of times, is observed by a crew member, who interprets it as Chaplin stealing the lady’s money. Therefore, the crew member’s so prefacing his question can only be motivated extralinguistically. It is interesting, though, that the crew member in the movie does not utter this question. Not only is the movie a silent one but there is also an intertitle showing up rendering the crew member’s reaction as You pickpocket! Thus, the motivation expressed through so in line 38 is a matter of how the speaker presented it. (41)
34 ...(2.3) um .. I guess you can say one of the & 35 & crew members thought he was stealing money, 36 (H) ... so-37 and he was getting ready to leave he & 37a & called him and said you know, 38 .. (H) so why are you pick-pocketing, (9C, NS)
Extract (42) is, in one aspect, a more typical one. Speaker B confused the characters of the young lady and her mother, apparently thinking that the person the young lady cared for was a child of about ten years of age. This obviously irritated his partner. For the question in line 201, so what did you think of it, two motivations are conceivable:
Chapter 2. So
either the irritation expressed by speaker A in lines 192–193 which obviously he could not quite resolve, or the first of the questions on a sheet in front of him which were part of the instructions and meant to help the participants discuss the movie. This first question read, referring to the movie, Did you like it? Why? (cf. Section 1.5.2 for the full text and purpose of the questions). These questions on the sheet actually motivated a considerable fraction of the questions and requests prefaced with so which were categorized as speech act marker. (42)
192 193 194 195 196 197 198 199 200 201 (44B,
A: B: A: B: A: B: A: B: NS)
he wants to get married with the mother? not with the daughter? yeah with the mother .. with the older-mhm. but the daughter was no more there any more. she was .. just gone. ...(3.2) mhm. it’s that’s what happened in the B part, mhm. ...(2.8) so what did you think of it.
An example of a so-prefaced question which was motivated by the previous discourse is given in extract (43). The old lady appeared in the first half of the movie, on the boat, up to the point where speaker A had to leave the room. Speaker B, in her narrative, had not mentioned the old lady again, since she does not appear in the second half of the movie, which is set for the most part in a restaurant in (presumably) New York. The fact that speaker B does not mention the old lady incites speaker A to inquire what happened to her (line 216). Other speakers for example ask about the marriage scene (so do you think they .. got married? (31A, line 392) or so he just married like the first person he saw? (125A, line 336)), being motivated more by what was told than by what was not told. (43)
214 that’s when that movie ended. 215 A: mhm. 216 so what happened to the old lady? 217 B: (H) I don’t know, 218 the old lady was [never seen ... again]. 219 A: [she was ... disappeared]? (125A, NS)
A further feature of so in this function which has not been mentioned so far is that often, though not always, so occurs at a transition relevance place. More than half of the instances of so in this category come with a speaker change with self-selection; of those which do not, a third of the instances of so are preceded by a long pause (longer than one second), thus allowing for a potential change of speakers. So as a marker of questions and requests occurred with the same frequency (0.040 and 0.038 times, respectively) per total words in both speaker groups, American and German.
Discourse markers in English discourse
... Speech act marker – opinion In the process of defining categories for the functions of so, I extended the notion of it as a speech act marker beyond questions and requests to include opinions as well. That is, whenever a speaker used so to introduce an expression of his or her opinion, this was categorized as ‘speech act marker – opinion’. Sometimes, as in line 138 of extract (44), the speaker explicitly declared the utterance to be an opinion, by using I assume, I think, I guess, etc.; sometimes, as in extract (45), no such explicit declaration was given. (44)
136 (H) an’ so they ended the movie with ... & 137 & Charlie an’ the lady walking inside & 137a & of the marriage license door. 138 (H) so I assume they got married. (138C, NS)
(45)
21 (H) the um .. the fish bit his nose. so that was kind of funny, 22 (142A, NS)
Like so functioning as a speech act marker for requests and questions, so marking opinions includes an element of result. The speaker presents his or her opinion as motivated by what s/he has said before. Unlike the former category, however, this one does not typically occur in connection with a change of speakers. Like so as a marker of result or consequence, so for marking opinions was used by the Americans one and a half times as much as by the Germans, who employed so for marking opinions as often as so for marking questions or requests. However, here the difference between natives and non-natives is not statistically significant.
... Marking implied result Another function of so which typically includes a (potential) transition relevance place is described by Schiffrin in the following way: “I suggest that so in (45), (46) and (47) [=her examples] does convey a ‘result’ meaning even if no result follows. (. . .) so indicates that a speaker has reached a point in the presentation of his/her ideas at which a hearer can infer what would come next even if it is not explicitly stated” (1987: 223). Exactly this happened repeatedly in my data. The ‘presentation of ideas’, though, does not need to be an elaborate one. One of the shortest examples is the following (extract (46)): (46)
320 .. what did you think was the funniest part. 321 A: oh, 322 I only saw half the film. 323 [so um], 324 B: [@@], 325 in your section. (22A, NS)
After her narrative of the second half of the movie, speaker 22B reads the first discussion question on the sheet. Speaker 22A immediately replies that he did not see
Chapter 2. So
the whole movie and ends his turn with so erm, implying that he cannot answer the question because he has seen only half of the film, and indicating that he turns over the floor to his partner. Speaker B takes up both the cue of so as a transition relevance place and the implication transmitted through so by adding to and modifying the question (i.e. funniest part ... in your section) in order to make it relevant for her partner. The ‘presentation of ideas’ may also consist of the presentation of scenes from the movie. In extract (47), for example, the speaker describes how Chaplin, who sits in the restaurant eating beans, observes that another customer is beaten up because he is ten cents short on his bill. Her so in line 205 indicates that she assumes her partner can see the implication this has for Chaplin. When her partner does not react, at least not verbally, within the next 1.5 seconds, she expresses the implication she had in mind, i.e. that Chaplin gets nervous. Such a procedure is found in about a quarter of all so implying a result, i.e. the speaker adds and expresses the implication later. (47)
199 B: .. a=nd so-200 ...(3.2) he’s eating up. 201 .. and there’s other guy, 202 ... in the restaurant, 203 ... it’s ten cent short, 204 ...(1.2) and so .. all the waiters beat him up, 205 ...(2.2) a=nd so. 206 ...(1.5) he gets kind of nervous, (131B, NS)
In the functions described so far, so always occurs at the beginning of the intonation unit, maximally preceded by another marker (e.g. and) or a filled pause. In contrast, so used to imply a result occurs not at the beginning but either at the end of the intonation unit or forming an intonation unit of its own. At the beginning of the section it was pointed out that this function typically includes a (potential) transition relevance point. We have seen that a change of speakers did indeed take place in extract (46), even though the intonation on so as transcribed in that extract – ending the intonation unit on a comma, which indicates continuing intonation – is rather uncommon in my data. Two thirds of all instances of so in this function are pronounced with falling intonation (which was transcribed as a period after so; see extract (47)). This intonation pattern gives an additional cue for the hearer that s/he may take over the floor (cf. Wennerstrom 2001: 170ff.). For another quarter, the intonation unit containing so was transcribed as being truncated (i.e. ending on two dashes). In all but two of these cases, a pause followed. In a third of all relevant instances, the hearer indeed takes the next turn, as in extract (46); in another third, so is followed by a pause after which the speaker resumes his or her turn. This concurs with Schiffrin’s findings, who writes in her section on so and participation structures that “a marker which is used in such locations must have a dual role: to show that a hearer has an option to take the next move, and to allow a speaker to continue if such an option is not taken. [. . .] I will suggest that this is precisely what so marks” (1987: 218).
Discourse markers in English discourse
If I said before that so as a speech act marker functions at the interactional level because a speech act needs a hearer to be felicitous, the interactionality is even more obvious in so as an indicator of implied result. So here directly addresses the hearer and challenges her/him to figure out what the speaker implies. So as an indicator of an implied result occurred with equal frequency (0.025%) in the data of the American and the German speakers.
... Marker of a transition relevance place Finally, there is a number of clearly interactional instances where so neither marks a speech act nor an implied result, but still occurs at a transition relevance place: it is used for the self-selection of a speaker (cf. Schiffrin 1987: 119ff.), as in extracts (48) and (49), but it may also, at the same time, indicate that the speaker turns the floor over to the partner again, as in extract (50). This is possible because, in contrast to the marker of implied result, so in this function is positioned either at the beginning of an intonation unit or forms one of its own. In (48), speaker 10B has finished her part of the experiment task, i.e. retelling the second half of the movie, and makes that clear by saying and that was the end. At this point, the floor is open to both partners, since the second part of the task, discussing the movie, is directed at both equally. However, her partner does not take the floor, and so speaker B continues herself after a pause of six seconds, beginning her new turn with so. (48)
265 266 267 268 269 (10B,
(H) and Charlie Chaplin uh picking her up, .. a=n’ they went inside. and that was the end. ...(6.0) so now we talk about what was-... what did we like? NS)
In a third of the instances of so marking a transition-relevance place, the previous speaker continues with so after a pause of between one and six seconds. More often (in half of the instances), the other speaker takes the floor starting with so, as in extract (49). Here, the participants are discussing “the challenges of making a silent movie”, as the question on the instruction sheet formulated it. Speaker B obviously does not have anything else to say (line 515), so speaker A offers a turn. (49)
509 510 511 512 513 514 515 516 517 (22A,
A: B: A: NS)
when they’re talking, they-right? ... <SV they were SV>-...(1.8) they were moving their mouths SV>, <SV OK SV>. u=h, ...(2.6) so everything was .. over-emphasized, what you were saying a little earlier there.
Chapter 2. So
However, only a third of the turns thus taken are immediately followed by an utterance containing a proposition, which usually consists of more than one intonation unit. In almost half of the instances, so is followed by a pause, during which the floor is open again to both partners. It is interesting to see that many of the instances of so in this function occur in stages of the discussion part where the participants did not really know what to say or how to start the discussion. It seems as if they take turns in passing on the floor to the other. Extract (50) illustrates this (lines 405–407): (50)
403 ...(2.0) <SV @@ SV> that was the [film]. 404 A: [OK]. 405 (H) ...(1.5) so. 406 B: so. 407 A: ...(1.2) shall we look [... at] the & 407a & <SV first half of .. X SV>? (41A+B, NNSs)
B has finished retelling the movie and remains silent. After a pause, A takes a turn with so (line 495), but with falling intonation. Thus, so is not the beginning of an utterance; rather, it indicates the end of the turn, the speaker passing the floor to the interlocutor. By also using so with falling intonation, B immediately signals that she passes the floor back to her partner, who indeed takes it after another pause. The Germans used so to mark transition relevance places 0.021 times per 100 words on average, while the Americans used it only 0.011 times. However, in the Median test the frequencies of this function in the two speaker groups did not show a significant difference.
... Unclassified instances It is an undisputed fact that language as it is used in the real world, and spoken language in particular, is not always an orderly matter. Therefore, it should not be surprising that there were a number of cases in my data which could not be classified. They amounted to 195 tokens, i.e. about a tenth of all instances of the linguistic item so. The majority of them were not classified because the context was not clear enough due to unfinished utterances or due to unintelligible passages, as the following extracts show. (51)
21 22 23 24 25 A: 26 B: 27 (4B, NS)
so anyway he is walking the streets. and he finds this shiny coin, (H) I don’t know how much it’s worth. but they find a shiny coin. ((coughing)) so there-he goes into a restaurant .. to eat.
Discourse markers in English discourse
(52)
519 B: they didn’t really mea=n .. that much. 520 and you .. you could tell that the young & 521 & woman an’ an’ Charlie Chaplin have & 521a & fallen in love with one another, 522 but she’s sort of played the um-523 ...(1.4) I don’t know. 524 uh passive .. woman, 525 so the XXXXXXX that was about it. 526 A: ...(2.8) and there is no really difference, (69B, NS)
In extract (51), the fact that the syntactic structure of the utterance started with so in line 26 is not finished, neither in the same nor in the following intonation unit, makes it impossible to tell which function so was supposed to have. The same effect is given in extract (52), line 525, due to several unintelligible syllables. Truncated structures are most often the reason for non-categorization; unintelligible passages were less frequent. (53)
374 B: [so he tried] to get it. 375 but the waiter’s standing there, 376 so the waiter’s foot goes [over it], 377 A: [uh huh], 378 B: and he keeps trying to get it. 379 and then look up, 380 and the waiter’s like? 381 [you know], 382 A: [uh huh], 383 B: (H) so anyway. 384 it goes on where he’s trying to get it, 385 [trying to] get it. 386 A: [mhm], 387 B: he can’t. (142B, NS)
A small number of cases (14 instances) occurred which remained unclassified for other reasons. In these cases, the context was relatively clear, but the function of so fitted none of the categories described above. However, these functions did not occur in sufficient numbers to justify a separate category. In eight instances, for example, the function of so might possibly be understood as marking a section end. This is done in two versions: either as so anyway or as so that’s it. The latter was used only by German speakers and only twice, so I will just present an example of so anyway (extract (53)). First the speaker describes a scene in which Chaplin tries to pick up a coin that has fallen on the floor through a hole in the waiter’s trouser pocket. When she uses so, she has not really finished with the scene, as the later lines show; therefore, a categorization as sequential so would not be justified.
Chapter 2. So
Half of the fourteen instances were produced by native speakers, the other half by Germans. The instances unclassified due to truncation or unintelligible passages occurred with more or less equal frequency in both speaker groups.
.. Summary of the functions of so In the previous sections, I have described the functions of so which I recurrently found in the data extracted from the Giessen Long Beach Chaplin Corpus. Five of them are not to be considered discourse marker functions, since so in those contexts usually cannot be omitted without leaving the grammatical structure and/or the propositional content of the utterance incomplete. As a discourse marker, so can be used at two levels. At the textual level, so structures the content of the interaction: it marks propositions as the result or consequence of previously uttered propositions and leads back from a digression to the main thread of the story or to the main idea of an argument. It may also be used to sum up a description of a scene or an opinion, put it into different words or give an example. Furthermore, so effects transitions in the narrative from one scene to the next and marks the boundary between different text types, in my data particularly between instructions for the experiment given by the researcher and the fulfillment of these instructions. At an interactional level, so functions as a marker of speech acts such as requests, questions, and expressions of opinion. Used at the end of an intonation unit, it also implies an unspoken result. The resultative notion is also present in the speech act marker function. However, I consider these functions of so to work at the interactional level because their main effect is not structuring textual content but organizing the participation of the speakers. As marker of speech acts and implied result, so is used for the self-selection of a speaker on the one hand and as indicator of a (potential) transition between speaker’s turns on the other. This perhaps becomes most evident in those instances which mark neither speech act nor implied result but a transition relevance place.
. Quantitative results for so Let me begin the quantitative analysis with an overview of so as a linguistic item. Figure 2.1a shows the distribution of non-discourse marker (non-dm) so, discourse marker (dm) so and unclassified instances. The last group, unclassified instances, has the same frequency in both the American speakers’ and the German speakers’ data. Non-discourse marker so is used more than twice as much by the Germans as by the US natives, while, in reverse, discourse marker so is used almost twice as much by the native speakers as by the German non-native speakers. However, the difference in non-discourse marker frequency does not achieve statistical significance (p = 0.074);
Discourse markers in English discourse 1.6 discourse marker non-dm unclassified
avg. % of total words per group
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0
Germans
Americans
Figure 2.1a Distribution of the linguistic item so – American versus German speakers
0.7 0.6
avg. % of total words per group
Germans Americans
0.5 0.4 0.3 0.2 0.1
TRP
IMR
SA op.
SA q+r
BOM
SEQ*
SRE*
MIU
0.0
R/C*
Figure 2.1b Distribution of the discourse marker functions of so – American versus German speakers (an asterisk indicates a significant difference between the speaker groups)
Chapter 2. So
Table 2.2 Discourse marker functions of so and their abbreviated labels Full label
Abbreviated label
Marking result or consequence Main idea unit marker Summarizing/rewording/giving an example Sequential so Boundary marker Speech act marker – question or request Speech act marker – opinion Marking implied result Marker of a transition relevance place
R/C MIU SRE SEQ BOM SA q+r SA op. IMR TRP
in contrast, the difference in discourse marker frequency is significant not only in the Median test but also in the t-test. Figure 2.1b gives a more detailed overview. It graphically represents the quantitative results given at the end of each section above for the nine discourse marker functions of so. The columns show the average frequency of the functions per 100 words in each of the two groups, American and German speakers. The asterisk indicates for which functions the difference between the groups was statistically significant. The labels of the functions had to be abbreviated; Table 2.2 correlates full and abbreviated labels. Only three functions show statistical differences between the US speakers and the German speakers. For so marking result and consequence (R/C) and for so used to summarize, reword or give an example, the frequency among the Germans is, on average, only two-thirds of the frequency among the Americans. For sequential so, the difference is considerably greater. The Americans use it more than eight times as often as the Germans. All three functions are situated at the textual level. In contrast, all functions at the interactional level were more or less equally frequent in both groups; there are no statistically significant differences. All native speakers and all but one of the German speakers used discourse marker so, and all its functions were employed in both groups by at least 15% of the speakers; none of the functions were exclusively found in one or the other group. The function of marking result or consequence was most widespread; only five Germans and one American did not use it. In contrast, so as a marker of a transition relevance place was employed by the least number of speakers – only four US speakers and fifteen Germans – but up to three times per speaker.
.. Non-linguistic factors In Chapter 1, I mentioned a variety of non-linguistic factors that might potentially be correlated to the frequency of a discourse marker. Of these factors, three can be analyzed on the basis of the demographic data for the Giessen Long Beach Chaplin Corpus: age, gender, and relationship between the partners. Each of these three have
Discourse markers in English discourse
been checked for a correlation with the frequency of individual functions and groups of functions of so.
... Age As explained in Section 1.7.1.2, the speakers of my data were divided into three age groups: from 18 to 21 years, 22 to 24 years, and 25+ years. While the Germans have between 17 and 37 members in each group, the middle group is empty for the Americans (the single person aged 23 is ignored for statistical purposes). The younger and the older group of Americans have 21 and 9 members, respectively. A statistically significant relationship could be found only for so as a marker of questions and requests. This function was used by the younger American group 3.5 times as much as by the older American group. For the Germans, frequencies were compared pair-wise (younger vs. middle aged group, middle aged vs. older group, younger vs. older group), which resulted in some statistically significant differences. The youngest group used the textual function of so to summarize, rephrase, or give an example almost four times as much as the middle group; they used the marker of questions and requests slightly more on average than the middle group. The distribution for the latter function is fairly even among the middle group of speakers; sixteen of the 37 members of that group used it once or twice. In contrast, seven of the eight instances in the youngest group come from a single speaker. For both functions, the difference between the youngest and the oldest group is not significant, in contrast to the Americans’ result. Instead, the youngest German speakers used so for implied result significantly more than the oldest speakers. The middle group in comparison to the oldest group employed the interactional functions in general and one non-discourse marker function more (adverb of degree or manner) and one non-discourse marker function less (so expressing purpose). ... Gender Regarding gender, the statistical analysis yielded some interesting differences between the sexes. In the following, I will only report those differences which were statistically significant. In the group of the US speakers, the females used textual so (particularly result/consequence and sequential so) significantly more than the males. So marking result or consequence was approximately 1.6 times as frequent in the females’ data, and sequential so almost three times as frequent as in the males’ data. This tendency, however, cannot be shown for the German speakers, even though the percentage of females in each group was approximately the same (71% of the US, 75% of the German speakers). Instead, the German female speakers used the interactional functions, particularly so as a speech act marker, both for questions and requests and for opinions, on average between four and six times more than their male colleagues. The only function of so used (3.7 times) more by the German male students is so in fixed expressions, i.e. and so on and or so.
Chapter 2. So
... Relationship between the partners The relationship between the participants, friends or strangers, was not found to have any significant influence on the frequency of so. This result stands in contrast to Redeker’s (1990: 376) findings. In her data, “speakers who were talking to a stranger used more ideational markers than those who were talking to a friend”. ‘Ideational markers’ in her definition include “semantically rich connectives” such as so (cf. also Section 1.7.1.5). However, the different results may be due to the fact that her ‘semantically rich connectives’ cover a wide range of markers, from question words to temporal connectives and causal conjunctions (1990: 372). Another reason might be a tighter restriction on who counted as friend or stranger. Of her participants she says that “eight of the 16 speakers were describing the film to a friend, who was also a student. The other eight speakers met their listener for the first time in the experiment” (1990: 370). The participants in the GLBCC were just asked in the questionnaire how they rated their relationship: friend, acquaintance, stranger, without any further explanations. Furthermore, for the influence of the partner relationship she compared only monologic sequences (1990: 371), while I compare both monolog and dialog sections. Jucker and Smith (1998) also found an influence of the relationship between the speakers on discourse marker frequency, but since their range of markers under investigation include like and you know but not so I will come back to their study in the relevant chapters of this work. ... Role In Section 1.7.1.6, I argue that the role of the speaker could possibly play a role in the distribution of discourse markers, including so. Speakers in role B always, i.e. for both narrative and discussion, had a hearer present face-to-face, who also reacted to what they said. This was the case for speakers in role A only for the discussion part. They did not receive any oral feedback to their narrative of the first part of the movie, even if they asked whether they could look at the researcher during their narrative. Speakers in role C had to make do without an active listener for both narrative and discussion part. Thus, it might reasonably be assumed that interactional functions are most frequent in the group of speakers in role B. In fact, this expectation is not quite met by the actual results. It turned out, for the American native speakers, that the speech of those in role A did not differ significantly from the speech of those in role C with respect to any of the functions of so. Surprisingly, a significant difference could not be found for speakers in role B versus speakers in role C either. A comparison between roles A and B, however, yielded some significant differences for textual functions of so: result/consequence, main idea unit marker, and sequential so. In each case, a considerably higher average frequency was found in the data of the speakers in role B. Above, I also reported higher frequencies for result/consequence and sequential so for American female speakers as compared to American male speakers. The gender distribution within the roles indeed is not exactly equal: 30% of the speakers in role A and 27% of the speakers in
Discourse markers in English discourse
role B are male. However, such a difference does not necessarily mean that there is an interrelationship between role and gender in this case. The statistical analysis of the German speaker data also revealed differences between the roles. Like their American counterparts, the speakers in role B used the main idea unit marker and sequential so on average much more than those in role A. Instead of result/consequence, though, they employed another textual function, summarizing/rewording/giving an example, also more than speakers in role A. Unlike the results in the American data, these three textual functions were not only significant for the comparison between role A and role B, but also for the comparison between role A and role C. Additionally, an interactional function, the speech act marker for opinions, showed a significant difference between A and C. Its frequency was the same for A and B, but higher for speakers in role C.
.. Linguistic factors More interesting seem to be the linguistic factors covered in the GLBCC. As mentioned before, we asked the German students whether they had spent time in an Englishspeaking country, and if so, where and for how long. Furthermore, we inquired about their use of English as a primary means of communication in several formal and informal contexts: with family members of the older and the same generation, with friends, and at school. The final question related to the participants’ acquisition of English: in which country they had learned English, and in which context(s) for how long, e.g. in school, at university, with friends or in a family, etc. For my analyses, I distinguished between formal and informal contexts as described in Section 1.7.1.7. In this division of contexts, the acquisition and use of English at school, university, and at a language institute count as formal, whereas the acquisition and use of English through or in the family and with friends count as informal. Periods of time spent in an English-speaking country also count as informal contexts. According to my hypothesis, Germans with a higher degree and longer time of either usage or acquisition of English in these informal contexts should use discourse marker so and/or some of its individual functions more than those students with less use or acquisition in informal contexts. An influence of learning English in a family or with friends could be shown only for so to summarize/rephrase/give an example, which was used four times more by those who had such informal learning environments. Those who at least sometimes used English with friends and/or family members committed only half as many errors in transferring constructions from German into English. Thus, my hypothesis proved right in saying that informal contexts for learning and employing English do have some influence on individual functions of so, but not as much as might have been expected. I said before that, for the quantitative analysis, I would not consider the British speakers due to their small number (four participants). I also said that I suspected differences in discourse marker use between the Americans and British speakers, even though I do not have sufficient data to provide evidence for this. However, in
Chapter 2. So
the GLBCC project, we asked the German participants whether they had spent time abroad, and if so, where. This type of information invites a number of comparisons. First, I compared those who had been abroad with those who had not. Again, we find a considerable influence of native speaker contact on the number of translation errors, in the sense that the errors are less frequent on average in the data of speakers who have contact to native speakers. Presumably, this contact helps the learners to replace the expressions transferred from German to English with native English expressions. Furthermore, German speakers apparently learned abroad to use so as a boundary marker. Finally, the textual functions as a group were also more frequent in the group of students with experience in an English-speaking country. Let me now bring the assumed British-American difference into the analysis. Of those who had spent time abroad, some lived only in one country for some time, others spent time in both Great Britain and the United States of America. Usually, though, one stay was much longer than the other. If I divide the German speakers into a group which was not abroad or spent more time in Great Britain and a group who had spent more time in the US, differences for two of the textual functions become significant in addition to textual functions as a group: so marking result or consequence, so to summarize (etc.). Sequential so misses significance with p = 0.054. All of them are clearly used more by the US-influenced group. However, if I leave out those who had not been abroad from the first group, so that it contains only those who had spent more time in Great Britain, all but one difference become insignificant. Only so to summarize reaches significance level, again with the American-influenced speakers using it more than four times more than the British-influenced ones. In a pair-wise comparison between students who never went abroad and (a) those who spent more in Britain and (b) those who spent more time in the US revealed that the time in Britain improved only their skill in avoiding translation errors (reduced from 0.068 to 0.013 per 100 words), while the time in the US significantly increased the frequency for textual so in general and for the main idea unit marker and the boundary marker in particular. Finally, I combined all this information to create a factor entitled “native speaker contact”. So far, even short periods of time abroad or speaking English with some friends had been sufficient for an inclusion in the respective group. The selection of the group with native speaker contact, however, was based on contacts which I deemed extensive enough to allow for the German speakers to pick up native competence features. That is, the speakers in that group had either spent more than four weeks in an English-speaking country and/or used English as primary means of communication more than just occasionally with friends and/or learned English in an informal environment. It turned out that both textual and interactional functions as groups were clearly and significantly used more by those who had extensive contact with native speakers. Besides the function groups, so as a boundary marker was also more frequent in that speaker group, while translation errors were rare, in comparison to those Germans without sufficient native speaker contact. Since the same difference in translation errors was also found between those who had been abroad and those who had not, it can be assumed that the substitution of incorrect, transferred expressions by
Discourse markers in English discourse
correct, native expressions is not so much effected by contact with native speakers in the L1 country (i.e. Germany) than by immersion in the target language, i.e. English.
0.07
avg. % of total words per group
0.06 0.05 0.04 0.03 0.02 0.01 0.00 age 18–21
age 25+
Figure 2.2a Distribution of so as speech act marker – question or request – according to age (American speakers)
0.14 0.12
avg. % of total words per group
age 18–21 age 22–24 age 25+
0.10 0.08 0.06 0.04 0.02 0.00 SRE
implied result
interactional
Figure 2.2b Distribution of some functions of so according to age (German speakers)
Chapter 2. So
.. Summary of the quantitative results for so The following section summarizes the quantitative results for the individual functions and function groups of so according to various linguistic and non-linguistic factors and illustrates them with charts.
Non-linguistic factors. The analysis according to age groups was not very enlightening. While the younger American speakers used more so for marking questions and requests than their oldest counterparts, the youngest German speakers employed so more for one textual function – summarize/rephrasing/giving an example – than the middle group, and more to mark implied result. The medium group in turn had a higher frequency of interactional functions of so than the oldest participants (see Figures 2.2a and b). More interesting are the results regarding gender distribution, which are illustrated in Figures 2.3a and 2.3b. The American female participants used textual functions in general and result/consequence and sequential so in particular more than the American males. The German females, in contrast, employed more interactional functions of so than the male participants, particularly for marking speech acts (both categories) and implied result. The relationship between the partners apparently did not have any influence on the frequency of the functions of so, while the role of the speakers had some impact. Among both native and non-native participants, the textual functions were more frequent in the data of speakers in role B than in those of speakers in role A (see Figures 2.4a+b). In the American speaker group, there were no significant results for the comparison role A vs. role C; in contrast, all of the significant differences between German speakers in roles A and B were also (and in part even highly) significant for role A vs. role C, plus a further difference in so marking an opinion. Speakers in role C employed so in all of these functions even more than speakers in role B. Linguistic factors. Learning English in an informal context had an impact only on so to summarize/rephrase/give an example; its frequency in the respective speaker group was closer to the native speaker frequency. In contrast, using English in an informal context did not affect the discourse marker functions of so, but it helped the German speakers to avoid translation errors. So did some time spent in an English-speaking country, particularly in Britain. Time abroad apparently also was responsible for a higher frequency in using so as a boundary marker and generally in its textual functions. Whether this time was spent in Great Britain or in the United States of America made a difference only for so to summarize/rephrase/give an example; it is possible (but cannot be proved for lack of the respective information) that this result is related to the result for learning English in an informal context, i.e. that the speakers who acquired English informally were surrounded by American English. Figure 2.5 presents the statistically significant differences between participants who never went abroad, those who spent more time in Britain and those who spent more time in the US. An
Discourse markers in English discourse 1.4 US male US female
avg. % of total words per group
1.2 1.0 0.8 0.6 0.4 0.2 0.0 R/C
SEQ
textual
Figure 2.3a Distribution of some functions of so according to gender (American speakers) 0.16 German male German female
0.14
avg. % of total words per group
0.12 0.10 0.08 0.06 0.04 0.02 0.00 SA q+r
SA op.
IMR
interactional
Figure 2.3b Distribution of some functions of so according to gender (German speakers)
analysis of all linguistic factors taken together to form a new factor ‘native speaker contact’ showed that both textual and interactional functions of so were more frequent in the data of the German speakers with such contact, as Figure 2.6 illustrates.
Chapter 2. So 1.4
avg. % of total words per group
1.2 US role A US role B
1.0 0.8 0.6 0.4 0.2 0.0 R/C
MIU
SEQ
textual
Figure 2.4a Functions of so according to role (American speakers)
avg. % of total words per group
0.10 German role A German role B German role C
0.08
0.06
0.04
0.02
0.00 MIU
SRE
SEQ
Figure 2.4b Functions of so according to role (German speakers)
SA op.
Discourse markers in English discourse 0.08 not abroad more GB more US
avg. % of total words per group
0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00 transl.
SRE
MIU
BOM
Figure 2.5 Distribution of some functions of so according to time abroad (German speakers) 0.7
avg. % of total words per group
0.6 no contact contact
0.5 0.4 0.3 0.2 0.1 0.0 textual
interactional
Figure 2.6 Distribution of so according to native speaker (NS) contact (German speakers)
Chapter 3
Well
. Functions of well in the literature The discourse marker well, unlike so, has received a lot of attention in the literature and has been covered in a number of different ways. One of the earliest descriptions of well as a discourse marker is found in Lakoff (1973). In a study on the appropriateness of answers and the reasons for this appropriateness, she noted that answers might be prefaced by well, but only under certain conditions: if the answer is an indirect one, i.e. if the respondent “is not giving directly the information the questioner sought” (Lakoff 1973: 458) or if the information supplied with the utterance prefaced by well is only part of the answer (1973: 459). Closely related is the case of well being used when “the speaker senses some sort of insufficiency in his answer” (1973: 463). In narrative, well may also be used “to indicate that details have been omitted” (1973: 464). Seven years later, Svartvik published his fairly comprehensive article “Well in conversation”, in which he treats “well in its function of sharing device” (1980: 168). He agrees with Lakoff ’s (1973) analysis of well in answers (Svartvik 1980: 172, 174), but he also finds evidence for a number of other functions, which he subsumes under “two major uses of well: as a qualifier and as a frame” (1980: 173). The following list summarizes his findings: As a qualifier, well indicates or marks ... – – – –
“Agreement, positive reaction or attitude” “reinforcement” “the non-straight and incomplete answer to the wh-question” “a non-direct or qualified answer”
(1980: 173f.)
As a frame, normally occurring non-initially, well ... – – – –
“shifts the topic focus to one of the topics which have already been under discussion” “introduces explanations, clarifications, etc.” indicates “the beginning of direct speech” functions as “editing marker for self-correction” (1980: 174f.)
In addition, he claims that well also functions at the level of discourse techniques: “as floorholder, hesitator, or initiator” (1980: 176). Despite this considerable number of functions he identified for well, he still states that “well in fact displays a much wider range of functions than those discussed here” (1980: 176).
Discourse markers in English discourse
Another seven years later, Schiffrin’s influential book Discourse Markers appeared (1987), in which she dedicated one chapter to well.10 Her focus lies on its interactive function in the participation framework (one of the levels in her discourse model) and the role of well regarding the coherence of discourse (1987: 102f.). As she accords in the introduction to this chapter, “[m]any of the specific results of my analysis parallel previous analyses of well” (1987: 102); thus, in many of the instances she describes, the use of well as a marker of contributions which are “not fully consonant with prior coherence options” (1987: 103) parallels Lakoff ’s and Svartvik’s analyses of well marking indirect, insufficient, qualified, non-straight and incomplete answers. The idea of a qualified answer is extended to cases “when respondents begin their answers with minimal tokens of question acknowledgement, and then add more information” (1987: 108), and indirect answers may even consist of a whole story. She also claims that her analysis does not only apply to question/answer pairs but also to request/(non)compliance pairs (1987: 114ff.). In the second section of her chapter, Schiffrin describes cases which do not fit the pattern of pairwise interaction, which she assumed for the first section. Among them are three subtypes of what she calls orientation shifts: “adjustments to the ideational content of talk (self-repair; [...]), changes in the deictic center of talk (reported speech), and alternations in the objectivity of the talk (reflexive frame breaks)” (1987: 123). Marking self-repair and reported (direct) speech are functions of well that Svartvik lists as frame uses of well (see above); the “reflexive frame breaks”, though, refer less to topic shifts but rather to a “shift in orientation from description to evaluation of events” (1987: 125). In addition, Schiffrin found that well may also preface contingent answers and requests that refer to previous utterances (1987: 120ff.). Concerning her model of discourse with five planes of talk (cf. Section 1.4.5), well “has its primary function in the participation framework because it anchors a speaker into an interaction as a respondent” (1987: 316). However, it also works at the four other levels: As a marker of self-repair, it functions in the ideational structure; since it regularly occurs at turn-transitional points, it also functions in the exchange structure. And “since individuals can respond to anything in talk which temporarily disrupts their expectations for upcoming coherence – ranging from unexpected knowledge [...] to a request with which they cannot comply – well also functions in information states [...] and action structures” (1987: 316). Three authors find that well is used to signal and mitigate some sort of confrontation. Starting from an investigation into conversational units and sequences based on Sack’s and Schegloff ’s Conversation Analysis, Owen (1981) reviews Lakoff (1973). Owen finds fault with Lakoff ’s usage of intuition-based examples rather than real data and with the lack of a systematic treatment of turns. Accordingly, Owen herself analyzes tape-recorded conversations, but limits her study to “instances of well which preface the second pair-part of an adjacency pair” (1981: 108). She finds that well is used at the beginning of the following second pair-parts:
Chapter 3. Well
– – – –
disagreement or qualified agreement with a statement of the conversational partner, responses to questions which cancel a presupposition of the question, non-compliances with requests, and rejections of an offer. (1981: 109)
All of these moves are seen as face-threatening acts (in the sense of Brown and Levinson 1978). Well then has the task of indicating that such a face-threatening act is about to occur and thus attends to the hearer’s face wants. However, Owen also presents an example of an adjacency pair with two inserted question-answer pairs, in which both the insertion and the return to the original adjacency pair are prefaced by well. She admits that the idea of face-threatening acts does not seem appropriate in this example, but does not present a solution. Watts also interprets well as “a move minimizing the face threat” in a facethreatening act (1986: 44). While Owen employed the notion of face without any differentiation, Watts uses Brown and Levinson’s distinction between positive and negative face and combines it with the notion of relevance with reference to Sperber and Wilson’s early version of Relevance Theory (1978). Thus, he is able to explain Owen’s last example as follows (1986: 44f.): The first well, which introduces the inserted question-answer pair, can easily be seen as minimizing a face-threatening act, since its speaker (B) does not give the expected answer to the initial question by his/her conversational partner (speaker A). The second well, which introduces the return to the original adjacency pair, i.e. which prefaces the answer to the initial question, can be seen as attending to the positive face of speaker B. Here is where the notion of relevance comes in: Generally, the positive face of a speaker is threatened whenever his/her contribution might be judged as not being relevant to the ongoing conversation. Well shifts the focus back from the inserted sequences to the original adjacency pair and thus indicates that speaker B’s contribution is nevertheless relevant. In Watts’ own words, well thus is “a device available to the speaker for the minimization of a possible face-threat contingent on a failure (whether real or assumed) to abide by the axiom of relevance” (1986: 58). This interpretation of well is also applied to a number of other examples from real data, with turn-initial as well as turn-internal well. Moreover, Watts claims that well in narratives is used for the same reason. When narrator uses it at points in the narrative where s/he shifts the topic focus (e.g. between background information and complicating action or between action and final comment/coda), its function is to forestall any possible challenges that s/he is not being relevant, i.e. possible threats to his/her positive face. Finell’s (1989) paper has a totally different focus, even though she also limits her discussion to well in responses. She is interested in the historical background of well: The OED cites a source from 1300 in which well “expresses a wish to be in good standing with a person” (Finell 1989: 654). In another citation from 1560, well contains a concessive element, i.e. in the cited case, it signals a partial agreement with the opinion of somebody else with an objection following (1989: 655). In both cases, “well is a
Discourse markers in English discourse
signal that the speaker wants to be on friendly terms with the interlocutor, in spite of the fact that the latter might disagree with him/her about the issue in question. These uses of well obviously have been maintained until the present day. Like Watts, Jucker (1993) employs Relevance Theory as a descriptive framework, but he does so more consistently and does not mix it with other frameworks (such as Politeness). In his paper, he takes up examples presented by other researchers and offers a re-interpretation in relevance-theoretic terms. The uses of well illustrated by these examples include well as a marker of insufficiency, as a face-threat mitigator, as “a frame marking device indicating a topic change or introducing direct reported speech” (1993: 438) and as a delay device. In all of these cases, as Jucker argues, well can be seen as a signpost signalling to the hearer that the context created by the previous utterance – whether produced by the current speaker or the current listener – is not the most relevant one for the interpretation of the impending utterance. (Jucker 1993: 440)
This is easy to see in the case of well as a marker of insufficiency of answers, since the insufficiency is defined in terms of the expectations the hearer (or speaker of the question) has regarding the answer. These expectations are based on explicit or implicit assumptions. By prefacing his/her answer with well, the speaker signals that “at least some of these [...] assumptions do not hold” (1993: 441). A similar reasoning can be applied to well as a face-threat mitigator (1993: 444f.). When a speaker A asks another speaker B to agree with a statement or makes an offer or request, s/he assumes (among other, more context-dependent things) that B is able and willing to do so. Well prefacing the response by B indicates that one or more of these assumptions is not correct. This is also true in turn-internal instances of well in which the assumptions of the interlocutor are not as explicitly denied as in disagreements, refusals and rejections of offers/requests. Thus, situations in which well is used “are often face-threatening for one of the participants, but well does not directly signal the face-threatening act but the shift in the relevant context” (1993: 451). In narratives, changes of the topic or focus are often introduced with well, as well as direct reported speech. According to Jucker, this is so because at these points “it is particularly relevant to reorient the interactants to a new set of background assumptions” (1993: 446), which are not as explicit as in questions, requests or offers, but are still present and influential for the hearer’s interpretation of the story. Finally, Jucker re-interprets well in an example by Svartvik (1980) in which speaker A wants to correct B’s assumption concerning a specific location (“B: on the floor”) and does not immediately find the appropriate word for that location (“A: on on [ði:] well on [ði:] you know on [ði:] hatchway there”, Jucker 1993: 447). Much less convincingly than in the other examples, he tries to apply a relevance-theoretic explanation and concludes that “the renegotiation of cognitive environments includes that of the speaker A herself ” (1993: 448). Schourup (2001) claims that there are certain cases of well which neither Schiffrin’s coherence-based approach nor Jucker’s relevance-theoretic approach can explain. In
Chapter 3. Well
one type of cases, there is no divergence from prior assumptions (according to Schourup), as in the sequence Marie looks lovely tonight. – Well so she does! (2001: 1027). Another type are discourse-initial uses, which are not covered by either approach because both assume an utterance prior to well. Therefore, Schourup rethinks well based on its treatment by Bolinger (1989), who considers well to be epistemic. In the sense in which Bolinger uses this term, it means that “well is used to ‘invoke a norm’ (1989: 315) and, more specifically, indicates ‘deliberate comparison’ with a norm” (Schourup 2001: 1030). Schourup then works through eleven types of well which Bolinger had set up, including exclamatory well, repeated well, and discourse-initial well, and critically reviews them. Even though Schourup cannot agree in all instances with Bolinger’s view that well refers to a norm, he generally supports his analysis. In an attempt to find a comprehensive description for all uses of well, Schourup arrives at the following, very general one: “the speaker who uses well is taking into account what is (or, more accurately, what is relevant), whatever the speaker takes that to be under circumstances” (2001: 1036). He then makes a considerable effort to show that well can usefully be viewed as a ‘mental state’ interjection, indicating that the speaker is in the mental state of considering. This idea of well indicating consideration is applied to a number of intuited examples and replaces Bolinger’s “notion that well evokes a norm” (Schourup 2001: 1057). In terms of individual functions of well, Schourup observes that a speaker using well can be “heard as considering what to say next” (2001: 1039, 1043). Thus, well functions as a continuative. In contrast to so or after all, well ensures that the statement it introduces is heard as the speaker’s own thought rather than the thought of his/her interlocutor in examples such as I’m always tired. – Well, you need to sleep more. (my example). Furthermore, well as a mental state interjection at the beginning of a quotation “conjures up the quoted speaker as a living presence giving on-the-spot thought to the communicative business at hand” (2001: 1046). Well is also said to have a structuring function. As we have seen above, in Svartvik’s (1980) view well can be used to change to a previously mentioned topic. Norrick (2001), however, claims that discourse markers in oral narratives, including well, are invested with special organizational functions, since “[s]torytelling differs significantly from regular turn-by-turn conversation in its sequential implications” (2001: 851). He investigated but and well in oral narratives, based on the narrative framework proposed by Labov and Waletzky (1967) and Labov (1972). In this framework, a narrative is divided into six parts: an abstract, an orientation (containing background information), the complicating action, the result or resolution, a coda, and evaluation at various points. Along this framework, Norrick found a number of organizational functions of well: – – –
“well serves as a discourse marker signaling the beginning of a story following a digression or interruption” (2001: 850), “well can signal the beginning of a new episode” (2001: 854), it can be used to return to the main story after a dialog part (2001: 855) because
Discourse markers in English discourse
– – –
“it signals a primary element of the story line, setting it off from digressions, interruptions and topical turn-by-turn talk” (2001: 854), it is “oriented toward furthering the main action and formulating the point of the story” (2001: 851), it may preface the coda of the story (2001: 854).
It may seem as if Norrick’s account provided the ideal basis for an analysis of well in the Chaplin narratives. However, this is not quite the case, for reasons that lie in the way the data were collected. Labov and Waletzky’s narrative framework is applicable to personal, free and spontaneous narratives in which the speaker aims at capturing the hearer’s attention at the beginning and throughout the story. In the GLBCC, in contrast, retelling the movie was a task that did not depend on personal interest. Some of the participants said or showed that they found the movie boring and that they were not really interested in retelling it or in listening to the narrative. Even for those narrators who liked the movie it was clear that the hearer would not try to divert the topic of the conversation to something else. This does not mean that the narrators did not use dramatic elements such as, for example, direct reported speech in the present tense. However, there was no need neither for an abstract or for a climax in the narrative or for a coda in order to meet the hearer’s expectations. Moreover, the participant hearer of the narrative (i.e. the one in role A) had already seen half of the movie and thus did not need much of an orientation (background information) to it. Accordingly, Labov and Waletzky’s framework can hardly be applied to the Chaplin narratives of the GLBCC, and since the functions of well found by Norrick rely on this framework, they are only indirectly transferable to the functions of well my data. The individual functions of well which were identified by other researchers and presented in this review formed the basis on which I build up a categorization system that fit the uses of well in my corpus, although my category descriptions do not follow any specific framework. As we will see, not all of the functions described in the literature are found in the corpus. Due to the experiment setup, for example, potentially face-threatening requests and offers do not occur, nor does discourse-initial well, since the narrative of the speakers is always preceded by the researcher’s instructions (which were not transcribed). On the other hand the uses found in the corpus are not limited to the descriptions given in the literature.
. The functional categories of well in this study For the linguistic item well, it is relatively easy to establish a dividing line between discourse marker and non-discourse marker uses. On the non-discourse marker side, we find it described in dictionaries (e.g. Cobuild 1987; cf. also OED 1989) as a verb (“tears welled up in my eyes”) and as a noun (“a hole in the ground from which a supply or water is extracted” or “an oil well”). As an adjective, it means that something or somebody is “in a good or satisfactory condition” (e.g. Webster’s 1994). Well also
Chapter 3. Well
functions as an adverbial, mostly equivalent to the adjective good. Furthermore, it is found in a wealth of hyphenated compounds (such as well-behaved, well-being, wellborn, well-done; Webster’s 1994) and in a number of fixed expressions, for example in as well (as). This list of non-discourse marker functions is by no means complete; however, not even all of those functions listed here occur in the GLBCC. Below, I will treat only the functions found in my data. In these functions, well is generally not omissible without rendering the syntactic construction it appears in ungrammatical. Well as a discourse marker has become what Ariel (1994: 3251) calls an “opaque operator”, in contrast to so. This is reflected in how dictionaries deal with it: What is generally considered discourse marker well appears in several dictionaries as interjection or exclamation (e.g. OALD 1974; COED 1976; Cobuild 1987, 1995; OALD 1989; Webster’s 1994; OALD 2000; CIDE 1995; Macmillan 2002) with long lists of what it is used for. I do not want to present these lists here, but I will refer to them where appropriate in the sections on the individual functions. All of them have in common that well is syntactically optional; thus, well conforms to the major criterion for discourse markers. Table 3.1 gives an overview of the discourse marker and non-discourse marker functions of well in my data. In the GLBCC, the linguistic item well occurs 547 times, which is relatively infrequent in comparison to so, occurring 1819 times. Well was used for only two non-discourse marker functions, but for no less than twelve discourse marker functions. The discourse marker functions are further divided into those at the textual level and those at the interactional level. I will briefly explain and illustrate the non-discourse marker functions first. Table 3.1 Functions of the lexical item well as used by German and American speakers in the GLBCC Non-Discourse Marker Functions: – adverbial use – “in addition” Textual Level: – searching for the right phrase – rephrasing/correcting – quotative well – move to the main story – introducing the next scene – conclusive well Interactional Level: – indirect answer – direct answer – response to self-raised expectations – contributing an opinion – continuing an opinion/answer – evaluating a previous statement
Discourse markers in English discourse
.. Non-discourse marker functions of well ... Adverbial use With few exceptions, well as an adverb (cf. Cobuild 1987; OED 1989; Webster’s 1994; CIDE 1995; and others) occurred in the discussion sections of the recordings, when the participants expressed their view of the movie, how they liked it and what they considered the major challenges of making a silent movie. The most frequent collocation of well as an adverb was (very) well done, as in extract (54), followed by played/acted well, as in extract (55). (54)
520 I thought .. that was pretty well done. (123B, NS)
(55)
350 all I wanted to say is that .. she plays & 350a & her part really well. (75B, NNS)
Several instances of hyphenated well (well-dressed, well-educated, well-off, etc.) all came from German speakers, while the other forms occurred in both native and non-native speech. However, less than a third of all adverbial uses came from the native speakers and more than two thirds from the Germans.
... “In addition” One of the uses of well in fixed expressions, also called idiomatic (OALD 1989) or phrasal (Cobuild 1987) use, had twice as many instances as other adverbial uses in the data and therefore deserved a category of its own. This phrasal use is generally paraphrased in dictionaries as “in addition”. In principle, it occurs in two forms: as well and as well as, but there was only one instance of the latter in my data (see extract (56)). All other instances occurred in the shorter form as well, illustrated in extract (57). (56)
316 ... as well as as that on=e fat man who & 316a & actually stole the money. (71B, NNS)
(57)
98 and she got uh she got beans as well, (40B, NNS)
The native speakers used well in expressions meaning ‘in addition’ only eight times, which corresponds to eleven percent of all its occurrences in the data. That is, almost 90% came from the German speakers.
.. Discourse marker functions of well In the section of the GLBCC that I used for my research, I found twelve distinguishable discourse marker functions of well. As with the functions of so, I checked the uses in the data against whatever functions have been described in the literature. This will
Chapter 3. Well
be reflected in my categories; however, I also discovered functions that had not been described before. In unified accounts of well, this marker can be seen as “a signpost signaling to the hearer that the context created by the previous utterance – whether produced by the current speaker or the current listener – is not the most relevant one for the interpretation of the impending utterance” (Jucker 1993: 440). Another view expresses that “the speaker who uses well is taking into account what is (or, more accurately, what is relevant), whatever the speaker takes that to be under the circumstances” (Schourup 2001: 1036). Unlike the accounts provided by Jucker and Schourup, my description of the functions of well does not intend to say what well means nor to formulate a core meaning of it. Rather, its functions are based on recurrent patterns of where well occurs in the narrative or interaction between the conversational partners, and what element in the narrative or discussion it marks. These elements can be textual, or they can be interactional, as we will see in the following sections. Accordingly, the discourse marker functions of well can be subsumed under one of the two levels, textual or interactional. Half of them serve to mark difficulties of the speaker with certain expressions or mark shifts within the narrative or between text types. The other half serves dialogic functions, indicating the type of relationship between the current utterance and a previous one: responding to questions asked by the other partner, evaluating a previous statement, contributing an opinion. Let me illustrate and explain my categories in detail, beginning with the textual level.
... Searching for the right phrase Schourup (2001: 1039) in his detailed account of well briefly mentions that the speaker in two of his examples “can be heard as stopping to reflect on how she should refer to herself ”. A similar idea is expressed in the Longman Grammar of Spoken and Written English (Biber et al. 1999: 1087), in the section on discourse markers: “Well can also occur in the middle of an utterance as a signal of self-correction or deliberation over the choice of expression”. According to the Cobuild dictionary it can be used “just before or after you pause, especially to give yourself time to think about what you are going to say” (Cobuild 1987, 1995). To my knowledge, these are the only references to instances in which well indicates that the speaker is searching for the right phrase. However, this function is by no means infrequent in my data. Typically, it is combined with other means of expressing ‘deliberation’, for example filled and unfilled pauses, truncated words and intonation units, repetition, or other markers. (58)
112 113 114 114a 115 116 117
this’s the supposed .. Land of Liberty, .. um .. at which point the film takes & & a dramatic twist with their liberty & & being taken away by & & being roped into a side of the ship. and forming an orderly cue, .. of uh--
Discourse markers in English discourse
118 119 120 121 122 (63A,
well. uh .. an orderly cue, @, I suppose you can call it, (H) um, NS)
In extract (58), the speaker tries to explain what he has seen: The passengers of the ship are pushed into one corner of the ship behind a rope, in a rather rough manner, and have to wait for being called out to pick up their papers. Speaker 63A in lines 117 and 118 is looking for an expression that corresponds to what he has in mind, but apparently does not find anything appropriate. So he just repeats the phrase he had used before, an orderly cue. Well here is surrounded by a filled pause, truncation of the intonation unit (line 117), another filled pause, an unfilled pause and a repetition. Truncation and multiple pauses around well in this function are frequent, particularly in the non-native speakers’ speech, but not necessary conditions for well to be categorized as a marker of searching for the right phrase. Everything is possible, from only one such other indicator, to several of the same kind (e.g. several filled pauses), to multiple combinations, as in extract (58). This extract shows one extreme of a continuum; extract (59) illustrates the other extreme, where well is only accompanied by the marker I guess: (59)
337 B: [and then] he just ends up on the streets & 337a & of well I guess New York? (4B, NS)
The two extracts above do not only differ in the number and types of deliberation markers but also in the outcome of the search. While speaker 63A did not find an appropriate expression of his thought, despite some seconds of attempting it, the search process of speaker 4B did not take long and was successful. Again, the two examples form the ends of another continuum describing the success of the search process. More frequently are cases in which the speaker, after some thinking, finds a phrase that s/he is satisfied with for the moment, as in extract (60) describing a scene between Chaplin and his lady-love: (60)
242 242a 243 244 245 (22B,
so .. you know he tries to propose and & & all of this stuff, and .. she’s being real shy=, but you know .. she lets him-well he kind of kisses her, NS)
A final extract may be allowed to illustrate a further difference within this category. So far, I have talked about ‘the right phrase’ or ‘the appropriate expression’. In many cases, the speaker is indeed searching for a linguistic phrase that conveys his or her thought or mental image. In other cases, however, the speaker is searching for content: What happened next in the movie? This is not surprising, given that the movie they
Chapter 3. Well
have seen basically consists of a series of slapstick scenes. Extract (61) gives a perfect illustration of a speaker who has explained one scene and is now searching for the content of the next: (61)
97 98 99 100 101 102 (27B,
B: A: B:
I think you can imagine how. <SV mhm SV>. and yeah well, what was then, I think-...(2.2) um, NNS)
This function of well was significantly more frequent in the German speakers’ data. Only nine percent of all its instances come from native speakers. This may not be surprising, since the Germans as non-native speakers can be expected to have difficulties expressing themselves in English; the more so, since the terminology required for the first half of the movie (which is set on a ship) is not necessarily a familiar one. Lexical difficulties of the Germans also became obvious when speakers asked, in English or German, for the English term for something, or simply used the German term.
... Rephrasing/correcting Well can also be used to rephrase or correct an utterance. This function has not received much more attention than the previous one in research into discourse markers. Svartvik in his treatment of well included the “use of well in its framing function [...] as editing marker for self-correction” (Svartvik 1980: 175) and found that it “frequently occurs in a context of hesitation phenomena” (p. 176). Schiffrin distinguishes between well marking self-repair, which she defines as “adjustments to the ideational content of talk” and background repairs “after which speakers return to the development of what they had been talking about prior to the addition of background material” (Schiffrin 1987: 123). This background repair seems to be less a repair proper, but rather a modification and extension of the background. As such, it plays a minor role in my data. The citation from Biber et al. (1999: 1087) above also contains the reference to well as a “signal of self-correction”; the Cobuild dictionary suggests its use “because you want to rephrase sth that you have just said” or “when you are modifying or correcting sth that you have just said” (1987, 1995). The majority of the cases found in my data are corrections in the sense that the first phrase or expression is incompatible with the one that substitutes it, that is, if one is true, the other cannot be true. In extract (62), for example, the speaker starts saying they say that-- (line 177), stops in the middle of the intonation unit, then corrects herself to he says that using well to introduce (or frame, in Svartvik’s terminology) that correction, so that we assume that there was only that one male person who said what follows in lines 178 and 179. This extract also represents the majority of the corrections in two other aspects: the apparently faulty phrase or expression is followed by truncation, and well im-
Discourse markers in English discourse
mediately follows the truncation. However, as other cases (e.g. extract (63)) show, a truncation need not be the case. (62)
176 177 178 179 (69B,
and then they go outsi=de, they say that-well he says that tomorrow he’ll engage & & them in some sort of a .. job, NNS)
(63)
91 91a 92 93 (47A,
they sit down and they h- have this game & & of .. uh cards. Charlie Chaplin um ... decides to shuffle, well is chosen to shuffle, NS)
In extract (63), the speaker decides that his utterance in line 92 is not quite appropriate. It was not Chaplin’s own decision to shuffle; rather he was chosen to shuffle. Since the speaker had finished the sentence structure before he decided to correct it, no truncation was necessary. The pattern that well immediately follows the intonation unit which contains the expression to be substituted, however, is realized in all cases of correction but one. The object of correction varies. Above, we have seen a pronoun and a verb being substituted by another. Other objects can also be nouns (e.g. this guy-- ...(1.4) well the priest or whatever, 47B lines 333f.) or noun phrases. In a minority of the cases classified as rephrasing/correcting, the emphasis is more on “rephrasing”, that is, there is not necessarily a sense of incompatibility between the utterances before and after well. Extract (64) illustrates such a case. Speaker 57B describes one of the first scenes in the restaurant, in which Chaplin orders and eats a bean dish and a French loaf in a particular manner. In the speaker’s opinion, he ate it apparently does not seem to be a sufficiently exact expression to depict what Chaplin did with the food. Therefore, she rephrases this expression as well he had one bean at a time to describe his way of eating more exactly. (64)
70 B: and ... (H) then Charlie Chaplin & 70a & ordered <@ beans @>, 71 and .. a big roll. 72 .. or a big <@ French loaf thing @>, 73 (H) and then ...(1.2) (H) .. he ate it. 74 well he had one bean at a time, (57B, NS)
This category of well also includes one case which resembles Schiffrin’s ‘background repair’. It does not exactly contain a background extension, but at least background information that the speaker had forgotten and adds later. However, as in Schiffrin’s example and definition, the speaker then returns to what she had started to say before.
Chapter 3. Well
(65)
175 yeah after they have paid, 176 they ...(1.4) get out of the restaurant, 177 a=nd he and the young lady, 178 they go to= .. they go around the streets, 179 and they are=-180 well he they get a job, 181 from the artist, 182 and get paid for it, 183 and the first thing they do is they & 183a & .. they marry. (208B, NNS)
In extract (65), the speaker keeps to the chronological sequence of events in the movie in the first two lines of the extract. Then she skips the scene in which the artist confirms that he engages Chaplin and his lady-love for a job, and in which Chaplin asks for and receives some dollars on account. This scene still takes place close to the restaurant. As the speaker is about to retell what Chaplin and the lady do towards the end of the movie (i.e. they marry), she suddenly remembers the scene she skipped before and deems it important enough to be reported as background knowledge to the marriage scene. The rephrasing/correcting function achieved about the same frequency as the search function in the native speaker group. The German speakers, however, used the same number of tokens, which means that the frequency of rephrasing/correcting well per 100 words in their group was only half as high as its frequency in the native speaker group. In both functions I have described so far, rephrasing/correcting and searching for the right phrase, there is a focus on lexical expressions and thus on very basic textual units. Therefore, these two functions clearly work at the textual level. In the following four functions, well is used to indicate shifts: shifts between the speaker’s voice and quoted material and shifts between parts of the narrative or argument.
... Quotative well (66) and he’s like
... (131A, line 88; NS)
The function of well illustrated in extract (66) above is a very salient one, not only in its syntactic characteristics but also in its pragmatic function. As a frame, to use Svartvik’s terminology, well here functions “as a signal indicating the beginning of direct speech, parallel to that of quotation marks in writing” (1980: 175; cf. also Macmillan 2002). In this function, it is typically preceded by a verbum dicendi or any other construction that conveys speech, and is followed by the quoted utterance. Extract (66) illustrates the most frequent collocation for quotative well in the native speaker data, that is, the quotative construction BE+like functioning as verbum dicendi (cf. Blyth et al. 1990; Ferrara and Bell 1995; Jucker and Smith 1998; Tagliamonte and Hudson 1999; Winter 2002) plus well. Other possibilities are go, say (which is particularly frequent in the Germans’ speech), ask, or think. Extract (67) shows that the construction conveying
Discourse markers in English discourse
speech can also be substituted by a particular voice quality which, for the quotation, is clearly different from the way the speaker normally talks. This is indicated in the transcript by setting the quoted text between angled brackets and the capital letter Q (i.e.). The speaker here imitates a talk show host: (67)
319 A: what’s like studious. 320 @@@@@@, 321 ((noise)) 322 B:, (131B, NS)
The two extracts above may give the impression that all instances of quotative well were in such a different voice quality. However, this is the case only in five of the seventeen native speaker instances. Extract (68) gives an example of well in this function without quotative voice quality: (68)
B: and then he’s like well the .. lady wants a coffee. (131B, line 259; NS)
Quotative well has been treated or at least mentioned by Svartvik (1980), James (1983), Schiffrin (1987), and Jucker (1993). For Schiffrin, “well marks the orientation shifts created by reported speech, i.e. talk whose original time, place, and possibly author is concurrent with the ongoing conversation” (1987: 124). In her four examples, three include well in the quotation (indicated by an apostrophe before well and at the end of the quotation). Well in such cases may also replicate one of its other functions besides introducing reported direct speech. In contrast, her fourth example illustrates a case in which well apparently is just used for setting off the following words as those of somebody else, since what follows is reported indirect speech. However, Schiffrin does not explicitly ponder the question of whether well belongs to the quoted utterance. Did the quoted person also use well? Svartvik writes on that question, “[i]ts function can then be taken merely as a signal indicating the beginning of direct speech, parallel to that of quotative marks in writing. However, in such collocations, well can also have the same functions as a turn opener in regular conversation between two or more participants” (1980: 175). For James, “it may not be immediately clear from the context whether it belongs to the preceding reporting clause (RC, NS) or following direct speech (DS)” (1983: 33f.). According to James, phonological features do not consistently mark well as belonging to reporting clause or direct speech either (1983: 37). Jucker (1993) refers to both Svartvik and James and comes to the conclusion that “it makes little sense to distinguish between those instances of well that belong to the reporting clause and those that belong to the quotation” (1993: 447). He sees well as a indicating a shift in the relevant background. For quotations, this means that well indicates a shift in the deictic center, from here and now to some other time and place. This seems to be more important if the speaker quotes him/herself: Jucker finds that well before direct reported speech is more frequent after I said than after he/she/they said.
Chapter 3. Well
In the section of the GLBCC which I analyze, there is no instance in which a participant quotes his/her own speech prefacing it with well. Rather, my data seem to support the position that well is a part of the direct speech: when a quotation voice quality is used for the reported utterance, well is included in that voice quality; in several other cases, the pitch of the voice is raised on well or on another marker before well (e.g. oh or OK). Moreover, Schourup (1985: 33f.) conducted an experiment in which 58 subjects were asked to punctuate an utterance containing well plus quoted speech. The overwhelming majority of them (51 subjects) attributed well to the quoted speaker rather than the quoting speaker by inserting quotation marks before the marker. Nevertheless, Schourup still treats well as an enquoting device, that is, as means inserted by the current (quoting) speaker to indicate that what follows well is quoted material (1985: 33ff., 69ff.). The type of data presented here makes the question particularly intriguing: The participants retell a silent movie. There are only occasional inter-titles with direct speech, and they do not contain well, nor are they prefaced by well when they are quoted by the participants. Despite the fact that the movie is a silent one, the participants used a lot of direct speech to describe what happened, thus assuming what the movie characters said. Accordingly, there is no way for us to check whether the participants added well where no well existed in the original speech. However, the frequency of well in this position seems to speak against an interpretation of well as part of the original and for an interpretation of it as a structural device used by the narrator.11 Even though I do not have any figures to support my claim I do not consider it likely that it would have occurred as often in “real” speech.12 Schourup (2001: 1046) obviously takes the same stand: well “is used as the first element in quotations much more often than literal quotation would require”. Assuming that the major function of well at the beginning of direct speech indeed is to indicate a transition between two types of text within the narrative, that is, between what the speaker says and what s/he quotes, it clearly operates at the textual level. In the frequency list of the native speakers, well at the beginning of a quotation occupies rank two; in the German speakers frequency list, it occupies a middling position. The frequencies of the two speaker groups do not achieve a statistically significant difference (which would be assumed at a level of p = 0.05).
... Move to the main story Linguistic papers are not the only sources of comments on discourse marker functions. Several dictionaries, under the heading of well as an interjection, also include remarks that it expresses “resumption or continuation of talk” (COED 1976), which is a rather vague description. More specifically, the Cobuild (1987) dictionary claims that well is used “to indicate that you are changing the topic, and are either going back to sth that was being discussed earlier or are going on to sth new”. In the OALD (1974) we find that well is “used to resume a story, etc.”
Discourse markers in English discourse
As Norrick (2001: 851) claims, “[s]torytelling differs significantly from regular turn-by-turn conversation in its sequential implications, so that we might expect it to invest DMs with special organizational functions not found in other forms of talk”. Since the narratives in the GLBCC were not conversational narratives, embedded in a conversation and used to make a point, but delivered under special conditions, we might expect this statement to be even more true. The function of well to indicate a move to the main story, for example, has not been mentioned in any of the publications presented so far, even though Norrick provides a functional description which comes very close to this function of well in the GLBCC, as we will see. According to Aijmer (1996: 209), well signals “a change of focus to a new part of the discourse”. However, it seems that the focus need not be on a new part; Svartvik (1980: 174) maintains that well “shifts the topic focus to one of the topics which have already been under discussion”. In a narrative, well is “oriented to the main action or point of the story in progress” and “carries the promised story forward” (Norrick 2001: 853f.). Since the narrators in the GLBCC did not intend to make a point, that is, did not have any other intention than retelling the movie according to the instructions they were given, well in this latter function plays a minor role. More fitting to the data in the GLBCC is another observation Norrick made: “well serves as a discourse marker signaling the beginning of a story following a digression or interruption” (2001: 850), even though the digression or interruption does not seem to be a necessary condition for well to occur. In about half of the instances in the category ‘move to story’, well marked the transition from the instructions given by the researcher to the beginning of the narrative. This occurred mainly in two forms. Extract (69) illustrates the first form, where well is the very first word of the speaker after the instructions. Extract (70) illustrates the second form, where well is preceded by OK and/or a filled pause. (69)
42 B: well, 43 when you left they went in to America. (57B, NS)
(70)
4 A: OK, 5 ... um, 6 ...(1.5) well, 7 the film starts on a ship, (63A, NS)
(71)
79 80 81 82 83 84 85 86
B:
A:
OK, @@@. um .. you left when they divided the people up, and allowed only the um .. the old lady and & & her .. and her daughter .. to go in, every[body]-[I was left] um a few seconds before, but--
Chapter 3. Well
87 B: ... OK. 88 well they [they] they um reached the & 88a & United States, (69B, NS)
In some other cases, there was a “prelude” to well and the start of the narrative itself. One (German) speaker in role A first protested that there is not really much I can tell about it, because it was a silent movie. In three other cases, speaker B first negotiated the starting point of the narrative with the partner, as in extract (71). Well at the return to the story after a digression or interruption also occurred several times. Twice, the speaker inserted an evaluation of the movie or a scene before returning to the narrative proper; in other cases, the interruption consisted of additional instructions or the request of the partner for clarification. Another frequent subtype of the ‘move to story’ category does not involve the narrative proper but rather the fulfillment of the instructions for discussion. A number of speakers (who were almost exclusively German) turned or returned to the questions on the instruction sheet using well, as in extract (72): (72)
286 B: well um, 287 what about .. the acting. 288 what do you think of the manor (sic!) & 288a & character, (45B, NNS)
The participants have just discussed whether they liked the movie, why, and which scene was the funniest for them. Speaker B then returns to the instruction sheet; the next question on the sheet reads, “What did you think of the acting by minor characters?” Thus, it resembles Svartvik’s shift of focus to a topic already mentioned before, even though the focus is not exactly shifted to a topic in Svartvik’s sense but to (part of) the instructions. Well was used in the same way after the end of the narrative to approach the first of the questions for discussion. In all of the cases described above, well functions at the textual level by marking a transition between and within text types: from instruction or digression/interruption to narrative, from narrative to discussion, and from discussion back to dealing with the next question of the instructions. Well used to move to the story was one of the more frequent functions among the German speakers. They used it twice as often as the native speakers.
... Introducing the next scene While the well described above mainly introduced a move back to the (main) story, the function of well in this category is the introduction of the next scene in the narrative. This function was also mention by Norrick (2001: 854), although he, referring to free narratives rather than retellings of a movie, calls them ‘episodes’. Well here also works at the textual level. Let me illustrate this function with a somewhat lengthy extract which contains several instances of well introducing the next scene.
Discourse markers in English discourse
(73)
234 235 240 240a 241 242 243 244 245 246 247 248 249 249a 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 (47B,
A: B:
A: B:
A: B: A: B:
... well then .. Chaplin’s thinking oh & & you know I’d better .. find my money. .. and then this beggar comes in off & & the street, .. yeah, (H) and shining this nice coin up you know, and polishing it all up nice and he’s like, ... you know .. I know that’s my coin. well the guy has eaten, ... and would-.. no the guy-the waiter comes in ... to the ... beggar, an’ .. gives him his food and asks to & & give him his ... (H) coin, and then the WAIter puts it in his pocket, ... and turns, .. and @ <@ goes out @>. I’m just gonna give him change or what the & & .. coin falls out of the WAIter’s pocket. right. ... an’ then .. Chaplin sees that, so he .. slams his foot on the coin, an’ .. started playing games back and forth & & to ... to get up and get the coin. well [then] ... he picks it up, [yeah], .. and goes to pay with the same coin. uh huh, cause the guy didn’t know he dropped it. .. <@ a=nd @> the waiter looks at it, .. bites it and it bends. meaning it was fake. ... it’s probably gonna bust him good. you know he’s gonna get in trouble for it. well, ... and there he sees this lady.
NS)
The extract starts after the speaker has told his partner that Chaplin just witnessed another customer being beaten up because he was ten cents short on his bill. With the first well, speaker B moves from this scene to describing Chaplin’s thoughts. What speaker 47B does not say at this point is that Chaplin searched for his coin in vain until he found the hole in his pocket, which leaves him in a state of desperation. When the beggar comes into the restaurant from the street and shines a coin, Chaplin apparently assumes that it is the coin he himself found before and lost again. The second well, in line 245, marks the transition from Chaplin observing and trying to contact the beggar, to the next scene, in which the beggar pays for his meal and the waiter loses the coin through a hole in his pocket. This transition may not be absolutely clear, though,
Chapter 3. Well
because the speaker remembers in lines 246f that he has not said anything yet about the beggar having ordered something. The next well in line 260 marks the transition from the slapstick scene in which Chaplin tries to pick up the coin but is interrupted by the waiter, who becomes increasingly suspicious of Chaplin’s behavior, to Chaplin paying with this coin. After having described the payment scene, the speaker uses another well to move to the scene in which Chaplin meets his lady-love. As in all of the instances above, the transition between scenes typically involves a transition in the focus of the narrative. This is the case for example when a new character is introduced, or when the speaker changes the focus from Chaplin’s thoughts to what Chaplin sees and observes. Well to introduce the next scene in the German speaker data is about 1.5 times as frequent as in the native speaker data. Nine Germans (i.e. 11.6% of the group) use it between one and three times. In the native speaker group, it is not distributed evenly at all. The six instances in this group were produced by only two of the 34 speakers. One speaker used it once, another five times. The latter speaker was an American exchange student at the Justus Liebig University of Giessen and was recorded in Giessen. Unfortunately, we do not know how much time he had already spent in Germany at that point and whether any influence of either German or of the English of his German friends on his use of well in English is possible. We only know that another American (exchange) student recorded in Giessen apparently was so mentally involved in German that he described the coin which Chaplin found in the street as a “Fünf-Mark-Stück”.
... Conclusive well In my data, I found two functions of well that were only used by Germans, and not at all by the native speakers, neither American nor British. Conclusive well is one of them and accordingly, it is not described in the relevant literature in the way it occurs in the GLBCC. The Cobuild dictionary (1987, 1995) mentions that it is used “to indicate that you have reached the end of the conversation”. This does not quite apply to the instances in the GLBCC, since the conversation typically continues, even if with a different scene description or discussion topic (see below). Norrick (2001: 851) suggested that well can mark the transition to “the final summary of a story” as one of the primary sections of a narrative. Due to the specific conditions of the narratives in the GLBCC, however, such a “final summary” is missing. Rather, conclusive well was sometimes used to give a conclusive statement of one intonation unit that summarized (the attempt at) a description, a scene or an opinion, not the point of the story or the story itself. More often, it preceded a conclusive statement that did not contain a summary. In any case, the use of conclusive well meant the end of the current scene description or discussion (sub-)topic. It thus organizes the discourse at a textual level, in a way similar to well marking the transition to the next scene. Extract (74) illustrates a case of well which introduces the summary of an opinion. The partners in this extract had (previous to the extract) exchanged their opinion on how they liked the movie: it was not bad, but they would not pay money to watch
Discourse markers in English discourse
it. However, they know that Chaplin and his movies were popular at his time, and therefore, they try to find reasons for his popularity. Speaker A finally concludes the discussion with the summarizing statement in line 593, preceded by well. (74)
585 585a 586 587 587a 588 589 590 590a 591 592 593 (42A,
(75)
59 because um I don’t know if & 60 & it’s the captain or a man of the crew, 61 because he sees that Charlie Chaplin & 61a & um um .. does something h- @@-62 or looks for something in the pocket & 62a & of this girl, 63 and he of course thinks that um the & 63a & tramp is a pickpocket. 64 a=nd so um he um tries um to um clear & 64a & the situation, 65 but um the girl finds the money in her pocket, 66 and is s- so happy, 67 a=nd um um wants to thank um this tramp, 68 so that um the crew member-69 ... well. 70 now knows that this is no pickpocket, 71 ... well. 72 and then .. after that you= can see & 73 & there comes .. the scene when they & 73a & arrive at the port, (69A, NNS)
B:
but maybe .. in former times & & he was a .. more typical way to-A: .. yes in former times [you] have no .. & B: [mhm]-A: & other movies .. <X to see X>. B: yeah. [that’s]-A: [that’s] the only uh possibility & & to .. watch a movies at all, and he was very popular and-B: mhm. A: well I think .. people .. really liked him. NNS)
Many of the instances of conclusive well carried a notion of “I have said enough about this scene or topic”. Well then was followed by a move either to the next scene or to the next topic. The speaker in extract (75) had first described how Chaplin put money into the pocket of a girl whose mother had lost hers. Several times he takes it out, takes some notes to keep for himself, puts the rest back, but she does not notice what he is doing. Instead, a crew member observed him. This is where extract (75) starts. The first well that occurs in the extract, in line 69, was classified as search for the right phrase (see Section 3.2.2.1). After having retold the scene, the speaker adds well in line 71, thus conveying the idea that she thinks, or has thought, about whether
Chapter 3. Well
there is anything else to say about the scene, but does not find anything worthwhile. Accordingly, she then begins her description of the next scene. In the narrative part, the meaning of conclusive well also can often be paraphrased as “I’m lacking more appropriate words to describe the scene in more detail”. This is applicable to the discussion part as well, when speakers do not know the vocabulary necessary to say what they think. Sometimes, however, the interpretation was closer to “I don’t know what to say anymore”. The participants in extract (76) discuss the acting of the major characters and the difference it makes when actors cannot or do not have to talk. While the participants may lack linguistic means to express themselves, they apparently also lack further ideas for more discussion on the same topic. Therefore, well in line 327 ends the present topic and already projects the upcoming shift to the next topic (viz. discussion question), which is realized in line 329. (76)
312 it’s this uh silent movie thing. 313 [I don’t]-314 A: [yeah]. 315 B: [[I]] don’t think we are used to it. 316 A: [[uh huh]]. 317 [uh huh]. 318 B: [because] we have all these & 318a & (H) [[...]] media stuff, 319 A: [[mhm]]. 320 B: uh=-321 A: mhm. 322 B: and and .. that’s just uh some & 322a & [...(1.2)] of the basics [[I]] think. 323 A: [mhm] [[mhm]]. 324 [[[not touching but]]]-325 B: [[[it’s acting without]]]-326 yeah. 327 A: .. well. 328 .. mhm. 329 ...(1.6) the minor characters? (209A, NNS)
In almost half of the cases, well formed an intonation unit of its own and was pronounced with falling intonation, as extracts (75) and (76) illustrate; this is indicated in the transcripts by a period at the end of the intonation unit. In another third of the cases, the statement which was preceded by well ended with falling intonation. Thus, intonation patterns strongly contributed to well being perceived as conclusive. As I mentioned at the beginning of this section, conclusive well was only found in the German speaker data and not in the native speakers data. With 0.026 instances per 100 words, it was one of the more frequent uses of well among the Germans. The six functions described so far all work at the textual level. Let me now turn to another six functions which structure the dialog between the partners and thus work at the interactional level.
Discourse markers in English discourse
... Indirect answer Well indicating an indirect answer is probably one of the functions described earliest in the history of discourse marker research (Lakoff 1973; see the beginning of this chapter), and the author perhaps one of the most cited in the research on well (cf. for example Östman 1981; Owen 1981; Holmes 1986; Finell 1989; Jucker 1993; Greasley 1994; Schourup 2001). Well here prefaces answers to questions or requests, and the speaker indicates that this answer is not a direct one or not a complete one. For Svartvik (1980: 174), this function falls under the use of well as a qualifier. ‘Indirect answer’ as a category can be understood here as including four subtypes: – – – –
answers which do not directly supply the information required, but from which this information can be deduced (cf. Lakoff 1973: 457), replies13 in which the answer to the question is delayed, i.e. given after some kind of acknowledgement or prelude (cf. Schiffrin 1987: 108), qualified replies which provide only a partial answer to the question (cf. Svartvik 1980: 174; Lakoff 1973: 459), insufficient answers, which do not provide the required information (cf. Lakoff 1973: 463),
and combinations of these subtypes. I will illustrate each of them in turn. Extract (77) twice illustrates the first subtype, indirect but deducible answers. Speaker 63A in both cases answers a question from the instruction sheet. The first question he answered read, “What did you think of the acting by major characters?” Speaker A does not reply directly, by saying for example, “I think the major characters did a bad job”, but by answering indirectly with well he’s only wandering in there, implying that there is nothing better to say about the major character(s). Thus, the speaker’s opinion is deducible from this utterance. To the next question on the sheet, “What did you think of the acting by minor characters”, he answers in a very similar way: well the minor characters didn’t do anything. Again, his opinion is deducible from this indirect answer to the question. (77)
318 319 320 321 322 323 324 325 (...) 333 334 335 336 (63A,
B: A: B: A:
A: B: A: NS)
[um], ...(1.1) ac[[ting by major]] characters. [[acting .. mhm]], ... well he’s only wandering in there. @@, @ all he does is waddle around with his & & trousers around his <@ knees @>, <@ waiting to stick ’em up @>. [minor] characters, well the minor characters didn’t do anything. .. no=. they just <@ kind of sat there @>.
Chapter 3. Well
Extract (78) gives two examples for delayed answers. The question answered here read, “What did you think was the funniest part?” After having dealt with the first question, speaker A uses well to move to the next question (line 316), even while speaker B start to say something possibly still referring to the first one. B takes up the change of topic, but instead of immediately giving an answer, he repeats part of the question. By doing so, he simultaneously gains time to think about what is his opinion on the matter and acknowledges the question. The first well indicating a delayed answer follows (line 319), but apparently, the speaker has not made up his mind yet; well here does not yet precede the answer proper but another utterance and another repetition of part of the question in order to gain even more time. The second well (line 322) follows, finally introducing the answer to the question at stake. (78)
316 317 318 319 320 321 322 322a 323 324 325 326 (66B,
<SV [well] the funniest part SV>. [a=nd uh], the funniest part-well, let me see @. <SV the funniest part SV>. well I think ...(1.8) the ... the & & dinner .. on the ship, A: [XXXXXX on the ship]. B: [on on the on on on] on the rolling sea, it was-yeah I think that was a funny part. NNS)
(79)
226 227 228 229 230 (45B,
B: A:
A: B:
... oh did you like it? ...(26.4) let’s ask .. did YOU like it. [@@], B: [well I] liked the second part. the first part was not funny. NNS)
Qualified answers which provide only a part of the information required, are exemplified in extract (79). Speaker 45B does not want to give an unqualified reply to the question whether she likes the movie because it would not correctly represent her opinion. Rather, she expresses what it was that she liked (and did not like) and prefaces this reply with well, to indicate that she takes the expectation of a direct answer into consideration. Finally, we find instances in which speakers did not really provide the information asked for. However, the speakers are usually aware of the fact that their replies are unsatisfactory and do not meet expectations, and therefore start them with well, as in extract (80): (80)
173 174 175 176
R: C:
what happened to the mother of the girl? ... o=h yeah. um, .. well,
Discourse markers in English discourse
177 I don’t really know, 178 but I guess she’s sick. (206C, NNS)
In the case of speaker 206C, the retelling of the movie had been rather short and without many details. In order to make the speaker talk more, the researcher asked about what happened to one of the characters from the first part of the movie. Speaker C remembers the character, but she is not able to give the relevant information and indicates this with well. At the same time, of course, this instance is also a case of a delayed (but insufficient) answer, as are others of this subtype. In applying this category, the type of question (for example yes/no question or wh-question) did not matter. However, Svartvik (1980: 174) comments that “[w]ell is particularly common in responses to wh-questions but also occurs after yes/no questions”. As observed before, the conditions under which the data of the GLBCC were elicited were not natural conditions, even though the task of telling a friend (or stranger) about a movie seems to be natural enough for young adults. Nevertheless, Svartvik’s observation still holds. All instances of well marking indirect or insufficient answers occur in the discussion part, and most of them are answers to questions from the instruction sheet. The first, double question read, “Did you like the movie? Why?” All other questions started with “What did you think ...”. Therefore, only about a quarter of the well’s (i.e. eight instances) in my data answered to yes/no questions. Of these, five referred to the first question on the instruction sheet for discussion, “Did you like it?” Only one yes/no question concerned a clarification of something in the narrative: speaker 5A wanted to know whether Chaplin and his lady-love went to the marriage office right after the restaurant scene. His partner replies, well they’re like walking, and there is right-- like they walked around the corner, and there it was. Neither a straight ‘no’ nor a straight ‘yes’ seemed to be appropriate to her, and so she prefaced her answer with well to indicate that her answer was not a direct one. The other two yes/no questions were combined questions/request: is anything else coming to your mind and can you compare it to other Chaplin movies. A simple ‘yes’ would be regarded as cheeky, since the expectation in case of a positive answer to the question is that the respondent says what is on his/her mind and does not hold it back. A simple ‘no’ might be more acceptable, but even then, the respondent would be accused of not trying to give a more elaborate answer. In all of the subtypes and examples discussed above, well shows the speaker’s consideration of the expectations that come with the question. Thus, the speaker interacts with the questioner. Even though the questioner in my data was the researcher in most cases and was not available as an interactant, the participants often treated the questions on the sheet as their own questions, so that the interactive force contained in the use of well was directed towards their partners. Therefore, well marking indirect answers as described in this section functions at the interactional level. In this function, well was found mostly in the data of the German speakers. They used it with an average frequency of 0.029 instances per 100 words:more than three
Chapter 3. Well
times more than the native speakers. Well marking indirect answers was one of the two interactive functions for which the difference between native and German speakers reached statistical significance.
... Direct answer As mentioned before, Lakoff was convinced that well occurred under certain conditions. Even though she does not explicitly say so, she seems to assume that straight, direct answers to a question are not prefaced with well. Concerning the conditions for well, she writes: well is used under two conditions: (1) when the answer sought can only be obtained by the questioner by deduction from the response given, and (2) when the reply is directed toward a question other than the overt one: that is, when a different question can be deduced by the respondent because of the conversation situation. (Lakoff 1973: 460; her emphasis)
Yet, I found several cases in my data where well prefaced even direct answers to which these conditions did not apply. Consider extract (81): (81)
143 please discuss ... questions such as, 144 did you like it and why. 145 A: .. (H) [well I] personally did not .. & 146 B: [<X OK X>] 147 A: & like the movie cause I [[don’t like]]-148 B: [[are we gonna]] & 148a & write on this? 149 A: no. (124A, NS)
Speaker 124A answers the (double) question in detail, directly, and without delay: did you like it? – I personally did not like the movie; why? – cause I don’t like--. Yet, she still prefaced it with well. One might perhaps object that I personally could count as a qualification of the answer, but it would not be a qualification in the sense that it provided only a partial answer, as in the category ‘indirect answer’. Furthermore, there are more examples in which there is nothing like a qualification: (82)
263 264 265 265a (141C,
.. and that was it. ...(3.4) um, ...(2.5) well I liked it cause it was & & just very comical, NS)
In extract (82), the speaker – a single participant – ends his narrative (and that was it) and then reads the instruction sheet including the questions for discussion. He had been instructed to express his opinion on the movie, following those questions, if he wanted. After reading the sheet, he therefore responds to the first question (Did you like the movie? Why?) by using its wording and giving exactly the information required, without any delay. These examples demonstrate that well is indeed possible
Discourse markers in English discourse
at the beginning of answers even when the conditions laid down by Lakoff are not met. However, Schiffrin notes that “well is more frequent when a larger set of answer options is encoded through the form of the question” (1987: 105). She found that well much more frequently preceded answers to wh-questions than to yes-no questions (1987: 107). This also seems to be true for my data; all of the questions in my data which were given a direct answer preceded by well were either wh-questions or combined yes/no- plus wh-questions. This category has the lowest frequency in both speaker groups. On average, it is only slightly more frequent among the Germans (0.008%) than among their American counterparts (0.006%).
... Response to self-raised expectations As we have seen, expectations that have to be met or at least acknowledged can be raised by questions from the conversational partner, or by questions from the researcher, i.e. on the instruction sheet. There are two other possibilities which occurred in my data, and which are similar to what Schiffrin commented about some of her examples: “In these examples, speakers are treating their own prior talk as something to be responded to” (1987: 123). This can also be said about the cases categorized as ‘response to self-raised expectations’. The following extract illustrates one of the two possibilities: (83)
321 B: well did you .. like it or <X why X> not, 322 A: well, 323 yes I like it because-324 ... well I like Charlie Chaplin, (69A, NNS)
In this and the parallel cases, the speakers like 69A above gave their opinion on something and continued with cause or because, after which they stopped short. In the transcripts, this is marked with two dashes for a truncated intonation unit. By starting an explanation of their reasons with (be)cause, the speakers raised the expectation that the explanation would indeed be given. The well following the truncation indicates their awareness of this expectation. The second possibility for creating such an expectation is using phrases such as what happened then, spoken with falling intonation as if the speaker were thinking aloud. In several of the relevant instances this might indeed be the case, since they come from single speakers without a partner. But compare line 33 in extract (84); this speaker role-played talking on a friend’s answering machine as vividly as if it were a real situation. She poses a question and subsequently answers it herself prefaced with well.
Chapter 3. Well
(84)
33 33a 34 35 36 (25C,
(H) .. well what else would you wanna & & know about it. ... hm, ...(1.2) well you gotta like Charlie Chaplin & & movies in order to like this movie, NNS)
Other instances are found in partner discussions, such as extract (85). They are half directed towards the partner and half towards to the speaker him or herself. This time, the utterance raising the expectation is not a question but a statement with the form of an imperative and the pragmatic force of a promise (line 98). Again, well (line 100) addresses and acknowledges the expectation raised, without any interference from the hearer. Nevertheless, the hearer is of course involved, since s/he waits for the promise to be fulfilled, the question to be answered, or the explanation that started with (be)cause to be completed, so that this type of well functions at the interactional level as well. (85)
96 97 98 99 100 100a (55A,
there are some COmic touches of course, (H) .. um-...(2.9) let me see if I remember them. ... (H) um-...(4.6) well the card game I guess & & with the .. the evil .. man. NS)
All of the instances with cause or because came from German speakers; the instances in which a question (or a formulation such as in line 98 of extract (85)) by the speaker him/herself raises the expectation come from both native and non-native speakers. On average, well marking the response to self-raised expectations is almost twice as frequent in the German group (0.01%) as in the native speaker group (0.006%). However, with a total of only 13 instances, it is too infrequent overall to show a statistically significant difference in the Median test.
... Contributing an opinion While the last three functions of well had to do with responding to questions, the next two have to do with expressing an opinion. For these two function, Schiffrin’s interpretation of well does not seem to hold. Schiffrin sees well as “a response marker which anchors its user in an interaction when an upcoming contribution is not fully consonant with prior coherence options” (1987: 102f.). Jucker’s conclusion is that well functions as “a signpost signaling to the hearer that the context created by the previous utterance – whether produced by the current speaker or the current listener – is not the most relevant one for the interpretation of the impending utterance” (Jucker 1993: 440). While the expression “not the most relevant one” might be interpreted in a way that even fits the examples from my corpus (see the extracts below), Schourup rightly comments that “both proposals [Jucker (1993) and Schiffrin’s] run afoul of examples [...] in which the second speaker does not diverge from an assumption created
Discourse markers in English discourse
by the prior utterance” (2001: 1028). Such a divergence indeed need not be the case, as several instances of well in my data show. Consider extract (86): (86)
622 622a 623 624 625 626 626a 627 628 629 629a 630 631 632 633 634 635 636 637 637a (47B,
B: A:
B: A: B: A: B:
.. but I think I think wha- wha’ interested & & me about this, when I watched it <X before X> is the way & & you have to like really concentrate. [<SV mhm SV>], [it’s not] like when you watch like & & .. you know Hollywood movie & & you like [[you SIT]] there, [[yeah]], and you don’t actually concentrate & & on [..] what’s going on it & [mhm], & jus’ uh comes into your head. .. (H) .. and [then she actually had to think]-[well .. usually it’s]-...(1.5) with with uh .. um a text it’s a & & lot easier to be interested, but ... (H) it was actually-I found it ... pretty interesting to & & [sit there and] watch,
NS)
Speaker 47A expresses her opinion of the movie and what it was like to watch it in contrast to watching a Hollywood movie: you have to like really concentrate. What coherence options does speaker B now have? He can agree with A, partially agree, or disagree. The option he chooses is to express his own opinion (beginning in line 633), which is in the same vein as A’s opinion and thus agrees with A. Nevertheless, he starts his contribution with well. (87)
168 168a 169 170 171 172 173 174 174a 175 176 177 178 179 180 181 182
B: A:
B: A:
they had like WAX or something all over & & the floor, (H) and all over the tables, so the things would slide back and forth more, (H) at least it looks like-.. it looked like it [was] WAX, [mhm] .. like it looked like it was covered & & with something. ... so I thought they did like-.. (H) yeah, pretty little .. uni=que ways of like-.. filming it? filming it. yeah. ...(1.2) but uh-[X]--
Chapter 3. Well
183 B: 183a 184 184a (1B, NS)
[well] also like when .. when they & & rocked back and forth, they could be only be moving the camera & & and not [<SV the people SV>].
Extract (87) is a similar case. Speaker 1A muses over a scene in the dining room of the immigrant ship in which Chaplin and other people slid back and forth according to the movement of the ship. Speaker B fully supports her partner, even by suggesting an expression to put A’s thoughts into words. When she starts to give her own opinion (line 183), she introduces it with well. In her contribution, there is no notion of contrast to what her partner said; rather, it fully supports her idea that the producers worked with certain tricks to achieve the effects in the movie. Thus, well here has the sole function of marking that the speaker begins to contribute his or her own opinion, which regularly supports the hearer’s opinion expressed before. Like the previous functions, it works at the interactional level. One description I have found in the literature which comes close to this function is Schourup’s assessment of well as a continuative. It effects that “the speaker is heard as considering what to say next”, but also indicates that “the speaker is actively taking into account what is already known or assumed” (2001: 1043). Somewhat more specific is one of the items in the Cobuild dictionary (1987): well can be used “to indicate that you are about to say something”. In the same vein are the functional features Fischer (1998: 124f.) attributes to well in a semantic test which combines utterances prefaced with well and supposed meanings or their negations, such as “well he’s been president of it for a long time, and indeed I wanted to say more about it”. She tested and confirmed three features of well than can be paraphrased as follows: “I have heard what you said”, “I wanted to say something now”, and “I wanted to say more about it”. The second paraphrase can indeed be applied to the current category of well, while the second is more fitting for the next category, as we will see. Well marking the contribution of the speaker’s opinion is slightly more frequent in the data of native speakers, who use it 0.01 times per 100 words. In the German speaker data, it is (with 0.008 times per 100 words) one of the three least frequent functions.
... Continuing an opinion/answer Like the previous function, the current one occurred only in the discussion part of the data. Svartvik names as one of the frame functions of well that it “shifts the topic focus to one of the topics which have already been under discussion” (1980: 174). However, he also mentions that, apart from those functions he described in detail, “we may detect a variety of specific, additional and optional, uses related to discourse techniques: as floorholder, hesitator, or initiator” (p. 176). Fischer (1998: 124f.) includes among the features of well also that it communicates “and indeed I wanted to say more about it”, as we have seen above. This seems to apply here. In the Cobuild (1987, 1995) dictionary, we find that well can be used “to indicate that you intend or want to carry on speaking”. Schourup’s (2001) ‘well as a continuative’ (see above) also comes into play
Discourse markers in English discourse
here. Biber et al. (1999: 1087) remark that “well marks continuation but with something of a contrast”. In the instances found in my data, however, this notion of contrast is absent. In the GLBCC, a number of German speakers used well not to reintroduce a topic that had been discussed before, but rather to continue expressing their opinion on a topic or their answer to one of the discussion questions, as speaker 209A does in the following extract: (88)
278 A: yes what I said before, 279 and I think .. um the scene in which he & 279a & manages to escape, 280 ... um .. after he’s won all the money, 281 B: mhm. 282 A: he um-283 I think that’s quite funny. 284 B: yeah with the revolver, 285 A: uh huh, 286 B: yeah. 287 A: well and .. the hat, 288 and he’s very elegant, 289 and (H) uh gentleman-like, 290 ...(1.2) although this doesn’t fit the & 290a & situation. (209A, NNS)
In the conversation before extract (88), speaker 209B had answered the question about the funniest scene. Speaker A does not like the movie, but she tries to find a scene that she also found funny and comes up with one in which Chaplin manages to escape from a threatening fellow dice and card game player by threatening him in turn with a gun that Chaplin had just acquired from him during the game. Speaker B signals that he remembers the scene, by supplying a detail A has not mentioned (line 284). A’s uh huh serves as an acknowledgement that B remembers the “correct” scene. This exchange is completed by B; he uses yeah to express his agreement with A’s assessment of that scene as a funny one. Finally, A takes the floor again starting with well and describes more details which she finds funny, thus continuing her answer to the question at stake. It is not necessarily the case that well is the first word in the utterance; the floor can also be taken using other words, e.g. and, as in extract (89). Speaker A has expressed her surprise at the serious face Chaplin keeps even when he is playing those funny parts. Speaker B cooperates with and supports her partner, not only with positive backchannels (lines 370, 372, 374) but also by suggesting just did it maybe once or twice as a supplement to A’s I don’t think they have filmed it too often. The last backchannel in line 374 also serves as a potential closing point for this topic. Speaker A, however, resumes his turn with a=nd and adds another remark to the topic/his opinion. Well here obviously is not the means to get the floor but a continuation marker.
Chapter 3. Well
(89)
367 368 368a 369 369a 370 371 371a 372 373 374 375 (42A,
and uh I don’t know if they-.. how often they had to .. (H) to uh & & try these scenes. and I don’t think they have .. filmed it & & too often, B: mhm. A: because they hadn’t so much money in & & earlier times. B: mhm they [just did it maybe once or twice], A: [XXX right mhm]. B: mhm. A: a=nd well he cuts .. this very very .. good. NNS)
(90)
347 348 349 350 351 352 353 354 355 356 (71B,
it was .. for that time I guess well done, but if you did it did this nowadays, where you can actually talk, that would be so eg- .. exaggerated. li- just like .. (H) um theatre or something. ...(2.2) so-.. I think-well it was well done, you could see= like um .. gestures, um or a .. a mime. NNS)
In another case (extract (90)), the partner of the speaker does not even take a turn, and does not even give a backchannel. B expresses her opinion about the movie and provides a potential transition relevance place by ending the utterance on falling intonation (line 351). Since A does not take the floor, even after B gives her more than one chance to do so (lines 352, 353), B herself continues and adds more to her opinion starting with well. As mentioned before, only German speakers used this function of well. With 0.012 instances per 100 words, it is more frequent in this group than well used to contribute one’s own opinion.
... Evaluating a previous statement One of the more generally acknowledged functions of well (besides marking indirect answers) is that of a mitigator of some sort of confrontation (Svartvik 1980; Owen 1981; Watts 1986; Finell 1989; see also the discussion at the beginning of the chapter). Nikula (1996) includes well in her long list of what she calls pragmatic force modifiers, which can intensify or decrease the force of an utterance. Smith and Jucker (2000: 216) mention in their paper on actually that, in their protocols, “a typical function of well was to downgrade a claim”. In both cases, well serves to make the following utterance less face-threatening. In the context of our experiment, there was not much occasion for confrontations or face-threatening acts at a personal level. There were basically no offers to be ac-
Discourse markers in English discourse
cepted or rejected, no requests other than asking the partner to begin the discussion; the narrative itself did not require any emotional involvement. Only in the discussion part, personal opinions were asked for. However, even when the two participants had different opinions about the movie, they treated them rather lightly, as just one part of the assignment within the experiment. Nevertheless, well was used in such cases; it introduced utterances which evaluated a previous statement or a conjunct of statements. This evaluation often referred to a statement that the partner had made, but even more frequently to the speaker’s own previous utterance(s) or to a conjunct of statements to which both had contributed. Of course, the evaluation of the partner’s utterance rarely was expressed bluntly (or ‘bald on record’, to use Brown and Levinson’s 1987: 69 terminology), in a “what you’re saying is nonsense” style. Since opinions were given solely for the purpose of the experiment and not for the defense of a personal conviction, there was no need to carry out face-threatening acts which would have endangered the relationship between the partners. There was only one pair in which well evaluating the partner’s statement prefaced simple no’s. Speaker A constantly tried to anticipate his (female) partner’s narrative, so that she sometimes had to tell him that his assumptions of what happened next were incorrect. Two of these instances started with well no as one intonation unit; the correction of the assumption was given in the next intonation unit. More often, speakers of well disagreed with their partners only partially, or at least made it seem that way, as in extract (91). (91)
182 183 184 184a 185 186 187 188 189 189a 190 191 192 192a 193 (46B,
A:
you can show everything like-.. I mean it’s dinner time, you don’t have to write .. & & [the sentence] it’s dinner time, B: [yeah], A: an’ .. then that the woman lost all her & & money or that it .. somebody stole it, I mean ... you can show it somehow. so I don’t think .. that is the SENSE & & of making a silent movie. B: well uh yeah, but .. uh when they guy came up to to them, an’ um ...(1.9) they uh he said ... I’m & & an artist, nobody would have .. guessed [that]. NNS)
Speaker A, who has seen only the first part of the movie, does not see the use of the inter-titles which were shown in the first part. Indeed none of them would have been necessary to understand what was happening. Speaker B expresses with well that she does not totally agree; however, since she can see the point speaker A makes, she hesitantly confirms A’s statement first (line 190). Then she gives the reason for her own opinion, i.e. that at least some of these inter-titles are necessary: nobody would have guessed that one of the characters in the second part of the movie was an artist.
Chapter 3. Well
Extract (92) twice illustrates the evaluation of the partner’s statement without a positive backchannel. Speaker B does not like Chaplin movies and does not find them funny. Speaker A, in contrast, finds them interesting. Her evaluation of speaker B’s opinion is given in line 140, with a short pause and oh well. The well alone already expresses her disagreement with B’s statement, not only in line 140, but also in line 148. The first well is followed by a statement of her own opinion, in a form which parallels the statement it contradicts. The second well overlaps with B’s speech; even though there is a longer pause after the overlap, A does not take the floor again until later. (92)
136 B: I hate Charlie Chaplin <SV movies SV>. 137 A: why? 138 B: .. I don’t know. 139 I don’t think they are funny. 140 A: .. oh well-141 I think they are interesting. 142 B: ...(1.5) because? 143 A: ... it’s just s- so different what the & 144 & people at these times could laugh about, 145 B: yeah. 146 OK. 147 maybe interesting but not funny ... & 147a & for ... today. 148 A: ...(1.8) [well], (202A, NNS)
The speakers in my data evaluated their partners’ utterances, but even more often they evaluated and modified their own. Speaker 1B (extract (93)) gives an example of an evaluation without modification, speaker 59B (extract (94)) a typical example of a combined evaluation and modification. Speaker 1B first expresses her dislike for silent movies. When she must assume that her partner has seen more than this one silent movie, she evaluates and qualifies her own statement about not liking silent films (in general) by saying that she has never seen others. Therefore, her dislike must rather be taken to refer to this one film instead of silent movies in general. This interpretation is enforced by the added so at the end of the intonation unit; the hearer is expected to draw the appropriate inference herself (cf. Section 2.2.2.8). (93)
229 B: 230 A: 231 232 233 B: 234 (1B, NS)
I don’t really like .. um .. silent films. I don’t either. ...(1.8) but-... it was OK for .. as .. a silent film goes, yeah. well I’ve actually never seen one <@ so @>.
Discourse markers in English discourse
(94)
156 157 158 159 160 (59B,
(H) but then, .. um= the= artist saidand-... well .. <@ he didn’t say it @> @@. he showed it, .. that he was going to take over the bill, NNS)
Instances of the type illustrated in extract (94) often occurred in the narrative section of the recordings. Speaker 59B here reports something that happened in the movie and then declares this something to be (at least partially) incorrect: Since the movie is a silent one, she cannot have heard anybody say anything. Nevertheless, she can make reasonable assumptions about what he must have said (he showed it). In the intonation unit(s) after the evaluation introduced by well, the evaluated statement is therefore modified. In another instance of self-evaluation, the speaker (45B, NS) said about Chaplin’s lady-love she didn’t do anything. well, .. her part was not to do anything. Well here conveys that the speaker is not satisfied with her own formulation and that she considers the modified statement after well to be more correct or perhaps fair. (95)
498 (...) 505 505a 506 507 508 509 509a (55B,
...(1.4) and Charlie was the good guy. A:
...(1.2) did you .. not mind the .. & & character being like that? B: ...(1.7) um, A: .. didn’t matter, B: ...(3.1) yeah well, (H) .. i- in the second part Charlie was & & not only the good guy. NNS)
In the German speaker group, a third option occurred. Speakers used well to evaluate not a single statement but a jointly produced context. In extract (95), for example, speaker B first states that Charlie was the good guy and (in the lines omitted) explains his good deed. A then asks whether she did not mind that and, when B hesitates to answer, suggests that it didn’t matter. B finally answers with a well that is at the same time an evaluation of her own statement, an evaluation of A’s suggestion, and a response to A’s question; it is followed by a modification of her assessment of Chaplin’s role. Despite some instances of evaluative well in monologic narratives (by speakers in role A or C), it occurs overwhelmingly in the narrative by speakers in role B, that is, those who had a partner for their narrative, or in the discussion part. When it refers to the speaker’s own statement, it gives directions to the hearer on how to interpret the speaker’s utterances. When it refers to the partner’s statement, it has a mitigating effect. Either way it is interactional. In both cases, it conveys that the speaker does not totally agree with a previous statement. On average, the German speakers used well for evaluating a statement more often than the American speakers (0.043 vs. 0.028 times per 100 words). However, there is no statistically significant difference between the groups. An interesting variation can
Chapter 3. Well
be found within the category: while the native speakers used well overwhelmingly to evaluate their own utterances, the German speakers used it with about equal frequency for their own and for their partners’ statements.
... Unclassified instances In the parallel section of Chapter 2, I said that a number of instances of so were not categorized due to unintelligible or truncated passages. This kind of cases also occurred for well, even though the problem here did not lie as much in passages that we could not acoustically understand but rather in the fact that the speakers did not finish the utterance started with well. Sometimes this was due to an overlap between the speakers and the partner carrying on, sometimes the speaker him/herself stopped in the middle of an intonation unit and then went on with a different grammatical construction and/or topic. Extract (96) illustrates the first case. Speaker B describes the beginning of the second part of the movie, which is set somewhere in the streets of (presumably) New York. When she uses well, it overlaps with the beginning of a question A asks. B reacts to the question and thus does not continue with whatever she was going to say after well. (96)
165 and he’s walking around an’ he has & 165a & no ... m- .. money, 166 so he can’t get food he’s ... hungry, 167 ... a=nd [well], 168 A: [why cause] he gave all his & 168a & money to that girl? 169 B: yeah. (0) 170 A: (0) what a moron. (131B, NS)
Extract (97) shows a case in which a speaker changes her mind without finishing a sentence or idea after well. Line 298 might be considered a continuation of line 297, but the speaker finally continues with a different sentence structure (line 301). (97)
296 A: ...(1.4) @@ they can’t really talk. 297 well u=h (Hx)-298 ... Charlie Chaplin-299 B: mhm. 300 A: it’s-301 ... the woman was quite good I think. (209A, NNS)
More interesting perhaps are the instances in which the context seems to be clear, but not the function of well, or in which well has an identifiable function which does not occur elsewhere in the corpus. In extract (98), speaker A was given instructions by the researcher and acknowledges them with oh .. well before beginning the narrative. Well is used here in a place where typically OK was employed in my data.
Discourse markers in English discourse
(98)
5 A: oh .. well. 6 @@, 7 ...(2.4) the film’s uh playing on a ship, (49A, NNS)
(99)
514 B: 515 516 516a 517 518 519 519a (5B, NS)
(H) <SV what else what else what else SV>, u=m, ...(1.8) ((sucking and hissing sounds)) <SV trying to think SV>, tshtshtsh, well almost. ... what about the .. beginning of & & <X our X> part,
In the last extract, (99), however, the function of well is not clear at all. Speaker B is trying hard to think of the funniest scene and uses well when he thinks he has found one that he liked. On average, unclassified instances due to truncated intonation units and unfinished utterances were about equally frequent in the Americans’ and the Germans’ data (0.008% and 0.009%, respectively). Well with an unclear function or a function without repetition in the corpus, however, occurred more than twice as often in the German group as in the native group (0.02% vs. 0.009%).
.. Summary of the functions of well In this section, I have described the functions of well in four groups: non-discourse marker functions, discourse marker functions at a textual level, discourse marker functions at an interactional level which have to do with responses, and discourse marker functions at an interactional level which have to do with the expression of an opinion or evaluation. In the non-discourse marker group, I illustrated well being used as an adverb derived from the adjective good (e.g. in expressions such as he played well), and as part of the construction as well (as), meaning ‘in addition’. As a discourse marker, well has a number of textual functions; it gives indications of textual problems or regarding the structure of the text. Most often it is used in my data when the speaker has difficulties expressing him/herself and is searching for the right phrase. Well is also employed when the speaker wants to correct an expression and substitute it with a more appropriate one. It structures narratives (and to a lesser extent also discussions) by marking quotations, leading (back) to the main story or to the instructions, marking the transition between scenes, and concluding a topic or description of a scene. At the interactional level, well can mark replies which do not directly answer the question asked, delayed answers, and partial and/or insufficient answers. However, direct answers can also begin with well, as my data showed. A third interactional function which concerns answers is well marking the response to expectations which the speaker him/herself raised, either by cutting off an utterance after (beginning an explanation
Chapter 3. Well
with) (be)cause (see extract (83)) or by asking questions such as what happened next. Other interactional functions of well have to do with the expression of an opinion or an evaluation: speakers can begin their contribution of an opinion with well, and they also use it in my data to indicate that they want to continue with their contribution (to the same topic). Furthermore, well serves to introduce an evaluation of a previous statement which might have been uttered either by the speaker or by the conversational partner.
. Quantitative results for well When we compare well as a linguistic item, we find that the German speakers on average use the discourse marker (dm) well more than twice as frequently as the American speakers (see Figure 3.1a). For non-discourse marker (non-dm) well, this differences increases considerably; the Germans use it five times more than their American counterparts. Unclassified instances occur also more often in the Germans’ speech. Of these three functional differences in averages, however, only non-discourse marker well achieves significance in the Median test. Both functions in this group, well as an adverb and well as ‘in addition’, are employed more frequently by the non-native speakers in my data. With 0.068 instances per 100 words (compared to 0.026 instances of adverbial well), the non-native speakers particularly seem to favor well in the construction as well (as); in the native speaker data we find only 0.006 instances per 100 words. Within the discourse marker group, the differences between the two speaker groups varies greatly. Figure 3.1b illustrates the relationship between the frequencies of the individual discourse marker functions of well in the native speaker group with 0.35 Germans Americans
avg. % of total words per group
0.30 0.25 0.20 0.15 0.10 0.05 0.00 non-dm
discourse marker
unclassified
Figure 3.1a Distribution of the linguistic item well – American versus German speakers
Discourse markers in English discourse 0.14 Germans Americans
avg. % of total words per group
0.12 0.10 0.08 0.06 0.04 0.02
EPS
CTA*
COO
RSE
DIA
IDA*
CCL*
NSC
MTS
QUO
RCP
SRP*
0.00
Figure 3.1b Distribution of the discourse marker functions of well – American versus German speakers (an asterisk indicates a significant difference between the speaker groups)
those in the German speaker group. Since the full labels of these functions would be too long for the graphical representation in this figure, Table 3.2 shows their corresponding abbreviations which are used in the figure. The first six perform their task at the textual level, the other six at the interactional level. In Figure 3.1b, functions for which the difference between native and non-native speakers achieved statistical significance in the Median test are marked with an asterisk. This is the case for four functions, two each at the textual and at the interactional level: The Germans searched for words or expressions using well to mark this search Table 3.2 Functions of well – full labels and their abbreviations Full label
Abbreviated label
Searching for the right phrase Rephrasing/correcting Quotative well Move to the main story Introducing the next scene Conclusive well Indirect answer Direct answer Response to self-raised expectations Contributing an opinion Continuing an opinion/answer Evaluating a previous statement
SRP RCP QUO MTS NSC CCL IDA DIA RSE COO CTA EPS
Chapter 3. Well
almost nine times more often than the native speakers. This was by far the most frequent function of well among the Germans. Thirty-three Germans employed it, twelve of them once, the others between two and seventeen times. One might ask at this point whether these Germans’ linguistic competence in English is particularly low. However, language proficiency was not an issue in the current study, so that this question cannot be answered here. With an average of 0.029 instances per 100 words, indirect answers were also more frequently prefaced by well in the non-native speaker group; the American speakers used it 0.0057 times per 100 words. In addition, the Germans employed well for two functions which did not occur in the native speaker data: to conclude a description or topic (textual level) and to continue expressing their opinion (interactional level). These instances came from a number of students; only two used conclusive well three and four times, respectively. All other users employed well for these two functions only once or twice. Overall, slightly more than two thirds of the speakers in each speaker group used discourse marker well.
.. Non-linguistic factors ... Gender Well as a discourse marker and in its non-discourse marker functions seems to be fairly neutral in terms of the sex of its user. The statistical analysis yielded only two functions for which the difference between male and female participants was significant. Among the American speakers, this was well at the beginning of a response to self-posed expectations (RSE). This was the only function which the American women did not use at all, while their male counterparts as a group used all functions of well (except for the two found only in the non-native data). Two of the ten men used the function under discussion, one once, the other twice. In my opinion, the low number of occurrences does not allow generalizations about male and female behavior in general, despite the statistical significance. In the non-native group, the men employed well to mark a direct answer on average four times as often as the women. This corresponds to five men out of nineteen who used it once, and three of the 58 women who also used it once. If we additionally consider the sex of the group or participant pair, we find that the American student who used well in the response to self-raised expectations twice was a single participant; the other student had a female partner. For the Germans, this additional factor is more interesting: All instances of well indicating direct answers came from the same-sex participant pairs, eleven of which were all-male and forty-five all-female (see Figure 3.2a). None occurred in the fifteen mixed pairs, and none in the data of the six single participants. The last statement may seem unnecessary, but the single participants as much as the participants with a partner answered to the questions on the instruction sheet and four times used well to begin an indirect answer. It is not clear whether the differences found are due to gender or perhaps to culture. A study done by Deborah Tannen (1994: 186ff.) suggests that American women
Discourse markers in English discourse 0.035 male female
avg. % of total words per group
0.030 0.025 0.020 0.015 0.010 0.005 0.000 DIA
unclassified
Figure 3.2a Distribution of some functions of well according to gender (German speakers) 0.16
avg. % of total words per group
0.14 all male all female mixed
0.12 0.10 0.08 0.06 0.04 0.02 0.00 DIA
non-dm
COO
RCP
EPS
Figure 3.2b Distribution of some functions of well according to the factor “sex of the group” (German speakers)
tend to be more indirect than men. Other authors found that Germans are more direct in conversation than English people (for example House and Kasper 1981; Edmondson and House 1991; cf. Clyne 1999). The question for the Germans would then be whether German men are indeed more direct, or whether they more often feel the necessity to hedge direct answers with well.
Chapter 3. Well
As can be seen in Figure 3.2b, the function ‘direct answers’ was not the only one which showed significant differences among the Germans regarding the sex of the group. The all-male pairs contrast with the all-female pairs also in the function ‘contributing an opinion’ and with the mixed pairs in the function ‘evaluating a previous statement’. At the same time, the all-female pairs contrast with the mixed pairs in the textual function ‘rephrasing and correcting a phrase’. Non-discourse marker functions in general are used least frequently by the all-male groups and most often by the mixed groups.
... Age In a comparison of the two American and three German age groups (see Sections 1.7.1.2 and 1.8.3), no statistically significant differences between Americans aged younger than 22 and those aged older than 24 could be found. For the Germans, in contrast, a number of differences were significant, which are illustrated in Figure 3.3. The youngest speakers, for example, used non-discourse marker well the least frequently, while those aged 22 to 24 used it most frequently, both with a significant difference from the oldest age group. In contrast, the youngest speakers indicated searching for the right phrase with well significantly more often than the speakers aged 22 to 24. This latter group employed the textual functions of well in general the least frequently. For the textual level and three of the textual functions, this difference was significant between this group (age 22–24) and the oldest group: searching for the right phrase (SRP), rephrasing/correcting (RCP), and the move to story function (MTS). The same difference applies to evaluative well (EPS), one of the interactional 0.35
avg. % of total words per group
0.30 age 18–21 age 22–24 age 25+
0.25 0.20 0.15 0.10 0.05
IDA
CTA
EPS
MTS
RCP
textual
SRP
non-dm
0.00
Figure 3.3 Distribution of functions of well according to age (German speakers)
Discourse markers in English discourse
functions. At this level, it is the youngest speakers who use well for continuing an opinion/answer more than the middle age group and well as a marker of an indirect answer more than the oldest speakers.
... Relationship between the partners The relationship between the partners, i.e. whether they were friends or strangers, did not seem to play any role in the frequency of well, neither in the native nor in the non-native group of speakers. ... Role In a pairwise comparison of speakers in role A (saw first half of the movie), role B (saw the whole movie), and role C (without a partner), there were no significant differences found among the native speakers. In the non-native speaker group, quotative well was less frequent among speakers in role A than among speakers in either role B or role C (see Figure 3.4). This should not be surprising: While there are some scenes of verbal interaction in the first part of the movie, much of this first part shows actions. The second part of the movie, in contrast, is full of verbal interaction and thus gives the narrator ample opportunity to report direct speech. Speakers in role C made much more frequent use of conclusive well than either of the other two speaker groups; the six of them used it seven times.
avg. % of total words per group
0.10
0.08
role A role B role C
0.06
0.04
0.02
0.00 QUO
CCL
Figure 3.4 Distribution of two functions of well according to speaker roles (German speakers)
Chapter 3. Well
.. Linguistic factors Tests for a variety of linguistic factors did not yield much more insight. Apparently, it did not have any influence on the German speakers’ frequency of well whether they learned English only in a formal environment, i.e. in school or in a language institute, or whether they also learned it in an informal environment, i.e. through friends or family members. This was one of the factors tested. Another concerned the speakers’ use of English as primary means of communication in informal contexts (conversations with older family members, family members of the same age, and friends). The results are displayed in Figures 3.5a and 3.5b. A significant difference was found for the function ‘introducing the next scene’ between those participants who never used English in the informal contexts listed above and those who used it at least sometimes. The latter group used well in this function only 0.007 times per 100 words, even less than the American native speakers, while the former group (who never used English informally) employed it more than three times as much (0.024 tokens per 100 words). Interestingly, the frequency of adverbial well seems to increase with the amount of English spoken in informal contexts. The group of students which used English more than just occasionally with friends and family employed adverbial well significantly more than the rest of the Germans and more than the Americans.
avg. % of total words per group
0.030 0.025 0.020 0.015 0.010 0.005 0.000
no
lish ng lE a orm inf
e som
lish ng lE a orm inf
er Am
s ker pea s ve ati nn a c i
Figure 3.5a Distribution of “well introducing the next scene” – German speakers according to the use of English in informal contexts versus American speakers
Discourse markers in English discourse
avg. % of total words per group
0.05
0.04
0.03
0.02
0.01
0.00
for e in l t t li
lish ng E l ma
or inf re o m
lish ng E l ma
s ker pea s ive nat n a c eri Am
Figure 3.5b Distribution of adverbial well – German speakers according to the use of English in informal contexts versus American speakers
Apart from the classroom, students frequently acquire additional competence in English through high school exchange programs or semesters abroad. Such times spent in an English-speaking country give the opportunity and even require the student to practice and hear English outside the classroom, in informal contexts, and therefore are a factor to be considered in this study. For well, however, time abroad played a role only for unclassified instances. Their number was considerably lower in the data of the students who had been abroad (0.011 per 100 words) than in the data of those who never spent time in an English-speaking country (0.068 instances per 100 words). In comparison, the native speakers used unclassified well 0.017 times in 100 words (see Figure 3.6). Biber et al. (1999: 1095) found that well is slightly more common in American English than in British English. Due to the low number of British speakers in the GLBCC, I cannot provide a direct comparison between the two varieties in my data. What I can do, however, is to compare German students who spent time in Great Britain with those who spent time in the USA. It turned out that they differed only with regard to one function of well, to reformulate and correct phrases. The frequency in the US-influenced group for this function was 0.002 per 100 words, while the British-influenced group used it 0.018 times. This is particularly interesting, since the American native speakers themselves used it 0.013 times (also shown in Figure 3.6); thus, the native frequency is closer to that of the British-influenced speakers than to that of the US-influenced speakers.
Chapter 3. Well
avg. % of total words per group
0.08
not abroad abroad in GB abroad in US American native speakers
0.06
0.04
0.02
0.00 unclassified
RCP
Figure 3.6 Distribution of some functions of well according to time abroad (German speakers) and comparison to American speakers
avg. % of total words per group
0.05
0.04
no contact contact American native speakers
0.03
0.02
0.01
0.00 DIA
unclassified
Figure 3.7 Distribution of some functions of well according to native speaker contact
A final test combined the above-mentioned factors into one which I called ‘native speaker contact’. That is, every participant who spent time abroad, used English with friends or family, or learned English in informal contexts was counted as having native speaker contact. The result of this test (given in Figure 3.7) basically confirmed the previous results:
Discourse markers in English discourse
The German participants with native speaker contact showed fewer instances of unclassified well than those without native speaker contact. In addition, the difference between the two groups was also significant for well initiating direct answers. Like unclassified well, this function was less frequent in the group with native speaker contact, while the ‘no native speaker contact’ group used it about four times as often.
.. Summary of the quantitative results for well Overall, the Germans used well much more often on average than the American speakers in my data, in its discourse marker functions and also in non-discourse marker functions, even though this difference is minor for unclassified well (see Figure 3.1a). Figure 3.1b, however, illustrates that the picture becomes more diversified if we look at the individual functions of well. Only four functions, two at the textual level and two at the interaction level, show a statistically significant difference between the two speaker groups. Analyses of the potential relationships between the quantitative results and a number of linguistic and non-linguistic variables result in some significant differences but show few unified patterns. Of the non-linguistic variables, the different age groups of speakers exhibit significant differences for several functions, with the oldest group using most of the textual functions more than the age group in the middle. The role of the speakers in the experiment is important only for quotative and conclusive well, while their gender influences the frequency of well to introduce direct answers. The sex of the group / participant pair is relevant for this latter function as well as for some others, though without a clear pattern. Besides unclassified instances of well, its functions as an adverb, to introduce the next scene, to rephrase or correct, and to introduce a direct answer are influenced by linguistic variables. Adverbial well increases with a more extended use of informal English, while some informal English use, time abroad and contact with native speakers decrease the frequency of the discourse marker functions and unclassified well. It seems that German speakers learn, through (informal) interaction with other (native) speakers, to reduce the non-native-like frequency of at least some functions of well. Comparisons with the American speakers, however, show that this learning process yields a slightly excessive effect: the Germans with more contact and informal English use the discourse marker functions mentioned above and unclassified instances even less frequently than the American students.
Chapter 4
You know
. Functions of you know in the literature It seems that the discourse marker you know is one of the most versatile and notoriously difficult to describe. In the research literature on you know, we find almost thirty functions accorded to it; more than half of them are supported by at least two authors. I will describe the most important of these below. In my data, I could distinguish only twelve distinctive functional patterns, summarized into ten categories, which I will explain and illustrate in the subsequent sections. The Concise Oxford English Dictionary (COED) in the 1976 edition tells us that you know is a colloquial expression reminding the hearer “that he knows or should know a thing” or serving “as a mere gap-filler in conversation”. The Oxford Advanced Learner’s Dictionary (OALD) of 1989 puts it very similarly. Apart from serving as a reminder, you know is “used as an almost meaningless expression when the speaker is thinking what to say next”. Even though researchers in the area usually claim that discourse markers do have a function and a pragmatic meaning, we find the idea of you know as a filler more than once. Östman (1981: 27f.) mentions that there is an idiolectal sub-function of you know when it is used as a speech habit, Crystal (1988: 47) claims that some people use it “to avoid the effort of having to think”, and Erman (1987: 175, 2001: 1344) finds that it may serve as a staller for time. Both, however, emphasize that these instances make up only a minor percentage of all occurrences of you know. Östman’s (1981) early study of you know deals with it in a politeness framework. In contrast to the two dictionaries mentioned above, he maintains that “the fact that the speaker uses you know is as such evidence that it is not at all clear that the ‘You’ knows” (1981: 17). Rather, the speaker wants the hearer to cooperate and accept the content of the speaker’s utterance, without arguing. Since the hearer’s opposition would be a threat to the speaker’s face, you know thus works as a face-saving device (1981: 17, 22). However, there is another way in which you know fulfills politeness functions. By using it, the speaker pretends shared knowledge, and this achieves intimacy. To illustrate this, Östman gives the example of the professor who, after an hour of teaching, starts to repeatedly use you know to indicate that he turns to a more informal style and promotes a (momentarily) more egalitarian relationship. After ten minutes, the students do the same, showing that they accept this style. Furthermore, “the use of you know implies that the speaker wants to give the addressee the feeling of greater power” and thus expresses deference (1981: 19).
Discourse markers in English discourse
Apart from its politeness functions, you know also serves some other functions, at the interactional as well as at the textual level. When it is accompanied by an interrogative intonation contour, for example, it typically requires a response. In contrast, when it is spoken with falling intonation, it conveys “I’ll say no more”, according to Östman (1981: 26f.). In both cases, it yields the floor to the hearer and thus assumes a function in turn-taking. It also works towards structuring a narrative; often, it introduces background information or digressions (1981: 19, 26). In connection with pauses, it may serve to indicate lexical search or content search, with the latter being “indicative of hedginess and indirectness” because the speaker’s knowledge is fuzzy (1981: 30). According to Schourup (1985), you know presumes some kind of common ground between hearer and speaker. At the same time, it expresses uncertainty about this common ground. It may, for example, question whether the speaker’s words have been understood, or ask for an acknowledgement that the hearer is familiar with what the speaker talks about or can appreciate it. If the speaker says, “I feel a chill, you know?”, s/he expects that the hearer either also feels a chill or at least knows that feeling and thus can appreciate what the speaker said (1985: 103f.). This kind of appreciation is particularly required when you know does not refer to the content of the utterance itself but to the implications of that content. In Schourup’s example (1985: 105), the speaker says that a streetlight goes on, adding It’s six o’clock, you know as an explanation for why the streetlight goes on. Since s/he does not make the relationship explicit, you know serves to ensure that the hearer has caught the significance of the explanation. Schourup also cites several (unpublished) works by Goldberg (1976, 1980, 1981) and agrees with her that you know may occur within a sentence structure in “positions of hesitation” (Schourup 1985: 106) and in repairs, even though he does not consider you know in such places to be basically different from you know in other slots (1985: 106, 124). He also agrees with Goldberg that you know often occurs at the beginning of a new topic, but again, his interpretation is different. In his opinion, you know in pre-topical position predicts common ground. However, it does not refer to common ground concerning a certain proposition or its implications, but to a general understanding between the conversational partners that will extend to the topic being introduced (1985: 109). Thus, you know conveys intimacy. Like Östman, Schourup includes sentence-final you know, though only with interrogative intonation, which invites some kind of response. Typically, this response is a (verbal or non-verbal) token of understanding or agreement. However, Schourup does not consider you know as a facilitator of turn-taking, for two reasons. Firstly, you know in itself does not yield the floor; this function is dependent on you know occurring “at a point of possible utterance termination” (1985: 130). Secondly, half of the instances of interrogative you know in his data received backchannel responses (for example mhm, uh huh, yeah), and were not followed by “a full change of turn” (1985: 135). In Holmes’ (1986) approach, you know fulfills two apparently opposite broad functions: expressing speaker confidence or certainty and expressing uncertainty of various kinds. Each of them has several sub-functions. On the certainty side, you know may be used to “introduce what s/he regards as incontestable mutual knowledge” or
Chapter 4. You know
conjoint knowledge (1986: 8). Even though Östman (1981) found that the use of you know does not necessarily mean that the hearer knows, and even though Biber et al. (1999: 1077) claim that you know sometimes imparts new information, it can also be the case that the information introduced by you know is indeed known to the hearer (cf. Jucker and Smith 1998: 193). This information may either be knowledge specifically shared between speaker and hearer (cf. Stubbe and Holmes 1995: 69), or general knowledge (cf. Schiffrin 1987: 274). More in the vein of Schourup’s ‘chill’ example (see above) is the attributive use of you know. According to Holmes, it can be paraphrased as “I’m confident you know the kind of thing I mean” (1986: 9). By assuming (and letting the hearer know about it) that the hearer has the relevant kind of background, the speaker expresses positive politeness. Also as an expression of certainty, you know can be used emphatically. Holmes paraphrases this function with “let me assure you”. When you know is used in this manner, there is no assumption that the hearer knows the information contained in the utterance which you know asserts. When you know expresses uncertainty, it has two main functions, which are further divided into sub-functions. In certain contexts, such as accounting an embarrassing experience, sharing personal information or making a critical comment, you know serves as an “appeal for reassurance” (1986: 10). In the first two contexts, the speaker is in a vulnerable position and asks for sympathy from the hearer, to be assured that s/he will not take advantage of that vulnerable position. In the third context (critical comment), you know serves to soften the force of the utterance (cf. also Crystal 1988: 47) and makes the speaker appear less blunt or bold. The second function of you know is the marking of linguistic imprecision. The linguistic imprecision relevant here takes on three forms, according to Holmes. When a speaker is uncertain about the choice of a word or phrase, you know signals lexical imprecision; when the speaker feels the need for further information to clarify some point or provide more precision, you know introduces qualifying information; when a speaker stops in the middle of a sentence and continues with a different syntactic structure, you know indicates a false start, which Holmes also counts as linguistic imprecision (1986: 10f.). All three sub-functions have also been included by other authors. Thus, both Östman (1981) and Erman (1987) mentioned you know in content and lexical search. The idea of you know indicating false start and repair appears in Erman (1987 and 2001). She also describes you know as introducing exemplification and clarification (1987), which are both subsumed under Holmes’ qualification. Erman published her PhD thesis (1987) on the pragmatic expressions, as she calls them: you know, you see and I mean. For Erman (1987 and also 2001), you know does have a turn-yielding and a turn-taking function, but she does not consider them the main functions of you know. Her data showed relatively low figures for you know in final position; rather, in the majority of her examples, pragmatic expressions (including you know) occur turn-medially. You know can also serve as a topic-shifter. More generally, Erman divides the functions of pragmatic expressions (PEs) into PEs as connective elements, as intrusive elements, and in incomplete structures (i.e. in repairs). For con-
Discourse markers in English discourse
nective elements, Erman distinguishes two levels at which pragmatic expressions may function: a micro-level and a macro-level. Most of the time, though not always, the levels correspond to whether the pragmatic expressions occurred between or within a macrosyntagm, which “typically consists of a main clause with subordinated clauses (if any)” (1987: 77). You know is found to function with equal frequency at both levels. Erman conveniently sums up the functions of you know at the micro- and macro-level as follows: 5.4.1 You know at the Micro-level When you know occurred within the sentence, it was most often used in order to: 1. introduce the consequence of or the reason for a fact presented in the previous discourse [...] 2. connect the theme and the rheme in the thematic structure [...] 3. introduce an exemplification of some part of a previous statement [...] 4. introduce a clarification of some part of a previous statement [...] 5.4.2 You know at the Macro-level In this context you know was used in order to: 1. introduce background information [...], which was sometimes in the form of a parenthetic comment [...] 2. mark the boundary between two modes of discourse [...] 3. terminate an argument in descriptive discourse [...] (Erman 1987: 114f.) The first two functions at the micro-level and the last of the macro-level remain solitary in that neither Erman herself in later publications nor other researchers support them. In contrast, the other four functions found more support. For example, Crystal (1988: 47) mentions that you know can be used to clarify or amplify meaning. The Cobuild dictionary (1987) states that it is used “when you are trying to explain more clearly what you mean”, and House (1996) terms you know a ‘Clarifier-Cajoler’. Furthermore, I have found ‘exemplification’ and ‘clarification’ to be useful categories for my own data. That you know is used to introduce background information is acknowledged also, as we have seen, by Östman (1981) and Schourup (1985) and repeated in Erman’s later publications (1992, 2001). The function of you know to “mark the boundary between two modes of discourse” (1987: 115) unfortunately is not further explained here, but judging from her 2001 publication, it is meant to refer to the boundary between direct and reported speech (2001: 1342). This function also occurred in the data of Schiffrin (1987) and Redeker (1991), as well as in my own data. As intrusive elements, pragmatic expressions occur “within the boundaries of the clause” (Erman 1987: 121). In this position, you know may serve as a hesitation marker and indicate that the speaker is searching for words either at a lexical level or at the content level, as Östman also noted (Östman 1981, see above). According to Erman, lexical hesitation is frequently marked by pauses and other pragmatic expressions in
Chapter 4. You know
addition to you know, while you know stands alone when the speaker searches for a content word (1987: 136f.). Within a constituent, you know mainly serves “as a staller for time and as an appeal to the listener to have patience” (1987: 137). The latter idea (appeal to have patience) is also supported by Holmes (1986: 10). Pragmatic expressions also occur in incomplete structures, which Erman describes as “structures where there is a break in the syntactic structure” (1987: 143) due to repair-work or interruption. The pragmatic expression is regarded as a marker of repair only if it occurs at the break itself. The function of you know again lies mainly in “stalling for time in connection with word search” (1987: 163). In Erman’s data, it frequently resulted in a lexical shift; the speaker either stopped in mid-sentence and then repeated part of the sentence prior to you know, or s/he chose another lexical expression after you know (for instance, another verb, as in he borrowed them from the- oh you know chose them from the porn, Erman 1987: 160, example (165) simplified). Such instances will also play a role in my own categorization system. Erman investigates the usage of you know in two further papers, published in 1992 and 2001. The first deals with pragmatic expressions in male and female interactions, while the second compares you know in adult versus adolescent usage. Both publications include several of the functions mentioned in Erman (1987), but they treat you know within different models. Erman (1987) classifies pragmatic expressions according to both functional and syntactic properties. There seems to be no explicit model which accommodates all functions. Erman (1992), in contrast, places the functions of pragmatic expressions along a lexical-pragmatic continuum (1992: 220f.). This continuum is divided into a textual level (towards the lexical end) and an interactional level (towards the pragmatic end). Both contain three functional areas: At the textual level, these are the decoding of information, orientation in the discourse, and the regulation of turns (conversation structure); at the interactional level we find hesitation, repair, and appeal. In the course of the paper, Erman mentions specific functions of you know within this framework. Thus, in the information structure, you know is used to signal to the hearer that s/he should “interpret part of the information as common ground” (1992: 222). You know also structures the discourse; it is frequently used to insert a parenthetic comment in the middle of an utterance, or to shift between indirect and direct speech. Her extract on the latter issue (1992: 223) is meant to illustrate the shift from direct to indirect speech (or, more exactly, from a quotation of another speaker’s words back to the current speaker’s own words). However, this example could as easily be interpreted as an interactional appeal for understanding and agreement. When you know occurs at turn transitions, it frequently does not just regulate turns but has other functions as well, according to Erman. At the interactional level, you know marks hesitation and repair/appeal. Erman sees the difference between repair and appeal in that the speaker comes up with a linguistically different expression or syntax in case of the repair, whereas when “the speaker simply giving up in mid-utterance” (1992: 226), you know is used to signal an appeal.
Discourse markers in English discourse
In terms of gender, Erman finds that pragmatic expressions in general “are much more frequent in same-sex as opposed to mixed-sex interaction (Erman 1992: 228). Unfortunately, she does not give separate data for the three expressions under investigation on this issue; in the remaining quantitative analysis, the three pragmatic expressions are kept separate, but without regard to the sex-combination of the group. It turned out that women use you know more often in discourse organization and for hesitation, while men use it more often for information decoding, turn regulation and for marking repair/appeal. At a more general level, Erman observes that men tend to use you know for rhetorical purposes and between incomplete propositions; women, in contrast, employed you know more often for turn-holding purposes and between complete propositions. Her results, however, present only percentages without indicating whether there are any statically significant differences. The same is true for her 2001 publication (see below). Erman (2001) views pragmatic markers as “monitors” in three domains: textual/ideational, social/interpersonal, and meta-linguistic. The sub-functions of you know within these domains, however, are mostly the same as in the previous works. New is the (explicitly expressed) idea that the most salient function of pragmatic markers “as discourse markers” (2001: 1340) is to create coherence. In order to help the addressee decode the message, discourse markers highlight certain elements in the discourse (cf. also Cobuild 1987). In contrast to Erman (1992), Erman (2001) counts the marking of hesitation and repair as belonging to the textual rather than the pragmatic/interactional level (see above). Apart from repair (which involves breaking off in mid-structure), markers may also be used to correct information content, for example in the form of a specification. In the teenage language corpus she investigated (the Corpus of London Teenage Language, COLT), the most frequent textual function is “linking propositions in narrative and descriptive discourse” (2001: 1343). As social monitors, pragmatic markers have clearly interactional functions. They involve the audience by asking for confirmation of understanding or of a previous claim, and asking for agreement with a statement or reference. Thus, they can have a comprehension-securing function. Turn switches also belong into this category, both as turn-taking and turn-yielding. Erman claims that adolescents tend to use you know more often for turn-taking, while adults tend to use it more for turn-yielding. The general function of pragmatic markers as a meta-linguistic monitor is a modal one, according to Erman. She calls this function “exclusively speaker-oriented” and describes it as follows: [I]n the present study markers with a modal function are assumed to be directed towards the speaker’s subjective appreciation of the illocutionary force of the utterance as a whole. Their most characteristic function is as emphasisers typically underscoring and rounding off an evaluative utterance. (Erman 2001: 1341)
The emphasizing function turned out to be relatively common in the COLT data, but absent in two adult corpora Erman refers to, the London-Lund Corpus and the data recorded by Schiffrin in the 70s (Schiffrin 1987). In addition, pragmatic markers in
Chapter 4. You know
the meta-linguistic domain can function as hedges and approximators. Erman lists you know as a candidate for both functions, but in her corpora, you know is not used as a hedge. As an approximator, you know is used only once in the adolescent data (“She said you’re, you’re nice, you’re pretty, you know whatever”, Erman 2001: 1348). However, as Erman says, “approximators can also be used with an appealing function, the speaker appealing to shared knowledge of the world, general truths, or otherwise ‘uncontroversial’ issues” (Erman 2001: 1348), which is particularly true for you know. Despite this double function, you know as an approximator/appealer is also scarce in the London-Lund Corpus. Schiffrin’s (1987) treatment of you know is detailed, but not unproblematic. As she states right at the beginning of the relevant chapter in her book, in her opinion you know and I mean are markers “whose literal meanings directly influence their discourse use” (1987: 267). In principle, there is nothing to say against this opinion. As we have seen in the discussion so far, many of the functions of you know indeed have to do with whether or not the hearer knows. However, this opinion is very unfortunately reflected throughout the chapter by a lacking distinction between discourse marker uses and non-discourse marker uses. The examples she presents often contain utterances in which you know is one of the sentence constituents and not syntactically optional. Thus, she includes extracts such as “y’know what Hasidic is?” (1987: 269),“Y’know these tea rooms?” (1987: 274), and “You know why?” (1987: 280) and mixes them with other extracts such as “[My job is to] See that the kids behave a little bit. Y’know.” (1987: 273) or “we’re all not perfect, y’know.” (1987: 276). In the first type of examples, the auxiliary “do” has been omitted and could be re-introduced without changing the grammaticality or meaning of the utterance. In the second type, this would render the utterance at best extremely awkward. Therefore, Redeker (1991: 1166) rightly criticizes Schiffrin’s approach on the grounds that “(Do) you know questions are not discourse markers but rather a particular type of conversational move”. Biber et al. (1999: 1077) comment on this problem: “Whereas in some cases the grammatical function of you know is unclear, in the majority of cases where it is initial and is not followed by something which definitely has to be a complement, it takes on the role of a discourse marker”. In contrast, Schiffrin deliberately deals with ‘(do) you know’ questions which take a variety of complements (1987: 288). Since one of my conditions for considering a linguistic item a discourse marker is syntactic optionality, I fully agree with Redeker and Biber et al. Therefore, most of Schiffrin’s conclusions about the functions of you know are not relevant for you know as a discourse marker in the present study. Nevertheless, Schiffrin does include some examples in which you know can easily be seen as a discourse marker (see above) and draws conclusions from them. Probably the most literal meaning of you know as a discourse marker can be found in cases in which it is used (utterance-final) as a reminder of shared knowledge (1987: 273, her extract (6)). This is a function which also occurs in my data. In her extract (7) on the same page, the speaker answers a question and thus brings about shared knowledge. Both are interpreted by Schiffrin as example of you know eliciting a display of information reception by the hearer. She also claims that you know “prefaces back-
Discourse markers in English discourse
ground explanatory clauses in narratives” to mark it as “to-be-shared information” (1987: 274). In several further examples (Schiffrin 1987: 276f.), you know marks statements expressing consensual truths and requests the hearer’s endorsement of these truths. Two of the examples, in which you know follows the statement, resemble Holmes’ (1986) emphatic use of you know. In the other three extracts, you know precedes the statement. In arguments, you know uttered after presenting a reason or a statement supporting the speaker’s position can serve to check on the progress of the speaker’s effort to win the hearer over to his/her side (1987: 279ff.). Three examples are given in which you know occurs at the beginning of direct reported speech (cf. Erman 1987, 2001), but Schiffrin does not see its primary function in marking quoted material; rather, she explains what other functions you know (also) has in the examples. Apart from the reference to shared knowledge, you know “in narrative occurs with internally evaluative events and with external evaluation” (1987: 284). External evaluation consists of meta-linguistic clauses through which the speaker comments on the importance or relevance of the current piece of narrative; internal evaluation is achieved within the narrative frame by presenting a particular part of the narrative as complicating action, for example. As I argued in chapter three referring to Norrick’s (2001) account of well in oral narratives, the narratives in the GLBCC are “ordered” narratives and not narratives occurring in a free conversation. They were not told to convey the speaker’s emotions or to support a point in an argument and therefore, in general, lack certain characteristics which we find in free narratives. Accordingly, we may not find the same functions of you know in the Chaplin narratives as Schiffrin did in her data. Watts (1989), who studied the attitude of speakers (in this case, of his own family) towards discourse markers, criticizes Schiffrin’s analysis of you know harshly: “Schiffrin has not dealt with some of the commonest occurrences of you know, and those she has omitted are precisely those which are perceptually salient” (1989: 221). According to Watts, you know is salient “when used to signal a transition in the information state which is either not achieved or achieved ineffectually” (1989: 221). Another salient use occurs when you know is not used to refer to common knowledge but rather to seek an acknowledgement of the hearer’s understanding, or, more generally, as a right-hand bracket. Then it can be interpreted as an evaluative cue. Watts also found that you know can be used to focus the hearer’s attention to upcoming new information (cf. also Erman 2001). Stubbe and Holmes (1995) explicitly focus on the social variation of you know and other “exasperating expressions” in a section of the Wellington Corpus of Spoken New Zealand English. Not much space is devoted to a detailed description of the functions of you know, in comparison to other papers; Stubbe and Holmes basically follow Holmes (1986). Thus, you know is taken to be an addressee-oriented device which may express both certainty and uncertainty: certainty when the speaker is confident that the hearer shares the relevant knowledge or reassures the hearer that the proposition
Chapter 4. You know
in question is valid, and uncertainty when the speaker is unsure about the hearer’s attitude or the linguistic precision of his/her own utterance (1995: 69). In terms of social variation, Stubbe and Holmes found that age and gender do not have any significant effect on the frequency of you know, at least not in the data of middle class speakers. Social class, in contrast, has a definite impact particularly in the group of young speakers; young male working class speakers use you know much more than their female counterparts. You know is generally more frequent in the working class; however, Stubbe and Holmes claim that “its relative frequency in the speech of the middle-class informants means it cannot be characterized as a vernacular feature in the same way as eh” (1995: 81). It should be mentioned here, though, that the samples of speech are not distributed equally across the variables checked, since there are no data from working class mid-age females. This lack somewhat limits the range of conclusions that can be drawn. He and Lindsey’s (1998) data consist of eleven academic counseling encounters, which are analyzed with regard to you know and its positions within the intonation unit. They found that almost eighty percent of all instances of you know occurred at the beginning of the intonation unit, eight percent within and thirteen percent at the end. You know forming an intonation unit of its own, however, is briefly mentioned but neglected in the quantitative analysis. In terms of the function of you know, they claim that it is used to mark information which is new, important, and salient, for the benefit of the hearer. This is, according to He and Lindsey, the main function of you know, even when it appears in the context of hesitation, truncated sentence structures, reported speech, etc. However, they do raise the question in the conclusion how they will account “for those instances of “you know” which do not belong to any of the categories of meaning or function identified in this paper” (1998: 151). Even though you know may be used for marking new information, this function does not lie at the core of what you know accomplishes, according to Jucker and Smith (1998), since it may also precede information which is known to the hearer. Jucker and Smith recorded conversations between pairs of students on a given set of topics and analyzed them with regard to a range of discourse markers including you know. Based on the relevance-theoretic framework they use, Jucker and Smith claim that “you know invites the addressee to recognize both the relevance and the implications of the utterance marked with you know” (1998: 194). This implication is often not expressed, but it is nevertheless “critical to a point being made” (1998: 196) if the utterance is marked with you know. The implication left unexpressed thus invites the hearer to complete the current stretch of discourse, at least mentally, by drawing the appropriate inferences. In this way, you know is used to exploit the common ground between the conversational partners. Fox Tree and Schrock (2002) in their paper on the “Basic meanings of you know and I mean” fully agree with Jucker and Smith’s (1998) account of you know. With reference to the functions presented by other researchers, they claim that discourse markers are not doing as much as these researchers suggest. In the light of Jucker and Smith’s account, Fox Tree and Schrock try to reinterpret the examples (and illustrated
Discourse markers in English discourse
functions) given by other researchers, even though in my opinion their reinterpretations are not very convincing. In several cases, for example, their explanation simply consists of saying that you know may co-occur with the phenomenon in question (e.g. repair, change of turn, etc.) but does not mark the phenomenon. Fox Tree and Schrock admit that formulations of basic functions such as Jucker and Smith’s formulation of the function of you know may seem too general, but at the same time they claim that they can at least predict when it is unlikely to occur: whenever the speaker has no interest in the hearer drawing his or her own inferences from what has been said. This may be the case with large audiences (and the potential range of inferences), in situations where incorrect inferences must not occur, or when a person of higher hierarchical rank does not want to suggest common ground between speaker and hearer. Also interesting though very general is their alternative formulation of the functions of you know and I mean: “Another way of viewing this is that you know encourages listeners to focus more on their own thoughts, and that I mean encourages listeners to focus more on speakers’ thoughts” (2002: 744). Finally, Macaulay (2002) presents a social stratification study similar to that by Stubbe and Holmes (1995), but with data from Scottish speakers and a more quantitative than qualitative approach. Though cases in which you know forms part of the syntactic structure of the utterance are excluded, no distinction is made between the functions of you know as a discourse marker. Adding a precaution, Macaulay conveniently summarizes his results as follows: From the figures presented above, some tentative conclusions can be drawn, though their significance may not extend beyond the boundaries of Scotland. (1) The use of you know varies among individuals from similar backgrounds. Some rarely use it, if at all, while others use it quite frequently. (2) Speakers are more likely to use you know in conversations with an acquaintance than in interviews with a stranger. (3) Women are more likely to use you know than men. (4) Adolescents have not yet developed the use of you know as a characteristic of their discourse style. (5) The use of you know is not more common in one social class than the other. [...] (6) The use of you know does not appear to be primarily based on assumptions of shared knowledge, but rather to form part of the speaker’s discourse style and the rhythmic organization of utterances, particularly when it is used at the end of an utterance. [...] (Macaulay 2002: 765) I will come back to some of these issues, such as age, gender, and relationship between the partners, in the course of this chapter. First, however, I will describe the functions of you know as used by the American and German speakers in the GLBCC.
Chapter 4. You know
Table 4.1 Discourse marker functions of you know as used by German and American speakers in the GLBCC Textual Level: – marking lexical or content search – marking false start and repair – marking approximation – introducing an explanation – quotative you know Interactional Level: – “imagine the scene” – “see the implication” – reference to shared knowledge – appeal for understanding – acknowledge that the speaker is right
This description will include references to the discussion of the relevant literature presented above.
. The functional categories of you know in this study The sequence you know occurs 324 times in my data and thus is less frequent than well. Unlike so and well, you know has only one non-discourse marker function (see below). Its discourse marker functions can be divided into two groups: five functions work at the textual level, another five functions work at the interactional level. While Table 4.1 gives an overview of these functions, the following sections will provide a detailed description of each.
.. Non-discourse marker function of you know In the literature, we find that certain cases of the lexical sequence you know are generally excluded from the analysis, often without any comment. Some, however, explicitly address the matter. Redeker (1991), for example, criticizes Schiffrin (1987) for including do you know questions. You know in such questions is not a discourse marker, in her opinion. Biber et al. (1999: 1077) take the sequence you know to be a discourse marker only if it does not have a complement. Macaulay gives a number of examples in which “the construction you know forms part of the syntax of the clause and could not be omitted” (2002: 752) and excludes these from his analysis. As I argued in Chapter 1, syntactic optionality is the most important feature in my description of discourse markers. Accordingly, I will not treat the sequence you know as a discourse marker if it is not syntactically optional. The extracts below illustrate both you know as part of a question (extract (100)) and you know taking a complement (extract (101), line 309):
Discourse markers in English discourse
(100)
192 do you know where they went. (47A, NS)
(101)
305 (H) .. a=nd I think he’s trying to get & 305a & the girl to marry him. 306 but she’s like, 307 she’s really shy=, 308 (H) but you can-309 you know that she likes him, (140B, NS)
.. Discourse marker functions of you know ... Marking lexical or content search According to the Macmillan Dictionary for Advanced Learners (2002), you know is “used while you think about what to say next”. Quite similarly, the Collins Cobuild Dictionary (1987) advises its readers that “[y]ou say ‘you know’ [...] to fill a gap in a conversation, for example when you are uncertain about what you are saying or what you are going to say next”. “Filling the gap” apparently is taken literally, since the examples given do not show any other sign of uncertainty or hesitation such as truncated sentence structures or pauses. However, Holmes (1986: 11) finds that, in this function, you know “is generally preceded by a pause or a verbal hesitation such as erm or er”. Erman (1987: 124) noted that there may even be several hesitation markers in succession, such as filled and/or unfilled pauses (Erman 1992: 225). Pauses also play a role in Östman’s (1981) description of you know indicating search. He distinguishes between lexical search, in which case you know is being followed by a pause, and content search, in which case you know is potentially preceded by a pause. In my data, both types of search occurred, but I think it is a bold move to maintain that the two types can (always) be distinguished and that the pause is decisive for the distinction. This can be seen in some extracts from the GLBCC. Extracts (102) and (103) seem to confirm Östman’s idea. In (102), the (non-native) speaker obviously lacks the English word to express what he has in mind, which is a little plate. You know here (line 254) is followed by a short pause and marks lexical search. (102)
251 252 252a 253 254 255 256 257 258 (28B,
.. and .. he gave back him some money, and he left the tip on the .. & & on the little um-...(2.1) um yeah, you know .. where you’d put on the bill, A: yeah. B: I don’t know .. the English word, A: [yeah], B: [@@@] NNS)
Chapter 4. You know
(103)
144 ...(1.8) so they just .. you know, 145 I guess got their names checked or whatever, 146 and then they got onto Land o’ Liberty. (141C, NS)
In extract (103), we find a pause before you know. At the end of the voyage on the immigrant ship, two officials set up a table with paperwork on it; one of them sits down, while the other obviously calls up individual immigrants. It is not totally clear what the official at the table does, so that speaker 141C adds I guess. It can reasonably be argued that he is not sure about the content of what he is going to say after you know; thus, Östman would be right about suggesting content search here. However, even though these are not the only examples in my data which seem to support his claim, there are others which question it. (104)
147 148 149 149a 150 150a 151 (10B,
(H) .. Charlie Chaplin got worried, and the waiter was telling him ... don’t-you know um uh um didn’t tell him & & anything obviously. (H) but .. the waiter ... uh rejected & & the coin obviously, because it was fake, NS)
In extract (104), the speaker describes a scene in the restaurant. Chaplin pays for his food with a coin, but the waiter bites on the coin and finds out it is fake. This is where the extract sets in. In the flow of speech, the speaker starts saying the waiter was telling him ... don’t-- and then realizes that the movie is a silent one. Therefore, she hesitates and for a moment does not know what to say next. You know here is followed by three filled pauses. If filled pauses count as pauses in Östman’s sense, this should be a case of lexical search. Comparing it to extract (103), however, it seems more like content search. Extract (105) likewise raises the question of content versus lexical search: (105)
96 97 98 99 99a 100 101 102 103 103a (63A,
um ...(1.1) the .. the captain of the & & ship accuses him of being a pickpocket, (H) .. and um .. the scene-the the situation is then revealed to the & & young girl who realizes that Charlie’s & & put the money in her pocket, and (H) um (H) whether-uh I don’t think she .. you know, .. really considers whether he taken it & & in the first place, NS)
In the scene preceding the one described in this extract, Chaplin had put some money in the girl’s pocket without her noticing it, taken it out again to keep a little bit for himself, put it in again, etc., and had been observed doing this by a member of the ship’s staff. The girl, whose mother had lost money, is called to check her pockets, and
Discourse markers in English discourse
speaker 63A describes her reaction to finding money in her pocket in lines 100–103. You know is both preceded and followed by a pause. Do we then have to assume lexical or content search? I would like to argue that in this and in a number of other cases, it is not possible to draw such a distinction. I doubt that the speakers themselves would always be able to tell whether they had the content in mind and could not think of the appropriate wording or whether they were trying to think of the content of what they are going to say next. Therefore, the category ‘lexical or content search’ contains a continuum of instances with clear lexical search at one end (as in extract (102)), apparent content search at the other, and cases with elements of both in between. In any case, you know marking search operates at the textual level, focusing on lexical choices and/or propositional content. Like most of the functions of you know, the frequency of you know marking search was significantly higher (p < 0.001) in the group of the American native speakers of English than in the group of the German non-native speakers. 17 of the 34 Americans used it between one and five times per person, but only 6 of the 77 Germans (up to three times per person).
... Marking false start and repair Erman (2001) tells us that one of the functions of you know as a ‘textual monitor’ is to signal repair. According to her definition, it occurs in places “where the speaker stops in mid-structure to make a restart” (2001: 1342). In her example, the speaker continues after you know with a syntactic structure different from the one before you know. Holmes (1986: 11) found that you know can indicate a false start, or, in other words, mark “a change of syntactic direction”. The examples she gives all present extracts in which the speaker starts a sentence, breaks off with you know and continues with a different syntactic structure. Thus, Holmes and Erman apparently talk about the same phenomenon, even though they use different terms. Quite generally, both “repair” and “false start” are often taken for granted; they are adopted without any further explanation of what these terms are used for (cf. for example Schiffrin 1987; Redeker 1990; Schiffrin 1994; Core and Schubert 1999; Andersen 2001; Cameron 2001). However, there is some literature which deals explicitly with the problem. Rieger (2000), for example, reviews Jefferson’s early description of what constitutes repair, or, in Jefferson’s own terminology, ‘error correction’. This error correction takes on various forms which can basically be summarized as [WORD1 + Hesitation + WORD2] (Jefferson 1974: 186). Word1 and Word2 have to be different; ‘hesitation’ includes cutoffs of words and sentences and filled and unfilled pauses. A comprehensive typology of repairs is presented by Fox and Jasperson (1995), also reviewed by Rieger. This typology ranges from the simple repetition of a word or two within the sentence structure to cut-offs followed by a different sentence structure, as mentioned by Erman and by Holmes. In between are cases in which a (truncated or complete) word is replaced by another word; this is frequently accompanied by the repetition of one or more words surrounding the repaired one. Also possible is a repetition of a phrase to which a word
Chapter 4. You know
is added. False starts are not treated separately but are said to be part of most of the repair types. Even though Fox and Jasperson do not talk about discourse markers, their typology describes a number of cases which are parallel to those marked by you know in my data. Among them are the cases which correspond to Fox and Jasperson’s first type, i.e. repetition without any correction. However, I would prefer to call these not ‘repair’ but simply ‘false starts’, since there is nothing which is being repaired. The repetition in these cases typically spans only one or two words. (106)
488 489 490 490a (5B, NS)
he .. he was sort of dressed like-you know, like you knew he didn’t have no .. & & not too much money,
(107)
390 B: 391 A: 392 B: 393 394 395 396 (4B, NS)
... and the second part was funny. .. yeah. but I didn’t-you know, I didn’t know what to expect, I just thought they would stay on the boat, and I just be telling you about boat stuff.
Extract (106) shows a case in which just one word, like, is repeated, while (107) illustrates the repetition of more than one word (I didn’t, lines 392 and 394). In both cases, the speaker cuts off in mid-sentence (indicated by two dashes), inserts you know, and then starts again with the repetition. In the other half of the instances in the current category, you know marks a repair. Unlike Erman and Holmes, I do not include cases here where the speaker cut off in mid-sentence and continues with a different syntactic structure after you know (but see extract (111) below). Rather, I use the term ‘repair’ to refer to cases in which a truncated or completed word is substituted by another word after you know within the same syntactic structure. Extract (108) shows such a case: (108)
34 it was rocky. 35 .. rocky so they were kind of roll-36 you know rocking back and forth. (142A, NS)
The speaker describes the movement of the immigrant boat and the passengers’ difficulties remaining standing or going straight. Two of the passengers, Chaplin and a corpulent lady, do not manage to do so and roll back and forth on the floor. This explains the speakers first choice, kind of roll--. However, for some reason she decides that rocking is the more appropriate expression for what she has in mind. In extract (108), nothing has been repeated, as in half of the repair instances. The other half is illustrated by extracts (109) and (110), with repetitions of one and more words before the corrected word:
Discourse markers in English discourse
(109)
17 he has no money=, 18 no f-19 you know no job or nothin’ like that, (8B, NS)
(110)
362 when he’s eating the beans, 363 ... he is eating with this-364 .. you know he’s eatin’ with the knife, (5B, NS)
Let me add one final example which seems to contradict my usage of the term ‘repair’. The speaker in this extract starts with I’m I’m, but continues with I’ve after you know, which requires a different syntactic structure. However, she still repeats the subject of this structure, while at least some of the speakers in Erman’s and Holmes’ examples restarted with a completely different syntax, e.g. but fortunately w- you know there’s been more recent research (Holmes 1986: 11). Moreover, extract (111) is the only case of this kind. (111)
410 I’m I’m-411 you know, 412 I’ve like never seen like a silent film & 412a & [really like]-(134A, NS)
I mentioned above that Erman considers you know in repairs to function as a textual monitor (2001, cf. also 1992). Even though I use the term ‘repair’ for slightly different cases, I agree with her assessment of you know as a textual marker in this function, since it focuses on lexical expressions. Only 3 of the 77 German speakers used the function of you know to mark repair and false start (once each); in contrast, 12 of the 34 American native speakers employed it between one and three times. Not surprisingly, this difference was statistically significant with p < 0.001.
... Marking approximation The function of you know as a marker of approximation is one of the less frequent functions. Nevertheless, it occurs as a distinctive pattern. It has also been mentioned briefly by Stubbe and Holmes. They classified you know as a pragmatic device when it expressed uncertainty regarding “the linguistic precision of the message” (1995: 69). In the example they give, it is not the complete message whose precision is uncertain but rather a single word. In my own data, you know was employed with a word or phrase or even a clause which apparently did not exactly render what the speaker had in mind, or whose exactness was not important. In other words, the linguistic expression is only an approximation of what would have been the most appropriate or correct expression. In this function, you know obviously also works at the textual level. Let me give some examples:
Chapter 4. You know
(112)
393 ... (H) the main characters was & 393a & that girl and the guy. 394 A: mhm, 395 B: and don’t think the woman did much, 396 .. but .. like .. Charlie Chaplin had & 396a & all these (H) ... you know. 397 .. big movements to show like what’s & 397a & going o=n, 398 [an’] his facial expressions that was cool. (140B, NS)
In extract (112), the pair of students is discussing the movie using some questions they have been given on an instruction sheet. They did not have to answer these questions; rather, they were meant to help the participants get into a discussion. Nevertheless, like most participants, the pair in this example answered them, including the question, “What did you think of the acting by major characters?” Speaker 140B in the extract tries to describe Chaplin’s acting, which, apparently, is not easy. The expression she finally settles on is not a very precise one and presumably does not render exactly what she had in mind. Therefore, it is preceded by you know to indicate this lacking exactness. (113)
14 B: 14a 15 16 17 18 19 (2B, NS)
um .. they let Charlie Chaplin through & & the gate, but everybody else had to stay. so he walked in the next scene, (H) he is this. they flash you know .. all cold an= you know & & no money or something like that.
Extract (113) contains even two instances of you know indicating approximation (line 18). Speaker 2B here describes how the movie continues after speaker A had to leave the room. The immigrant boat, on which the first part of the movie took place, has arrived in America, and the immigrants are let off one by one. Then there is a cut, and we see Chaplin walking in a street with houses on either side. An inter-title (i.e. a black screen with words printed in white) comes up which says “Later ... broke and hungry”. Speaker 2B does not exactly remember the words of this inter-title. She renders them as all cold and no money, but since she is aware of the fact that these were not the exact words, she places you know in front of both expressions. Additionally, she marks the imprecision with or something like that. What characterizes this stretch of speech besides this imprecision is the fact that the words chosen are sufficiently exact for the purpose of the narrative. They give the impression of Chaplin feeling miserable, and they inform the hearer about the fact that Chaplin does not have any money, which is relevant for the ensuing scenes. The first of these scenes is the topic of the following extract:
Discourse markers in English discourse
(114)
11 he= .. was walking, 12 and .. he’s found this uh .. coin. 13 an’ .. must’ve been .. worth more than & 14 & you know just a coin you know, 15 but anyhow. 16 .. he= .. found some money. (10B, NS)
The speaker here tells her partner that Chaplin found a coin, but she does not leave it at that. She remembers at this point that Chaplin later tries to pay two meals and two or three cups of coffee with just this one coin, and wonders whether a single coin could have had so much purchasing power. Accordingly, she concludes that it must’ve been .. worth more than you know just a coin by present-day standards. She does not know exactly how much it was worth, nor can she spontaneously think of the highest value of a coin she knows; therefore, she contents herself with the rather imprecise expression just a coin, preceded by you know to mark the lacking exactness. (115)
127 B: [s=o] it was chocolate, 128 so ... he was like-129 I guess the waiter’s like-131 you know whatever, 132 (H) so they um-133 ...(1.3) Charlie Chapman um .. & 133a & kind of got nervous, (125B, NS)
The last extract in this sub-section, (115), describes part of the scene in which Chaplin attempts to pay – with one coin. Unfortunately, the waiter tests the coin by biting on it and finds that it bends. Speaker 125B tries to describe the waiter’s reaction; she can see that he talks to Chaplin and starts to report what he says (line 128). However, since the movie is a silent one and no inter-title is shown at this point, she can only guess what he might have said (lines 129ff.). This is marked in two ways: by using the construction BE+like in place of a quotative verb, which implies a less than literal quotation (cf. for example Blyth et al. 1990; Andersen 1998; Winter 2002), and by adding you know whatever. Similar to you know plus or something like that in extract (14), you know whatever serves to indicate both the lacking precision of the quotation and the fact that this precision is not necessary at this point; the speaker indicates that, within the story frame, it is sufficient for the hearer to know that the waiter complains about and rejects the coin as a payment. You know marking approximation or lacking exactness was used once or twice by seven American speakers and only one German speaker.
... Introducing an explanation While the functions discussed so far concentrated in various ways on the appropriateness of lexical expressions (with the exception of content search, cf. Section 4.1.2.1),
Chapter 4. You know
we are turning now to functions which help to structure the contents of what is being said. In his list of “situations which permit the use of you know”, Crystal includes the following: “In the middle of a sentence, it is often used to clarify or amplify the meaning of what one has just said. It warns the listener that the next words are particularly important. He’s just got a new BMX – you know, one of those tough little bikes...” (Crystal 1988: 47). Similarly, Erman found that when “you know occurred within the sentence, it was most often used in order to [...] introduce a clarification of some part of a previous statement” (Erman 1987: 114). The example given for this function shows a modification of things up on my walls into just small things (1987: 109). Under the heading “Introducing qualifying information”, Holmes writes: “you know precedes an optional clause constituent which introduces more specific or clarifying information” (Holmes 1986: 11). None of the terms (‘amplify’, ‘more specific’, ‘clarifying’) seems to have a technical meaning here. However, the word ‘clarifying’ implies that the hearer may misinterpret what the speaker said. There was only one instance of you know in my data where the speaker had to add clarifying information to prevent such misinterpretation: (116)
440 B: an’ he’s like441442 A: [uh huh]. 443 B: (H) he’s like. 444 he’s like. 445 they go back and forth back and forth, 446 (H) ... yeah I know actually I & 446a & wanna pay for it. 447 (H) so then he’s like. 448 I-449 you know Charlie Chaplin is like & 449a &. 450 you know he has no money. 451 A: uh huh. 452 B: so the the guy goes, (142B, NS)
Speaker 142B tells her partner about the scene in which another customer in the restaurant in which Chaplin is sitting comes over to Chaplin’s table, talks to him and his lady-love, and offers to pay Chaplin’s bill. Chaplin refuses repeatedly. Speaker B enlivens the scene in which the bill is pushed back and forth between Chaplin and the other customer by using direct speech. After a while, she realizes that her use of the pronoun he (lines 440, 443, 444, 447) is ambiguous to the extent that her partner might not be able to tell who said no no no (line 447). Therefore, she clarifies this in lines 449f.: you know Charlie Chaplin is like.
Discourse markers in English discourse
(114)
11 12 13 14 15 16 (10B,
he= .. was walking, and .. he’s found this uh .. coin. an’ .. must’ve been .. worth more & & than you know just a coin you know, but anyhow. .. he= .. found some money. NS)
If the term ‘amplification’ is taken to mean that the speaker offers additional information (for example background information, cf. Östman 1981; Erman 1987, 1992, 2001), then amplification introduced by you know is also rather rare. Extract (114), repeated here, contains such a case. As speaker 10B tells the scene in which Chaplin finds a coin, she inserts a parenthetic comment (lines 13 and 14). This comment contains additional information about the value of the coin; it is marked with you know (end of line 14). Erman also claims that you know is used to introduce exemplifications (Erman 1986: 141, 1987: 114). This also occasionally occurred in my data. In extract (117), speaker 140A tries to give reasons why he finds it difficult to determine whether the movie was great or not. One reason is that there was a lot of rolling all over the place. However, this is not the only reason. This rolling all over the place is just an example of what he has in mind, and he indicates as much in line 414 (stuff like that), after he has given up on finding words for his thoughts. Since it is meant to be an example rather than the whole reason, rolling all over the place it is introduced by you know. (117)
407 A: 408 409 B: 410 A: 411 411a 412 413 414 (140A)
it is rough .. determining whether & & it was a great or <@ not @>. [(H) I mean]-[now I mean] this-it was it was a lot of & & you know rolling all over the place, an’-...(1.1) you know, (Hx) [stuff like that so].
Rather than for clarifications, amplifications, or exemplifications, then, you know was used most of the time for marking an explanation: typically, the speaker mentions something, a concept or an idea, or gives his/her opinion, and then decides that s/he has to express it in different (and perhaps more) words to make it plain what s/he meant. In extract (118), speaker 6B expresses his opinion of the first part and then feels the need to explain exactly why the seasick thing made him dislike the first part – because it made him a little nauseous. Here, you know is added right after the explanation. The much more common position, however, is at the beginning of an intonation unit, as in extracts (119) through (122), where you know introduces the explanation. These extracts give typical examples of this function of you know.
Chapter 4. You know
(118)
200 201 202 A: 203 B: 203a (6B, NS)
(119)
102 an’ a little .. box came up, 103 <SV you know some words came up and said & 104 & broke and hungry or .. whatever SV>, (22B, NS)
(120)
276 if I if I found it on TV, 277 you know [zapping] around I wouldn’t & 277a & watch it. (45B, NNS)
(121)
322 B: [[but then again there’s nothing else]]. 323 you know it’s like-324 there was no other kind of & 324a & entertainment though. (124B, NS)
(122)
146 and the conductor is like. 147 .. you know gives like an I’m sorry bow & 147a & and walks away, (131A, NS)
... I didn’t like the first part. because of the= the seasick thing? yeah. (0) (0) making me a little & & [a little nauseous you know ],
I have labeled all the cases presented above as ‘you know marking explanation’, firstly because exemplification, clarification, and amplification were far too infrequent to form categories of their own, and secondly because, in a wide interpretation, all three can be seen as special types of explanations. In any case, they all function at a textual level: you know is used to indicate ideational relationships between statements or concepts which precede it and those which follow it. Only occasionally is you know added to the end of a statement to indicate its relationship to the previous statement or concept. Marking explanations is, with a total of 60 tokens, the most frequent category of you know in my data. 53 tokens indicate explanations in the narrower sense. 17 instances occur in the speech of 8 German speakers; the remaining instances are distributed across more than half of the American speakers. The difference between the two groups is clearly significant.
... Quotative you know Four studies mention that you know occurs with reported speech. Schiffrin apparently found only two such instances in which you know prefaced reported speech; therefore, she just mentions it and does not devote any further space to it (Schiffrin 1987: 282). The term ‘reported speech’ used by Schiffrin is a bit misleading. Redeker describes the function of you know in this context more exactly as “to mark the transition from the speaker’s own talk to a direct-speech report” (Redeker 1991: 1163). She uses this
Discourse markers in English discourse
and other “enquoting devices”, which occur not only in dialog but also in monologic narratives, to support her claim that these should be considered discourse-segment transitions. The quotative function of you know in itself does not play any role in her publication. Erman lists you know being used to “mark the boundary between two modes of discourse” as one of the functions at the macro-level (1987: 115). The corresponding example shows you know after reported (direct) speech and before the direct speech of the current speaker. In Erman (1992: 223), she calls the former “indirect speech” and the latter simply “direct speech”. You know is said to mark transitions between the modes in both directions; however, the example given again shows you know after “indirect” and before “direct” speech. You know in this function orients the hearer in the discourse at the textual level (1992: 221, 223). Accordingly, Erman (2001) categorizes it as a discourse marker within the group of textual markers. In this publication, Erman compares quotative you know with quotation marks in written text and gives an example which contains quoted speech introduced by well you know; the transition to direct speech is performed by anyway. He and Lindsey (1998: 143) also list you know at the beginning of reported speech in their description. However, they do not limit the term “reported speech” to directspeech reports; indirectly reported speech is also included. The two types together made up 18.7% (17 instances) of all instances of you know in their data, which consist of university counseling sessions. This latter information is important in the assessment of the inclusion of indirectly reported speech. What other people, perhaps other members of staff, have said may be important in such a counseling session, but it is unlikely to be reported as direct speech. In other contexts, such as narrating some events, encounters, discussions, etc., to a friend, quoting somebody directly to enliven the narrative is more likely. Like Redeker, He and Lindsey do not describe you know introducing reported speech just in its own right. Rather, since only newsworthy information is reported, this function serves to support the argument that you know “introduces new, important, salient information” (1998: 143). In my data, quotative you know was, with thirteen tokens, one of the less frequent functions. Nevertheless, it is frequent enough to deserve a detailed description. The category ‘quotative’ was applied only when it was followed by quoted speech. This excludes indirectly reported speech as described and included by He and Lindsey. It also excludes you know after a direct-speech report and before the current speaker’s speech, as included by Erman. Predictably, indirectly reported speech does not play any major role in a movie narrative like this; you know following (and not preceding) quotations will be covered mostly in the category described in Section 4.2.2.6. You know as an enquoting device, which I call quotative you know, occurs in two sets of environments. One set concerns voice quality. When a speaker quotes the speech of one of the characters in the movie, s/he may do so in his/her normal voice, or s/he may change the voice quality, for example to a higher or lower pitch, to indicate that the words spoken in this voice quality are not the speaker’s own but somebody else’s words. In the transcripts, this is indicated by enclosing the respective stretch of words
Chapter 4. You know
in angled brackets and a capital Q. With you know as a quotative, this happens four times. It is illustrated in extract (123). The quotation in line 68 actually is not quite what Chaplin said. Chaplin and his lady-love had watched another customer being beaten up and thrown out of the restaurant by several waiters. As they come back and pass Chaplin, he asks one of them, “What’s the trouble, waiter?” These words are shown by means of an inter-title, a black screen with text printed in white. However, as we will see in Chapter 5, the quotative construction BE+like, which the speaker uses here, serves to indicate that the following quotation is less than literal (cf. for example Jucker and Smith 1998: 184; Andersen 1998). (123)
67 so Charlie Chaplin’s like you know & 68 &, 69 and (H) the waiter says he was & 69a & ten cents short. (6B, NS)
(124)
168 there is a big door that says & 168a & marriage licenses, 169 so um Charlie asks the girl you know shall & 170 & we get married [and this sort of thing]. (4B, NS)
Extract (124) illustrates a case in which the quotation is not spoken in a different voice quality. However, it is clearly recognizable as a quotation through the use of a quotative verb (asks) and a change of the deictic center. Other quotative verbs which have been used with you know are think, say, or the functionally equivalent construction BE+like, with BE+like being the most frequent. In the present extract, the speaker reports the content of an utterance which she assumes has been uttered, rather than reporting the exact words. She indicates this by adding and this sort of thing. Apart from that, she could not have been sure of any exact wording, since there was no inter-title shown in this scene. So far, I have talked about the matter of quotation voice quality as if there were a clear boundary line between this special quality and normal voice quality. This is actually not the case. Rather, there is a continuum; clear and extreme cases of changed voice quality for quotations have been marked with angled brackets and capital Q, while less clear cases have not been marked. However, this does not mean that the speaker does not set off the quotation at all by vocal means. Intonation also serves to mark direct-speech reports. In addition, speakers frequently start the quotation in a (at least slightly) changed voice quality and end in their own normal voice quality. Such cases have not necessarily been marked. The second set of contexts is made up by the position of you know. I said above that I excluded you know which is not followed by a quotation. However, quotative you know may, at the same time, be followed and be preceded by a quotation, as we can see in extract (125). The speaker talks about a scene in which another customer in the restaurant offers to pay for Chaplin’s bill. Both stretches of quotation, before and after
Discourse markers in English discourse
you know, are said in quotation voice quality, and both quote the same speaker. This position of you know makes up almost half of all quotative you know. (125)
92 93 94 94a (5B, NS)
(126)
461 A: and he’s like here’s my suit, 462 (H) you know but I’m poor. (134A, NS)
.. he’s like, you know, an’ Charlie Chaplin is sayin’ & &,
Extract (126) shows another case of you know embedded between two stretches of quotation, but in normal rather than quotation voice quality. The participants are discussing the discrepancy between Chaplin’s clothes on the ship and the fact that he comes to America among obviously poor immigrants, ending up poor himself. The quotations here are a case in point for the problem of quotative voice quality discussed above. The voice does not change pronouncedly enough to justify marking it as different, and yet it becomes clear through intonation and slight pitch differences that here’s my suit and but I’m poor are meant to be Chaplin’s words. The extract is taken from the discussion between the participants, not from the narrative part. Speaker 134A does not report any scene from the movie; rather, she tries to express the contrast between Chaplin’s clothes and cane and the fact that he is an immigrant. The words reported in the extract are therefore hypothetical speech. The introduction to it, he’s like, again warns the hearer that what follows need not have been words actually spoken by the person “quoted”. The same is true for the quotation in line 244 of the last extract, (127). Here the speaker does indeed refer to a scene from the movie. Even though the movie is a silent one, it is usually possible to tell when a character says something because of the lip movement. However, in this scene, there is no lip movement, and line 244 “quotes” something which Chaplin is apparently thinking, judging from his body gestures and facial expression. (127)
239 240 240a 241 242 243 244 (47B,
B:
A: B:
he didn’t have his money. .. and then this beggar comes in & & off the street, .. yeah, (H) and shining this nice coin up you know, and polishing it all up nice and he’s like, ... you know .. I know that’s my coin.
NS)
Quotative you know was used by American native speakers only (once or twice by 11 speakers). The Germans apparently did not know about this function, since they did not use it at all. I fully agree with Erman (1992, 2001) that you know as a quotative functions at the textual level. This is particularly true for you know between a quotative verb and
Chapter 4. You know
the quotation itself. When you know is embedded between stretches of reported direct speech, however, it also carries an element of appeal to the hearer to imagine the scene which has been presented by the speaker as lively as possible through oral means. This appeal is described in the following section.
... “Imagine the scene” We are now moving from the textual to the interactional level. Functions at the interactional level have in common that the speaker, with using you know, tries to involve the hearer in some way. This can range from an appeal for mental involvement only to an appeal to actively participate in the conversation (cf. Schiffrin 1987: 295). Östman writes about the latter: “If accompanied by an interrogative intonation contour, you know almost explicitly requires a response from the addressee” (1981: 26). With such an interrogative intonation contour, you know functions like a tag question, as Crystal (1988: 47) notes in his list of “situations which permit the use of you know: [...] At the end of the sentence, it often acts as a kind of tag question – as a check that the listener is understanding what is being said”. Even though Crystal does not explicitly say that the listener is required to respond, this would be necessary for the check to be performed successfully. Both eliciting a response and securing the comprehension of the hearer plays a role in Erman’s (2001) account. The former, which she also explains as confirmation-seeking function, occurs mainly in the British adult corpus she uses (London Lund Corpus, LLC). The comprehension-securing function, in contrast, is more frequent in the British adolescent corpus (COLT). However, she defines this function rather narrowly: “the speaker making sure that the listener has correctly understood specific references made in the text, usually to people but also to objects and other phenomena” (2001: 1346). How do the functions described by these researchers relate to you know being used in the Chaplin Corpus? Here, we find that the type of corpus, i.e. a corpus consisting of movie narratives, influences the functional types of you know and their frequency. Much of the appeal for the hearer’s mental involvement concerns the details of the narrative. Thus, in one of the most frequent functions, you know can be paraphrased as “you can imagine the scene, can’t you?”, “I’m sure you can imagine the scene”, or “Please imagine the scene!”, depending on the intonation contour. I will begin with you know accompanied by an interrogative or “tag question” intonation contour. In the transcripts, this intonation was indicated by a question mark at the end of the intonation unit. Such an “interrogative” you know occurred four times in my data, out of 39 which have been labeled “imagine the scene”. Extract (128) illustrates a relevant instance: (128)
309 310 311 312 313 314
B:
A: B: A:
(0) before they even flashed, that he was an artist who wanted to paint them, I knew he was an artist. how did you know? because he was= [.. going] like thi=s. [he have a hat]?
Discourse markers in English discourse
315 B: 316 A: 316a B: 317 A: 318 B: 319 A: 320 321 (2B, NS)
like .. he was [[pic]]turing them & [[I]]-& on a canvas, (0) (0) oh see [I was]-[you know ]? I thought-@@ I thought when you said that, I thought he had .. an uh artist hat o=n,
The students in this session talk about one of the characters in the movie who was a customer in the restaurant where Chaplin dined and who later offered to pay Chaplin’s bill. This customer appears only in the second half of the movie, the part which speaker A did not see. Yet, she plays an active part in this extract, contributing questions, suggestions, and her own thoughts. Speaker B explains that she knew this man was an artist because of his gestures, which she tries to imitate (line 313) and describe. It seems, however, that she is not entirely sure whether her partner can imagine the gesture. Therefore, she adds you know with interrogative intonation to check on her partner’s comprehension. A does indeed respond, as Östman would have predicted, though not with a direct answer to B’s question. Rather, she responds by explaining what she herself had thought during B’s narrative. Two other examples of interrogative you know also seem to question Östman’s claim, or, to put it more exactly, the realization of the response to interrogative you know, at least in my data, is not as unambiguous as might be expected from his formulation. In extract (129), speaker 5B tries to explain and interpret the young lady’s behavior: When she meets Chaplin in the restaurant, he looks at her hands, and she reacts by lowering her head.14 B’s comprehension securing you know (“can you imagine the scene?”) after her description overlaps with A’s expression of recognition (line 263). Does the interrogative you know then elicit the hearer response? In the third case of interrogative you know by native speakers, the hearer does not provide any backchannel (extract (130)); he does not even nod. Instead, the speaker simply continues with the narrative. Only one German participant pair provides the expected pattern, with you know? being duly followed by the hearer’s yes (extract (131)). (129)
258 259 260 A: 261 B: 262 263 A: 264 B: 264a (5B, NS)
and she was sort of embarrassed by that. but-.. she was embarrassed? or something like shy, [you know ]? [o=h], ...(1.4) tryin’ to see what else has & & happened in there.
(130)
190 191 192
after .. that ordeal’s .. over Charlie & & Chaplin asks the waiter (H)-the host?
Chapter 4. You know
193 (H) you know ? 194 .. he’s like. (140B, NS)
(131)
181 but he had his hat on his .. head. 182 <@ you know @>? 183 A: <@ yes @> @. (0) (41B, NNS)
However, we find one case of you know with terminating intonation and another with continuing intonation, both of which are followed by the hearer’s acknowledgement (yeah). Interrogative and terminating intonation on you know are infrequent (four and two instances, respectively). The remaining instances of you know classified as “imagine the scene” generally fall into four groups in terms of their position: They may either occur at the beginning of the intonation unit (IU; four instances), in the middle of the intonation unit (nine instances), at the end of the intonation unit (eleven instances), or form an intonation unit of their own (nine instances). Most IU-final instances are found after a quotation; these are described below. (132)
202 203 204 205 (47B,
...(1.6) made it look like he wasn’t too & & obvious <X in X> like he’s picking up-picking coins up off the ground for fun, you know, NS)
Extract (132) presents a case of you know forming an intonation unit of its own. The speaker here tries to convey the manner in which Chaplin picks up the coin he found in the street. This seems to be a slightly difficult undertaking; therefore, you know follows the description. The transcription conventions are not detailed enough to convey the exact intonation of you know in line 205. However, judging from the tape, it can be paraphrased as “I’m sure you can imagine the scene”, spoken with a certain confidence that the hearer can imagine, and yet also casually enough to tell the hearer that the matter is not overly important. The following extract is interesting because it contains three instances of IUinternal you know. They are all preceded by an item included in Schiffrin’s list of discourse markers (so, but, and), and they all convey the same sub-variant of “imagine the scene”. (133)
238 239 240 241 241a 242 242a 243 244 245
and they go around a corner, and there is a blank door that just says-it just says marriage licenses on it, and now he’s got a couple of dollars & & from the ARTIST, so .. you know he tries to propose and & & all of this stuff, and .. she’s being real shy=, but you know .. she lets him-well he kind of kisses her,
Discourse markers in English discourse
246 247 248 (22B,
and .. she says no no, but she’s smiling, and .. you know she’s pretending to be shy, NS)
Starting from Chaplin and the lady finding a door with a sign marriage licenses, speaker 22B assumes that her partner can imagine what happens in a movie like this in such a situation. You know thus is equivalent to saying “I’m sure you can imagine what happens now”: Chaplin trying to propose, the first kiss, the lady pretending to be shy. While you know at the end of an intonation unit or forming an IU of its own typically refers back to the scene just described, extract (133) shows that you know within the intonation unit (or, for that matter, at its beginning), can also refer forward to the scene the speaker is going to narrate. In fourteen cases, you know followed a quotation. According to Erman’s (1992) analysis, these cases function like you know introducing a quotation in that you know provides the transition between direct and indirect speech. Pragmatic markers “may serve several functions simultaneously” (Erman 1987: 121), but they rarely serve two or more functions to the same extent in a single instances. Thus, although you know following a quotation could also function as a transitional device, in these cases it seemed primarily to be an appeal to imagine this quotation being spoken (or thought). It is interesting to note certain similarities within this sub-function of “imagine the scene”. Except for two of the quotations in this group, all are spoken in quotation voice quality (see previous section). This would perhaps speak for you know as a transitional device; however, there is a second common characteristic: All but three of the quotations consist of exclamations rather than sentences, such as in the following two extracts: (134)
80 ...(1.5) and then he was really like & 80a & gosh you know, (9C, NS)
(135)
292 they walk out of the restaura=nt, 293 and are, 294 you know, 295 that’s all over. (140B, NS)
Speaker 9C (extract (134)) describes the moment in which Chaplin realizes that his money may not be sufficient for all the food which he has ordered for himself and for his lady-love. Speaker 140B (extract (135)) describes the couple’s reaction when they finally leave the restaurant without having been beaten up for a shortage of payment. Also noticeable is the fact that all quotations in this group are introduced either by the construction BE+like or simply by was/are as a quotative verb. As I have mentioned above, several authors found that BE+like is used for less than literal quotations. Thus, the utterances in the narratives marked as coming from one or more of the movie characters do not represent direct-speech reports in the traditional sense. Instead, they try to invoke the image of an emotion, reaction, or attitude. This becomes also clear
Chapter 4. You know
for example when speaker 143B has the artist sayas a reaction to seeing Chaplin and his lady-love, who apparently seem to make a perfect model for his paintings. These emotions or reactions could in most cases be described with an attribute (‘he was stunned’, ‘they were relieved’), but the speakers preferred a more dramatic, but also perhaps more ambiguous variant to present the same content. Therefore, it makes perfect sense to add you know as an appeal to imagine the scene, including the emotion portrayed. After you know introducing an explanation, “imagine the scene” was the most frequent category, together with ‘lexical and content search’. It occurred 39 times, only four of these from German speakers. Obviously, this difference was also statistically significant.
... “See the implication” The category which I have entitled “see the implication” is related to the previous category. Both present an appeal to the hearer for an involvement ranging from mental to active participation in the narrative. Both can be most easily described by paraphrasing you know; while the you know of the previous category asked the hearer to “imagine the scene”, the you know of the present category asks the hearer to “see the implication”. In most cases, this is an implication of what the speaker just reported for the characters in the movie. In both categories, the intonation can convey that the speaker is sure of the hearer’s ability to do so, asks the hearer to do so, or checks whether s/he is able to (“tag-question” intonation, see above). (136)
72 73 73a 74 75 A: 76 B: (2B, NS)
and he goes to pay the bill, .. and the waiter .. bends the coin & & it’s no good. [you know ], [o=h], (H) so then he doesn’t know what he’s gonna do.
Extract (136) illustrates a typical case. The speaker has told the hearer that Chaplin did not have the money to pay his bill in the restaurant, but that he managed to grab a coin with which another customer had paid and which the waiter had lost. With this coin, he goes to pay the bill, but it turns out that it is a fake coin. Speaker 2B’s you know asks the hearer to imagine what implications this had for Chaplin (i.e. that he was shocked because, after believing he got out of his fix of not having money to pay the bill, he finds himself in this situation again). The hearer provides a backchannel (a lengthened o=h) indicating that she understood the full significance of what speaker B has just told her. This backchannel overlaps with B’s you know, so that we can assume that it was not you know itself which triggered it. After the backchannel, B expresses part of the implication she had in mind, i.e. what she says follows from two intermediate steps which her partner had to see for herself: Since the coin is fake, Chaplin still has not paid his bill and he does not have any other money; so then he doesn’t know what he’s gonna do.
Discourse markers in English discourse
(137)
233 ...(1.3) and when he asked for coffee .. & 234 & from the woman this other guy comes & 234a & in and he’s like looking at his coin & 235 & likeI found a coin you know, 236 A: mhm, (0) 237 B: (0) he sits there an’ orders a coffee, (131B, NS)
You know was followed by an immediate oral reaction from the hearer only in a third of the instances in this category. Extract (137) shows such an instance. After Chaplin had realized that he must have lost the coin he found in the street through the hole in his pocket, he tries to procrastinate paying by ordering another coffee for his ladylove. When this other guy comes into the restaurant who apparently just found a coin, the implication would be that this coin is Chaplin’s coin, the one he found and lost again. This implication is supported by the way Chaplin reacts in the movie. However, speaker 131B does not immediately mention Chaplin’s reaction nor the implication about the coin. Instead, she just adds you know at the end of her (long) intonation unit and leaves it to her partner to figure out what the scene just described means for Chaplin. A’s backchannel, mhm, can be an indicator that she does see the implication as B intended, but it could also be just a signal that she has received the information provided by B. The intonation of mhm is neutral (a short rise) and thus does not give us an unambiguous clue either. The content of extract (138) builds on the content of the scene described in extract (136), the implication which the hearer is asked to see with you know can only be drawn if the hearer knows about the scene with the fake coin. The scene in extract (138) takes place much later, when another customer, an artist, had come to sit at Chaplin’s table and pays his bill. The waiter brings back change (one coin) and leaves it on the table, so that Chaplin is able to take it surreptitiously and pay his own bill with it. Still suspicious, the waiter bites on that coin as he did with the previous one, but it doesn’t bend. So it must be real money and not fake. You know follows the description of the suspicious waiter’s behavior and its outcome and invites the hearer’s involvement, i.e. to think of the implication this outcome (it doesn’t bend) has. As we have seen in Chapter 2, so with final intonation also conveys an implication which the speaker does not utter at this point (category ‘implied result’). Thus, the two markers here have similar functions, and when the hearer finally provides a backchannel, he may have done so as a reaction to either marker or to both. Similarly to speaker 2B, but unlike 131B, speaker 4B then expresses the implication, to make sure it is clear and understood by the hearer. This is typical of about half of the instances in the present category. (138)
146 147 148 149 149a 150 151
and he pays the bill with that, and the guy puts it in his mouth, and it doesn’t-the waiter puts it in his mouth and it & & doesn’t bend, you know, so.
Chapter 4. You know
152 A: mhm. 153 B: ... obviously it’s not .. fake. (4B, NS)
(139)
111 B: and he goes in the in the restaurant, 112 .. thinking that he’s got money, 113 but he hasn’t got money. 114 you know ? 115 um, 116 and he orders some food, (75B, NNS)
I said above that only in a third of the cases in this category the speaker received an immediate oral reaction to you know from the hearer. Not even interrogative intonation prompted such a reaction, as extract (139) shows. In this extract, B had described that Chaplin found a coin, put it in his pocket, and, due to a hole in that pocket, lost it again without noticing. The unspoken implication invoked by you know here would be that he will certainly have a problem in the restaurant because he will not be able to pay for his food. Despite the interrogative intonation of you know (which is, admittedly, not very strongly interrogative), A does not react in any way, not even by nodding. It seems here that intonation alone does not evoke such a reaction if it is not accompanied by body language. The participants 75A and 75B are both male and, during the narrative, only occasionally look at each other. Thus, they show a behavior which seems to be typically male, according to an analysis by Tannen (1991). When B says you know?, he does not look at A, so we may assume that B does not expect an answer. Since A does not look at B either, A cannot have detected any expectancy, unless he feels that B is pausing to allow him to respond. The category “see the implication” was applied to fifteen cases, i.e. less than half as often as the category “imagine the scene”. Six instances were produced by German speakers, nine by American speakers. However, this did not result in a statistically significant difference. Thus, “see the implication” is one of only two discourse marker functions of you know which are not used significantly more frequently by the native speakers than by the non-native speakers in my data.
... Reference to shared knowledge The literal meaning of you know seems to imply that, when it is used as discourse marker, the hearer does indeed know. However, we have seen above that the marker you know has a number of functions where the knowledge of the hearer is not an issue; if it is presupposed at all, then in the form of the ability to imagine a scene or see an implication. Several researchers even explicitly claim that you know is rather – or at least can be – used for new information (Brinton 1990; He and Lindsey 1998; Biber et al. 1999; cf. Jucker and Smith 1998). Nevertheless, this does indeed not exclude you know from being employed for what the speaker considers general information (Schiffrin 1987; Jucker and Smith 1998), common ground (Erman 1992), “incontestable mutual
Discourse markers in English discourse
knowledge” (Holmes 1986: 8; cf. Östman 1981), and shared knowledge (Schiffrin 1987; Stubbe and Holmes 1995) or experience (Biber et al. 1999). In the GLBCC, you know in this realm takes on a rather specialized function. It was frequently used to focus the hearer on information which was actually shared, and it was exclusively used to refer to characters or scenes from the movie. ‘Shared’ here means that both participants knew the character or scene, either because they had both seen it, or because partner B had told partner A about it. Thus, the speaker always knew that the hearer had (or should have) the relevant knowledge (cf. Schiffrin 1987: 268). You know served to make sure that the hearer was able to access this knowledge at this particular moment because it was critical to understanding the ongoing narrative or to understanding the point the speaker was making (cf. Jucker and Smith 1998: 196). Extracts (140) and (141) exemplify this: (140)
(141)
184 185 186 187 188 188a (63B,
A: B: A:
while .. this guy is getting beaten up, there’s the bird. .. [yeah]. [you know that] he meets on the [[ship]]. [[yeah]] and and yeah.
NS)
139 and ... (H) then he met in the restaurant & 139a & a woman, 140 you know the young woman whom he had & 140a & given all .. his money. 141 you know on [the ship]. 142 A: [<SV mhm SV>], (42B, NNS)
Extracts (140) and (141) are very similar; one comes from a native speaker participant pair, the other from a German pair. In both cases, the speakers in role B re-introduce the character of the young lady. She appears for the first time on the immigrant boat in the first part of the movie. Chaplin falls in love with her, and he gives her money because her mother’s money had been stolen. This is what both partners know. The young lady then re-appears in the second half of the movie (which only the participant in role B has seen), where Chaplin meets her unexpectedly in the restaurant. Virtually all speakers in role B (and role C, for that matter) consider it important that this is not any lady but the young lady from the boat. Some speakers chose to make sure that the hearer understood this detail by employing you know, for example speaker 63B and speaker 42B. When speaker 63B reports that there’s the bird, his partner (63A) responds, after a pause, with yeah. Apparently, however, the pause shows 63B that his partner has not immediately understood. Therefore, he reinforces by using you know and giving additional information which successfully helps his partner to access the image of the young lady. This becomes clear with the repeated yeah yeah of speaker 63A, which means that speaker 63B can continue his narrative, building on this understanding.
Chapter 4. You know
It is interesting to see here how the two speakers, 63B and 42B, first introduce the young lady in their narrative. In a study on referring expressions also based on the GLBCC, Smith et al. (2001) argue that the form of a referring expression is influenced by two factors: common ground and saliency. Saliency here refers to the importance of the entities for the discourse; in the context of the GLBCC, this means that referents which play a major role in the movie are highly salient, while supporting characters and stage props are less salient (2001: 532). Thus, the referring expression will indicate whether the referent is part of the common ground and will also give clues as to the saliency of the referent. Smith et al. found that “[s]alient entities from the first half of the movie [...] are almost invariably introduced with a definite or possessive NP” (2001: 533), as would be expected (cf. Lyons 1999). This is also the case for the introduction of the young lady in extract (140). Thus, speaker 63B indicates that this character is part of the common ground, even though overestimates his partner’s ability to recognize the expression bird as referring to the young lady. By using a definite expression, 63B also indicates that the character will play an important role in the rest of the narrative. However, the participants in the GLBCC did not all share the view of the young lady as a major character. Several speakers thought that she is not very important; they considered her a supporting character or, as 42B expressed it in the discussion part, a second main .. character. When 42B first introduces the young lady (line 139 in extract (141), she does so by using an indefinite noun phrase. This type of referring expression can be seen as a reflection of her opinion about the saliency of the young lady. However, the referring expression after you know in line 140, a definite noun phrase, reflects the recognition that this character is part of the knowledge 42B and her partner share. After providing information about the lady which should enable her partner to identify the character, she expects a reaction showing that he knows who she is talking about. When this reaction does not come, she provides additional information, again introduced by you know. This time, her partner utters a mhm, which may or may not be a token of recognition.15 However, much more common reactions to the appeal expressed by you know are yeah, uh huh, and oh yeah. The two extracts above show references to shared knowledge which are important for the narrative. However, such references may also be crucial for arguments in the discussion part. In the next extract, number (142), the participants are discussing the acting of the minor characters. When speaker A expresses his opinion about one of the characters, it is important for him that his partner knows which character he is referring to. This desire is expressed through you know. Like the speakers discussed above, he therefore keeps adding information about the character (lines 532, 534) until he is convinced that his partner has recognized the referent.
Discourse markers in English discourse
(142)
531 the= ... the guy the thief was & 531a & really convincing. 532 .. you know the big guy. 533 B: [yea=h]. 534 A: [<SV at the] beginning SV>. (47A, NS)
In the group of the native speakers, it was rare for the hearer not to respond to you know. Those speakers who did not receive a token of recognition included a speaker in role C and one in role A during the monolog part, both of which did not have an active hearer. In the group of the German speakers, almost half of the instances of you know remained unanswered, at least verbally. So far, all extracts have shown you know as reference to a character. However, these make up only half of the cases in the native speaker group (though two thirds among the German speakers). The following extract illustrates a case in which the reference is to a feature of the movie: (143)
215 216 216a 217 217a 218 A: 219 B: 220 A: (6B, NS)
because like .. when they are playing cards, and all of a sudden it went to a shot & & at the dining room, .. you know like this random shot & & [in the dining] room, [yeah like]-[[yeah]], [[like]] cut in.
Before the beginning of the movie proper, a text informs the viewer that it took Chaplin several days and nights to cut the movie. Speaker 6B thinks that this procedure, without sleep at any night, negatively impacted the making of the movie. One of the consequences in his opinion is this random shot (lines 217f.). Again, when the hearer does not show immediate recognition of the issue referred to (lines 216f.), the speaker reinforces the reference, using you know to introduce a repetition. This time, the hearer remembers the random shot. More typical non-person references are to particular scenes or to actions of characters, for example you know how he was going through her pocket (a scene which underlines speaker A’s surprise at Chaplin not having money, because he kept some money for himself), or when .. you know with the <@ Star Spangled Banner playing @> (as an explanation when speaker A left the movie room). All of them were followed by responses from the hearer. However, it seems to depend not only on the presence of you know but also on the speaker’s expectations as expressed with intonation and body language whether such a response was given (see Sections 4.2.2.6 and 4.2.2.7). If the response was not expected, you know served more as a reminder of facts which both partners know, as in extract (144).
Chapter 4. You know
(144)
105 and then he’s walking down the street, 106 with no money, 107 cause you know he gave her the money an’ stuff, 108 an’ he’s hungry an’ everything, 109 ((clears his throat)) 110 and then uh. (143B, NS)
Here it seems that it is no question for the speaker whether the hearer remembers the correct scene. Therefore, the speaker proceeds with the narrative without waiting for a reaction. Not surprisingly, you know as a reference to shared knowledge was significantly more frequent in the native speaker group; they used twice as many tokens as the German speakers, who employed it 9 times.
... Appeal for understanding The last three categories, “imagine the scene”, “see the implication”, and the ‘reference to shared knowledge’ had in common that you know in these categories basically asked the hearer to get involved with the content of the movie/narrative. The next two categories go one step further, i.e. you know appeals to the hearer on behalf of the speaker rather than on behalf of the narrative. The category treated in this section, ‘appeal for understanding’, is related to the ‘lexical and content search’. In the latter, you know accompanied the search for appropriate expressions or correct content (within the frame of the narrative); in the former, the speaker cannot find the right words, but asks the hearer to understand anyway. We find a classic case in extract (145): (145)
372 373 373a 374 375 376 377 378 (51A,
.. remember [this] dress and this & [yeah], & um ...(1.1)? [[ ]]? B: [[yeah I don’t know]]. [yeah]. A: [this] you know, and he he was falling down, NNS) B: A:
Speaker 51A tries to find the English word for German Kopftuch, which would be ‘scarf ’. But she cannot think of it, nor is her partner able to help her. So she finally gives up searching for the English translation and appeals to her partner’s understanding. In this case, this is not a problem for the partner at all, since both have German as their first language and therefore know what A meant. In extract (146), the situation is slightly different. When Chaplin pays his bill with what turns out to be a fake coin, he orders some more coffee for his lady-love to procrastinate, to gain more time to think about a solution for his problem. After having described what Chaplin did when he discovered the coin was not a real one, speaker B sets off to say why he did
Discourse markers in English discourse
that. But apparently she cannot think of the right expression (line 196). Rather than making an effort searching for this expression, she immediately appeals to the hearer’s understanding of what she meant to say (line 197) with you know. (146)
191 B: and he paid for it, 192 but it was fake, 193 ... cause it could bend, 194 ... so then he goes& 195 & I want .. some coffee .. for her, 196 so to-197 .. you know. 198 [<X final X>]-199 A: [he knows] he knows it’s faked? (130B, NS)
The appeal for understanding often, but not always concerns lexical choices. It may also concern a deficiency in the argument of the speaker or in the way s/he expresses his/her opinion. In Chapter 1, I described the task of speakers in role C, i.e. speakers without a partner, as watching the whole movie, retelling the whole movie and giving their opinion along the lines of the questions all participants were given as discussion questions. Some speakers in role C (and also some in role A during their monolog narrative of the first part of the movie) asked whether they were allowed to look at the researcher while they talked. This request was always granted, with the caveat that the researcher would not say anything. Occasionally, however, the researcher could not help uttering a backchannel, as in extract (147): (147)
174 .. using hand motions and stuff like & 174a & that it was pretty good. 175 (H) um ...(1.6) the minor characters & 175a & I mean they did good too. 176 <SV I mean .. you know SV>. 177 R: mhm <SV @@@ SV> (137C, NS)
The speaker apparently did not know much to say about the movie. The first line in the extract refers to the acting by major characters; the second refers to the acting by minor characters. The speaker seems to be aware that her statements are very general and not overly meaningful, but she cannot think of anything better to say. Therefore, her you know in line 176 appeals to the understanding of the researcher, to whom she feels responsible, and serves as a kind of apology. Thus, you know does not substitute the search for a word but rather the search for content. In the present case, the appeal expressed through you know apparently functioned effectively at the interactional level: It incited the researcher to give a minimal token of appreciation. Finally, you know in this category was also occasionally extended by German speakers to appeal for sympathy for the speaker (extract (148)). Speaker A expresses his opinion on Chaplin’s clothes at some length and with some conviction; speaker B does not seem to be overly convinced, however. She “never looks for what people wear”
Chapter 4. You know
(lines 250f.) and therefore cannot join in the judgment of whether Chaplin’s clothes fit the situation or not. Although she is expressing an opinion, speaker B is admitting that she does not really know, and her you know in line 246 is an appeal for speaker A to understand this. Like the you know of speaker 137C above, her you know also carries an apologetic note. (148)
241 A: uh one of the reasons was that Charlie Chaplin & 242 & his usual clothes doesn’t really fit, 243 ...(1.1) on the ship. 244 they seems to be uh wrong placed. 245 <X it’s X> all those other guys & 245 & are wearing rags, 246 .. and so on, 247 B: mhm. 248 ... it’s true but-249 I don’t know .. I don’t know you know, 250 I never look for the clothes or what & 251 & happened wha- what people wear, (42B, NNS)
In terms total number of tokens, you know as an appeal for understanding appeared more times in transcripts from German speakers than from American speakers, although the average number of instances per speaker was higher for the Americans. There is no statistically significant difference between the two groups.
... Acknowledge that the speaker is right Another kind of appeal involving the hearer is found in extracts such as (149). In this extract, the participants have finished the narrative part and now turn to the discussion part. This is usually done by reading the first question on the sheet in front of them, which reads “Did you like it? Why?” Speaker 47B expresses his opinion about the movie in line 372, and then adds you know. (149)
371 .. did you like it. 372 B: ... I thought it was pretty good, 373 .. you know, 374 for no words, 375 ... it was amusing, (47B, NS)
This you know fulfills the function of asking the hearer to agree with him. The category occurs exclusively in the discussion part of the experiment (with one exception, in which the speaker re-enacts a discussion in the movie). Typically, the speaker expresses his/her opinion on some aspect of the movie, and does so with a certain confidence. Here lies one of the essential differences between the appeal for understanding and the appeal to acknowledge that the speaker is right. In cases of the former category, the speaker is unsure either of the appropriate lexical expression (and gives up searching) or of his/her own opinion or of what to say about the movie; therefore, s/he asks for
Discourse markers in English discourse
the hearer’s understanding. In cases of the latter category, the speaker has a definite opinion and voices it. You know here perhaps could be paraphrased as “This is my opinion. Don’t you agree?” Even when there is hesitation involved and searching for lexical expressions, the speaker is not hesitant about his/her opinion. Extract (150) shows such a case: (150)
444 I ... DON’T want to be FORCED to feel & 444a & PITY for him. 445 because it’s just so-446 it’s too much. 447 because sometimes he’s just dumb. 448 .. he’s not-449 it’s-450 .. he’s dumb. 451 .. [@ <@ that’s what I think @>]. 452 A: [@@@@@], 453 B: (H) you know, (71B, NNS)
Speaker 71B thinks that the behavior of Chaplin is exaggerated, in that he acts in a way which is more stupid than she finds acceptable (it’s too much; he’s just dumb; lines 446f.). She tries to find a better way to express this for a while (lines 445–450), but she does not manage. However, she is still very clear about her opinion (that’s what I think); with you know, she appeals to her partner to acknowledge this opinion. Like the appeals to imagine the scene or to see the implication, there is a range involved in the appeal to acknowledge that the speaker is right, or rather, two types of ranges. One concerns the urgency of the appeal which is expressed through the intonation contour. Sometimes, you know is said in a very casual manner, implying that there is not much of a debate whether the speaker is right in his/her opinion, or at least has the right to be of that opinion. In other cases, it is important for the speaker to obtain an explicit agreement from the hearer; I will present such a case below. The other, related type of range concerns the question whether speaker and hearer are of the same opinion, or of opposing opinions. If they are not of the same opinion, it is of course important for the speaker to show that his/her arguments are better than the hearer’s, and to have the hearer accept this. Since the discussions about the Chaplin movie rarely touched personal or even intimate levels, there was usually no need to insist on or contradict an opinion with vehemence. However, speaker 4B (extract (151)) re-enacts a scene from the movie with a clash of opinions. In this scene (as told by 4B), the waiter does not want to give Chaplin food, but Chaplin insists that he has to because he, Chaplin, has money to pay for it. Chaplin’s you know (line 37)16 functions as a means to make the waiter acknowledge, either in word or in deed, that he has to serve Chaplin.
Chapter 4. You know
(151)
29 29a 30 31 32 (...) 36 37 38 (4B, NS)
.. he= sits down at a table next to a & random man. (H) and um the waiter comes over, and he’s like-an’ he’s like I don’t want to feed you. and um .. Charlie Chaplin’s like um .. like & & yeah y- you have to you have to you know, I have money here dada,
I mentioned above that, with you know, the speaker asks the hearer to acknowledge the speaker’s opinion. However, “asking the hearer” does not necessarily mean that the speaker expects an oral reaction. In most of the cases, there is no change of turns or even backchannels following you know. In extract (149) above, for example, speaker 47B continues explaining why he thought the movie was pretty good, rather than waiting for a response. Likewise, speaker 71B (extract (150)) continues her turn after you know in line 453. In most of the cases, you know is said with continuing intonation (marked with a comma in the transcripts). However, there is also a smaller number of cases in which you know has interrogative intonation. All but one of them are duly followed by some kind of response. Extract (152) shows two instances, in line 321 and 331. Speaker 125B can see absolutely no sense in what happens in the two parts of the movie. For her, there is no logical relation between the events (get off the boat, eat, get married; lines 319f.). She underlines her opinion with you know in interrogative intonation and thus asks for her partner’s acknowledgement that this opinion is justified (i.e. that she is right). Speaker A promptly reacts; she half-heartedly agrees, trying to mediate with a reference to the American dream. Speaker B then repeats her opinion arguing from a different angle: marrying does not change the immigrant status of Chaplin and his lady-love (lines 326–328), so that there is no point in getting married (line 330). Again, she asks for a verbal reaction with interrogative you know. A’s turn starts with an answer to the question in line 330 (what’s the point); the second line of her turn, however, is a response to B’s you know. Again it seems that she does not want to fully agree with her partner (I don’t know), so that B finally gives up trying to elicit the acknowledgement that she is right with her opinion (it was alright). (152)
317 318 318a 319 320 321 322 323 324 325
B:
A:
B:
so what was the-... I mean the first part and the & & second one make no sense. they get off the boat, they go eat and get married. ... you know ? yeah, I guess that’s ... the American dream, [<@ I don’t know @>]. [I don’t know].
Discourse markers in English discourse
326 ... I mean they’re both still-327 .. aren’t they both still like ...(1.2) & 327a & immigrants, 328 even though they marry each other & 328a & they’re both on the same boat. 329 A: yeah. 330 B: what’s the point of getting married. 331 ... [you know ]? 332 A: [that they’re married] in America, 333 I don’t know. (125B, NS)
The present category was, with 32 tokens, one of the more frequent functions. A third of the instances came from 8 German speakers; the 10 American speakers who produced the rest of the instances used you know in this function up to 6 times. The difference between the two speaker groups reached a statistical significance of p = 0.015 (required: p = 0.05).
... Various functions There were twelve instances in my data which I have subsumed under the category of “various functions”. The functions of you know in these cases were clear, but none of them occurred more than three times, i.e. not regularly enough to form a category of their own. Three of the functions are described in the literature on you know, and I will describe each in turn. For the remaining instances, it was possible to describe the function which you know served, but this function was found only once or twice in the data. Both Östman (1981) and Erman (1987, 1992, 2001) mention that you know may be used to introduce relevant background information, for example in a parenthetic comment, as Erman calls it. In my data, you know is employed only once for this purpose: (153)
270 271 A: 272 B: 273 273a (4B, NS)
reminded me a lot of like Laurel and Hardy? uh huh. because um-.. you know I just finished watching a & & bunch of those.
In the discussion of the movie, B remarks that this silent Chaplin movie reminded her of (also silent) movies with Stan Laurel and Oliver Hardy. Since, at age 23, B would not have seen those movies when they were popular, she explains that she just watched some of them (without mentioning the occasion or reason). This is background information which the hearer could not have accessed by himself. Östman also lists as a function of you know that it introduces digressions, while for Erman (2001), it can form the boundary between topics. These two functions are related; in both cases, you know introduces some content which does not belong to the
Chapter 4. You know
current topic. If the speakers return to the topic they talked about before you know, this other content remains a digression. If the speakers continue their conversation with what was introduced by you know, it becomes the new topic. Extract (154) below was preceded by a discussion between the speakers (a German female in role A and an American male in role B) about whom Chaplin married in the end, since A finds B’s description of this matter unlikely and confusing. A then switches to a different topic, talking about the scenes in general (line 504). B at first seems to take up the topic (I think so too, line 505), but then he moves on to a discussion of the minor characters with although you know, which subsequently becomes the new topic of the conversation. (154)
503 503a 504 504a 505 506 506a (44B,
B: A: B:
he married somebody that looked his age & & and that was fairly young. ...(3.7) all the scenes .. were very & & strange to me. ...(2.9) I think so too even-... (H) although you know the minor & & characters--
NS)
In the sub-section on you know being used as a reference to shared knowledge (see Section 4.2.2.8), I said that you know had been found by other researchers to introduce general knowledge. The speakers in my data only used it three times for this purpose. Two cases refer to the acting of the characters in the Chaplin movie, the third to the type of character Chaplin plays in most of his movies. In all three cases, the statement following you know is presented as if the proposition expressed were absolutely natural and incontestable general knowledge. This is true in the non-native (German) speaker’s use of you know (since the movies she is talking about are indeed silent ones), but not necessarily in the other two cases. Nevertheless, you know helps to make it seem that way: (155)
586 B: yeah the acting has to be brilliant otherwise-587 [you know if it’s not perfect it can’t be & 587a & funny]. (47B, NS)
(156)
(talking about characters portrayed as evil) 469 compared or contrasted with Charlie, 470 who is .. you know the gentle nice guy & 470a & who always wins in the end. (55A, NS)
(157)
336 well .. since .. you know they didn’t & 337 & really have .. um .. sound in that films, 338 (H) they exaggerated everything=, (71B, NNS)
Discourse markers in English discourse
Only two of the twelve cases in this mixed category were produced by German speakers, so that there is an obviously significant difference between German and American speakers.
... Unclassified instances The final category for you know is made up of unclassified instances. Fortunately, this category is the most infrequent one, but still necessary, since there are cases where it is impossible to judge which function you know fulfills. There are basically two reasons for this: In recordings of the speech of more than one person at a time, speech is not always intelligible, and speakers do not always get to say what they wanted to say. Without an appropriate amount of context for you know, its function cannot be determined with confidence. (158)
241 242 243 A: 244 245 B: 246 A: 247 248 B: (6B, NS)
<X wi= X> really good plot, and you know it’s [like]-[exactly], it wasn’t like [[..]] overly-[[um]]-.. you know, some stuff wasn’t overly played out, [mhm],
Speaker 6B in extract (158), for example, starts saying and you know it’s [like]-- (line 242). His like already overlaps with his partner’s [exactly], after which the partner continues and speaker 6B does not manage to re-gain the floor. Thus, it is not clear whether B’s you know was meant to introduce an explanation, for example, or perhaps a reference to shared knowledge. Where a lack of intelligibility is the cause for applying the category ‘unclassified’, it is either the you know itself which is not clear, or the syllables following you know. A quarter of unclassified instances come from Germans; the rest occur in native speaker data.
.. Summary of the discourse marker functions of you know We have seen in this section that you know fulfills a number of textual and interactional functions. At the textual level, it marks the speaker’s search for lexical expressions and/or the content of what s/he is going to say next. It also occurs after truncations, when the speaker repeats part of an utterance with a word or two corrected, or when the speaker simply starts anew. You know is also, though less frequently, used to indicate that a word, phrase, or clause lacks exactness and thus is only an approximation to what the speaker had in mind. By far the most frequent function of you know in my data is to introduce an explanation. Explanation here can also include occasional exemplifications. The last text-structuring function which I have described is you know as a quotative, i.e. at the beginning of reported direct speech. ‘Direct speech’ in this context also refers to potential or probable direct speech and to thoughts. Typically,
Chapter 4. You know
you know in this spot is not taken as being part of the quotation but as a means to introduce or set off the quotation. At an interactional level, you know is used to involve the hearer in various ways and to various degrees. Two of the functions can best be described by paraphrasing the meaning of you know: “you can imagine the scene” and “you can see the implication”. Both can range from “I know / I’m sure you can” to the equivalent of a tag question (“You can, can’t you?”), depending on the intonation of you know. However, “imagine the scene” is more than twice as frequent as “see the implication”. As you know already seems to imply, it can also be employed when the hearer does know and simply has to be reminded or asked to access the shared knowledge at that particular point in the narrative or discussion. Furthermore, it serves to express two types of appeal: When the speaker gives up searching for an expression or does not know what to say about the topic at hand, s/he may use you know to appeal to the hearer for understanding despite the deficit. However, the other appeal occurs twice as often, and almost exclusively in the discussion part. Here, the speaker asks the hearer to accept and acknowledge the speaker’s opinion, even if this acknowledgement often is not expected in the form of oral response.
. Quantitative results for you know In each of the sections on the individual functions of you know, I added a paragraph on their distribution between the American and the German group of speakers. This distribution has been illustrated in Figures 4.1a and 4.1b. Since the full labels of the categories of you know would be too long for such charts, Table 4.2 assigns abbreviations to the full labels. The first five categories denote functions at the textual level, the next five functions at the interactional level. The last category contains, as the label says, various identifiable but infrequent functions at both levels. Table 4.2 Functions of you know – full labels and their abbreviations in the charts Full label
Abbreviated label
Lexical/content search False start/repair Approximation Explanation Quotative you know “Imagine the scene” “See the implication” Reference to shared knowledge Appeal for understanding Appeal to acknowledge that the speaker is right Various
LCS FSR APP XPL QYK ISC SIM RSK AFU ASR VAR
Discourse markers in English discourse
avg. % of total words per group
0.08
Germans Americans
0.06
0.04
0.02
0.00 LCS
FSR
APP
XPL
QYK
Figure 4.1a Distribution of the textual functions of you know – American vs. German speakers 0.07
avg. % of total words per group
0.06 Germans Americans
0.05 0.04 0.03 0.02 0.01 0.00 ISC
SIM°
RSK
AFU°
ASR
VAR
Figure 4.1b Distribution of the interactional functions of you know – American vs. German speakers (the circle ◦ indicates that there is no statistically significant difference)
On average, the Americans used discourse marker you know more than five times as much as the Germans. Also, most of the individual functions showed a statistically significant difference. This is true for all textual functions, three of the five interactional functions, and for the mixed category labeled “various”. None of the Germans seemed
Chapter 4. You know
to know that you know can introduce direct speech reports, and only one of them employed it to indicate that the following expression lacks exactness (i.e. is an approximation). Also relatively infrequent in the non-native speaker group were you know to mark false starts or repairs and you know asking the hearer to “imagine the scene”. Only two interactional functions, “see the implication” and the appeal for understanding, were employed by the Germans with a frequency similar to that in the native speaker group. In the chart (Fig. 4.1b), they are marked with a superscript circle. It is not immediately obvious why “see the implication” should be frequent among the Germans. However, it makes sense that the appeal for understanding occurs in their data almost as often on average as in the native speaker data, and thus relatively more frequent than other functions of you know: We would expect them as non-native speakers of English to have difficulties with finding English expressions occasionally or even regularly, which would make an appeal to understand them despite these difficulties more likely to occur.
.. Non-linguistic factors ... Gender We have seen above that the results concerning the relationship between gender and the use of you know are not uniform. Erman (1992) claims that the different functions of you know tend to be employed to different degrees by men and women: Males use it more for turn regulation and repair/appeal, women use it more for discourse organization and hesitation. Moreover, pragmatic expressions in her data are more frequent in recordings with speakers of the same sex than in recordings of mixed-sex interactions. It must be added, though, that Erman’s quantitative analysis obviously is based on (average) percentages only and not on more in-depth statistical analyses. For Scottish speakers and you know in general, i.e. without a distinction of individual functions, Macaulay (2002: 765) discovered that “[w]omen are more likely to use you know than men”. Stubbe and Holmes (1995) found that it is more frequent in the speech of young male working class speakers than in the speech of young female working class speakers; however, they did not find any statistically significant influence of gender on the frequency of you know in their data of middle-class speakers. We do not have any information on the social background of the participants in the GLBCC, but since all of them are students, it is reasonable to compare them to Stubbe and Holmes’ middle-class speakers rather than to their working-class speakers. A test of the frequency of you know, in both the native speaker and the German speaker group, did not show any significant effect of the sex of the participant in either group. This applies to you know in general as well as to the individual functions of you know. Likewise, the difference between having same- versus opposite-sex partners apparently had no influence either. Of the Americans, there were only two US participants in allmale recordings. No significant difference could be found between the participants in all-female pairs and the participants in mixed-sex pairs. For the Germans, I carried out pair-wise comparisons between speakers in all-male vs. mixed-sex, all-female
Discourse markers in English discourse
vs. mixed-sex, and all-male vs. all-female recordings, without finding any significant differences either.
... Age Concerning age, the picture drawn in the literature is similarly complex. In contrast to the adults, Scottish adolescents apparently have not yet incorporated you know as a discourse marker in their discourse style, according to Macaulay (2002: 765). Erman (2001) claims that adolescents tend to use you know more often for turn-taking, while adults tend to use it more for turn-yielding. In Stubbe and Holmes’ (1995) data, you know was not related to age. Their speakers came from “a range of different age groups” (1995: 66), without any further definition of these age groups. Even though a number of the participants in the GLBCC were of age 18 or 19 at the time of the recording, an age which can still be considered (late) adolescence (Andersen 2001), I group the participants differently, due to reasons I have explained in Section 1.8.3. Thus, there are two groups for the Americans, age 18 to 21 and age 25 to 40; the Germans are divided in the same age groups plus one consisting of age 22 to 24. In the native speaker data, the two age groups used you know in all functions statistically with the same frequency. In the German data, the only significant difference showed up for the medium age vs. the older group for you know as a marker of lexical or content search. The students aged 22–24 were the only ones who used this function at all. To assume from this result that this age group’s data shows more instances of searching for lexical expressions or for content, however, would be a rash conclusion. Rather, they use you know more often to mark the search instead of well, for example, as the younger age group does (see the quantitative results for well, Section 3.3.1.2). ... Relationship between the partners The relationship between the partners in the experiment can potentially exert an influence on the use of discourse markers. However, neither Redeker (1990) nor Jucker and Smith (1998) found a statistically significant difference between strangers and friends for you know. The result for my data is the same, in the native speaker as well as the German speaker group. ... Role The participants in the GLBCC encounter three different situations, depending on which role they take on. Speakers in role B always have a partner when they talk, while speakers in role C do not have a partner at all. They maximally have a silent listener. Speakers in role A perform the narrative as a monolog, but for the discussion, their partner is present. It would seem likely that monolog versus dialog situations have an influence on at least the interactional functions, since they address the hearer directly and sometimes require a response. This is partially the case, but not consistently so. In the native speaker group, those in role B use the function “imagine the scene” significantly more than those in role A (average role B: 0.085; average role A: 0.008). This result confirms our expectations; however, it is the only significant difference between
Chapter 4. You know 0.07 role A role B
avg. % of total words per group
0.06 0.05 0.04 0.03 0.02 0.01 0.00 XPL
textual
ISC
interactional
Figure 4.2 Distribution of you know according to speaker roles (German speakers)
the two roles. Moreover, speakers in role C, who by definition do not have a partner, use “imagine the scene” almost as often (average: 0.08) as speakers in role B, who always have a partner. The results from the non-native (German) data fit the picture slightly better. Speakers in role B use interactional you know thirty times as much as speakers in role A, with a statistical significance of p = 0.022. However, again the only individual function for which such a statistically significant difference can be found is “imagine the scene”, which the speakers in role A do not use at all. Somewhat surprisingly, the two roles also differ considerably with respect to textual functions in general and you know introducing explanations in particular (see Figure 4.2). Even though the textual functions mainly serve to structure the narrative or the argument in the discussion part, German speakers seem to feel that a discourse marker like you know, which inherently addresses the hearer (you), requires somebody who will know or at least understand. Accordingly, speakers in role C do not employ discourse marker you know at all, and fourteen of the fifteen instances of discourse marker you know in the data of speakers in role A come from the discussion part, not from the monolog narrative.
.. Linguistic factors Regarding the non-native participants, there are a number of linguistic variables for which we gathered information through the questionnaire which they were asked to answer as part of the experiment. In Chapter 1, I quoted a number of scholars who are convinced that discourse markers are not taught in formal language education and therefore can only be learned through the contact with speakers who use them,
Discourse markers in English discourse
preferably native speakers of English. Arguably, this also applies to language education in Germany viz. German learners of English. Thus, one of the variables concerned the question whether or not the non-native participants had learned English in an informal environment, i.e. outside the (school or university) classroom or language institute. For ten students, this was the case. However, they did not use any of the functions of you know significantly more or significantly less than their fellow students. Another variable divided the non-native participants into those who never use English in such informal contexts and those who do at least to some extent; these two groups did not differ significantly in their usage of you know either. While, for the German students, the usage of English in informal contexts referred to this usage in Germany, another source of acquiring discourse markers can be times spent abroad in an English-speaking country. These constituted a separate variable with three groups: students who never went abroad, students who spent more time in Great Britain, and students who spent more time in the USA (cf. Sections 1.7.2.4 and 1.7.2.5). These three groups were compared pair-wise. It turned out that the students who went to the USA for some time did not use you know differently from either of the other two groups. However, significant differences in frequency exist between the “never abroad” group and those who spent more time in Great Britain for two textual and one interactional function: Those who did not go abroad did not use you know for marking lexical or content search or to convey a request to “see the implication”, while they did use it to begin explanations with a relatively high frequency. In contrast, the students who had been to Britain employed you know for lexical or content search and for conveying “see the implication”, but not for introducing explanations. These results are illustrated in Figure 4.3. When students who never went abroad are compared to 0.035 not abroad more GB
avg. % of total words per group
0.030 0.025 0.020 0.015 0.010 0.005 0.000 LCS
XPL
SIM
Figure 4.3 Distribution of you know according to time abroad (German speakers)
Chapter 4. You know
the group of those who did go to either the US or Britain, the former group still uses you know introducing explanations significantly more than the latter. So far, I have described only the differences which achieved statistical significance at the five per cent level. Besides these differences, it might be mentioned that the students who never went abroad only used four functions of you know: explanation, reference to shared knowledge, appeal for understanding, and the appeal to acknowledge that the speaker is right. Those students who had been to the US employed all functions except for the quotative function, while the data of the group who had spent more time in Britain lacked instances of you know for explanations, quotations, and the appeal to acknowledge that the speaker is right. Finally, I tested a combination of the above-mentioned factors: All students who had spent some time abroad (even if it was less than four weeks), learned English in an informal environment, or used it with friends and/or family members counted as students with native speaker contact.17 This applied to sixty participants. It turned out that they used you know to introduce explanations only half as often as those participants without native speaker contact (p = 0.044). Considering the results from the variable ‘time abroad’, this would have been expected. Interestingly, the result changes when a stricter evaluation is applied and only those are considered as having (had) native speaker contact who spent more than four weeks abroad or used English as the main means of communication not just occasionally. This restriction resulted in the group with native speaker contact consisting of only forty-eight participants and a significant difference between the groups not for you know introducing explanations but for “see the implication”. It was used only by those participants with this stricter definition of native speaker contact.
.. Summary of the quantitative results for you know On average, the American native speakers of English used you know in each of its functions in my data more than the German EFL speakers. For most of the functions, this constitutes a statistically significant difference at the five per cent level. Only two interactional functions do not achieve such a difference between the two speaker groups: you know conveying the request to “see the implication” and you know as an appeal for understanding. While Germans employ these two functions relatively frequently, they do not seem to know that you know may introduce direct speech reports (quotative you know). Of the non-linguistic variables, neither gender nor the relationship between the partners seem to have any influence on the usage of you know. Age does not play a role either in the native speaker group, while among the German speakers, only the medium age group uses you know to mark lexical or content search to a significant extent, in comparison to the younger and the older group. The role of the participants influences the frequency of you know more: The Americans in role B use it with the function “imagine the scene” ten times more than those in role A; Germans in role B also employ this function, you know introducing explanations, and textual and
Discourse markers in English discourse
interactional functions in general much more than their counterparts in role A. However, while the data of single Americans (role C) show all but two of the ten discourse marker functions of you know, the data of single Germans does not contain any discourse marker you know at all. It seems that Germans, unlike Americans, generally feel the need for a partner in order to employ you know. Three functions are significantly affected by linguistic variables in the non-native speaker data. You know marking lexical or content search and you know conveying “see the implication” are employed by students who spent some time in Great Britain, but not by those who never went abroad. In contrast, the latter group (not abroad) used you know to introduce explanations, which the students who went to Britain did not employ. These differences are partially replicated when the German participants are grouped into students with and students without native speaker contact. If “native speaker contact” is defined more generously, the Median test returns a statistically significant difference for ‘explanation’ (i.e. those without contact use it more); if it is defined more narrowly, the test returns a difference for “see the implication”, which those students with native speaker contact use more. Thus, we can venture to say that a higher exposure to (native) English, particularly in Great Britain, leads to a reduced frequency of you know introducing explanations and an increased frequency of you know conveying the appeal to “see the implication”.
Chapter 5
Like
(159)
142 and then like this guy walks in. 143 and I thinks he was like homeless or whatever, 144 a=nd .. he’s like-145 he orders a cup of coffee or whatever, 146 and he has a quarter. 147 or like the money whatever it was. 148 and he’s like rubbing it, 149 he’s all happy, 150 and Charlie is like-151 he’s thinking. 152 (H) and so like u=m what happens is, (8B, NS)
With seven non-verbal instances of like in eleven intonation units and a total of more than one hundred instances in her transcript, speaker 8B definitely leads the group of American native speakers regarding the frequency of like. However, her usage is by no means a rare phenomenon. For almost half of the group, the linguistic item like ranked among the ten most frequent words in their speech. This high frequency suggests that like plays an important role in the communication of native speakers, or at least in the communication of California students. The present chapter therefore is dedicated to analyzing the role of like in the GLBCC data. Firstly, we have to distinguish between three major functions / function groups: (a) like as a discourse marker, (b) like as a quotative, and (c) like as a verb, noun, adjective, adverb, preposition, conjunction, and in fixed expressions (cf. for example OED online; OALD 2000; and others). Publications on like generally either concentrate on the quotative function of like (such as Blyth et al. 1990; Romaine and Lange 1991; Ferrara and Bell 1995; Tagliamonte and Hudson 1999; Winter 2002) or include this function as part of the discourse marker like (e.g. Schourup 1985; Helt and FosterCohen 1996; Andersen 1997; Jucker and Smith 1998; Fuller 2003). Few dedicate any space to the third functional group (c)), except to mention that this is the sort of instances which is excluded from the research (e.g. Hasselgren 2002a). In this chapter, however, I will, for the sake of completeness, give short descriptions and examples of non-discourse marker, non-quotative, like. The focus will be on the discourse marker like; following the researchers listed above, I will also include quotative like, though not in much detail.
Discourse markers in English discourse
. Functions of like in the literature One of the first researchers who wrote more than just a short passage about like was Schourup in his treatment of Common Discourse Particles in English Conversation (1985). He describes like as what he calls “evincive”; as an element which “is used to express a possible unspecified minor nonequivalence of what is said and what is meant” or, in other words, as “a device available to speakers to provide for a loose fit between their chosen words and the conceptual material their words are meant to reflect” (Schourup 1985: 42, his emphasis). He finds this to be true in most instances of like. For like introducing direct discourse, it can mean that the quoted material was not actually spoken but thought or felt. When used after questions, it also refers to a thought, in this case to the thought of the hearer of the question. Like can also be read as “for example”. Again, it conveys that what is said is not identical with (i.e. expresses only a part of) what the speaker has in mind. Like as an interjection, after prefatory material, shows less of the evincive reading. This type of like often follows conjunctions and is followed by filled and unfilled pauses. According to Schourup, in this position it “clearly indicates intended continuation” (1985: 54). As we will see, much of what Schourup said about like can be found in later publications as well. Another frequently cited paper was written by Underhill (1988). Even though he does not refer to Schourup, he indirectly contradicts his interpretation of like: “I claim that like functions as an approximator [meaning ‘sort of ’] much less often than one might think”, though such instances do exist (Underhill 1988: 240f.). For Underhill, the central function of like lies in focusing the hearer on the following piece of information as the “most significant new information” or the point of the sentence (1988: 238). It introduces new concepts or entities, which in some cases may additionally be unusual and “not meant to be taken literally” (1988: 241). Here, the approximator reading comes in. Underhill also found like in questions and in answers to questions; in both cases, he interprets it as a focuser. Neither of the readings seems to apply in examples where it functions as a hedge; likewise, Underhill is not able to provide an interpretation for like introducing entire sentences. In addition, he not only excludes grammatical like from his analysis but also like with numbers. Noteworthy, however, is his contention that like cannot be randomly placed anywhere in a sentence. Rather, its position is rule-governed. Following this statement, Underhill lists a number of syntactic environments which can follow like. Meehan (1991) takes a totally different approach. In her opinion, like has been and still is undergoing a process of grammaticalization.18 She also works through the various (sub-) functions of like, but in the chronological order of their first appearance. Thus, she begins with like in the sense of ‘similar to’ (as in “... now we’re like brother and sister”, p. 40), followed by like meaning ‘approximately’, for example as in “I wrote it in like ten minutes” (p. 41). For both senses, she found examples dating back to the 18th century, although like in the sense of ‘approximately’ was not necessarily used with numeral expression at that time. Like as a conjunctional equivalent to ‘as if ’ was already in use in the 19th century. This is a usage which she also claims to have developed
Chapter 5. Like
from the ‘similar to’ reading. Like Schourup, she then lists like meaning ‘for example’. However, not all of her text extracts illustrate the same kind of like Schourup describes. Some of the instances of like in her illustrations of ‘for example’ would not qualify as discourse markers in the sense defined in Chapter 1. She herself states that “[i]n these examples as in all of those given up to this point, like cannot be removed” (1991: 43). As I stated in Chapter 1, optionality is a criterion which I consider essential for the definition of discourse markers. In contrast, like can be omitted when its function is “to focus the listeners attention on specific information” (1991: 44). (This is listed as the fifth function of like by Meehan and is similar to Underhill’s interpretation.) The final function is like with quotations, mostly in the construction BE+like, but also as FEEL+like. While the first meanings of like in the list still are more lexical, quotative like is a form which has already become grammaticalized. The studies discussed so far all analyzed like in American English, with most of the data coming from conversations. The study published by Miller and Weinert (1995) is different in two aspects: Firstly, their data consist partly of map-task dialogs between children, teenagers and young adults, and partly of spontaneous conversations between (young) adults. Secondly, all data were recorded in Scotland. Obviously, like in Scottish English is not the same as like in American English, since the Scottish data do not support several of Schourup’s findings concerning the functions of like. Instead, Miller and Weinert describe like as “a non-introducing, non-contrastive focuser that can focus on new or given information and entities” (1995: 379) and compare its function to that of it- and wh-clefts. They also distinguish between clause-initial like and clause-final like. Concerning the former, they agree with Schourup that it can be paraphrased as ‘for example’. About the latter they write that “[t]he most general term for the role of like is ‘clearing up misunderstanding”’, countering “objections and possible false assumptions on the part of the listener” (1995: 386, 388f.). A number of publications within the last ten years view like from a relevancetheoretic perspective. Helt and Foster-Cohen (1996) investigate both like and about in the speech of fifth-graders in free conversation, guided narratives, and story-retelling, with an emphasis on quantitative analysis. They found that about tends to be followed by nominal structures, quantitative expressions and direct objects, while like tends to be followed by verbal or “other” structures and the subject of a clause, thus having a wider scope than about. Relevance-theoretic notions come into play in their proposal of an explanation for speaker choices. These choices are influenced by the fact that conceptual properties dominate in the meaning of about, whereas like has procedural meaning: “We believe that like marks the speaker’s attitude or stance toward the proposition, namely that she feels it is important in its entirety, so that the procedural message would say something like, ‘Pay special attention to the whole clause, not just the new information in it”’ (1996: 316). Relevance theory also plays a role in Jucker and Smith’s (1998) account of like and several other discourse markers. In their data, a series of conversations with predefined topics between pairs of California university students, like was, after yeah, by far the most frequent discourse marker. It was found to have four individual func-
Discourse markers in English discourse
tions in the data: as an approximator (with numerical expressions), as an exemplifier, as a hedge (which might be paraphrased as ‘kind of ’), and in quotative constructions. All of these functions have in common that they can be explained in terms of the relevance-theoretic notions of procedural information / processing instructions and ‘loose fit’. Thus, like in all its functions gives processing instructions to the addressee, namely, that s/he should expect a loose fit between the utterance and the thought it represents. Or, in other words, it instructs the addressee not to take the following too literally. Thus, like indicates a similarity relation between utterance and thought. This applies to all of the functions listed above: The speaker who uses like as an approximator indicates that s/he does not aim at any exact figure. Exemplifications by nature represent only part of what the speaker has in mind. As a hedge, i.e. when like is used with descriptions, these also are only an approximation to what would be an exact description. ‘Hedge’, in the way Jucker and Smith use this term, also includes a search for words. They explain it as an indicator of a very loose fit between utterance and thought due to the fact that the thought does not have a definite form yet. When like occurs clause-initially after “prefatory material”, as Schourup calls it, the speaker can be seen as indicating “that the following phrase or clause may not be the most successful representation of the thought to be communicated but only a sufficient approximation” (1998: 189). As a quotative, like instructs the addressee to understand the quoted material as potential utterances rather than actual words. Thus, the quotation may represent an approximate version of spoken words, thoughts, or feelings. In all cases, like introduces less than literal or exact representations which are nevertheless literal or exact enough to serve the communicative purpose, i.e. to be as relevant as required. Andersen (1997, 1998, 2000, 2001) explains the function of like in the same vein. The monograph published in 2001 constitutes a detailed examination of The Bergen Corpus of London Teenage Language (COLT). It concentrates on three specific linguistic items (innit and is it as invariant tags and follow-up questions and the pragmatic marker like), analyses their pragmatic functions and sociolinguistic distribution, and compares the adolescents’ usage in COLT with adult usage in the BNC. Andersen brings together two different frameworks, Relevance Theory and Grammaticalization; the first is used to explain the functional characterization of the linguistic items, while the second serves to explain their diachronic development. On like, he comments that the way the London teenagers use it corresponds mainly to its description in American data; British usage as found in the northern dialects (cf. Miller and Weinert above) does not occur in the COLT data. In addition to the subfunctions of like analyzed by Jucker and Smith (1998), Andersen found that like sometimes puts meta-linguistic focus on the following expression. Thus, a speaker may use like to distance him/herself from this expression, for example to indicate that it is not part of the speaker’s typical repertoire. In this sub-function, like can be paraphrased as ‘so-called’. Another group of sub-functions is not characterized by non-identical resemblance but is rather defined formally, namely like as a hesitational/linking device. Three of the four functions in this group are related to truncated utterances (false start, self-repair, cut-off without resuming); the fourth links “syntactically (sometimes even
Chapter 5. Like
thematically) unrelated structures” (2001: 255). This group of functions may also be accomplished by it’s like. In terms of social distribution, Andersen finds that the females in COLT used all functions of like except for quotative BE+like and it’s like as a discourse link slightly more than their male counterparts. The frequency of like, again with the exception of BE+like and it’s like, increases also with age, within an age range of ten to nineteen years. Beyond the description of the linguistic items he investigates, Andersen draws the conclusion that his hypothesis, namely that innit, is it, and like are undergoing a process of grammaticalization, is vindicated. In his view, the linguistic differences which he found between teenage and adult language are beginnings of language change in an attitudinal, pragmatic area. In connection with this hypothesis, he claims that these linguistic differences are stages of grammaticalization as described by Hopper and Traugott (1993). These stages again are defined in relevance-theoretic terms. Items such as like, innit, and is it first have conceptual meaning only. Then, they begin to be used with procedural meaning in addition, i.e. they are also used to give the addressee processing instructions. In a later stage, the procedural meaning becomes conventionalized, while the linguistic item can still be used with its conceptual meaning. Andersen can indeed convincingly show that the necessary conditions for this process are given; however, he cannot prove that these conditions and the particular usage of like by the teenagers will persist into their adulthood and thus accomplish diachronic change. The process of grammaticalization is most obvious in like as a quotative. There seem to be no data on when like was first used in this way, its origin or evolution (cf. Romaine and Lange 1991: 244). Meehan (1991) does not give a date to her examples of BE+like, but they seem to come from her own data. Blyth et al. (1990) call it “a new quotative in American oral narrative”; they refer to an Editor’s note by Butters, dating from 1982, as one of the first publications to deal with this quotative. We do not know how like came to be used as a quotative, but it is obvious that it has become conventionalized to a high degree. Typically, it is found in the construction BE+like, substituting a verbum dicendi. As such, it cannot be omitted without rendering the utterance it contains ungrammatical; it has a fixed position in the syntax of the utterance. It is therefore not surprising that the paper by Romaine and Lange (1991) is entitled “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress”. Further evidence of its grammaticalization is the fact that, in the meantime, the construction BE+like has also been acknowledged by at least some lexicographers; the sixth edition of the Oxford Advanced Learner’s Dictionary (OALD 2000) explains it as meaning say in American usage, while there is still no mention of it in the Cambridge International Dictionary of English (CIDE 1995). The second edition of the OED (OED 1989) does actually contain two examples which might count as instances of quotative like: “Like wow...wonderful” and “And I thought like wow, this is for me”. However, they are listed as “B. adv. (quasi-prep., conj.)” under the explanation “d. N. Amer. colloq. Followed by an adj.: in the manner of one who is. [...]Also in less analysable constructions”.
Discourse markers in English discourse
Concerning the meaning or function of like as a quotative, researchers agree that it can be used not only to introduce speech which has actually been uttered but also thoughts or feelings, as we have already seen above. Blyth et al. (1990) compare like with two other quotative verbs, say and go. They find that the ability to introduce an inner monolog, a state of mind, or a thought is particular to BE+like. In addition, BE+like differs from say in three grammatical features: it tends to go with present tense rather than past tense, with continuous rather than punctual aspect, and with the first person singular rather than the third person singular. This last feature distinguishes BE+like also from go. Romaine and Lange moreover discovered that BE+like can be used in variation with go and say to maintain the various roles in the report of some interaction (1991: 242). In an analysis of three corpora of spoken American narratives and conversations, Ferrara and Bell (1995) noticed that it’s like is used slightly differently from general third person singular BE+like. Males mainly employed it to “report the collective thoughts of a group”, and “both males and females to report a habitual style of thought or speech for one individual (1995: 278). All three studies mentioned above dealt with American data. However, as the research by Tagliamonte and Hudson (1999), Andersen (1997, 1998, 2000, 2001), and Winter (2002) shows, quotative like is spread in the speech of Canadian, British, and Australian youth as well. Tagliamonte and Hudson note that “the linguistic trajectory of the innovative form be like is remarkably parallel, not only across the British and Canadian corpora, but is also comparable with previous reports of this form in the United States” (1999: 147). Yet, this quotative is more restricted in its use outside the USA: “in contrast to American English, in both British and Canadian English be like is still highly localized, being used for non-lexicalized sound or internal dialogue and for first person subjects” (Tagliamonte and Hudson 1999: 166), while Winter reports for Australian English that BE+like “has a relatively limited distribution but appears to be grammaticalized for the third person” (2002: 6). Age generally is found to have an influence on the frequency of BE+like and also on the frequency of discourse marker like. Blyth et al. (1990) report that the group of speakers aged 20–24 used quotative like with the highest frequency by far, although it was also present in the speech of the speakers aged 27–32. Those older than 38 years did not use it at all, neither in Blyth et al.’s data nor in a corpus analyzed by Ferrara and Bell (1995) which was recorded in 1992. Andersen (2001) confirms this for the adult speakers in the BNC as well. The age groups in Dailey O’Cain’s study (2000) are separated slightly differently: speakers aged 14–29 employed quotative like frequently, speakers aged 30–49 did so rarely, and speakers over age 50 did not use it at all. For the discourse (or pragmatic) marker like, or focuser like, as Dailey-O’Cain (following Underhill 1988) calls it, the results are very similar in both American and British English (Dailey-O’Cain 2000; Andersen 2001). The results on the influence of gender are not as consistent across the various studies. Blyth et al. (1990) found that, in contrast to their hypothesis, quotative like was used more by men than by women. One year later, Romaine and Lange (1991) claimed that it was employed more by women. The data analyzed by Ferrara and Bell (1995)
Chapter 5. Like
suggest that the relative distribution changes over time: while BE+like was more common in the speech of females in the corpus recorded in 1990, the corpus recorded in 1992 showed equal frequencies of this construction between the sexes. Dailey-O’Cain’s study (2000) did not yield any significant differences between male and female either, neither for focuser nor for quotative like. However, in the perception of her participants, young women are still seen as using it most often of all speaker groups; in particular, like is associated with “Southern Californian ‘Valley Speak”’ (2000: 70). Both results, those by Ferrara and Bell (1995) and those by Dailey-O’Cain (2000) seem to fit with the general assumption in research on language change that women are the forerunners of innovative language use (cf. Andersen 2001). According to the results published by Tagliamonte and Hudson (1999), this appears to be the case for British speakers: quotative like is significantly favored by British females. For Canadian speakers, however, this is not the case; no significant difference was found between the sexes, even though quotative like is still considered innovative usage in Canada (1999: 160). Andersen (2001) reports that adolescent females from London used the pragmatic marker like significantly more, but he found no male-female difference for quotative like. In accordance with the assumption mentioned above, this might be taken to indicate that quotative like has already been established among London teenagers, while marker like has not. Finally, there are two papers testing the influence of other variables on the frequency of like and a number of other discourse markers, which might be relevant for my research. Jucker and Smith (1998) divided the participants in their data into pairs of students who were friends and students who did not know each other before the experiment. They found that like was significantly more frequent between friends than between strangers. Fuller (2003) carried out an experiment setup in which six students were recorded both in casual conversation with friends or family members and being interviewed by a research assistant. Thus, she created a speech context which differed in three dimensions: speaker role, formality of the event, and relationship between the speakers. Following Jucker and Smith (1998), she divides discourse markers in presentation markers and reception markers (cf. also Section 1.7.1.5). She argues that, in an interview context, interviewees mostly present information and do not have as much occasion to react to information they receive. In a conversation, however, presenting and receiving information are presumably balanced. Therefore, she expected a higher frequency of presentation markers such as like in the interview, despite the fact that the interview partner are strangers, assuming that the influence of the speech context overrides marker preferences which are due to the relationship between speakers. However, her assumption turned out not to be correct for like; there was no significant difference between speech contexts. In the following sections, I will examine the functions of like and their frequencies as they occur in the data from American and German speakers in the GLBCC.
Discourse markers in English discourse
Table 5.1 Functions of the lexical item like as used by German and American speakers in the GLBCC Non-Discourse Marker, Non-Quotative Functions: – (verb) – preposition – conjunction Discourse Marker Functions: – searching for the appropriate expression – marking an approximate number or quantity – introducing an example – marking lexical focus Quotative Function: – quotative like
. The functional categories of like in this study As I mentioned above, like appears in three functional groups: as a quotative; as a discourse marker; and as verb, noun, adjective, adverb, preposition, conjunction, and in fixed expressions. While quotative like and the individual functions of discourse marker like are treated in publications on linguistic research, non-discourse marker, non-quotative like is typically described in dictionaries. However, not all of its possible uses will be found in data such as the GLBCC. Like as a noun, adjective or adverb, for example, did not occur in my data and therefore will not be described below. In contrast, verbal like was very frequent, since the participants were asked whether they liked the movie. It will nevertheless be excluded from the analysis for a very simple reason: The program which returned all instances of like did not search for conjugated forms such as liked or likes. Since these forms were frequent but not counted, verbal like was abandoned from the analysis altogether. I will begin the description of the functional categories of like with non-discourse marker, non-quotative like. Discourse marker instances form the main part of this section, while quotative like will be treated shortly as the last function. All functions are presented in Table 5.1.
.. Non-discourse marker, non-quotative like Non-discourse marker, non-quotative like (excluding the verb) occurred mainly in two functions, either as a preposition expressing a comparison of some kind, or as a conjunction. Each of them has further sub-functions, as we will see.
... The preposition like When like did not function as a quotative, discourse marker or verb, it was most frequently used as a preposition. As such, it made up more than half of the instances in this function group. The elements which are compared using like can take on a variety
Chapter 5. Like
of forms. While dictionaries such as the OALD (e.g. OALD 2000) or the Cambridge International Dictionary of English (CIDE 1995) often sort the usages of like according to their meaning, I have sorted the examples below according to the syntactic elements which like connects. Each of the patterns presented occurred between twenty-five and forty times. (160)
360 he’s like a little kid, (45A, NNS)
(161)
278 it wasn’t like ha ha laughter, (4B, NS)
(162)
237 he looked like a pig. (41B, NNS)
(163)
342 B: yeah the [waiter looks] like him. (63B, NS)
(164)
54 the silver ware and anything like that. (8B, NS)
(165)
152 cause I have never seen a movie like that. (9C, NS)
(166)
216 .. and then carries on like that, (63B, NS)
(167)
659 B: of course you could see it like that. (67B, NNS)
(168)
264 but ... she like went like that, (1B, NS)
(169)
313 B: because he was= [.. going] like thi=s. (2B, NS)
(170)
85 B: so all of a sudden .. they are .. & 85a & treated like third class people, (48B, NNS)
(171)
662 he’s dressed like a lumberjack. (63A, NS)
In one of the frequent patterns, like compares two elements expressed in a personal pronoun or a noun phrase (extracts (160) and (161)), so that the pattern has the form NP/Pron BE like NP. Speaker 45A compares Chaplin (he) to a little kid, while speaker 4B tries to describe her reaction to the Chaplin movie. Both use a pronoun before like and a noun phrase after it. In another pattern, the auxiliary BE is substituted by LOOK, as shown in extracts (162) and (163). Again, we find pronouns instead of noun phrases, both before and after LOOK like. However, the variant in (162) occurs most often, since it is typically Chaplin (he) who is compared to something or somebody. This pattern is very frequent in the speech of the German participants; the five native speakers who use it do so not just once but several times.
Discourse markers in English discourse
The noun phrase after like in such a comparison can be substituted not just by a pronoun but also by a demonstrative, as we see in extracts (164) and (165). Typically, this demonstrative is that, even though this is possible, too. Instead of a noun phrase, speakers also tend to employ anything and whatever as pronouns before like. Like does not only compare things or people but also actions. However, it is rare to have a verb on either side of like. In most of the cases, like is preceded by a verb phrase and followed by one of the demonstratives this and that, as in extracts (166) and (167). That tends to stand for complex propositions. Speaker 63B first describes how Chaplin rejects the artist’s offer to pay Chaplin’s bill, and then says and then carries on like that, referring to Chaplin’s actions. That in the data of speaker 67B (extract (167)) refers to an interpretation of the movie which sees it as being critical of capitalism; it in line 659 stands for the movie itself. That and, much more often, this are frequently accompanied by gestures and demonstrations of what the speaker refers to with the demonstrative pronoun, since in many cases this is obviously the easiest way of conveying what happened in the movie. Speaker 1B, for example, tries to explain how the young lady behaved when she met Chaplin in the restaurant, a behavior which includes head and hand movement and facial expression (extract (168)). Speaker 2B explains why she knew that the customer who offered to pay for Chaplin’s bill was an artist: he was making hand movements (plus head movements and facial expression) as if he were drawing something onto an invisible canvas (extract (169)). Last but not least, we find instances in which like is preceded by a verb phrase and followed by a noun phrase. These are illustrated in extracts (170) and (171). This type of comparison was particularly frequent in the data of the German speakers and rather infrequent in the native speakers’ data. The most conventionalized form of a comparison with prepositional like certainly is the expression something like that, with its variants something like this and stuff like that/this. This expression occurred so frequently that it seemed justified to treat it separately in the quantitative analysis. On average, measured against the total number of words per group, the German speakers used it slightly more often than the native speakers, and only they (but not the natives) used the variants with this instead of that.
... The conjunction like Like as a conjunction is generally considered informal (Cobuild 1987; OALD 2000), colloquial (COED 1976; Webster’s 1994), or non-standard (OALD 1974; CIDE 1995). The Oxford Advanced Learner’s Dictionary in its third edition (1974) comments on like replacing as that it is “common use among those who have not been taught to avoid it; considered incorrect, but found in many good writers”, and its fourth edition (OALD 1989) attributes like replacing as if to US American usage. The two types of the conjunction like are illustrated in extracts (172) through (176). Like in the function of as if is by far the more common one. It occurs repeatedly with SEEM, as in extract (172), or with LOOK, as in extract (174). The last two examples ((175) and (176)) show like replacing as.
Chapter 5. Like
‘as if ’: (172)
114 and uh ...(1.1) it seemed like she .. & 114a & liked .. Charlie Chaplin, (31A, NNS)
(173)
91 and uh .. even Charlie Chaplin like of & 92 & occasions .. (H) made ... an eMOtion like & 92a & he was going to get sick. (55A, NS)
(174)
13 cause it looks like her mother is sick, (134A, NS)
‘as’: (175)
237 B: ...(2.3) yeah they weren’t so advanced & 237a & like today, (27B, NNS)
(176)
122 C: .. like I said I know why it’s famous now. (121C, NS)
The quantitative results for non-discourse marker, non-quotative and non-verbal like are summarized in Figure 5.1a. The German speakers seem to be familiar with using like as a preposition in a variety of comparisons (PCN); their data show a particular preference for the sub-type something/stuff like this/that (SLT). In contrast, relatively 0.18
avg. % of total words per group
0.16 Germans Americans
0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 PCN°
SLT°
CJT
Figure 5.1a Distribution of like as preposition and conjunction – American vs. German speakers (the circle ◦ indicates that there is no statistically significant difference)
Discourse markers in English discourse
few Germans employ like as a conjunction, while this function occurs almost 0.1 times in 100 words in the native (US) speaker data.
.. Discourse marker functions of like ... Searching for the appropriate expression (177) 50 B: and so um .. he’s all scared, 51 trying to find the money, 52 and then .. all of a sudden this .. like a= & 53 & .. guy who just found th- .. the money also, (5B, NS)
Like the discourse markers well and you know, like can also be used when the speaker is “thinking about what to say next” (Cobuild 1995; Macmillan 2002) or is searching for the appropriate expression, as extract (177) shows. Speaker 5B obviously has difficulties finding a fitting term to describe the male person who enters the restaurant and plays a role for the next scene. like here is preceded by a pause and followed by a lengthened article plus pause. Schourup (1985) describes this type of like as “pausal interjection” and proposes a reason for the usage of like in such places: “A speaker saying like during a pause to formulate a continuation subtly suggests a reason for the pause: the material about to follow is difficult to formulate appropriately or precisely” (1985: 53, 56). Jucker and Smith (1998: 187f.) report similar instances. However, despite the fact that they describe the speaker’s linguistic behavior as “he searches for the right word” (1998: 187), they prefer to see like as an indicator of a very loose fit between the speaker’s thought and his utterance “because the thought itself is at this stage fairly fuzzy in the mind of the speaker” (1998: 188). In the data of the COLT, Andersen, too, found “a more general hesitational use [of like] which signals that ‘I have something on my mind, but I don’t know how to put it”’ (2001: 249), but apparently these instances occurred mainly with false starts and repairs. Unlike Andersen, I do not consider the search for the appropriate expression and false starts or repairs to belong to the same category. When an instance of like is taken as a marker of search in my categorization system, this is done only if there are other hesitational markers involved, too, such as filled and unfilled pauses. In contrast, the impression of ‘search’ does not have to be present at an instance of false start. While you know occurs in a number of such cases, I found like in my data only six times with false starts, i.e. directly following the truncation of an intonation unit and preceding a new start. Three of these six cases contain corrections of a word or phrase already uttered, the other three are mere repetitions. All six will be found in the category ‘Various functions’ (Section 5.2.2.6). Let me present some more examples of what I have categorized as like indicating the search for the appropriate expression. Extracts (178) through (180) show that the notion of searching for words is strongly present, even though not everybody who uses like for that purpose finds the right words. Speaker 22A, for example, starts expressing
Chapter 5. Like
his opinion about the movie in general. Then he searches for words; as he does so, it does not become entirely clear whether he tries to think of what he could say about the movie or whether he has something definite in mind which he cannot express. In any case, he does not finish his effort and continues, after a long pause (line 280) with a particular scene, which he compares to another movie he has seen. (178)
277 278 279 280 280a 281 282 283 (22A,
A: B: A:
B:
it was OK. yeah it was OK. I really do mean .. like .. that um-...(2.7) the rocking of the boat was worse & & than the Blair Witch Project in making & & you wanna [.. get sick]. [oh really]? I didn’t see the Blair Witch.
NS)
The German speaker 206C in extract (179), in contrast, knows exactly what she wants to express: the gestures with which Chaplin ordered a long French loaf of bread. To find the English words for what she has in mind, however, presents itself as a somewhat difficult task (line 76). I said above that, for like to be classified as an indicator of search, at least one other hesitational marker had to be present. Such markers can be filled and unfilled pauses, lengthening, truncated words, or hedges such as kind of . (179)
72 .. (H) but anyway he assumed he had the money, 73 went into the .. restaurant, 74 and ordered beans and a flute, 75 .. um .. it was pretty funny how he ordered it, 76 cause he= .. s- he made like .. playing the & 76a & flute. 77 he’s .. he’s pretending to playing the flute, 78 and that’s how he ordered the .. French stick. (206C, NNS)
Extract (179) shows that these markers did not have to occur in the immediate neighborhood of like, but most often they occurred at least in the same intonation unit. In extract (180), the description of the artist’s initial behavior (who, at the sight of Chaplin and his lady-love, started to draw them on invisible paper with an invisible pencil) illustrates how pauses, like, the hedge kind of, a self-correction (from doing to going), and very general expressions (going like this) work together to create a strong impression of searching for words. (180)
170 (H) an’ he started .. like kind of doing=-171 you know going like this .. with his .. & 172 & fingers you know like th-(10B, NS)
Fifteen of the 77 German speakers used like up to three times when they were searching for the appropriate expression, creating an average of 0.024 instances per 100 words in the German speaker group. Of the American speakers, slightly more than half em-
Discourse markers in English discourse
ployed like in this function, but several of those who did, used it more than once. The speech of one American even contained twelve instances. On average, this amounts to 0.1 instances per 100 words in this speaker group, with a clear statistically significant difference from the Germans.
... Marking an approximate number or quantity There is not much debate about the function of like when it is used with numeral or quantitative expressions. Even though Underhill (1988: 234) excludes this function from his analysis because he concentrates on a different kind of like, he clearly acknowledges it as meaning ‘approximately’. Schourup found that it occurred frequently in the data he analyzed (1985: 38). It is also included in the treatments by Meehan (1991), Jucker and Smith (1998) and Andersen (e.g. 2001). Like preceding a numeral expression indicates that the number is not an exact one; thereby, it reduces the speaker’s commitment to the literal truth of his/her utterance.19 There are several examples of this in my data: (181)
196 and they go back and forth like two & 196a & or three times, (8B, NS)
(182)
186 (H) he got like I guess ten cents back, (125B, NS)
(183)
8 ...(1.4) because it’s like his eleventh comedy, 9 .. out of the twelve, (141C, NS)
(184)
96 like five big guys. (206C, NNS)
Speaker 8B in extract (181) describes the scene in which the artist takes Chaplin’s bill and offers to pay for it. Chaplin takes it back and rejects the offer, the artist repeats the offer, Chaplin rejects it again. At that stage of the movie, no spectator counts exactly how many times the artist makes his offer. Therefore, in using “like two or three times”, the speaker gives the hearer an idea of what was going on without committing herself to the exactness of her utterance. The same is true for speaker 125B in extract (182). She refers to the change which Chaplin gets back after paying his bill. The spectator of the movie can see that the change consists of small coins, but does not get to know about the exact amount. Speaker 141C makes a comment about Chaplin’s work at the beginning of his narrative. He knows that Chaplin made twelve comedies, and he apparently thinks that The Immigrant is his eleventh one. Like indicates, however, that he is not quite sure about that; it could also perhaps be the tenth comedy, or the twelfth, or any other of the twelve. Thus, it again reduces the speaker’s commitment to the exactness of the utterance. Extract (184) shows that like as a marker of approximate numbers is also known among the German speakers. Here, speaker 206C refers to the scene in which a number of waiters come and beat up a customer who has not fully paid his bill. As with the scene in which the artist offers to pay the bill, normally no
Chapter 5. Like
spectator counts how many waiters are involved in the beating. Therefore, the speaker does not know for sure and adds like to the number of waiters. Usually, like in this function is described in the literature as preceding numbers. However, Meehan (1991: 40) uses the term “quantity phrases”, which potentially includes more than just numbers. Following her, I applied the present category to cases in which like precedes expressions of quantity, time period, or frequency which do not contain a cardinal or ordinal number. One example is the expression of monetary value. Speaker 5B (extract (185)) talks about like a silver dollar, which describes the approximate value of the coin which Chaplin found in the street. Similarly, Chaplin gave like a penny to the waiter, according to one German speaker (201B). Another type of expressions refer to time periods, such as the like .. twenties (extract (186)). This is roughly the time period for which the outfit and make-up of the characters in the movie, especially of the men, are typical, in the opinion of speaker 63B. It could also have been like erm in the forties or perhaps thirties, as a German speaker (68B) assumed in the discussion part. Both speakers additionally express their uncertainty about the correct time period, speaker 63B by adding or whenever it was, speaker 68B by adding I don’t know. And another native speaker completely avoids committing herself and simply explains that it was such .. like a long time ago. In all of these cases, the exact value or time period does not matter; it suffices for the present purpose of the speaker if the hearer gets a rough idea of it (i.e. of the value or time period). A third type of expressions has to do with frequencies in the widest sense. Speaker 5B first described the funny details of what happened in the restaurant, in the second part of the movie. When he, in extract (187), says that it was funny like the whole time, we can reasonably assume that there may have been moments in which he did not laugh. However, for him the restaurant scenes felt funny throughout. Speaker 134, in contrast, expresses a frequency of zero (extract (188)). She has like never seen a silent movie. Yet, the like tells the hearer that the term never is not quite correct. As the speaker herself adds a few lines later, the never is true only under certain conditions: she has never watched a complete silent movie, paying full attention as she did in the experiment. Since she does not know how long the movie was, she again uses like plus the rather indefinite expression such a long time (lines 417f.). (185)
17 I guess it was like a silver dollar & 17a & on the f=loor. (5B, NS)
(186)
652 like .. twenties or whenever it was. (63B, NS)
(187)
307 B: but like the whole time it was funny, (5B, NS)
(188)
412 412 (...) 416 416a
I’ve like never seen like a silent & & film [really like]-I I’ve seen like you know bits and & & pieces of ’em,
Discourse markers in English discourse
417 but I never like had actually watched & 417a & one for like-418 (H=) .. such a long time, (134A, NS)
These last two examples are somewhat different from the previous ones. Typically, like is used before numerical or quantity phrases to reduce the commitment of the speaker to the utterance and because the exactness of the information is not important for the speaker’s purpose. In extracts (187) and (188), however, like is used together with terms expressing the extreme ends of the frequency scale to convey the feeling of the speaker. Since the feeling of the speaker is emphasized rather than the quantity phrase, these cases can be considered peripheral cases of like marking approximate numbers. Even though Schourup found this function of like to be a frequent one in his data (1985: 38), it belonged to the least frequent categories in my data. Like preceding numeral or quantitative expressions made up only five percent of all discourse marker instances of like. Of these five percent, a quarter (nine instances) were found in the data of the German speakers, while the remaining three quarters came from American native speakers. As is the case in most categories, the difference here also is statistically significant.
... Introducing an example The OED in its second edition (OED online) gives one entry for like which says, “In mod. use (with following dat.) often = ‘such as’, introducing a particular example of a class respecting which something is predicated”. Illustrations for this usage are “A critic like you [...]” and “A birth like that of Keats [...]”. The illustration given in the Oxford Advanced Learner’s Dictionary (OALD 1989) for the preposition like meaning ‘for example’ is slightly different: “We could look at some modern poets, like Eliot and Hughes”. It coincides with the sample sentence we find in Webster’s New World Dictionary of American English for the same function of like: “fruit, like pears and peaches, for dessert”. All four examples have in common that like is syntactically immobile and cannot be omitted without rendering the sentence ungrammatical. However, like can also introduce examples as a discourse marker, as the following extract shows: (189)
477 B: uh the minor characters would be .. & 477a & like the guys on the boat I guess, (22B, NS)
Here, the speaker starts discussing the question “What did you think of the acting by minor characters?” In order to talk about their acting, she first has to say who are minor characters in the movie. Instead of presenting a definition or a list of all the characters who would count as minor, she gives an example, like the guys on the boat. There are certainly more minor characters in the movie, but the hearer now has an idea of whom she considers relevant to the question. In her sentence, like is grammatically optional, i.e. the sentence would still be well-formed without it.
Chapter 5. Like
The usage of like as an exemplifier has been widely acknowledged in the literature. Schourup presents a considerable list of extracts in which like “appears to have the meaning of ‘for example”’ (1985: 48ff.). Miller and Weinert, who take Schourup as their main reference for the functions of like, confirm that “Schourup’s analysis of LIKE as equivalent to FOR EXAMPLE is appropriate” also in their Scottish data (1995: 388). Illustrations of exemplifying like are also found in Meehan’s (1991) account; however, she does not distinguish between discourse marker and non-discourse marker uses. Her list of example sentences starts with one in which like is syntactically obligatory (“You go through a text like a Shakesperean (sic!) play”, 1991: 42); it continues with examples in which like can be omitted and which are similar to Schourup’s examples. From a relevance-theoretic perspective, exemplifying like, as other types of like, can be explained as indicating a discrepancy between the speaker’s thought and his/her utterance, as Jucker and Smith argue (1998: 184f., 188): The phrase preceded by like represents just one of the items or concepts the speaker has in mind, and not his/her whole thought. A similar line of reasoning can be found in Andersen (2001: 236). In my data, exemplifying like occurs 101 times. Two variants can be distinguished: like introducing an example of a general concept, and like introducing an example without a general concept being mentioned. The extract (189) above illustrates the first variant; the guys on the boat are examples of the concept minor characters. In extract (190), the speaker also talks about certain types of characters. He argues that he could predict the roles of the characters because they were kind of prescribed. A little bit later in the discussion he explains that the roles or characters he refers to are predictable because they have to frame and contrast the main character, Chaplin. One of these characters in speaker 55A’s opinion is the big guy with .. boots. Since this guy is only an example for the type of entities he has in mind, he prefaces line 460 with like. (190)
457 458 459 460 461 (55A,
(191)
670 because it’s so many little plots, 671 [like] (H) he’s in the restaurant fighting. (142B, NS)
(192)
361 (H) so they had to act extremely like & 361a & .. starting to vomit, (51A, NNS)
.. I XX <X predict X> the um-... the each of the roles of the characters. they are kind of prescribed. (H) uh like the the big guy with .. boots, .. uh on the boat, NS)
Extracts (191) and (192) illustrate more cases in which the example introduced by like is preceded by a general term or description. As we can see, these general terms or descriptions do not have to refer to characters of the movie, nor do they have to take the form of a noun phrase. Speaker 142B in extract (191) expresses the difficulties she had in understanding the movie because there were no sub-titles (and only occasional inter-titles) and because the movie consisted of so many little plots rather
Discourse markers in English discourse
than one line of action. One of these little plots is the scene in the restaurant where Chaplin (he) is fighting (line 671). It exemplifies the kind of entities the speaker has in mind; the description of this scene is therefore prefaced by like. Speaker 51A in extract (192) also has a general concept in mind, but she does not express this concept in the form of a noun phrase. Rather, it is a general description: they had to act extremely. Acting extremely for her would be, for example, that people were starting to vomit, an exemplifying description which she introduces with like. General concepts expressed in noun phrase and in descriptions, referring to characters, scenes, and other entities and ideas all occur in native and in non-native speech. However, examples introduced with like do not have to be preceded by the explicit mention of a general concept. Often, we will find that the speaker has such a concept in mind, but only expresses the example. In these cases, the role of like becomes even more important. Andersen (2001: 236) notes that the function of like here is “to suggest that the following noun phrases are to be construed as exemplifications of wider categories”. Extract (193) presents a typical case: (193)
152 you know he started to order like coffee & 152a & and other stuff. (22B, NS)
Within the restaurant frame, ordering things belongs to the licensed/typical activities (cf. Schank and Abelson 1977; Sanford and Garrod 1981; Smith et al. 2001). Also, the kinds of things a customer can order in a restaurant are limited. An additional limitation is set through the context of this utterance. Speaker 22B has told her partner that Chaplin (the referent of he) had just watched how another customer was beaten up because he did not have enough money. Charlie found that he did not have any money and became nervous. In order to delay leaving, he started to order like coffee and other stuff. A cup of coffee does not cost very much, and nobody would expect a person without money to order expensive food. When the speaker in extract (193) utters lines 152f., she can assume that her partner has these limitations in mind. Thus, he will be able to infer the wider category for which speaker 22B gives coffee as an example, namely “things which you can order in a restaurant and which are not costly”. (194)
274 .. they didn’t show like a ring or anything, (1B, NS)
(195)
678 A: [[sometimes it’s] like confusing, 679 so you need like dialog there. (0) 680 B: (0) yea=h. 681 A: even like .. subtitles will work but [still]-(142A, NS)
Extract (194) follows a discussion of the participant pair about a particular scene which 1B did not quite understand. Chaplin meets the lady; they are both happy to see each other again, but at one point the lady looks down and both seem sad. Speaker 1B assumes that she is married, but they didn’t show like a ring or anything. Even though
Chapter 5. Like
the speaker does not explicitly say it, the hearer can easily infer that a ring here is an exemplification of anything which may signal that the lady is married. Speaker 142A in extract (195) had difficulties understanding everything in the movie because it sometimes was confusing for her. In order to understand better, she would have preferred to have speech in the movie. Obviously, she thinks of spoken language when she uses the term dialog in line 679. But if spoken language is not possible, written language such as subtitles (line 681) would also be helpful. Subtitles here stands for any kind of language which is not spoken, or at least not of the kind which she has in mind when she uses the term dialog. Unlike the speakers in the extracts above, who add and other stuff (extract (193)) and or anything (extract (194)), speaker 142A does not give any additional clue besides like that the expression following it is an example, one out of several possibilities. Nevertheless, this is made perfectly clear through the use of like. American and German speakers used the same number of like to introduce an example. However, most of the instances of like in the German data follow a general term or description (39 out of 45), while such cases account for less than half of the instances in the American speakers’ data. Not all speakers employed like as an exemplifier; only two thirds of the Americans did so and one third of the Germans, though up to five times per speaker in each group. Measured against the total number of words per speaker group, the Germans used exemplifying like about half as often as the Americans. This average difference also achieved statistical significance (p = 0.013).
... Introducing an explanation (196) 190 B: ... they would show
the movie, 191 and then there would actually be someone & 191a & playing the piano there. 192 like a=long for the ac[companiment]. (6B, NS)
The discourse marker like has been studied in considerable detail, not only in intuition-based examples but also in corpus data (e.g. Schourup 1985; Underhill 1988; Meehan 1991; Miller and Weinert 1995; Andersen 1997; 1998; Jucker and Smith 1998; Andersen 2000, 2001). Yet, none of these researchers describes a function of like which seems to fit its usage in extract (196). Speaker 6B gives his partner information on the context of movie presentations at a time when the movies were still silent ones. While the film was shown, there was somebody in the theater who accompanied it playing the piano. The clause following like in line 192 can hardly be said to be new information, nor is it particularly focused information (cf. Underhill 1988; Miller and Weinert 1995). The speaker is not unsure about what he says, nor does he want to hedge it or indicate that it does not exactly represent what he had in mind (cf. Underhill 1988; Schourup 1985; Jucker and Smith 1998; Andersen 2001). Rather, he extends the information given in lines 191 and 191a to make it more understandable; when he said that somebody in the theater was playing the piano, this meant that the music accompanied the movie instead of being played before or after it. In other words, he gives an
Discourse markers in English discourse
explanation of the information. Like in this function, i.e. introducing an explanation, is by no means infrequent in my data. It occurs 137 times in the speech of the American and German participants. The following extracts ((197) and (198)) present some more examples. In the scene described in extract (197), Chaplin makes boxing movements. Previous to the scene he witnessed how another customer was beaten up because he did not have enough money to pay his bill. When Chaplin then checks his pockets, he finds a hole in his pocket through which he must have lost his money. Therefore, he already expects to be beaten up himself and prepares for the fight by practicing boxing moves; he’s doing all this stuff with his hands, as speaker 123B puts it (line 267). Realizing that this formulation may not be precise enough, she then adds as an explanation like getting ready to be beat up. (197)
263 before that he was ... thinking that & 263a & he was gonna get beat up. 264 A: ... oh [.. yeah yeah]. 265 B: [cause he had no money]. 266 <SV so he’s ... like SV> .. pretending & 266a & to punch, 267 and he’s doing all this stuff with his hands, 268 like getting ready to be beat [<@ up @>]. (123B, NS)
(198)
513 B: I saw that-514 I think I saw that real room, 515 that they were in, 516 ... the the tumbling room that they had? 517 A: oh yeah? 518 B: yeah like on on this-519 ... XX channel or something, 520 like they really showed the Chaplin room, 521 .. like how they were really tumbling. (143B, NS)
Extract (198) is preceded in the recording by the participants going through all of the discussion questions on the instruction sheet. After a long pause and speaker 143A’s statement so that’s it, Speaker 143B somewhat unexpectedly starts again with line 513 in the extract. On TV, he saw the room in which some of the scenes were filmed. In the part of the recording following the extract, he refers to one scene on the boat in the dining room where a lady rolled on the floor, Chaplin tripping over and rolling along with her. In this and other scenes from the first part of the movie, the floor moves as if the boat were rolling in a heavy sea. For speaker 143, it is clear that this was done in what he refers to as a tumbling room. In the course of his turn, however, he realizes that his initial utterance (lines 513 through 516) needs to be expanded in order to become understandable for his partner. Three times in extract (198), he uses like to introduce an explanation. The first like (line 518) precedes an explanation of B’s I saw (lines 513 and 514): He did not see the real room directly but on TV. The second
Chapter 5. Like
like (line 520) introduces an explanation of what he means by the real room, namely the Chaplin room. This expression is still opaque, but together with the context, it at least contributes to a better understanding. The third like (line 521) precedes the explanation that “they” (i.e. the producers of the TV program, cf. Smith et al. 2001) did not just show the room in which the scene was filmed but how the impression of a rolling boat was effected. Despite the fact that tumbling in this context may not be clear by itself, speaker 143A does not seem to be irritated by this term. The partners do not speak about the rolling of the boat at any other time of the recording, but it is possible that they discussed the question of how this effect was produced while they watched the first part of the movie together. Similar instances of like can be found in the data of the non-native (i.e. the German) speakers. Thus, speaker 45B describes the behavior of a man as he’s like <WH o=h WH> (WH= whispering voice quality); in the next intonation unit she explains this description as like looking as if he couldn’t believe it. Similarly, like in this function occurs several times when the participant who has seen the second half of the movie describes Chaplin’s eating manners. Speaker 55B, for example, tells her partner, he= eats them with the knife (them = beans), and then explains how he does that: like (H=) shoves them off. Two particular sub-types occurred several times in my data, though not with any considerable frequency. Therefore, they have been included in the category ‘introducing an explanation’. In one of them, like introduces a repetition of something the speaker has said before. This is done in almost, but not exactly the same words. Extract (199) illustrates such a case. The speaker, 1B, refers to the beginning of the second half of the movie, where we first see Chaplin walking along a town street. An inter-title is then shown which reads Later ... broke and hungry. Speaker 1B obviously does not remember the exact wording of the inter-title, but she remembers its content. Line 20 is an attempt at reproducing the words of the inter-title. Like (line 22) introduces the explanation that, whatever the exact wording may have been, the inter-title conveyed that Chaplin was poor and hungry. Like here converts what would sound like a mere repetition into an explanation. As a side effect, together with the change in the formulation, it indicates that the speaker distances herself to a certain degree from the previous line she uttered. However, this distance is not a characteristic of like introducing an explanatory reformulation, as extract (200) shows. (199)
20 it said like poor and hungry later or whatever, 21 A: [mhm], 22 B: [like] that he was poor and hungry, (1B, NS)
(200)
216 216a 217 217a (6B, NS)
and all of the sudden it went to & & a shot at the dining room, .. you know like this random shot & & [in the dining] room,
Discourse markers in English discourse
After the dinner scene in the first part of the movie, the following scenes are set on the deck of the boat. However, in between, a man is shown very briefly who is still sitting in the dining room, eating the leftovers. Speaker 6B first refers to this as a shot at the dining room, which then becomes this random shot in the dining room. You know here urges B’s partner to recognize this as part of their shared knowledge, while like marks the reformulation as an explanation, an explanation which is intended to help B’s partner to remember this shot. Such an explanatory reformulation does not necessarily involve a repetition of words; it may also be a reformulation of content. Both forms are found in native and non-native data, even though the following extract comes from a non-native speaker. Speaker 45A summarizes his opinion about the movie: (201)
266 267 268 (45A,
...(2.0) I I’d rather .. say it was & & neutral for me like not goo=d, not not really bad, NNS)
Obviously, he feels that neutral is not a typical term for expressing an opinion about a movie. Accordingly, he immediately adds an explanation of what he means by neutral: not good, not really bad. This explanation is introduced by like. Even though 45A does not repeat any words, the formulations following like express the same content as the formulation before like. This leads us to the second sub-type within the present category. It may be described as like introducing an alternative term. A typical example relates to money (extract (202)). At several points in the movie, we see characters handling coins: Chaplin finding a coin in the street, the beggar shining a coin in the restaurant, the artist leaving a coin as tip, Chaplin receiving several coins as change. We are never shown the coins in detail; however, from their size and from what can be purchased with them, we can guess at their kind and value. Thus, speaker 4B in extract (202) refers to the coin which the artist leaves as a tip as a big coin at first. Then she adds, like a silver coin. Big here does not only refer to the size of the coin, but also to its value. The alternative term silver coin, introduced with like, clarifies this. (202)
141 and then .. the .. artist takes his & 141a & change and leaves a big coin, 142 like a silver coin, (4B, NS)
(203)
103 104 105 (44B,
(H) and then he um spots a coin on the ground. so like a, like a .. five dollar piece or something, NS)
Extract (203) is a particularly intriguing case. The speaker, 44B, is an American participant who was recorded in Giessen, where he spent a semester or two as an exchange student. Line 104 shows how much he has been influenced by his surroundings: His first description of the coin Chaplin finds in the street is given in German. The alter-
Chapter 5. Like
native term he then chooses as an explanation (preceded by like) is an almost literal translation into English (line 105). (204)
191 B: what was your favorite part of it? 192 like the funniest part? (27B, NNS)
A final example for the ‘alternative term’ function comes from a German speaker (extract (204)). Speaker 27B asks her partner about her opinion of the movie. At first she wants to know which was her favorite part. Then she decides to substitute funniest for favorite. In the context of a discussion about this Chaplin movie, these two terms can be considered as expressing the same idea; the part which the speaker’s partner likes best would most probably be the part which was the funniest in her partner’s opinion. However, funniest is less ambiguous and more specific than favorite. Therefore, this alternative term, introduced by like, also clarifies and explains what the speaker wants to know. Like introducing explanations, including reformulations and alternative terms, was used significantly more frequently by the American speakers than by the German speakers (on average 0.177 vs. 0.045 times per 100 words). As exemplifying like, it was rather unevenly distributed among the native speakers. Some did not use it at all, while others used it up to twelve times. A similar statement can be made of the German speaker group: Most of those Germans who employed explanatory like did so more than once (though only up to six times). However, the majority of the Germans (58 out of 77) did not use it at all.
... Marking lexical focus We find the idea that like serves to focus on the following information in several of the publications presented at the beginning of the chapter. Underhill (1988) sets out to demonstrate that nonstandard (i.e. non-grammatical) like “functions with great reliability as a marker of new information and focus” (1988: 234). ‘New information’ and ‘focus’ coincide in the definition he adopts, which says that ‘focus’ is “the most significant new information in a sentence – often, the point of the sentence” (1988: 238). He convincingly shows in the first half of his paper that like in his examples precedes the focus, i.e. new information or the point of the sentence. In his discussion of further (sub-)functions of like, he argues that like can set off unusual notions. In these cases, “the speaker is distancing himself from the focussed segment” (1988: 241). Miller and Weinert (1995) call like a discourse organizer which has a noncontrastive highlighting/focusing function. As such, they claim, its task is comparable to that of wh- and it-clefts, with the difference that like is not deictic. Like may focus on new information, but it may also focus on given information. In this respect, Miller and Weinert disagree with Underhill (1988). They also present an analysis of the (additional) function of clause-initial like, which Underhill could not explain satisfactorily (1988: 244): “clause-initial like is concerned with the elucidation of pre-
Discourse markers in English discourse
vious comments, whereas clause-final like is concerned with countering objections and assumptions” (1995: 392). For Andersen (2001), focusing is one of several functions of like. Most of its functions have to do with non-identical resemblance between the linguistic expression and the thought this expression is meant to represent. Thus, like marks what Andersen calls “loose use”. As a focuser, however, like does not indicate a discrepancy in conceptual properties of thought vs. utterance. Rather, the speaker uses it to express a certain distance from the linguistic form s/he uses following like: “like can be construed as a signal that the chosen expression does not fit readily into the linguistic repertoire of the speaker, i.e. that the speaker feels a minor discomfort with its use” (2001: 243). Therefore, Andersen calls this a meta-linguistic focus. Such a usage is particularly plausible in the speech of adolescents, who do not use adult or formal language readily yet without feeling some discomfort. This seems to be the case for the speakers of The Bergen Corpus of London Teenage Language, Andersen’s data source. Other descriptions of focuser like are to be found in Meehan (1991: 44) and Dailey O’Cain (2000). Meehan includes focusing like as one of several functions of like, with reference to Underhill (1988). Dailey O’Cain writes that Romaine and Lange (1991) “distinguish two functions of like in English” (2000: 60): like as a quotative and like as a non-contrastive focuser. However, this is a misrepresentation of Romaine and Lange’s paper. They refer to Underhill (1988), among other authors, and contradict him to a certain degree: “[W]e will argue that the meanings of ‘approximative’ and ‘similarity’ as well as the focus function have contributed to both the discourse uses of like” (1991: 245). “both” in this quotation refers to like as a quotative and like as a discourse marker, as they term it. Thus, ‘focusing’ is just part of the functions of like. In my data, instances of like occurred such as the following: (205)
166 (...) 173 174 (10B,
(Hx) .. (H) a=nd there was this uh MAN, I thought it was-.. he was like a movie director or something. NS)
Speaker 10B makes a statement about the assumed profession of this man (who later offers to pay Chaplin’s bill in the restaurant). Like marks the following noun phrase, movie director. This is not a very precise description of the man’s job (an inter-title identifies him as an artist), and the speaker seems to be aware of that (or something). Thus, like may indicate that the following expression does not exactly present the facts. The notion that the speaker distances herself from the expression, as Andersen (2001) found it in his teenage language data, is not present. The same applies to extract (206): (206)
130 they walk around like a corner, (138C, NS)
In this extract, they refers to Chaplin and his lady-love, who, at this point in the movie, have just left the restaurant and said good-bye to the artist. Line 130 describes precisely what they are doing then: They walk around the corner of a house. In principle, there
Chapter 5. Like
would not be a need to mark a corner as imprecise; however, a number of students mentioned in the discussion part that they found it difficult not to miss any details in this movie. Thus, it is possible that the speaker was not sure whether she remembers this detail correctly. Another potential explanation lies in Underhill’s suggestion that like marks new information. In that case, we might find like in almost every intonation unit during the narrative part, since only repetitions within the narrative or explanations (which may also be marked with like) are not new for the hearer. For some speakers, this seems to apply (one American used like in this manner 47 times), while others employed this function only occasionally. As we have seen above, we cannot always tell whether a speaker felt unsure about the expression marked with like. In the application of the functional category system, it turned out to be very difficult to apply a distinction of subtypes consistently. However, it is clear that like in the cases presented above highlights the following expression, whatever the reason may be. It should also be noticed that the scope of like in this function is a noun phrase in extracts (205) and (206). As extracts (207) through (209) show, like can also focus on verb phrases and on adverbs and adjectives. However, it never has a full clause in its scope. Therefore, I have called this function of like ‘marking lexical focus’. (207)
268 (H) and he like .. picks her up, 269 .. a=nd takes her in there. 270 (H) and u=m .. that was like the end. 271 then they play .. like music. (8B, NS)
(208)
24 ... um he went into a restaurant, 25 it was like a fancy restaurant, (1B, NS)
(209)
677 and the axe since .. like .. & 677a & surreptitiously disappeared. (63A, NS)
The speaker in extract (207) describes Chaplin’s attempts to “persuade” his lady-love to marry him, or, more precisely, to enter an office through a door with the sign “marriage licenses” on it. The first like highlights a verb phrase which portrays Chaplin’s rather unusual action (line 268). Since these actions are not the speaker’s own actions, there is no need for distancing herself from them (cf. the quotation from Underhill above, 1988: 241). The second like, line 270, puts the end into focus, which for many participants came as a surprise. Finally, the third like marks the noun music. This seems to be a case of an imprecise expression; significantly, the music which is played at the end of the movie is not just any music, but a rendering of Felix Mendelssohn-Bartholdy’s wedding march. Speaker 1B in extract (208) tells her partner that Chaplin went into a restaurant (line 24) after he found some money in the street. Then she gives additional information about the restaurant: it is not a cheap one, but like a fancy restaurant. Again, it is not possible to tell whether she considers the term fancy to be imprecise, or whether
Discourse markers in English discourse
she sees the need to highlight or focus on it because it plays an important role for her narrative. (She later describes Chaplin as having bad manners.) Speaker 63A (extract (209)) refers to a curious difference in the scenery between two shots of the same scene. Chaplin is shown playing first cards, then dice games with some other men. In the background, you can see an axe hanging on the wall. When Chaplin repeatedly wins in the games, one of the players gets very angry and threatens Chaplin. The wall again comes into view; the axe is no longer hanging there, but it does not come into play either in the rest of the movie. It like .. surreptitiously disappeared. Here, like mainly highlights the adverb and to a lesser degree the verb. The extract is taken from the discussion part; the speaker remarks on his observation because, when he was watching the movie, he thought the angry player would use the axe against Chaplin. Instead, it disappeared. Here, like reinforces the impression that the speaker found this change in the scenery very curious. Let me finally add two more extracts from German speakers. (210)
172 (H) when they like .. take hands he realizes & 173 & that her handkerchief is all wet. (54B, NNS)
In the restaurant, some quick shots show the young lady Chaplin fell in love with on the boat. She is crying. When the two discover each other in the restaurant, they are very happy at first, until they take each other’s hands, as speaker 54B (extract (210)) describes. This action is not an unusual one; rather, it is possible to read like plus the pause before take as an indicator that the speaker considers the expression take hands as imprecise or as not conveying the mental picture she has in mind, since Chaplin and the lady do not stand side by side but facing each other. Speaker 71B in extract (211) comments on the pattern of repetition which can be found throughout the movie. The extract refers to a scene in which the waiter tells Chaplin to take off his hat. Chaplin at first does not understand. So the waiter repeatedly takes off Chaplin’s hat and Chaplin puts it back onto his head. Speaker 71B did not consider this funny. Quite to the contrary, she found it really annoying (line 377). Again, like functions as a focuser, here reinforcing the expression of the speaker’s displeasure at this scene. (211)
373 374 375 376 377 (71B,
the waiter actually put it off, and Charlie Chaplin put it on. this went on like-it it seemed for minutes. (H) .. it was like really annoying, NNS)
In the category of like as a focuser, I have included one sub-type which Andersen does not regard as a case of meta-linguistic use but rather as a sub-type of loose use (2001: 237ff.). Several times, American speakers employed like before an expression which is clearly exaggerated. When speaker 124B (extract (212)) says that Chaplin was eating like a whole plate of beans one by one, we know from the movie that this is not
Chapter 5. Like
the case: Chaplin started eating his beans in this manner, but he soon switched to a different method. Moreover, he never finished eating them because he was interrupted several times. Speaker 134A (extract (213)) is surprised about the fact that not only the female but also the male characters in the movie wear make-up. When she talks about tons of eyeliner, we do not need to have seen the movie in order to know that this statement is exaggerated. (212)
191 he was eating like a whole plate of & 191a & beans one by one, (124B, NS)
(213)
455 A: and he still had like (H) tons of eyeliner, (134A, NS)
Andersen calls this sub-type “hyperbolic use”. I agree with his claim that such hyperbolic use can be seen as a kind of “non-identical resemblance between the encoded and the communicated concepts” (2001: 237). To indicate that an expression is exaggerated (i.e. a hyperbole), a speaker may highlight it by using a particular intonation and stress, or by using like before the expression. Thus, like once more functions as a highlighting or focusing device. Focusing like occurs in the American speakers’ data 280 times, more than any other function of like. Among the German speakers, it is the most frequent discourse marker function with 57 instances. In percentages, this corresponds to an average of 0.53 tokens per 100 words for the Americans and 0.06 percent for the Germans. The figures also represent a statistically significant difference between the two groups. Like as a focuser is common among native speakers from the United States. This is shown not only in the average percentage of this function, but also in the fact that only two Americans in my data did not use it. However, as with other functions of like, we find an uneven distribution among the speakers. Frequencies per speaker range from one to 47 instances. Of the German speakers, 21 employ focuser like. Here, we find a range from one to ten instances per speaker.
... Various functions As you know, like was also employed in some other functions, though not frequently enough in any of them to form a separate category. With 15 instances the most frequent of these functions is like as a discourse link. Andersen (2001: 255) describes it as linking “syntactically (sometimes even thematically) unrelated structures”. This formulation seems rather vague; what it basically means is that like in a number of cases apparently has no identifiable function other than linking two sentences. However, we can say a little bit more about what I have labeled ‘discourse link’. All instances come from a single speaker, and all of them are used in the narrative part. The discourse link like occurs in two variants: either the speaker changes the focus of the narrative from one movie character to another, or the speaker returns from a digression to the main story line. Extract (214) shows a typical example of each variant. Before the like in line 142, the speaker has been talking about Chaplin and the fact that Chaplin does not
Discourse markers in English discourse
have any money. Like introduces a shift of focus from Chaplin to the homeless guy who enters the restaurant. This usage is similar to functions of two other discourse markers, sequential so (see Chapter 2, Section 2.2.2.4) and well to introduce the next scene (Chapter 3, Section 3.2.2.5), which also implicitly mark a shift of focus. Like in line 152 shifts the narrative focus from an aside (Chaplin’s thoughts, lines 150f.) back to the main story line (here the story of the homeless guy, and, eventually, its consequences for Chaplin). A similar shift is accomplished by so as a boundary marker (Section 2.2.2.5) and well moving to the main story (Section 3.2.2.4). (214)
140 B: 141 142 143 144 145 146 147 148 149 150 151 152 153 (8B, NS)
so (H) he kept u=m sitting there trying to & & figure out what he’s gonna do with the money. and then like this guy walks in. and I thinks he was like homeless or whatever, a=nd .. he’s like-he orders a cup of coffee or whatever, and he has a quarter. or like the money whatever it was. and he’s like rubbing it, he’s all happy, and Charlie is like-he’s thinking. (H) and so like u=m what happens is, the guy pays for his coffee,
Another recognized function of like occurs six times in the data. Schourup, in his data, found “several instances of like preceding a restart” and defines restart as “a point at which the present speaker stops an item under construction and recommences” (1985: 54). Andersen distinguishes between ‘false start’, which in his definition is given when “a speaker cuts off her utterance and resumes talk with a new syntactic structure” and ‘self repair’: “the speaker resumes talk with an item that is syntactically connected with the preceding material” (2001: 255). In contrast to Andersen, I have counted as ‘false start’ instances such as extract (215), in which a structure is cut off and resumed with the same words, instances in which one (or two) word(s) of the structure before like is changed after like, as in extract (216), and also an instance in which like precedes a structure which differs from the one started before the truncation. (215)
35 the waiter keeps ta-36 like keeps tapping him. (8B, NS)
(216)
282 (H) it was just funny how he got his-283 ... like how he sometimes leaned in there, (141C, NS)
Again, this function is only used by American speakers, though by several and not just one. Besides the two functions described above, like occurred twice in the expression to be all like into it. The remaining instances in the category ‘various’ (three of which were
Chapter 5. Like
produced by German speakers) did not show any recognizable pattern of the function of like, even though they were clearly instances of the discourse marker like.
.. Like – a textual discourse marker In Section 5.2.2, I have explained and illustrated all of the discourse marker functions for which like was used by the participants in the GLBCC. In my data, like frequently indicates that the speaker is searching for the appropriate expression to represent what s/he has in mind. When like precedes a number or a quantitative expression, it marks this expression as an approximate one which should not be taken too literally. A further well-known and acknowledged function of like can be paraphrased as ‘for example’. The concept following like represents only an exemplification of what the speaker has in mind. In one variant of this function, this exemplification follows the mention of a general concept for which an example is given, while in another variant, such a general concept is not mentioned at all. Twice as frequent as exemplifying like in my data is the use of like to introduce an explanation. Often, this explanation adds details to information given before, so that this information is more understandable. In other cases, the speaker expresses the same informational content in different words, which may achieve the same effect of making the information more understandable. Sometimes, the speaker just exchanges one term or expression for an alternative one. Like as a focusing device is by far the most frequent function. It has more tokens than the other four discourse marker functions taken together. As a focuser, like highlights the following word or expression for a number of possible reasons. In some cases, the speaker uses like to show that this expression is exaggerated; in others, the speaker may not be sure whether the expression is appropriate. In both cases, like indicates what is called ‘loose use’ in relevance-theoretic approaches (e.g. Jucker and Smith 1998; Andersen 2001). Other reasons may be that the expression highlighted by like describes unusual actions, ideas important for the narrative, surprising events, etc. Finally, like in this function can also reinforce feelings, for example displeasure. Whatever the reasons are, like focuses on the part of the utterance in its scope, which is typically just a word or a (noun or verb) phrase. All of these functions are set at the textual level. Like gives an indication of textual problems; it marks part of what the speaker utters and gives the hearer instructions on how to process and interpret it. It also structures utterances into statement and explanation, or (implicit) general concept and an exemplification of it. It provides an emphasis on certain expressions for a range of possible reasons. However, unlike the discourse markers so, well, and you know, it does not play a role in the interaction between speaker and hearer: it does not occur in transition relevance places (as so does), nor evaluate previous statements or introduce answers (as well does), nor does it appeal to the hearer to respond at least with mental involvement (as you know does). Like therefore is a purely textual discourse marker, while so, well, and you know function at the textual as well as at the interactional level.
Discourse markers in English discourse
.. Quotative like In Section 1.4.4 of Chapter 1, I have argued that like functioning as a quotative device is not a discourse marker, at least not in the sense I have defined it. The main reason for this is that it typically forms part of a construction BE+like which functions as a verb of saying. As such, it is integrated into the syntactic structure of a sentence and cannot be removed without rendering the sentence ungrammatical. Since this research focuses on discourse marker functions, I will present quotative like as it occurs in my data only briefly here. Like in its quotative use is the most frequent function after the like as a lexical focuser and almost twice as frequent as like introducing an explanation. However, while the American participants used it 246 times, the Germans employed it only 35 times.20 As with other functions of like, quotative like is also distributed very unequally within the groups. A third of the Americans did not employ it at all, while another third used it more than 8 times (up to 58 times). The number of ‘quotative like users’ among the Germans amounts to only 15 speakers who use it between one and six times. I have said above that quotative like typically combines with a form of BE. Of the instances of like which I classified as introducing a quotation, the great majority followed this pattern, as in extract (217). Speaker 2B describes Chaplin’s reaction when he detected that he had lost his money through a hole in his pocket. In extract (218), she describes how she felt at the beginning of the movie when the boat was shown rolling from side to side and people getting sick. Both extracts illustrate what other researchers found out about the usage of quotative like as compared to other quotative verbs or constructions: like is frequently used for introducing expressions of emotions and for introducing thoughts or potential utterances (cf. for example Blyth et al. 1990; Romaine and Lange 1991). Since the movie which the speakers retell is a silent one, with only occasional inter-titles, the speakers mostly cannot know what the characters in the movie said. Sometimes, we can see lip movements, so that the person retelling the scene can assume speech, but often enough, as in extract (217), the “quotation” represents feelings, reactions, and thoughts. Another example that like does not have to introduce a speech report is given in extract (219). Speaker 131A reports a nonlinguistic utterance which Chaplin (may have) made (“tse-tse-tse”) and then explains it. Her explanation contains another “quotation”, but the speaker makes it clear that this is nothing Chaplin actually said. Interestingly, speaker 2B uses a third person pronoun for two different purposes: he to represent Chaplin’s thoughts in extract (217), and it to represent her own feelings. While he/she/they+BE+like were the most frequent forms in the narratives, reporting the (assumed) speech of the movie characters, it+BE+like occurred rather infrequently. It should also be noted that quoted speech, whether it reports actual or assumed speech, thought, feelings, etc., is uttered with a particular voice quality in many cases. This ‘quotation voice quality’ is marked in the transcript byat the end of the quoted material. However, speakers do not always use such a
Chapter 5. Like
particular voice quality, as extracts (218) and (219) show. Setting off quoted material by changing the voice quality serves to make the narrative livelier; other than that, it does not seem to have any specific function. (217)
58 B: (H) and so [@ then he’s like@], (2B, NS)
(218)
239 A: [oh so] you at the beginning said you & 239a & were getting <@ si=ck @@@ @>, 240 B: from watching it it was like oh god, (2B, NS)
(219)
112 he’s like ((”tse-tse-tse” noise)) 113 like shaking his head like I can’t & 113a & believe that ha=ppened, (131A, NS)
Besides BE+like, the corpus shows other forms of introducing a quotation with like. Of these, LOOK+like is the more frequent one, occurring 17 times in the American speakers’ data and once in a German’s speech. Extracts (220) and (221) illustrate its two variants, looking/looks like and looking at (somebody) like: (220)
67 so he was looking like, (9C, NS)
(221)
384 and he’s looking at them like & 384a &, (123B, NS)
Other verbs may also be combined with like to form a quotative construction, such as thinking like, see each other like, meant like, asking like and others, followed by a quotation. However, each of them occurred only once, either in native or in nonnative speech.
.. Unclassified instances Not surprisingly, like also occurs in contexts which are acoustically unintelligible or incomplete. Thus, the first like in line 269 of extract (222) could possibly be interpreted as the start of an explanation, but we cannot judge the function of the second like in this line, due to the unintelligible passage before it, and due to the fact that speaker B starts a new syntactic construction after the overlap (indicated by square brackets). (222)
267 .. I don’t know. 268 they’re like focused on like her. 269 .. [like XXXX like], (1B, NS)
(223)
483 484 484a 485
cause she had this piece of handkerchief? ... and I don’t know if that belonged & & to her mom? .. (H) .. and she was holding it,
Discourse markers in English discourse
486 and the guy’s like-487 Charlie’s like looking at her like & 487a &, (140B, NS)
The truncation of line 486 in extract (223) is responsible for the fact that we cannot classify like in this line with any certainty. We may assume that the speaker intended to pronounce a sentence like the one in the following line (487) and stops short just to substitute Charlie for guy. However, this would be an assumption. The speaker could also have intended to continue line 486 with a quotation, for example. Cases like this make up the major part of the unclassified instances. (224)
98 B: 99 100 101 102 103 104 104a (4B, NS)
the coin he lost. ...(1.2) (H) so then the um the waiter takes it, and puts it in his pocket, but it falls out of the bottom of his pant leg. so Charlie tries to get it, and like totally and takes like something & & <X and like Laurel and Hardy tries to & & get it X>,
Sometimes, there is no truncation in the sense used in the transcription of the GLBCC, i.e. the speaker stopping short in mid-sentence and in the midst of an intonation contour. Unintelligibility does not seem to play a major role either. Still, the meaning of a part of the utterance may remain unclear, as happens in line 103 of extract (224). Following the general trend in the distribution of like between the two speaker groups, the American speakers produced four times as many unclassified instances as the German speakers did.
. Quantitative results for like The quantitative results of the comparison between American native speakers (US NS) and German EFL speakers (Germans) at the level of the individual functions of like are summarized in Figures 5.1a (see Section 5.2.1) and 5.1b. The first figure illustrates the distribution of the non-discourse marker, non-quotative functions of like, while the second figure shows the distribution of the remaining categories. As in the previous chapters, the labels used in the figures are abbreviations of the full category labels. Table 5.2 gives an overview of the full labels and the corresponding abbreviations. Of the non-discourse marker, non-quotative functions, only like as a conjunction shows a statistically significant difference (i.e. at a level of p < 0.05) between the two speaker groups. In contrast, like as a preposition indicating a comparison was used almost as often by the Germans as by the Americans on average. Like in the fixed expression something like that or one of its variants found particular favor among the Germans. The lack of a statistical significance for these two functions is indicated by a
Chapter 5. Like
Table 5.2 Functions of like – full labels and their abbreviations Full category label
Abbreviated label
Preposition like (comparison) something/stuff like this/that Conjunction like Searching for the appropriate expression Approximate number Introducing an example Introducing an explanation Lexical focus Various functions Quotative like
PCN SLT CJT SAE ANM XMF EXL LFC VAR QUL
Discourse marker like in general Non-discourse marker, non-quotative like
dm like non-dm like
superscript circle behind the labels in the chart (Fig. 5.1a, Section 5.2.1). All other categories of like showed a statistically significant difference in their frequency distribution between the two speaker groups at the five percent level. In Figure 5.1b, we can see that, on average, like as a lexical focuser was the most frequent discourse marker function among the Germans, closely followed by like introducing an example and introducing an explanation. Somewhat less frequent is quotative like. However, the highest average frequency only lies at 0.06 instances of like per 100 words. In contrast, the American native speakers in my data employed
avg. % of total words per group
0.6 Germans Americans
0.5 0.4 0.3 0.2 0.1 0.0 SAE
ANM
XMF
EXL
LFC
VAR
QUL
Figure 5.1b Distribution of like as discourse marker and quotative – American vs. German speakers
Discourse markers in English discourse
the lexical focuser like with more than 0.5 instances per 100 words and quotative like with 0.46 instances per 100 words. They used three functions (searching for the appropriate expression, marking an approximate number, and introducing an explanation) about four times as much as the Germans. The smallest average difference between German and American speakers exists for like introducing an example. All Americans used discourse marker like at least once, while this was the case for only 44 out of 77 Germans.
.. Non-linguistic factors ... Gender There are two views on the distribution of like which are relevant for the discussion of like and gender. One is the public and popular view that mainly women, particularly young women, use like most frequently (Dailey-O’Cain 2000). The other is the academic view that women are the forerunners in language change (cf. Andersen 2001), which can be used to explain some of the results by different researchers. Thus, Andersen argues that females in his data employed the marker like, but not the quotative like more than their male counterparts because the quotative has already been established, while the marker usage is still new. A similar argumentation can be applied to Ferrara and Bell’s (1995) findings that quotative like was more frequent in female speech in the 1990 data, but equally frequent for both sexes in the 1992 data. In the quantitative analysis of my data, I will only consider functions of like for which a Median Test yielded statistically significant results. For the American speakers, such a significant result with respect to the male-female distribution was given for only two functions: something like that and like introducing an example. In both cases, the male participants used like almost twice as often as the female participants. Obviously then, the public view on who uses like does not correspond to the actual distribution, or at least not to the distribution among students in Southern California. The analysis of the German speaker data does not yield more than two significant results either. German female and German male students differ only in the frequency of like introducing an example and of discourse marker like in general. However, in contrast to the American speakers, it is the German females who use the exemplifier like and the function group (including all six categories discussed in Section 5.2.2) more than three times as often as the males. When we add the gender of the participant’s partner and compare all-female pairs with all-male pairs, the result for discourse marker like in general is the same in the German speaker group. In addition, the German speakers in mixed pairs employed like to introduce explanations, while the German speakers in all-male pairs did not use this function at all. This result is significant at a probability of p = 0.024. The frequency of this function in the all-female pairs is similar to that in the mixed pairs, but not significantly different from the frequency in the all-male pairs (p = 0.053). A comparison including the gender of the participant’s partner in the American speaker group did not yield any significant results.
Chapter 5. Like
Can we now conclude that quotative like, which is equally frequent among German men and women, is an established function, while discourse marker like is not? The results presented above seem to support such a statement, in analogy to the results from the COLT. However, we must take into account that English is a foreign language for the German speakers; linguistic factors may also play a role, as we will see.
... Age The results from various studies coherently show that the frequency of like is highest in the speech of young adults and about zero in the speech of the older generation (age 40/50+). However, the definition of the age groups varies between the different studies. Thus, Blyth et al. (1990) defined three age groups: 20–24, 27–32, and 40+. The youngest group in Dailey O’Cain’s (2000) study, in contrast, comprises speakers aged 14–29, while the oldest group is 50 years and older. Ferrara and Bell (1995) define a young group and an old group more or less like Blyth et al., but do not present any results for an intermediate age group. Andersen (2001) finally divides the adolescents (aged 10 to 19) into three groups and compares them mostly to adults aged 30 and older. However, he also states that the discourse marker like “was not uncommon among speakers in their twenties and thirties” in his adult corpus, a section of the BNC (2001: 288). Of the adolescents, the speakers aged 17–19 used discourse marker like most frequently. The speakers in my data range from 18 to 40, but with an unequal distribution. Two thirds of the American speakers were 18 to 21 years old at the time of the recording, while almost half of the Germans were between 22 and 24 years old. Among the Americans, there was only one speaker in this age group. Participants aged 25+ made
avg. % of total words per group
0.30 0.25
young medium
0.20 0.15 0.10 0.05 0.00 LFC
non-dm
discourse marker
Figure 5.2 Distribution of like according to age (German speakers)
Discourse markers in English discourse
up twenty-six percent of the American speaker group and thirty percent of the German speaker group. A comparison between the younger and the older group of the Americans revealed no statistically significant differences, even though half of the older group is older than 30. While it is true that none of those over 30 years of age uses discourse marker like more than ten times and thus below the American average, it is also true that this applies equally to half of the young group. An effect of age-grading on the frequency of discourse marker and quotative like therefore cannot be confirmed by my data. Among the German speakers, there are no statistically significant differences between the middle and the oldest group, nor between the youngest and the oldest group. However, the youngest group used both the non-discourse marker (non-dm), non-quotative functions of like and the discourse marker (dm) functions of like more often than the middle age group (see Figure 5.2). In particular, they employed like as a focuser three times as much as their (middle group) peers. If we look at average frequencies only, we find that the figures for the middle age group and for the oldest age group are more or less equal, so that we cannot say that the frequency of like consistently decreases with age. However, we cannot confirm the opposite either, since a third of the speakers in the oldest group were aged 25 at the time of the recording, and only one speaker was older than 30 years.
... Relationship between the partners Does the relationship of the partners in the experiment influence the frequency of like? Redeker found this to be true in her study (which also included retelling a movie) for pragmatic markers in general; friends used them significantly more often than strangers (1990: 375). Redeker’s list of markers did not include like, but Jucker and Smith (1998:177) report that friends also used like significantly more often than strangers. Their data, however, consisted not of narratives but of discussions about a number of topics assigned by the researchers. The Median Test for the American speakers in my data yielded a statistically significant difference between friends and strangers only for like introducing an example, which the friends used more than twice as often as the strangers. In the German speaker group, the test identified significant differences for two categories and a category group, as shown in Table 5.3 and presented graphically in Figure 5.3. Friends were much more likely than strangers to use like to search for the appropriate expression, to Table 5.3 Significant differences for functions of like between friends and strangers within the German speaker group (average tokens per 100 words) Category
friends (average)
strangers (average)
searching for the appropriate expression (p = 0.031) lexical focus (p = 0.005) discourse marker like in general (p = 0.013)
0.035 0.088 0.245
0.007 0.003 0.090
Chapter 5. Like
avg. % of total words per group
0.30 0.25 friends strangers
0.20 0.15 0.10 0.05 0.00 SAE
LFC
discourse marker
Figure 5.3 Distribution of like according to speaker relationship (German speakers)
focus on the following expression, and discourse marker like in general. Thus, my data partially confirm the results of Jucker and Smith (1998).
... Role As explained in Section 1.5.2, the participants in the GLBCC acted in one of three roles. Speakers in role A retold the first part of the movie (without a hearer), acted as a hearer to the narrative of speaker B, and participated in the discussion of the movie. Speakers in role B retold the second part of the movie (with speaker A as hearer) and discussed it with speaker A. Speakers in role C did not have a partner; they saw and retold the whole movie (without a partner) and gave their opinion about it. Thus, the roles differed not only in having a partner or not, but also in the part of the movie which the speakers retold. In the first part of the movie, there is more action and less (oral) interaction between the characters than in the second part (see Appendix 1, Summary of the movie), while the second part focuses on the emotions of the characters and their interaction. Given the fact that quotative like can be used both to report speech and thoughts and to express emotions, we might expect a higher rate of like per 100 words in the speech of participants in role B. Indeed, quotative like is the only function which achieved statistical significance in the American speaker group. As expected, participants in role B used it (three times) more than participants in role A. If we look at individual frequencies, we find that only three out of ten speakers in role A employed quotative like at all, with two of these three using it ten and thirteen times, respectively. In contrast, almost all speakers in role B (and in role C) employed quotative like.
Discourse markers in English discourse
avg. % of total words per group
0.30 0.25 role A role B
0.20 0.15 0.10 0.05 0.00 EXL
XMF
discourse marker
QUL
Figure 5.4 Distribution of like according to speaker role (German speakers)
Among the German speakers, we find significant differences for three individual functions. In addition to quotative like, German participants in role B also showed a higher frequency of like to introduce an explanation, like to introduce an example, and also of the function group ‘discourse marker like in general’ than the participants in role A. The exact averages are given in Table 5.4 (see also Figure 5.4). The result concerning quotative like meets our expectations. The remaining results agree with those for you know: you know to introduce explanations (including exemplifications) and the function group ‘textual you know’ were also more frequently used by speakers in role B than by those in role A. (The function group ‘textual you know’ corresponds to discourse marker like, since like only functions at the textual level.)
.. Linguistic factors While non-linguistic factors are relevant for native and non-native speakers alike, we now turn to linguistic factors, which are relevant only for the non-native speakers. In Table 5.4 Significant differences for functions of like between roles within the German speaker group (average tokens per 100 words) Category
role A (average)
role B (average)
quotative like (p = 0.002) introducing an explanation (p = 0.017) introducing an example (p < 0.000) discourse marker like in general (p = 0.013)
0.005 0.031 0.023 0.121
0.065 0.063 0.067 0.251
Chapter 5. Like
Chapter 1, I claimed that the context of language acquisition – formal vs. informal – has an influence on the usage of discourse markers in general and at the level of their individual functions. Therefore, I compared those German speakers who learned English in a family or with friends with those who learned English only in school, at a language institute or at the university. It turned out that there was no difference between the two groups in terms of discourse marker like in general. Of the individual functions, only two showed a statistically significant difference: Participants who learned English in informal contexts used like marking an approximate number and like introducing an example much more often than participants who learned English only in formal contexts. These results do not completely correspond to my claim, which may be due to the small number of participants in the group with informal acquisition of English (ten speakers). Another reason could be that other factors are more influential, as the following comparison suggests. In response to our question about their primary means of communication (PMC) with friends or family members, half of the German participants said that they never use English in these contexts. 26 participants indicated that they used English with some friends. For another 9 participants, English is the primary means of communication either with all friends, or with some friends and with family members. Statistical comparisons between the three groups reveal no significant differences at all between those who only occasionally use English as primary means of communication and those nine speakers who do so regularly. However, the occasional users employ a number of functions and discourse marker like in general more frequently (with p < 0.05) than the speakers who never use English in informal contexts, as we can see in Table 5.5. If we combine the two groups of Germans who occasionally and who regularly use English as a PMC and compare them to those who never do, like marking the search for the appropriate expression disappears from the list of functions with significant differences. Presumably, this is due to a higher English language proficiency among those who regularly use English, which would reduce the need to search for words, and thus the need to employ like for this function. However, we can add two functions with significant differences: like introducing an explanation and like introducing an example (see Figure 5.5). The tendency remains the same; when Germans talk to friends (or family members) in English, they use like in several functions more frequently than those who do not have English language Table 5.5 Significant differences in the frequency of like with respect to the use of English as primary means of communication (average tokens per 100 words) English as PMC
occasionally
never
marking approximate numbers searching for the appropriate expression lexical focus discourse marker like in general
0.016 0.045 0.116 0.332
0.004 0.016 0.027 0.108
Discourse markers in English discourse
avg. % of total words per group
0.30 0.25 never occasionally/regularly
0.20 0.15 0.10 0.05 0.00 SLT
ANM
LFC
EXL
XMF
dm
Figure 5.5 Distribution of like according to the use of English as PMC
interactions. Indeed, the averages of the three groups suggest that the frequencies of all discourse marker functions and of quotative like increase with occasional usage of English as PMC and either remain the same or decrease again with a regular usage of English. Interestingly, the expression something like that follows the opposite tendency; its frequency decreases with the increased usage of English as PMC. The picture looks similar if we add as a factor whether the participants had been abroad in an English-speaking country. This is the case not only for those who still use English with friends and/or family members but also for some of those who do not use it as a primary means of communication any more. I have combined the factor PMC with the factor “abroad – yes/no” to create a distinction between speakers with native speaker contact and speakers without such contact. For this distinction, I assume that the friends and family members to whom the participants talk in English are native speakers of English; I also assume that students being abroad interact not only with speakers of a non-English language background but also with native speakers. Thus, in Table 5.6, speakers who never use English as primary means of communication and who have never been in an English-speaking country count as having no contact to native speakers. All others count as having at least some contact. Table 5.6 shows that there are significant differences between the “no contact” and the “some contact” group for almost all functions already represented in Figure 5.5 (English as PMC), except for like marking approximate numbers. Instead, the distribution of quotative like becomes statistically significant. Like for lexical focus, explanations, examples, as a discourse marker in general, and as a quotative are rare if not non-existent in the data of the German participants without any native speaker contact. The expression something like that, in contrast, is more than twice as fre-
Chapter 5. Like
Table 5.6 Significant differences in the frequency of like with respect to native speaker contact (average tokens per 100 words) native speaker contact
no contact
some contact
something like that lexical focus introducing an explanation introducing an example discourse marker like in general quotative like
0.085 0.005 0 0.01 0.025 0
0.032 0.074 0.057 0.057 0.232 0.046
quent in the data of the speakers without contact as in the data of those with at least some contact. The last linguistic variable to be analyzed is the difference between British and American English. The discourse marker like not only occurs in American English. Miller and Weinert (1995) show that it also exists in Scottish English. However, they also note that the placement and the functions of Scottish like do not quite coincide with Schourup’s (1985) description of American like. The like which Miller and Weinert analyzed seems to be restricted to the north of Britain, while American like is known throughout the United States and gaining ground elsewhere, too. Andersen (2001), for example, takes the view that like as it occurs in the Corpus of London Teenage Language (COLT) does not come from the north of Britain but rather from America. Regarding quotative like, it seems even less controversial that this function of like originated in the United States and is slowly spreading in other countries as well (cf. Blyth et al. 1990; Romaine and Lange 1991; Ferrara and Bell 1995; Tagliamonte and Hudson 1999; Dailey-O’Cain 2000). That there is a difference between American and British English also becomes obvious when we compare the frequency of like in my data (recorded in 1999) and the COLT (recorded 1993): instances of discourse marker like make up 1 percent of all words spoken by American native speakers of English in the GLBCC, while the tokens of like in the COLT, “markers and non-markers included” (Andersen 2001: 265), make up only 0.3 percent of all words. The question in the present research is whether this difference is reflected in the usage of like by the German speakers in the GLBCC. In order to investigate this question, I split up the Germans into three groups and compared them pairwise: those who had not been abroad, those who had spent more time in Great Britain, and those who had spent more time in the USA. It turned out that those who never went abroad used something like that and like to search for an appropriate expression significantly more often than those who spent more time in Britain. Other than these, there were no differences between these two groups. In contrast, the British-influenced participants showed a much lower frequency of all discourse marker functions of like individually and as a group and of quotative like than the speakers who had spent more time in the USA. The same significant differences were found in the comparison between participants who never went abroad and the American-influenced participants, except for
Discourse markers in English discourse 0.35 not abroad more GB more US
avg. % of total words per group
0.30 0.25 0.20 0.15 0.10 0.05 0.00 PCN
ANM°
SAE°
LFC
EXL
XMF
dm
QUL
Figure 5.6 Distribution of like according to time abroad (All functions are significantly different between “more GB” and “more US”. All functions except for ANM and SAE are significantly different between “not abroad” and “more US”.)
like marking approximate numbers and like searching for an appropriate expression. These relationships are illustrated in Figure 5.6. It also interesting to compare the German speakers who had spent more time in the United States with the Americans themselves. At the beginning of Section 5.3, I explained that the frequency of like was significantly different between Americans and Germans for all five individual discourse marker functions of like, discourse marker like in general, and quotative like. In the comparison between American-influenced Germans and American native speakers, however, this is only the case for like as lexical focuser, like to introduce explanations, discourse marker like in general, and quotative like.
.. Summary of the quantitative results for like Of the non-linguistic factors, gender proved to have opposite effects in the American speaker and in the German speaker group. American females used something like that and like introducing examples less frequently than their male counterparts. In contrast, the German females used like introducing examples and discourse marker like in general more often than the German males. Age played a role only among the Germans, and only between the younger and the middle groups of speakers (age 18–21 vs. age 22–24). The younger speakers employed both discourse marker like (particularly like as lexical focuser) and non-discourse marker, non-quotative like more than their slightly older peers.
Chapter 5. Like
The relationship between speakers affected the distribution of like more among the Germans than among the Americans, even though in all cases like was more frequent between friends than between strangers. This resulted in a significant difference between American friends and strangers for exemplifying like, and significant differences among the German speakers for like marking search, like as a lexical focuser, and discourse marker like in general. Finally, the speakers who retold the second half of the movie to a partner (role B) employed quotative like significantly more than their partners (role A) among both Americans and Germans. In addition, Germans in role B also showed a higher frequency than their partners in role A for like to introduce explanations, exemplifier like, and, again, discourse marker like in general. This latter result is not related to the results for gender, since the ratio of males and females in role B is the same as the male-female ratio in role A. In terms of linguistic factors, the statistical analyses showed that past or present interaction in English in informal situations boosts the frequency of discourse marker like, in its individual functions and generally. This is particularly the case for interaction in an American English context. While learning English in informal contexts made a difference only for like marking approximate numbers and as an exemplifier, using English at least occasionally as a primary means of communication led to increased frequencies of like in all discourse marker functions except for marking the search for the appropriate expression. When we add times abroad (which equals past interaction in English) to create the factor “some native speaker contact / no native speaker contact”, we find that like marking approximate numbers ceases to be significantly different, while quotative like becomes significantly different in the two groups. This factor presupposes that all English language interaction takes place with native speakers of English. The influence of American English, in contrast to British English, becomes obvious when we compare German participants who spent time abroad in Great Britain with those who spent time in the USA. The difference between the two groups is significant for all individual discourse marker functions, discourse marker like in general, and quotative like: American-influenced Germans used like more frequently in all these functions.
Chapter 6
Conclusion
The present study relates to previous and current research in several ways. Firstly, it is based on a very specific corpus, the Giessen-Long Beach Chaplin Corpus (GLBCC). This corpus was tailored to meet the needs of a contrastive study of native and nonnative speech including both qualitative and quantitative analyses. The GLBCC combines two text types, (movie) narratives and discussions: pairs of students watched a silent (Chaplin) movie, retold it and exchanged their opinions about it. In contrast to Chafe’s Pear Story experiment (Chafe 1980a), both narratives and discussion were (initially) restricted to the English language, which allowed us to use a movie with occasional language elements in written form (i.e. as inter-titles). However, the Pear Story project and the GLBCC share the basic idea that the movie should not pose any problems for the participants in the experiment in terms of language understanding. As it turned out, understanding the movie depended much more on paying attention to the scenes than on reading and understanding the inter-titles. The specific design of the GLBCC presents particular advantages. A number of studies on discourse marker usage investigate corpora containing spoken data from a range of contexts and on a range of topics, such as the London-Lund Corpus (e.g. Erman 1992; Aijmer 2002), the Santa Barbara Corpus of Spoken American English (e.g. Lenk 1998), or the Corpus of London Teenage Language (e.g. Andersen 1998). Due to the range of contexts and topics, such corpora can be expected to contain many different discourse markers used in all their possible functions. However, also due to the range of contexts and topics, these corpora are not particularly suitable for a comparative investigation such as the present study. Even if we compare data from one of the (native) English language corpora with a similar corpus in another language or by non-native speakers of English (cf. Romero Trillo 1997), we cannot ensure comparable conditions with respect to factors which might influence the use of discourse markers. Other studies rely on more restricted data, either collected in interviews (e.g. Blyth et al. 1990; Hays 1992; Fuller 2003) or conversations recorded in a lab setting (e.g. Nikula 1993; Jucker and Smith 1998). Such approaches have the advantage that various conditions and the topics can be controlled, but they also produce recorded material that does not necessarily represent natural speech. Therefore, generalizations can only be made to a limited extent. A similar problem occurs in interlanguage pragmatics. Studies in this field frequently use language tests, discourse completion tests, role plays, and meta-linguistic interviews in order to investigate linguistic behavior (see Barron 2003: 59ff. for an overview). However, these methods show what a speaker does lin-
Discourse markers in English discourse
guistically when s/he has time to think about it; they do not record spontaneous speech production. The design of the GLBCC combines approaches from both sides. Recordings were made in a lab setting, and the participants were given a specific task with a pre-defined topic, even though this task was a relatively natural one (telling a peer about a movie and exchange opinions about it; cf. Chapter 1, Section 1.7.1.7). Due to these conditions, we cannot generalize from the lab situation to real-life linguistic behavior in all aspects. For example, not all discourse markers of the English language may occur, and those which do occur may not be used in all their possible functions. However, it is important to emphasize that native and non-native participants were subjected to exactly the same conditions, which was essential for the purpose of the present study. Thus, we can compare the linguistic behavior of the two participant groups under perfectly congruent conditions. We can also assume that the findings concerning the relative frequency of discourse markers / discourse marker functions in the two groups extend beyond the lab situation whenever conditions are the same for native and non-native speakers. In other words, it is likely that German students who are speakers of English as a foreign language will use the discourse markers so, you know, and like less frequently than American students who are native speakers of English, and the discourse marker well with approximately the same frequency. This presumably applies at least to (movie) narratives and discussions, the two text types produced in the GLBCC. Further research in the same vein would have to establish whether more far-reaching generalizations are possible. On the basis of a substantial section of the GLBCC, I have given a detailed and comprehensive account of four discourse markers so, well, you know, and like, following a bottom-up approach to the data. This account includes a thorough description of all functions which these markers assume in the data. The resulting classification system consists of two levels, the textual and the interactional level.21 While so, you know, and well have a number of individual functions which are analyzed as working at the textual and at the interactional level, like only functions at the textual level. My description of functions was not restricted to native speaker data alone, as other, similarly detailed descriptions (e.g. Schiffrin 1987; Lenk 1998; Aijmer 2002), but extends to the data of non-native, German speakers, with the aim of comparing native and non-native usage of discourse markers. One view which guided my research has been shown to be valid: A sound comparison between native and non-native discourse marker use has to be carried out at the level of individual functions and not just of the discourse markers in general (cf. Romero Trillo 2002 or Nikula 1996). The results of my analysis reveal that the German speakers also employed the four discourse markers I have investigated; however, differences occurred in the usage of the individual functions. While some functions found in the native speaker data seem to be completely unknown to the Germans, some functions are employed only by the Germans. Based on the classification of discourse marker functions, I have further carried out extensive quantitative analyses in two main steps. The first step consisted of comparing the frequency of discourse marker functions in the speech of native and
Chapter 6. Conclusion
non-native participants, which revealed considerable differences for the markers so, you know, and like. The second step consisted of investigations into the relationship between a variety of non-linguistic and linguistic factors and the frequency of discourse marker functions. Such factors range from age and gender to times spent in an English-speaking country and native speaker contact in general. As the most crucial result of this step, I have been able to show that the interaction with and contact to native speakers of English contributes to a more native-like usage of discourse markers among the German speakers. The present study also makes an important contribution to interlanguage pragmatics, a branch of second language acquisition research (Kasper and Blum-Kulka 1993b: 3). As I mentioned above, information about the pragmalinguistic behavior of a second or foreign language learner is often gathered with the help of questionnaires, discourse completion tasks, interviews, and role plays. Most of these tasks focus the learner’s attention on the linguistic performance. In this respect, the GLBCC offers an alternative, since the task given to the participants focused their attention on the movie and on the communication between the partners. Furthermore, most of the studies in the area of interlanguage pragmatics investigate speech act performance, i.e. requests and refusals, thanking, apologizing, greetings, etc. (cf. Trosborg 1995; Cenoz and Valencia 1996; Hartford and Bardovi-Harlig 1996; Baba 1999; Gass 1999a; Golato 2002; Barron 2003). However, there is no doubt that, apart from speech acts, discourse markers play an important pragmatic role in the speech of native speakers. Aijmer comments: Native speakers use discourse particles with great precision as signposts in the interaction, for example to make it easier for the hearer to understand how the different parts of the text are related. If a non-native speaker uses discourse particles incorrectly or not uses them at all this may lead to misunderstandings. A structural and functional description of discourse particles will therefore be useful for learners [...]. (Aijmer 2002: 3)
Several authors agree that misunderstanding is only one negative effect of lacking or wrongly used discourse markers. Other effects include negative judgments on the nonnative speaker (Svartvik 1980: 171; Erman 1987: 1; Nikula 1993: 127) and a “giveaway” of the speaker’s “foreignness” (Hasselgren 2002b: 103). Therefore, it must be desirable in most cases to improve the non-native speaker’s usage of discourse markers. Before we can think about any means to achieve such an improvement, it seems appropriate not only to provide “a structural and functional description” of discourse markers, as Aijmer suggests, but also a description of marker usage at various interlanguage stages of non-native learners of English.22 In addition, it would be helpful to have at least preliminary suggestions as to which factors contribute to a more native-like usage of markers. The present study provides all three: a detailed functional description, a description of the state of affairs of discourse markers in non-native speech, and findings about the relationship between discourse marker use and potentially influential factors.
Discourse markers in English discourse 1.4
avg. % of total words per group
1.2 Germans Americans
1.0 0.8 0.6 0.4 0.2 0.0 so*
well
you know*
like*
Figure 6.1 Distribution of the four discourse markers (an asterisk indicates a significant difference between the speaker groups)
In Section 1.4, I formulated a number of sub-questions to the main research question, which I will use for my summary. The first of these sub-questions read, “How does the frequency of the selected discourse markers in German non-native speakers’ discourse compare to the frequency in native (American) discourse?” Figure 6.1 illustrates the answer to this question. The columns represent the number of tokens of all discourse marker functions of each item per 100 words spoken by the respective group; they do not present other functions of the linguistic items so, well, you know and like. So is by far the most frequent discourse marker in both groups, followed by like among the Americans and well among the Germans. For the native speakers, well is the least frequent discourse marker; for the EFL speakers, this is you know. The American participants used so almost twice as often as the Germans and both you know and like even more than five times as often. This difference is statistically significant for all three markers. The discourse marker well is an exception to the rule; statistically, there is no difference between the two speaker groups in terms of frequency. Which functions do the selected discourse markers have in the native speakers’ and in the non-native speakers’ discourse? Quite generally, the functions of so, well, and you know, identified in Chapters 2–4, can be divided into textual and interactional functions. Like, in contrast, only functions at the textual level. Textual functions organize the content of what is said or mark (parts of) utterances as specific types of utterances, while interactional functions address the hearer directly or organize the sequence of turns between the participants. Interactional functions are particular to each discourse marker in this study; in contrast, textual functions can be similar across different markers. Thus, like and you know can both indicate that the following
Chapter 6. Conclusion
formulation is an approximate one, even though like does this mostly for numeral expressions, while you know takes non-numerical phrases and clauses as its scope. Both markers can introduce explanations and exemplifications, functions which are also fulfilled to some extent by so. Well and you know introduce quoted material and occur at repairs or reformulations; well, you know and like all mark a search for the right expression. When a speaker wants to move to the next scene in the movie narrative, s/he has the choice between so and well. These two markers also lead from a digression back to the main story/idea and from instructions to the narrative. However, it must be emphasized that these are similar functions. When two different markers apparently serve the same purpose, there are differences either in the position of the marker, in its scope, or slight differences in pragmatic meaning (see Müller in prep. for a detailed analysis). In addition to the functions listed so far, so and like each have a (very frequent) textual function which is not found for the other two markers: So marks result or consequence, and like functions as a lexical focuser by highlighting the following expression. At the interactional level, each discourse marker shows its own profile. So occurs at transition relevance places and marks speech acts, such as questions, requests, or giving an opinion. When it occurs at the end of a turn, it often marks the implication of a result, i.e. it requires the hearer to infer the result which the speaker has in mind or has mentioned before. Well is frequent at the beginning of evaluations and answers. The latter can be direct or indirect answers or even responses to expectations which the speaker her/himself raised. A speaker can also contribute an opinion beginning it with well. This functional overlap between so and well is the only one between the markers at the interactional level. You know appeals to the hearer to get at least mentally involved. This appeal can take on a range of different forms: The speaker may ask the hearer to remember a piece of shared knowledge (such as a character from the first half of the movie); to imagine the scene or see its implication for the movie character; to understand the speaker, even if the speaker cannot express him/herself very clearly; or to acknowledge that the speaker’s opinion is right. Also, the force of the appeal ranges from “I know you remember / can imagine” to waiting for an explicit (oral) acknowledgement. Like is the only discourse marker in this study which does not have any interactional functions. These are the discourse marker functions found in the data of the American speakers. Do the German speakers also use them? The answer is yes for all functions except one: the use of you know to introduce quoted material does not occur in the German speakers’ data. However, it should be added here that, when we say that German speakers use the other functions listed above, this does not imply high frequencies. It means that there is at least one speaker who does, but not necessarily all or even most of them. Thus, none of the discourse marker functions of you know is employed by more than eight German participants. The same is true for four functions of well and one of like. On the other hand, there are two discourse marker functions which do not occur in the data of the American speakers, but in the data of the Germans: well to introduce
Discourse markers in English discourse
a summary/conclusion (sixteen speakers) and well to continue expressing an opinion (ten speakers). Obviously, it is not enough to just state that discourse markers are found in certain functions in both the native and the non-native data. It is also relevant to compare the frequencies of the markers/functions. As the most meaningful quantitative unit I have chosen tokens of functions per 100 words. Since mere averages per group can be misleading, Median tests have been used to see whether groups of speakers (here American versus German speakers) differ significantly in the frequency with which they use individual functions or groups of functions of the selected markers, such as all discourse marker functions of a marker. Table 6.1 lists all discourse marker functions according to their level; bold print indicates that there is a statistically significant difference between the frequency of that function among the Americans and the frequency among the Germans. The marker Table 6.1 All discourse marker functions according to their level. Bold print + *: Germans use this function significantly more often than Americans. Only bold print: Americans use the function (group) significantly more often than Germans. level / marker
TEXTUAL
INTERACTIONAL
so
– marking result or consequence – main idea unit marker – summarizing / rewording / giving an example – sequential so – boundary marker
– speech act marker – question or request – speech act marker – opinion – marking implied result – marker of a transition relevance place
well
– *searching for the right phrase – rephrasing / correcting – quotative well – move to the main story – introducing the next scene – *conclusive well
– *indirect answer – direct answer – response to self-raised expectations – contributing an opinion – *continuing an opinion / answer – evaluating a previous statement
you know
– marking lexical/content search – marking false start and repair – marking approximation – introducing an explanation – quotative you know
– “imagine the scene” – “see the implication” – reference to shared knowledge – appeal for understanding – acknowledge that the speaker is right
like
– searching for the appropriate expression – marking an approximate number or quantity – introducing an example – introducing an explanation – marking lexical focus
Chapter 6. Conclusion
itself in bold print indicates that such a difference exists for all discourse marker functions of that marker as a group. As we can see from the table, the relationship between American and German data in terms of the frequency of functions depends on the discourse marker. While German speakers employed four functions of well significantly more often than the American speakers, the latter employed all other functions marked as significantly different more than the Germans. This concerns only three of the functions of so, but most of the functions of you know and all of like. Thus, the average frequencies depicted in Figure 6.1 and particularly the ratios of native versus non-native averages reflect the results of the Median tests. Where one group uses the marker approximately twice as often as the other group (so and well), only a few functions of these markers show a significant difference; where one group uses the marker approximately five times as often as the other group, (almost) all functions show a significant difference. In the literature on discourse markers, several non-linguistic variables have been identified to have an effect on the frequency of markers, for example social class, age, gender, and the relationship between partners in a dialog recording. In the GLBCC, there is no information on the social class of the participants, but we have data on the age, gender, and partner relationship of the speakers. In addition, the role of the speakers has been included in the tests for non-linguistic factors, as about half of the speakers did not have a partner for their narrative. It turned out that neither gender nor age nor the relationship between the partners had the same effect on both the American and the German speaker group. Moreover, all of the four factors had more impact on the German data than on the American data. I will summarize the results for each factor in turn.
Gender. Female American participants used the textual functions of so significantly more often than their male counterparts. Female German participants, in contrast, used its interactional functions more often than the German males. American females further employed well to introduce responses to self-raised expectations and exemplifying like less often than the males, while German females employed well for direct answers less often than the German males, but used exemplifying like and discourse marker like in general more often. Age. The age range in the GLBCC is more limited than in most other studies which found an impact of this variable. It does not span generations; only 10 percent of the speakers included in this study are more than 28 years old, while the youngest speakers are aged 18. Predictably, then, age did not show major effects on the discourse marker use by the American speakers. Only so to mark questions or requests was used significantly more often by the youngest speaker group (age 18–22) than by the oldest speaker group (age 25+). However, we find an effect of age on several functions and function groups among the German speakers, even though there does not seem to be much of a clear tendency for most of them. The only statement that could be made is that the youngest
Discourse markers in English discourse
speaker group used a number of functions more than other speakers. For example, they employed two functions of so (‘summarizing/rewording/giving an example’ and ‘speech act marker – question or request’), two functions of well (‘searching for the right phrase’ and ‘continuing an opinion/answer’), like as a lexical focuser, and discourse marker like in general more frequently than the middle group. They also used so to mark implied result and well to begin indirect answers more often than the oldest group. The middle group, in turn, employed interactional so and you know to mark lexical/content search more frequently, but textual well and evaluative well less frequently than the oldest speakers.
Relationship between the partners. Both Redeker (1990) and Jucker and Smith (1998) found that friends use discourse markers more frequently than strangers. In the data for the present study, this can be confirmed only for like. American friends employed it more often than strangers to give an example, while German friends employed its discourse marker functions as a group and ‘search for the appropriate expression’ as well as the lexical focuser more often than German strangers. Role. The role of the speaker apparently was the non-linguistic factor with the highest influence on the frequency of the discourse markers. Among the Americans, speakers in role B used the textual functions of so significantly more often than the speakers in role A. The same is true for the German speakers. Germans in role A also employed them less often than Germans in role C. No differences could be found for well among the American speakers, but German speakers in role A used quotative well least, while those in role C used conclusive well most frequently. In both speaker groups, American and German, speakers in role A seemed to see little need for you know asking the hearer to imagine the scene (since they did not have a partner for their narrative), in comparison with speakers in role B. For the Germans, however, this lack of need also extended to you know in general, both at the textual and the interactional level. Concerning interactional you know, there was also a significant difference between German speakers in role B and those in role C. Germans in role A moreover showed a lower frequency of discourse marker like (and in particular like for explanations and examples) than their fellow participants in role B, and fewer instances per 100 words of like to mark approximate numbers than the participants in role C. Finally, ‘role’ made a difference for quotative like among both Germans and Americans; it was more frequent in the data of those who retold only the second part of the movie (role B) than in the data of the participants who only retold the first part (role A). Where or how do non-native speakers learn to use discourse markers? According to several authors (e.g. Hays 1992; Sankoff et al. 1997), discourse marker usage is not taught at school but can only be learned through the contact with native speakers. In the questionnaire accompanying the recording of the GLBCC, questions were included which aimed at testing this claim. One question asked about the context in which the
Chapter 6. Conclusion
non-native speakers learned English; another, about whether they used English as a primary means of communication. A third question asked for information about any time the participant had spent in an English-speaking country. Of the information gathered through the questionnaire, I considered only those parts relevant which referred to English being learned or spoken in informal contexts (e.g. with friends or family), since a number of authors claim or assume that discourse markers occur mainly in informal contexts (Crystal and Davy 1975; Holmes 1986; Andersen 1998, among others). Thus, I arrived at a number of linguistic variables, which will be described below. Statistical tests were used to find out whether any of these variables corresponded with the way the German speakers employed the four discourse markers in question, i.e. on the frequency of the individual functions of so, well, you know, and like. The first variable tested concerned the question whether the speakers had learned English in an informal context. This was the case for only ten out of 77 speakers; it affected only three individual functions significantly: so to summarize or rephrase, like marking approximate number, and like as an exemplifier. All three functions were used more often by those who had learned English with friends or in a family. Somewhat more influential was the current use of English as a primary means of communication in informal contexts. On one hand, it significantly reduced the frequency of (non-discourse marker) expressions with so which were direct translations from German expressions and the frequency of well to introduce the next scene. On the other hand, it significantly increased the frequency of all discourse marker functions of like except for ‘searching for the appropriate expression’. The impact of native speaker contact becomes even more obvious when learning and using English in informal contexts are combined with times spent abroad to form the factor ‘native speaker contact’. Those Germans which did have such contact employed well introducing direct answers and unclassified instances of well less often than their fellow participants without native speaker contact; in contrast, they used discourse marker so (particularly at the textual level), one function of you know (‘see the implication’), and discourse marker like in general and three of its functions (‘lexical focuser’, ‘introducing an explanation’, introducing an example’) more often. Thus, the claim mentioned above that non-native speakers learn to use discourse markers through contact with native speakers can be confirmed for the German speakers in the GLBCC for the two discourse markers so and like. Taking this study as a starting point, further research on several open questions seems to be promising. In my research, for example, I have not answered the question to which extent cultural differences play a role in the frequency of discourse marker functions such as well introducing direct answers and well introducing indirect answers. Also, on the basis of the corpus data alone, I have not been able to identify reasons for the relatively high frequency of well and the relatively low frequency of so, you know, and like in the Germans’ speech. However, according to Granger, the interlanguage of advanced learners such as the German participants in the GLBCC “is the result of a very complex interplay of factors: developmental, teaching-induced and
Discourse markers in English discourse
transfer-related” (Granger 2004: 135; cf. Odlin 1989). If this is the case, it would be useful to investigate these three areas with respect to discourse markers to shed more light on the linguistic behavior of the German speakers in this study. The sections below suggest directions which such investigations might take. As we have seen above, it has been claimed that discourse markers are not taught at school (Hays 1992; Sankoff et al. 1997) To my knowledge, there is no study on the usage of discourse markers in the classroom (yet), but at least, we can look at how discourse markers are represented in German textbooks of English as a foreign language. Such textbooks have been and still are the main source for English teachers and the basis of most classroom learning from grade five through grade ten. I examined three textbook series which were popular at the time when the German participants of the GLBCC learned English at school, Green Line, English G, and Notting Hill Gate. The first two were published in the 80s (and early 90s), the third was published in the 90s. The tendencies in the representation of discourse marker so, well, you know and like were similar in all three textbook series. So and well are explicitly introduced at the beginner’s level. So is translated as ‘also’, ‘darum’, ‘deshalb’; and well as ‘also’, ‘nun’, ‘naja’ (Beile et al. 1984; Schwarz et al. 1985; Edelhoff 1994). Well occurs up to 24 times and is more frequent than so in all three books. In contrast, you know occurs only once in each book, and like does not occur at all. The picture changes slightly in more advanced levels (Beile et al. 1987; Schwarz et al. 1991; Edelhoff 1997). The frequency relationship between so and well is more balanced; you know occurs up to six times and like up to three times. As a translation of you know we find ‘weißt Du’ and ‘nämlich’ in English G6, but no translations at all in the other two textbooks. Discourse marker like is not translated at all. Given this representation of the four discourse markers in German textbooks of English, it is not surprising that the German speakers in the GLBCC did not have much difficulty using well, but apparently were not used to employing you know and like as much as the native speakers did. It also fits the picture that so is less frequent in the German speaker data than in the native speaker data, since it seems to be underrepresented in the textbooks (compared to the frequencies of so and well in the native speakers’ speech). It could be objected that the language use of the teacher is at least as influential as the textbooks are. This is very likely. However, most of the English teachers in Germany are Germans, not native speakers of English. In addition, in most of Germany it is not obligatory for prospective teachers to spend time in an English-speaking country. Therefore, the question arises whether the teachers’ English is more native-like than the English of our participants, and whether they are able to illustrate nativelike discourse marker usage. A study of classroom interaction and teacher language focusing on discourse markers could shed more light on this issue. As my data show, so and well can be used to fulfill similar functions: shifting from instructions or interruptions to the main story line or idea, introducing the next scene, and summarizing/concluding. When we compare the frequencies for these functions, we find a tendency in the German speaker group to use well rather than so, compared
Chapter 6. Conclusion
to American usage. Why should that be the case? First of all, the distribution of the two discourse markers in the German textbooks mentioned above suggests that students might be more used to employing well than so. Furthermore, these textbooks and a popular English-German dictionary (Collins-PONS 1983) give ‘also’ as a possible translation for both so and well. Thus, the German translational equivalent does not help the non-native speakers to choose between the two markers for the functions listed above. In addition, there is a third factor which leads at least some speakers to prefer well over so. A number of students expressed the view that “well sounds more English than so”, that native speakers use well more often than so, and that the usage of well is related to “good English”. This brings us to another potential factor. Most of our German participants were students of English. In practical language classes, they are taught to avoid ‘false friends’ and to beware of L1 interference when they speak or write English. It is possible that the German speakers in my data suspect such a false friend in the case of so: In German, there is a particle ‘so’ which can be used at the beginning of an intonation unit, just like the English discourse marker so (cf. for example Burkhardt 1987; Ehlich 1987). If the students decide to use well rather than so, they can be sure to have chosen an English word. In this sense, the overuse of well is not born out of an “avoidance or underproduction of some ‘difficult’ structure” (Ellis 1994: 305), but rather out of an avoidance of ‘German English’ or a transfer from German that seems to be too easy. A similar argumentation might work for you know. It literally translates as ‘du weißt’; the German translational equivalent would be ‘weißt du’ or, more colloquially, ‘weißte’. Again, ‘weißte’ can be used in the same positions as you know and with some of the same functions, so that German students of English might suspect L1 interference and therefore use you know less often. Concerning the frequency of you know, a contrastive view is also possible. In a comparison of parallel corpora, Romero Trillo was able to show that the frequency of English discourse markers in the speech of Spanish EFL speakers was related to the frequency of the markers’ equivalents in the native Spanish corpus (Romero Trillo 2001). In order to test whether his findings could also apply to you know in my data, I calculated the frequency of ‘weißte’/‘weißt du’ and ‘nämlich’ (a translation given in one of the textbooks) in a small German-language corpus with the same experiment setup as the GLBCC. Indeed, both expressions were very infrequent and, taken together, even less frequent than you know among the German speakers in the GLBCC. However, besides the translational equivalents, you know (and other discourse markers) may have a number of functional equivalents in German. A thorough investigation of the relationship between discourse markers and particles would have to include a category system for German particles, similar to the system developed in the present study, followed by a comparison at the functional level. This still remains to be done. The present study combines a detailed, functional description of discourse marker usage with quantitative analyses and a cross-linguistic focus. Thus, it contributes to
Discourse markers in English discourse
traditional discourse marker research and interlanguage pragmatics research alike: it complements the former with a qualitative and quantitative account of discourse markers in non-native data, and it complements the latter with a description of discourse marker usage which forms part of a speaker’s pragmalinguistic behavior. Thus it may serve as a basis for further research to bridge the gap between discourse marker research and interlanguage pragmatics, and to formulate consequences of the findings for teaching English.
Notes
. Lyons (1981) talks about ‘descriptive meaning’, but he does not refer to discourse markers. . PERL is a programming language often used in computer linguistics. The PERL programs used for this research were written by myself. . Several of the participants did not have a partner. Others had partners with a different language background; in these cases only the discourse markers of the German or American speaker were counted. Additionally, in the initial recordings the participant who watched only the first half of the movie (role A) wrote down a summary of that part, rather than retell it orally. This led to the exclusion of the speaker in role A. . When a participant did not have a partner (referred to as a speaker in role C), s/he retold the whole movie and expressed his/her opinion on it. This took between five and fifteen minutes. . I am fully aware that intonation units can also be contour-defined, as in the London-Lund Corpus, or pause-defined, as in the Spoken English Corpus. . But see Cameron (2001: 39) on the problem of “the definite, ‘full and faithful’ representation” of the data. . Her 2002 publication, which deals with so among other markers, bears the subtitle, “The Semantics and Pragmatics of Discourse Markers”, and on page 2 she states, “In this book I do not make a distinction between discourse markers and discourse connectives. [...] my own use of the terms ‘discourse marker’ and ‘discourse connective’ is not intended to reflect a commitment to a class of discourse markers/connectives.” . The expression in brackets below the example indicates the number of the recording, the role of the speaker, and whether the speaker was a native (NS) or non-native (NNS) speaker. Despite the generality of the latter term, all NNSs cited here are Germans. . I have chosen to refer mainly to dictionaries which the German students may also have consulted. . This chapter is based on the same analysis as her paper on well which she published two years earlier (Schiffrin 1985) in Language. . See also James (1983) and Jucker (1993) on this topic. James (1983) claims that well can be unambiguously assigned to either the reporting clause or the direct speech based on phonological features such as precision, timing and key. However, phonological features have not been analyzed in as much detail yet within this project. . Even though the narrators did not hear what was actually spoken in the movie, they could see lip movements and thus tell that speech was meant to occur at those points. . Lakoff distinguishes the terms ‘answer’ and ‘reply’ and defines ‘answer’ as a response to the complement and ‘reply’ as a response to the performative (1973: 461). I do not follow this distinction but use ‘answer’, ‘reply’, and ‘response’ interchangeably.
Discourse markers in English discourse . Incidentally, B’s interpretation is completely wrong. Chaplin looks at a black-edged handkerchief in the lady’s hand, which, together with some previous shots of the lady crying, indicates that her mother died. Thus, it is a sign of grief that she lowers her head. . When 42B later uses you know again to refer to a character from the movie whom she had introduced earlier, the reaction again is a mere mhm. . Judging from the transcript, it would be possible to argue that this is a case of “imagine the scene”. However, the intonation contour of the utterance marks you know as belonging to Chaplin’s speech. . The assumption is made here that friends/family members with whom the students speak English are native speakers of English. This need not necessarily be the case, but it is most likely. . Hopper and Traugott (1993, 2003) define grammaticalization “as the process whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions, and, once grammaticalized, continue to develop new grammatical functions” (2003: xv). . For analyses of vague expressions of numbers and quantities and of varying speaker commitment, see for example Channell (1994); Moxey and Sanford (1997); Jucker et al. (2003). . The Macmillan Dictionary for Advanced Learners (2002), which is based on corpus data, already describes the quotative function of like. However, it was published after (most of) the recordings contained in the GLBCC, so that the German participants could hardly have consulted it at the time of the recording. . Aijmer (2002) describes the functions of discourse markers as working on a textual and on an interpersonal level. However, her definitions of the levels (2002: 40ff.) differ from the definitions I use. . Granger (1996a, 2002) advocates an “Integrated Contrastive Model” which includes the interplay between contrastive analysis (comparing two languages) and contrastive interlanguage analysis (comparing different interlanguages and comparing interlanguages with the native language). The present study mainly contributes to the latter type of comparison.
References
Abraham, Werner (1979). “BUT”. Studia Linguistica, 33, 89–119. Abraham, Werner (1991). “Discourse particles in German: How does their illocutive force come about?” In W. Abraham (Ed.), Discourse Particles. Descriptive and Theoretical Investigations on the Logical, Syntactic and Pragmatic Properties of Discourse Particles in German (pp. 203– 252). Amsterdam: John Benjamins. Ahrens, Rüdiger, Bald, Wolf-Dietrich, and Hüllen, Werner (Eds.). (1995). Handbuch Englisch als Fremdsprache. Berlin: Erich Schmidt Verlag. Aijmer, Karin (1996). Conversational Routines in English. Convention and Creativity. London: Longman. Aijmer, Karin (2002). English Discourse Particles. Evidence From A Corpus [Studies in Corpus Linguistics]. Amsterdam: John Benjamins. Andersen, Elaine S., Brizuela, Maquela, DuPuy, Beatrice, and Gonnerman, Laura (1999). “Cross-linguistic evidence for the early acquisition of discourse markers as register variables”. Journal of Pragmatics, 31, 1339–1351. Andersen, Gisle (1997). “They gave us these yeah, and they like wanna see like how we talk and all that. The use of like and other pragmatic markers in London teenage speech”. In U.-B. Kotsinas, A.-B. Stenström, and A.-M. Karlsson (Eds.), Ungdomsspråk i Norden (pp. 82–95). Stockholm: MINS. Andersen, Gisle (1998). “The pragmatic marker like from a relevance-theoretic perspective”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers: Description and Theory (pp. 147–170). Amsterdam: John Benjamins. Andersen, Gisle (2000). “The role of the pragmatic marker like in utterance interpretation”. In G. Andersen and T. Fretheim (Eds.), Pragmatic Markers and Propositional Attitude (pp. 17–38). Amsterdam: John Benjamins. Andersen, Gisle (2001). Pragmatic Markers and Sociolinguistic Variation: A Relevance-Theoretic Approach to the Language of Adolescents. Amsterdam: John Benjamins. Anping, He (2002). “On the discourse marker so”. In P. Peters, P. Collins, and A. Smith (Eds.), New Frontiers of Corpus Research (pp. 41–52). Amsterdam: Rodopi. Ariel, Mira (1994). “Pragmatic operators”. In R. E. Asher (Ed.), The Encyclopedia of Language and Linguistics. Oxford: Pergamon Press. Ariel, Mira (1998). “Discourse markers and form-function correlations”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers: Description and Theory (pp. 223–259). Amsterdam: John Benjamins. Austin, J. L. (1962). How To Do Things With Words. The William James Lectures delivered at Harvard University in 1955. Cambridge, MA: Harvard University Press. Baba, Junko (1999). Interlanguage Pragmatics: Compliment Responses by Learners of Japanese and English as a Second Language. München: Lincom Europa.
Discourse markers in English discourse
Bardovi-Harlig, Kathleen (1996). “Pragmatics and language teaching: Bringing pragmatics and pedagogy together”. In L. F. Bouton (Ed.), Pragmatics and Language Learning Vol. 7 (pp. 21– 40). Urbana-Champaign: University of Illinois. Bardovi-Harlig, Kathleen (2002). “Pragmatics and Second Language Acquisition”. In R. B. Kaplan (Ed.), The Oxford Handbook of Applied Linguistics (pp. 182–192). Oxford: Oxford University Press. Barron, Anne (2003). Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study abroad context [Pragmatics & Beyond New Series]. Amsterdam: John Benjamins. Bazzanella, Carla (1990). “Phatic connectives as interactional cues in contemporary spoken Italian”. Journal of Pragmatics, 14, 629–647. Beile, Werner, Beile-Bowes, Alice, Hellyer-Jones, Rosemary, Lampater, Peter, and Reisener, Helmut (Eds.). (1984). Learning English: Green Line 1. Unterrichtswerk für Gymnasien. Stuttgart: Ernst Klett. Beile, Werner, Beile-Bowes, Alice, Hellyer-Jones, Rosemary, Lampater, Peter, Posener, Alan, and Roth, Rolf W. (Eds.). (1987). Learning English: Green Line 4. Stuttgart: Ernst Klett. Bell, David M. (1998). “Cancellative discourse markers: A core/periphery approach”. Pragmatics, 8, 515–541. Bialystok, Ellen (1991). “Achieving proficiency in a second language: A processing description”. In R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood Smith, and M. Swain (Eds.), Foreign/Second Language Pedagogy Research. A Commemorative Volume for Claus Faerch (pp. 63–78). Clevedon: Multilingual Matters. Bialystok, Ellen and Hakuta, Kenji (1994). In Other Words. The Science and Psychology of SecondLanguage Acquisition. New York: Basic Books. Bialystok, Ellen (1998). “Coming of age in applied linguistics”. Language Learning, 48, 497–518. Biber, Douglas, Johansson, Stig, Leech, Geoffrey, Conrad, Susan, and Finegan, Edward (1999). Longman Grammar of Spoken and Written English. London: Longman. Biber, Douglas (2000). “Investigating language use through corpus-based analyses of association patterns”. In M. Barlow and S. Kemmer (Eds.), Usage-Based Models of Language (pp. 287– 313). Stanford, CA: CSLI Publications. Blackwell, Susan (2000). “Looking up look: Discourse markers in the Bank of English”. In J. M. Kirk (Ed.), Corpora Galore. Analyses and Techniques in Describing English. Papers from the Nineteenth International Conference on English Language Research on Computerised Corpora (ICAME 1998) (pp. 3–16). Amsterdam: Rodopi. Blakemore, Diane (1988). “‘So’ as a constraint on relevance”. In R. M. Kempson (Ed.), Mental Representations. The Interface between Language and Reality (pp. 183–195). Cambridge: Cambridge University Press. Blakemore, Diane (1992). Understanding Utterances. An Introduction to Pragmatics. Oxford: Blackwell. Blakemore, Diane (1996). “Are apposition markers discourse markers?” Journal of Linguistics, 32, 325–347. Blakemore, Diane (2002). Relevance and Linguistic Meaning. The Semantics and Pragmatics of Discourse Markers [Cambridge Studies in Linguistics]. Cambridge: Cambridge University Press. Blass, Regina (1990). “Constraints on relevance. A key to particle typology”. Notes on Linguistics, 48, 8–21. Blum-Kulka, Shoshana, House, Juliane, and Kasper, Gabriele (1989). Cross-Cultural Pragmatics: Requests and Apologies. Norwood, NJ: Ablex.
References
Blyth, Carl, Jr., Recktenwald, Sigrid, and Wang, Jenny (1990). “‘I’m like, ‘Say What?!”’: A new quotative in American oral narrative”. American Speech, 65, 215–227. Bolinger, Dwight (1989). Intonation and Its Uses: Melody and Grammer in Discourse. Stanford, CA: Stanford University Press. Breivik, Leiv Egil and Hasselgren, Angela (Eds.). (2002). From the COLT’s mouth ... and others’. Language Corpora Studies In Honour of Anna-Brita Stenström. Amsterdam: Rodopi. Brinton, Laurel J. (1990). “The development of discourse markers in English”. In J. Fisiak (Ed.), Historical Linguistics and Philology (pp. 45–71). Berlin: Mouton de Gruyter. Brinton, Laurel J. (1996). Pragmatic Markers in English. Grammaticalization and Discourse Functions. Berlin: Mouton de Gruyter. Brinton, Laurel J. (1998). “‘The flowers are lovely; only, they have no scent.’: The evolution of a pragmatic marker in English”. In R. Borgmeier, H. Grabes, and A. H. Jucker (Eds.), Anglistentag 1997 Giessen. Proceedings (pp. 9–33). Trier: Wissenschaftlicher Verlag. Brown, Penelope and Levinson, Stephen C. (1978). “Universals in language use: politeness phenomena”. In E. N. Goody (Ed.), Questions and Politeness (pp. 56–311). Cambridge: Cambridge University Press. Brumfit, Christopher (2001). Individual Freedom and Language Teaching: Helping Learners to Develop a Dialect of their Own. Oxford: Oxford University Press. Bublitz, Wolfram (1995). “Gesprächsprinzipien und gambits”. In R. Ahrens, W.-D. Bald, and W. Hüllen (Eds.), Handbuch Englisch als Fremdsprache (pp. 186–190). Berlin: Erich Schmidt Verlag. Burkhardt, Armin (1987). “SOSO? Kritik und weiterführende Überlegungen zu Konrad Ehlichs Aufsatz über die Funktionen des deutschen so”. In I. Rosengren (Ed.), Sprache und Pragmatik. Lunder Symposium 1986 (pp. 299–313). Stockholm: Almqvist & Wiksell International. Bürki-Cohen, Judith, Miller, Joanne L., and Eimas, Peter D. (2001). “Perceiving non-native speech”. Language and Speech, 44, 149–169. Bussmann, Hadumod (1996). Routledge dictionary of language and linguistics. Transl. and edited by Gregory Trauth and Kerstin Kazzazi. London: Routledge. Butters, Ronald R. (1982). “Editor’s note [on be like ‘think’]”. American Speech, 57,149. Byram, Michael (Ed.). (2000). Routledge Encyclopedia of Language Teaching and Learning. London: Routledge. Cameron, Deborah (2001). Working With Spoken Discourse. London: Sage. Canale, Michael and Swain, Merrill (1980). “Theoretical bases of communicative approaches to second language teaching and testing”. Applied Linguistics, 1, 1–47. Canale, Michael (1983). “From communicative competence to communicative language pedagogy”. In J. C. Richards and R. W. Schmidt (Eds.), Language and Communication (pp. 2–27). London: Longman. Caron-Prague, Josiane and Caron, Jean (1991). “Psychopragmatics vs. sociopragmatics: The function of pragmatic markers in thinking-aloud protocols”. In J. Verschueren (Ed.), Pragmatics at Issue: Selected Papers of the International Pragmatics Conference, Antwerp, August 17–22, 1987 (pp. 29–36). Amsterdam: John Benjamins. Catford, J. C. (1998). “Language Learning and applied linguistics: A historical sketch”. Language Learning, 48, 465–496. Cenoz, Jasone and Valencia, José F. (1996). “Cross-cultural communication and interlanguage pragmatics: American vs. European requests”. In L. F. Bouton (Ed.), Pragmatics and Language Learning Vol. 7 (pp. 41–54). Urbana-Champaign: University of Illinois.
Discourse markers in English discourse
Chafe, Wallace L. (Ed.). (1980a). The Pear Stories. Cognitive, Cultural, and Linguistic Aspects of Narrative Production [Vol. 3. Advances in Discourse Processes]. Norwood, NJ: Ablex. Chafe, Wallace L. (1980b). “Preface to volume III”. In W. L. Chafe (Ed.), The Pear Stories. Cognitive, Cultural, and Linguistic Aspects of Narrative Production (pp. xi–xviii). Norwood, NJ: Ablex. Channell, Joanna (1994). Vague Language. Oxford: Oxford University Press. Cherrington, Ruth (2000). “Interlanguage”. In M. Byram (Ed.), Routledge Encyclopedia of Language Teaching and Learning (pp. 307–309). London: Routledge. CIDE (1995). Cambridge International Dictionary of English. For Advanced Learners. Cambridge: Cambridge University Press. Clyne, Michael G. (1999). “Intercultural discourse”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 500–507). Amsterdam: Elsevier. Cobuild, Collins (1987). Collins Cobuild English Language Dictionary. London: Collins. Cobuild, Collins (1995). Collins Cobuild English Dictionary. London: HarperCollins. COED (1976). The Concise Oxford Dictionary of Current English (6th edition). Oxford: Clarendon Press. Collins-PONS (1983). Pons-Globalwörterbuch Englisch-Deutsch. Stuttgart: Klett. Cook, Guy (1998a). “Communicative competence”. In K. Johnson and H. Johnson (Eds.), Encyclopedic Dictionary of Applied Linguistics (pp. 62–68). Oxford: Blackwell. Cook, Vivian J. (1998b). “Interlanguage”. In K. Johnson and H. Johnson (Eds.), Encyclopedic Dictionary of Applied Linguistics (pp. 174–176). Oxford: Blackwell. Core, Mark G. and Schubert, Lenhart K. (1999). “Speech repairs: A parsing perspective”. Presented at the ICPhS Satellite Meeting on Disfluency in Spontaneous Speech, Berkeley, CA, July 1999 (pages 47–50 in the working notes). Cots, Josep Maria (1992). “Native and non-native speakers’ intuitions about conversational competence”. In C. Mair and M. Markus (Eds.), New Departures in Contrastive Linguistics. Neue Ansätze in der Kontrastiven Linguistik. Proceedings of the Conference Held at the Leopold-Franzens-University of Innsbruck, Austria, 10–12 May 1991 (pp. 169–181). Innsbruck: Verlag des Instituts für Sprachwissenschaft. Couper-Kuhlen, Elizabeth (2001). “Intonation and discourse: Current views from within”. In D. Schiffrin, D. Tannen, and H. E. Hamilton (Eds.), The Handbook of Discourse Analysis (pp. 13–34). Malden, MA: Blackwell. Crystal, David and Davy, Derek (1975). Advanced Conversational English. London: Longman. Crystal, David (1988). “Another look at, well, you know...” English Today, 13, 47–49. Crystal, David (1997). A Dictionary of Linguistics and Phonetics. Oxford: Blackwell. Dailey-O’Cain, Jennifer (2000). “The sociolinguistic distribution of and attitudes toward focuser like and quotative like”. Journal of Sociolinguistics, 4, 60–80. Davies, Alan (2003). Native Speaker: Myth & Reality. Clevedon: Multilingual Matters. De Cock, Sylvie, Granger, Sylviane, and Petch-Tyson, Stephanie (2003). The Louvain International Database of Spoken English Interlanguage – LINDSEI. http://www.lftr.ucl.ac.be/fltr/ germ/etan/cecl/Cecl-Projects/Lindsei/lindsei.htm, accessed 09-Jul-03. De Fina, Anna (1997). “An analysis of Spanish bien as a marker of classroom management in teacher-student interaction”. Journal of Pragmatics, 28, 337–354. de Haan, Pieter (1998). “How ‘native-like’ are advanced learners of English?” In A. Renouf (Ed.), Explorations in Corpus Linguistics (pp. 55–65). Amsterdam: Rodopi. Degand, Liesbeth (2000). “Causal connectives or causal prepositions? Discursive constraints”. Journal of Pragmatics, 32, 687–707.
References
Degand, Liesbeth and Sanders, Ted (2002). “The impact of relational markers on expository text comprehension in L1 and L2”. Reading and Writing: An Interdisciplinary Journal, 15, 739–757. Du Bois, John W. (1980). “Introduction – The search for a cultural niche: Showing the pear film in a Mayan community”. In W. L. Chafe (Ed.), The Pear Stories. Cognitive, Cultural, and Linguistic Aspects of Narrative Production (pp. 1–7). Norwood, NJ: Ablex. Du Bois, John W. (1991). “Transcription design principles for spoken discourse research”. Pragmatics, 1, 71–106. Du Bois, John W. (2002). “What is (natural) discourse? Implications for spoken corpus research”. Plenary lecture held at the 23rd ICAME conference, Göteborg, May 24, 2002. Edelhoff, Christoph (Ed.). (1994). Notting Hill Gate. Lehrwerk für den Englischunterricht. Textbook I. Frankfurt am Main: Moritz Diesterweg. Edelhoff, Christoph (Ed.). (1997). Notting Hill Gate. Lehrwerk für den Englischunterricht. Textbook 5 A. Frankfurt am Main: Moritz Diesterweg. Edmondson, Willis and House, Juliane (19910. “Do learners talk too much? The waffle phenomenon in interlanguage pragmatics”. In R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood Smith, and M. Swain (Eds.), Foreign/Second Language Pedagogy Research. A Commemorative Volume for Claus Faerch (pp. 273–287). Clevedon: Multilingual Matters. Ehlich, Konrad (1987). “so – Überlegungen zum Verhältnis sprachlicher Formen und sprachlichen Handelns, allgemein und an einem widerspenstigen Beispiel”. In I. Rosengren (Ed.), Sprache und Pragmatik. Lunder Symposium 1986 (pp. 279–298). Stockholm: Almqvist & Wiksell International. Ellis, Rod (1987). “Contextual variability in Second Language Acquisition and the relevancy of language teaching”. In R. Ellis (Ed.), Second Language Acquisition in Context (pp. 179–194). London: Prentice-Hall. Ellis, Rod (1994). The Study of Second Language Acquisition. Oxford: Oxford University Press. Ellis, Rod (1997). Second Language Acquisition. Oxford: Oxford University Press. Erman, Britt (1986). “Some pragmatic expressions in English conversation”. In G. Tottie and I. Bäcklund (Eds.), English in Speech and Writing. A Symposium (pp. 131–147). Stockholm: Almqvist & Wiksell. Erman, Britt (1987). Pragmatic Expressions in English. Stockholm: Almqvist & Wiksell. Erman, Britt (1992). “Female and male usage of pragmatic expressions in same-sex and mixedsex interaction”. Language Variation and Change, 4, 217–234. Erman, Britt (2001). “Pragmatic markers revisited with a focus on you know in adult and adolescent talk”. Journal of Pragmatics, 33, 1337–1359. Escudero, Paola and Sharwood Smith, Michael (2001). “Reinventing the native speaker, or ‘What you never wanted to know about the native speaker so never dared to ask.”’ In S. H. FosterCohen and A. Nizegorodcew (Eds.), EUROSLA Yearbook (pp. 275–286). Amsterdam: John Benjamins. Ferrara, Kathleen W. and Bell, Barbara (1995). “Sociolinguistic variation and discourse function of constructed dialogue introducers: The case of be+like”. American Speech, 70, 265–290. Ferrara, Kathleen W. (1997). “Form and function of the discourse marker anyway: implications for discourse analysis”. Linguistics, 35, 343–378. Finell, Anne (1989). “Well now and then (squib)”. Journal of Pragmatics, 13, 653–656. Fischer, Kerstin and Drescher, Martina (1996). “Methods for the description of discourse particles: Contrastive analysis”. Language Sciences, 18, 853–861. Fischer, Kerstin (1998). “Validating semantic analyses of discourse particles”. Journal of Pragmatics, 29, 111–127.
Discourse markers in English discourse
Fischer, Kerstin (2000). From Cognitive Semantics to Lexical Pragmatics: The Functional Polysemy of Discourse Particles. Berlin: Mouton de Gruyter. Flowerdew, John and Tauroza, Steve (1995). “The effect of discourse markers on second language lecture comprehension”. Studies in Second Language Acquisition, 17, 435–458. Fox, B. and Jasperson, R. (1995). “A syntactic exploration of repair in English conversation”. In P. Davis (Ed.), Descriptive and Theoretical Modes. The Alternative Linguistics (pp. 77–134). Amsterdam: John Benjamins. Fox Tree, Jean E. and Schrock, Josef C. (1999). “Discourse markers in spontaneous speech: Oh what a difference an oh makes”. Journal of Memory and Language, 40, 280–295. Fox Tree, Jean E. and Schrock, Josef C. (2002). “Basic meanings of you know and I mean”. Journal of Pragmatics, 34, 727–747. Fraser, Bruce (1988). “Types of English discourse markers”. Acta Linguistica Hungarica, 38, 19– 33. Fraser, Bruce (1990). “An approach to discourse markers”. Journal of Pragmatics, 14, 383–395. Fraser, Bruce (1996). “Pragmatic markers”. Pragmatics, 6, 167–190. Fraser, Bruce (1998). “Contrastive discourse markers in English”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers. Descriptions and Theory (pp. 301–326). Amsterdam: John Benjamins. Fraser, Bruce (1999). “What are discourse markers?” Journal of Pragmatics, 31, 931–952. Fuller, Janet M. (2001). “The principle of pragmatic detachability in borrowing: English-origin discourse markers in Pennsylvania German”. Linguistics, 39, 351–369. Fuller, Janet M. (2003). “The influence of speaker roles on discourse marker use”. Journal of Pragmatics, 35, 23–45. Gass, Susan M. (1999a). “Second Language Acquisition: Conversation”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 572–577). Amsterdam: Elsevier. Gass, Susan M. (1999b). Interlanguage Refusals: A Cross-Cultural Study of Japanese-English. Berlin: Mouton de Gruyter. Golato, Andrea (2000). “An innovative German quotative for reporting on embodied actions: Und ich so/und er so ‘and I’m like/and he’s like”’. Journal of Pragmatics, 32, 29–54. Golato, Andrea (2002). “German compliment responses”. Journal of Pragmatics, 34, 547–571. Goldberg, Julia A. (1976). “The Syntax, Semantics, Pragmatics and Sociolinguistics of Some Conventionalized Parenthetical Clauses in English: ‘you know’ and ‘I mean”’. Unpublished diploma dissertation, Department of Linguistics, Cambridge University (England). Goldberg, Julia A. (1980). “Discourse Particles: An Analysis of the role of ‘you know’, ‘I mean’, ‘well’, and ‘actually’ in conversation”. Ph.D. dissertation, Cambridge University (England). Goldberg, Julia A. (1981). “Hey, y’know, have I got a topic for you”. Unpublished paper. Goss, Emily L. and Salmons, Joseph C. (2000). “The evolution of a Bilingual Discourse Marking System: Modal particles and English markers in German-Amerian dialects”. The International Journal of Bilingualism, 4, 469–484. Granger, Sylviane (1996a). “Learner English around the world”. In S. Greenbaum (Ed.), Comparing English Worldwide. The International Corpus of English (pp. 13–24). Oxford: Clarendon Press. Granger, Sylviane (1996b). “From CA to CIA and back: An integrated contrastive approach to bilingual and learner computerised corpora”. In K. Aijmer, B. Altenberg, and M. Johansson (Eds.), Languages in contrast: Papers from a Symposium on Text-based Cross-linguistic Studies Lund 4–5 March 1994 (pp. 37–51). Lund: Lund University Press.
References
Granger, Sylviane (2002). “A Bird’s eye view of learner corpus research”. In S. Granger, J. Hung, and S. Petch-Tyson (Eds.), Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching (pp. 3–33). Amsterdam: John Benjamins. Granger, Sylviane (2004). “Computer Learner Corpus Research: Current Status and Future Prospects”. In T. Upton and U. Connor (Eds.), Applied Corpus Linguistics: A Multidimensional Perspective (pp. 123–146). Amsterdam: Rodopi. Greasley, Peter (1994). “An investigation into the use of the particle well: Commentaries on a game of snooker”. Journal of Pragmatics, 22, 477–494. Greenbaum, Sidney (Ed.). (1996). Comparing English Worldwide. The International Corpus of English. Oxford: Clarendon Press. Gregg, Kevin R. (1996). “The logical and developmental problems of SLA”. In W. C. Ritchie and T. K. Bhatia (Eds.), Handbook of Second Language Acquisition (pp. 50–83). San Diego: Academic Press. Gregg, Kevin R. (1999). “Second Language Acquisition: History and theories”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 577–584). Amsterdam: Elsevier. Hartford, Beverly S. and Bardovi-Harlig, Kathleen (1996). ““At your earliest convenience:” A study of written student requests to faculty”. In L. F. Bouton (Ed.), Pragmatics and Language Learning Vol. 7 (pp. 55–69). Urbana-Champaign: University of Illinois. Hasselgren, Angela (2002a). “Sounds a bit foreign”. In L. E. Breivik and A. Hasselgren (Eds.), From the COLT’s mouth ... and others’. Language Corpora Studies In Honour of Anna-Brita Stenström (pp. 103–123). Amsterdam: Rodopi. Hasselgren, Angela (2002b). “Learner corpora and language testing. Smallwords as markers of learner fluency”. In S. Granger, J. Hung, and S. Petch-Tyson (Eds.), Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching (pp. 143–173). Amsterdam: John Benjamins. Hasund, Kristine (1998). “Protecting the innocent. The issue of informants’ anonymity in the COLT corpus”. In A. Renouf (Ed.), Explorations in Corpus Linguistics (pp. 13–27). Amsterdam: Rodopi. Hasund, Kristine (2002). “Congratulation, like! – Gratulerer, liksom! Pragmatic particles in English and Norwegian”. In L. E. Breivik and A. Hasselgren (Eds.), From the COLT’s mouth ... and others’. Language Corpora Studies In Honour of Anna-Brita Stenström (pp. 125–139). Amsterdam: Rodopi. Hays, Paul R. (1992). “Discourse markers and L2 acquisition”. In D. Staub and C. Delk (Eds.), The Proceedings of the Twelfth Second Language Research Forum (pp. 24–34). Michigan: Papers in Applied Linguistics – Michigan. He, Agnes Weiyun and Lindsey, Brian (1998). ““You know” as an information status enhancing device: Arguments from grammar and interaction”. Functions of Language, 5, 133–155. Helt, Marie E. and Foster-Cohen, Susan H. (1996). “A relevance theoretic approach to older children’s use of discourse markers”. In A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski (Eds.), Proceedings of the 20th Annual Boston University Conference on Language Development (pp. 308–317). Somerville, MA: Cascadilla Press. Hölker, Klaus (1991). “Französisch: Partikelforschung”. In Lexikon der Romanistischen Linguistik (pp. 77–88). Tübingen: Niemeyer. Holmes, Janet (1986). “Functions of you know in women’s and men’s speech”. Language in Society, 15, 1–22. Holmes, Janet (1996). “The New Zealand spoken component of ICE: Some methodological challenges”. In S. Greenbaum (Ed.), Comparing English Worldwide. The International Corpus of English (pp. 163–181). Oxford: Clarendon Press.
Discourse markers in English discourse
Hopper, Paul J. and Traugott, Elizabeth Closs (1993). Grammaticalization. Cambridge: Cambridge University Press. Hopper, Paul J. and Traugott, Elizabeth Closs (2003). Grammaticalization (2nd edition). Cambridge: Cambridge University Press. House, Juliane and Kasper, Gabriele (1981). “Politeness markers in English and German”. In F. Coulmas (Ed.), Conversational Routine. Explorations in Standardized Communication Situations and Prepatterned Speech (pp. 157–185). The Hague: Mouton. House, Juliane (1982). “Gambits in deutschen und englischen Alltagsdialogen. Versuch einer pragmatisch-kontrastiven Analyse”. Grazer Linguistische Studien, 17/18, 110–132. House, Juliane (1996). “Developing pragmatic fluency in English as a foreign language. Routines and metapragmatic awareness”. Studies in Second Language Acquisition, 18, 225–252. Hymes, Dell (1972). “On communicative competence”. In J. B. Pride and J. Holmes (Eds.), Sociolinguistics. Selected Readings (pp. 269–293). Harmondsworth: Penguin Books. James, Allen R. (1983). “‘Well’ in reporting clauses: Meaning and form of a ‘lexical filler”’. Arbeiten aus Anglistik und Amerikanistik, 8, 33–40. Jaworski, Adam (1998). “Pragmatic competence”. In K. Johnson and H. Johnson (Eds.), Encyclopedic Dictionary of Applied Linguistics (p. 249). Oxford: Blackwell. Jefferson, Gail (1974). “Error correction as an interactional resource”. Language in Society, 2, 181–199. Jespersen, Otto (1940). A Modern English Grammar on Historical Principles, Part V: Syntax (Vol. 4). London: Allen and Unwin. Johansson, Stig (1998). “On the role of corpora in cross-linguistic research”. In S. Johansson and S. Oksefjell (Eds.), Corpora and Cross-linguistic Research. Theory, Method, and Case Studies (pp. 3–24). Amsterdam: Rodopi. Johnson, Keith and Johnson, Helen (Eds.). (1998). Encyclopedic Dictionary of Applied Linguistics. Oxford: Blackwell. Johnson, Keith (2001). An Introduction to Foreign Language Learning and Teaching. Amsterdam: Pearson Education. Jucker, Andreas H. (1988). “Review of Deborah Schiffrin, Discourse Markers. (Studies in International Sociolinguistics 5). Cambridge: CUP”. Multilingua, 7, 219–224. Jucker, Andreas H. (1993). “The discourse marker well: A relevance theoretical account”. Journal of Pragmatics, 19, 435–452. Jucker, Andreas H. and Smith, Sara W. (1998). “And people just you know like ‘wow’: Discourse markers as negotiating strategies”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers: Description and Theory (pp. 171–202). Amsterdam: John Benjamins. Jucker, Andreas H. and Ziv, Yael (Eds.). (1998a). Discourse Markers. Descriptions and Theory [Vol. 57. Pragmatics & Beyond New Series]. Amsterdam: John Benjamins. Jucker, Andreas H. and Ziv, Yael (1998b). “Discourse markers: Introduction”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers. Descriptions and Theory (pp. 1–12). Amsterdam: John Benjamins. Jucker, Andreas H., Smith, Sara W., and Lüdge, Tanja (2003). “Interactive aspects of vagueness in conversation”. Journal of Pragmatics, 35, 1737–1769. Kachru, Braj B. and Nelson, Cecil L. (2001). “World Englishes”. In A. Burns and C. Coffin (Eds.), Analysing English in a Global Context. A Reader (pp. 9–25). London: Routledge. Kasper, Gabriele and Dahl, Merete (1991). “Research methods in interlanguage pragmatics”. Studies in Second Language Acquisition, 13, 215–247. Kasper, Gabriele and Blum-Kulka, Shoshana (1993a). Interlanguage Pragmatics. Oxford: Oxford University Press.
References
Kasper, Gabriele and Blum-Kulka, Shoshana (1993b). “Interlanguage pragmatics: An introduction”. In G. Kasper and S. Blum-Kulka (Eds.), Interlanguage Pragmatics (pp. 3–17). Oxford: Oxford University Press. Kasper, Gabriele (1997). “Can pragmatic competence be taught? (NetWork #6) [HTML document]”. Accessed March 18, 2004 from the World Wide Web: http://www.nflrc.hawaii.edu/ NetWorks/NW(2006/. Honolulu: University of Hawai’i, Second Language Teaching & Curriculum Center. Kaufmann, Anita (2002). “Book review: The Santa Barbara Corpus of Spoken American English – Part I”. Journal of Pragmatics, 34, 1309–1316. Keller, Eric (1979). “Gambits: Conversational strategy signals”. Journal of Pragmatics, 219–238. Keller, Eric (1981). “Gambits: Conversational strategy signals”. In F. Coulmas (Ed.), Conversational Routine. Explorations in Standardized Communication Situations and Prepatterned Speech (pp. 93–113). The Hague: Mouton. Kramsch, Claire (2002). “Standard, norm, and variability in language learning. A view from foreign language research”. In S. M. Gass, K. Bardovi-Harlig, S. S. Magnan, and J. Walz (Eds.), Pedagogical Norms for Second and Foreign Language Learning and Teaching (pp. 59– 79). Amsterdam: John Benjamins. Kryk, Barbara (1992). “A crosslinguistic look at discourse particles”. In C. Mair and M. Markus (Eds.), New Departures in Contrastive Linguistics. Neue Ansätze in der Kontrastiven Linguistik. Proceedings of the Conference Held at the Leopold-Franzens-University of Innsbruck, Austria, 10–12 May 1991 (pp. 43–50). Innsbruck: Verlag des Instituts für Sprachwissenschaft. Kyratzis, Amy and Ervin-Tripp, Susan (1999). “The development of discourse markers in peer interaction”. Journal of Pragmatics, 31, 1321–1338. Lakoff, Robin (1973). “Questionable answers and answerable questions”. In B. B. Kachru, R. B. Lees, Y. Malkiel, A. Pietrangeli, and S. Saporta (Eds.), Issues in Linguistics. Papers in Honor of Henry and Renée Kahane (pp. 453–467). Urbana: University of Illinois Press. Lakoff, Robin (1975). Language and woman’s place. New York: Harper Colophon. Lamiroy, Beatrice (1994). “Pragmatic connectives and L2 acquisition: The case of French and Dutch”. Pragmatics, 4, 183–201. Lauerbach, Gerda E. (1993). “Interaction and cognition: Speech act schemata with but, and their interrelation with discourse type”. In R. A. Geiger and B. Rudzka-Ostyn (Eds.), Conceptualizations and Mental Processing in Language (pp. 679–708). Berlin: Mouton de Gruyter. Lee-Wong, Song Mei (2001). “Coherence, focus and structure: The role of discourse particle NE”. Pragmatics, 11, 139–154. Lenk, Uta (1995). “Discourse markers and conversational coherence”. In B. Wårvik, S.-K. Tanskanen, and R. Hiltunen (Eds.), Organization in Discource. Proceedings from the Turku Conference (pp. 341–352). Turku: University of Turku. Lenk, Uta (1997). “Discourse markers”. In J. Verschueren, J.-O. Östman, J. Blommaert, and C. Bulcaen (Eds.), Handbook of Pragmatics (pp. 1–17). Amsterdam: John Benjamins. Lenk, Uta (1998). Marking Discourse Coherence. Functions of Discourse Markers in Spoken English [Vol. 15: Language in Performance]. Tübingen: Gunter Narr. Lyons, Christopher (1999). Definiteness [Cambridge Textbooks in Linguistics]. Cambridge: Cambridge University Press. Lyons, John (1981). Language and Linguistics. Cambridge: Cambridge University Press. Macaulay, Ronald (2002). “You know, it depends”. Journal of Pragmatics, 34, 749–767.
Discourse markers in English discourse
Macmillan (2002). Macmillan English Dictionary for Advanced Learners. Oxford: Macmillan/ Bloomsbury. Magnan, Sally Sieloff and Walz, Joel (2002). “Pedagogical norms. Development of the concept and illustrations from French”. In S. M. Gass, K. Bardovi-Harlig, S. S. Magnan, and J. Walz (Eds.), Pedagogical Norms for Second and Foreign Language Learning and Teaching (pp. 15– 40). Amsterdam: John Benjamins. Maschler, Yael (1994). “Metalanguaging and discourse markers in bilingual conversation”. Language in Society, 23, 235–366. Maschler, Yael (2000). “Toward fused lects: Discourse markers in Hebrew-English bilingual conversation twelve years later”. The International Journal of Bilingualism, 4, 529–561. Matras, Yaron (2000). “Fusion and the cognitive basis for bilingual discourse markers”. The International Journal of Bilingualism, 4, 505–528. McCarthy, Michael (1993). “Spoken discourse markers in written texts”. In J. M. Sinclair, M. Hoey, and G. Fox (Eds.), Techniques of Description: Spoken and Written Discourse. A Festschrift for Malcolm Coulthard (pp. 170–182). London: Routledge. McCarthy, Michael and Carter, Ronald A. (1994). Language As Discourse. Perspectives for Language Teaching. London: Longman. McDonough, Jo (1998). “Foreign vs. second language learning”. In K. Johnson and H. Johnson (Eds.), Encyclopedic Dictionary of Applied Linguistics (pp. 133–134). Oxford: Blackwell. Meehan, Teresa (1991). “It’s like, ‘What’s happening in the evolution of like?’: A Theory of Grammaticalization”. Kansas Working Papers in Linguistics, 16, 37–51. Miller, Jim and Weinert, Regina (1995). “The function of LIKE in dialogue”. Journal of Pragmatics, 23, 365–393. Mosegaard Hansen, Maj-Britt (1998). The Function of Discourse Particles [Vol. 53: Pragmatics & Beyond New Series]. Amsterdam: John Benjamins. Moxey, Linda M. and Sanford, Anthony J. (1997). “Choosing the right quantifier. Usage in the context of communication”. In T. Givon (Ed.), Conversation: Cognitive, Communicative, and Social Perspectives (pp. 207–331). Amsterdam: John Benjamins. Mukherjee, Joybrato (2002). Korpuslinguistik und Englischunterricht. Eine Einführung [Vol. 14: sprache im kontext]. Frankfurt am Main: Peter Lang. Müller, Simone (In prep.). “Pragmatic functions across markers: A comparison based on the Giessen-Long Beach Chaplin Corpus (GLBCC)”. Nayar, P. Bhaskaran (1998). “Variants and varieties of English: Dialectology or linguistic politics?” In H. Lindquist, S. Klintborg, M. Levin, and M. Estling (Eds.), The Major Varieties of English: Papers from MAVEN ’97, Växjö, November 20–22, 1997 (pp. 284–289). Växjö: Växjö University. Nikula, Tarja (1993). “The use of lexical certainty modifiers”. In L. F. Bouton and Y. Kachru (Eds.), Pragmatics and Language Learning, Vol. 4, 1993 (pp. 126–142). Urbana-Champaign: University of Illinois. Nikula, Tarja (1996). Pragmatic Force Modifiers. A Study in Interlanguage Pragmatics: University of Jyväskylä. Norrick, Neal R. (2001). “Discourse markers in oral narrative”. Journal of Pragmatics, 33, 849– 878. Nunan, David (1999). Second Language Teaching & Learning. Boston: Heinle & Heinle. OALD (1974). Oxford Advanced Learner’s Dictionary of Current English, A. S. Hornby (3rd ed.). Oxford: Oxford University Press. OALD (1989). Oxford Advanced Learner’s Dictionary of Current English, A. S. Hornby (4th ed.). Oxford: Oxford University Press.
References
OALD (2000). Oxford Advanced Learner’s Dictionary of Current English, A. S. Hornby (6th ed., edited by Sally Wehmeier). Oxford: Oxford University Press. Odlin, T. (1989). Language Transfer – Crosslinguistic Influence in Language Learning. Cambridge: Cambridge University Press. OED (1989). The Oxford English Dictionary (2nd edition). Oxford: Clarendon Press. OED online The Oxford English Dictionary (2nd edition). Oxford: Clarendon Press. Olshtain, Elite and Cohen, Andrew (1991). “Teaching speech act behavior to nonnative speakers”. In M. Celce-Murcia (Ed.), Teaching English as a Second or Foreign Language (pp. 154–165). Boston: Heinle & Heinle. Olynak, Marian, d’Anglejan, Alison, and Sankoff, David (1990). “A quantitative and qualitative analysis of speech markers in the native and second language speech of bilinguals”. In R. C. Scarcella, E. S. Andersen, and S. D. Krashen (Eds.), Developing Communicative Competence in a Second Language (pp. 139–155). Boston: Heinle & Heinle. Östman, Jan-Ola (1981). You know: A Discourse-Functional Approach [Vol. II:7: Pragmatics & Beyond]. Amsterdam: John Benjamins. Östman, Jan-Ola (1995). “Pragmatic particles twenty years after”. In B. Wårvik, S.-K. Tanskanen, and R. Hiltunen (Eds.), Organization in Discource. Proceedings from the Turku Conference (pp. 95–108). Turku: University of Turku. Owen, Marion (1981). “Conversational units and the use of ‘well ...”’. In P. Werth (Ed.), Conversation and Discourse (pp. 99–116). London: Croom Helm. Paikeday, Thomas M. (1985). The Native Speaker is Dead! Toronto: Paikeday. Phillipson, Robert (1991). “Some items on the hidden agenda of second/foreign language acquisition”. In R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood Smith, and M. Swain (Eds.), Foreign/Second Language Pedagogy Research. A Commemorative Volume for Claus Faerch (pp. 38–51). Clevedon: Multilingual Matters. Phillipson, Robert, Kellerman, Eric, Selinker, Larry, Sharwood Smith, Michael, and Swain, Merrill (Eds.). (1991). Foreign/Second Language Pedagogy Research. A Commemorative Volume for Claus Faerch. Clevedon: Multilingual Matters. Piepho, Hans-Eberhard (1974). Kommunicative Kompetence als übergeordnetes Lernziel des Englischunterrichts (Communicative competence as the overarching goal of English language teaching). Dornburg-Frickhofen: Frankonius. Piller, Ingrid (2001). “Who, if anyone, is a native speaker?” Anglistik. Mitteilungen des Deutschen Anglistenverbandes, 12, 109–121. Poulisse, Nanda (1999). “Communication strategies”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 484–849). Amsterdam: Elsevier. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey, and Svartvik, Jan (1985). A Comprehensive Grammar of the English Language. London: Longman. Rampton, M. B. H. (1990). “Displacing the ‘native speaker’: Expertise, affiliation, and inheritance”. ELT Journal, 44, 97–101. Rathmayr, Renate (1989). “Zur Frage der Lehr- und Lernbarkeit von Partikeln am Beispiel des Russischen”. In H. Weydt (Ed.), Sprechen mit Partikeln (pp. 623–633). Berlin: Walter de Gruyter. Redeker, Gisela (1990). “Ideational and pragmatic markers of discourse structure”. Journal of Pragmatics, 14, 367–381. Redeker, Gisela (1991). “Review article: Linguistic markers of discourse structure. Review of Deborah Schiffrin, 1987, Discourse Markers, Cambridge: CUP”. Linguistics, 29, 1139–1172.
Discourse markers in English discourse
Richards, Jack C. and Schmidt, Richard W. (1983). “Conversational analysis”. In J. C. Richards and R. W. Schmidt (Eds.), Language and Communication (pp. 118–154). London: Longman. Rieger, Caroline L. (2000). “Self-Repair Strategies of English-German Bilinguals in Informal Conversations: The Role of Language, Gender, and Linguistic Proficiency”. Unpublished doctoral thesis. Department of Modern Languages and Cultural Studies, University of Alberta at Edmonton, Alberta. Rieger, Caroline L. (2003). “Repetitions as self-repair strategies in English and German conversations”. Journal of Pragmatics, 35, 47–69. Risselada, Rodie and Spooren, Wilbert (1998). “Introduction: Discourse markers and coherence relations”. Journal of Pragmatics, 30, 131–133. Ritchie, William C. and Bhatia, Tej K. (Eds.). (1996). Handbook of Second Language Acquisition. San Diego: Academic Press. Romaine, Suzanne and Lange, Deborah (1991). “The use of like as a marker of reported speech and thought: A case of grammaticalization in progress”. American Speech, 66, 227–279. Romaine, Suzanne (1995). Bilingualism. Oxford: Blackwell. Romero Trillo, Jesús (1997). “Your attention, please: Pragmatic mechanisms to obtain the addressee’s attention in English and Spanish conversations”. Journal of Pragmatics, 28, 205– 221. Romero Trillo, Jesús (2001). “The pragmatic “dis-functions” of discourse markers in nonnative speakers of English: A corpus-driven approach”. Paper presented at the 8th Biennial Congress of the International Association for Dialogue Analysis (IADA), Göteborg, April 19–21, 2001. Romero Trillo, Jesús (2002). “The pragmatic fossilization of discourse markers in non-native speakers of English”. Journal of Pragmatics, 34, 769–784. Rösler, Dietmar (1982). “Teaching German modal particles”. International Review of Applied Linguistics in Language Teaching (IRAL), 20, 33–38. Rouchota, Villy (1998). “Procedural meaning and parenthetical discourse markers”. In A. H. Jucker and Y. Ziv (Eds.), Discourse Markers: Description and Theory (pp. 97–126). Amsterdam: John Benjamins. Salmons, Joseph C. (1990). “Bilingual discourse marking: code switching, borrowing, and convergence in some German-American dialects”. Linguistics, 28, 453–480. Sanford, A. J. and Garrod, Simon (1981). Understanding Written Language: Explorations in Comprehension Beyond the Sentence. Chichester: John Wiley & Sons. Sankoff, Gillian, Thibault, Pierrette, Nagy, Naomi, Blondeau, Hélène, Fonollosa, Marie-Odile, and Gagnon, Lucie (1997). “Variation in the use of discourse markers in a language contact situation”. Language Variation and Change, 9, 191–217. Savignon, Sandra J. (2000). “Communicative language teaching”. In M. Byram (Ed.), Routledge Encyclopedia of Language Teaching and Learning (pp. 124–129). London: Routledge. Schank, R. and Abelson, R. (1977). Scripts, Plans, Goals and Understanding. Hillsdale, NJ: Erlbaum. Schiffrin, Deborah (1985). “Conversational coherence: The role of well”. Language, 61, 640–667. Schiffrin, Deborah (1987). Discourse Markers. Cambridge: Cambridge University Press. Schiffrin, Deborah (1994). Approaches to Discourse. Oxford: Blackwell. Schourup, Lawrence (1985). Common Discourse Particles in English Conversation. New York: Garland. Schourup, Lawrence (1999). “Tutorial overview: Discourse markers”. Lingua, 107, 227–265. Schourup, Lawrence (2001). “Rethinking well”. Journal of Pragmatics, 33, 1025–1060.
References
Schwarz, Hellmut, Taylor, Carl, and Vettel, Franz (Eds.). (1985). English G. Neue Ausgabe. Band A1 für das 5. Schuljahr an Gymnasien. Berlin: Cornelsen. Schwarz, Hellmut, Taylor, Carl, and Vettel, Franz (Eds.). (1991). English G. Neue Ausgabe. Band A6 für das 10. Schuljahr. Berlin: Cornelsen. Selinker, Larry (1972). “Interlanguage”. International Review of Applied Linguistics, 10, 209–231. Shumin, Kang (2002). “Factors to consider: developing adult EFL students’ speaking abilities”. In J. C. Richards and W. A. Renandya (Eds.), Methodology in Language Teaching. An Anthology of Current Practice (pp. 204–211). Cambridge: Cambridge University Press. Simon-Vandenbergen, Anne-Marie (2001). “Analysing text and discourse. A report”. The European English Messenger, X, 79–85. Sinclair, John M. (1996). EAGLES. Preliminary recommendations on Corpus Typology. http://www.ilc.cnr.it/EAGLES96/corpustyp/corpustyp.html. Accessed 08/08/(2003. Sinclair, John M. (2001). “Preface”. In M. Ghadessy, A. Henry, and R. L. Roseberry (Eds.), Small Corpus Studies and ELT. Theory and Practice (pp. vii–xv). Amsterdam: John Benjamins. Smith, Sara W. and Jucker, Andreas H. (2000). “Actually and other markers of an apparent discrepancy between propositional attitudes of conversational partners”. In G. Andersen and T. Fretheim (Eds.), Pragmatic Markers and Propositional Attitude (pp. 207–238). Amsterdam: John Benjamins. Smith, Sara W., Jucker, Andreas H., and Müller, Simone (2001). ““Some artist guy”: The role of salience and common ground in the formulation of referring expressions in conversational narratives”. In E. T. Németh (Ed.), Pragmatics in 2000: Selected Papers from the 7th International Pragmatics Conference (pp. 528–542). Antwerp: International Pragmatics Association. Souto Silva, Rosangela (2000). “Pragmatics, bilingualism, and the native speaker”. Language & Communication, 20, 161–178. Sperber, Dan and Wilson, Deirdre (1978). “On Grice’s theory of conversation”. Pragmatics Microfiche, 3. Sperber, Dan and Wilson, Deirdre (1986). Relevance: Communication and Cognition. Oxford: Blackwell. Spolsky, Bernard (1999a). “Second and foreign language pedagogy: ASTP”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (p. 603). Amsterdam: Elsevier. Spolsky, Bernard (Ed.). (1999b). Concise Encyclopedia of Educational Linguistics. Oxford: Elsevier. Spolsky, Bernard (2002). “Norms, native speakers, and reversing language shift”. In S. M. Gass, K. Bardovi-Harlig, S. S. Magnan, and J. Walz (Eds.), Pedagogical Norms for Second and Foreign Language Learning and Teaching (pp. 41–58). Amsterdam: John Benjamins. Stenström, Anna-Brita (1994). An Introduction to Spoken Interaction. London: Longman. Stenström, Anna-Brita and Andersen, Gisle (1996). “More trends in teenage talk: A corpusbased investigation of the discourse items cos and innit”. In C. E. Percy, C. F. Meyer, and I. Lancashire (Eds.), Synchronic Corpus Linguistics: Papers from the Sixteenth International Conference on English Language Research on Computerized Corpora (ICAME 16) (pp. 189– 203). Amsterdam: Rodopi. Stubbe, Maria and Holmes, Janet (1995). “You know, eh and other ‘exasperating expressions’: An analysis of social and stylistic variation in the use of pragmatic devices in a sample of New Zealand English”. Language & Communication, 15, 63–88. Svartvik, Jan (1980). “Well in conversation”. In S. Greenbaum, G. Leech, and J. Svartvik (Eds.), Studies in English Linguistics for Randolph Quirk (pp. 167–177). London: Longman.
Discourse markers in English discourse
Svennevig, Jan (1999). Getting Acquainted in Conversation [Vol. 64: Pragmatics & Beyond New Series]. Amsterdam: John Benjamins. Tagliamonte, Sali and Hudson, Rachel (1999). “Be like et al. beyond America: The quotative system in British and Canadian youth”. Journal of Sociolinguistics, 3, 147–172. Takahara, Paul O. (1999). “Pragmatic functions of discourse markers in English and Japanese”. In J. Verschueren (Ed.), Pragmatics in 1998. Selected Papers from the 6th International Pragmatics Conference. Vol. 2 (pp. 547–576). Antwerp: International Pragmatics Association. Tannen, Deborah (1991). You Just Don’t Understand. Women and Men in Conversation. London: Virago. Tannen, Deborah (1994). Gender and Discourse. New York: Oxford University Press. Tarone, Elaine (1999). “Interlanguage”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 507–512). Amsterdam: Elsevier. Tarone, Elaine (2001). “Interlanguage”. In R. Mesthrie (Ed.), Concise Encyclopedia of Sociolinguistics (pp. 475–481). Amsterdam: Elsevier. Tognini-Bonelli, Elena (2001). Corpus Linguistics at Work. Amsterdam: John Benjamins. Tottie, Gunnel (1997). “Relatively speaking: Relative marker usage in the British National Corpus”. In T. Nevalainen and L. Kahlas-Tarkka (Eds.), To Explain the Present. Studies in the Changing English Language in Honour of Matti Rissanen (pp. 465–481). Helsinki: Société Néophologique. Trim, John L. M. (2000). “Common European Framework”. In M. Byram (Ed.), Routledge Encyclopedia of Language Teaching and Learning (pp. 122–124). London: Routledge. Trommer, Sylvia (1990). “Die Vertextungsfunktion von gambits in Alltagsgesprächen (am Beispiel des Englischen)”. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung, 43, 243–258. Trosborg, Anna (1995). Interlanguage Pragmatics: Requests, Complaints and Apologies. Berlin: Mouton de Gruyter. Underhill, Robert (1988). “Like is, like, focus”. American Speech, 63, 234–246. Unger, Christoph J. (1996). “The scope of discourse connectives: implication for discourse organization”. Journal of Linguistics, 32, 402–438. Varonis, Evangeline Marlos and Gass, Susan M. (1985). “Miscommunication in native/nonnative conversation”. Language in Society, 14, 327–343. Vine, Bernadette (2000). “Getting things done: Some practical issues in a functional investigation of directives in spoken extracts from the New Zealand and British components of the International Corpus of English”. In C. Mair and M. Hundt (Eds.), Corpus Linguistics and Linguistic Theory. Papers from the Twentieth International Conference on English Language Research on Computerized Corpora (ICAME 20), Freiburg im Breisgau 1999 (pp. 370–374). Amsterdam: Rodopi. Watts, Richard J. (1986). “Relevance in conversational moves: A reappraisal of well”. Studia Anglica Posnaniensia, 19, 37–60. Watts, Richard J. (1988). “A relevance-theoretic approach to commentary pragmatic markers: the case of actually, really and basically”. Acta Linguistica Hungarica, 38, 235–260. Watts, Richard J. (1989). “Taking the pitcher to the ‘well’: Native speakers’ perception of their use of discourse markers in conversation”. Journal of Pragmatics, 13, 203–237. Webster’s (1994). Webster’s New World Dictionary of American English (3rd college edition). New York: Prentice Hall. Wennerstrom, Ann (2001). “Intonation and evaluation in oral narratives”. Journal of Pragmatics, 33, 1183–1206.
References
Weydt, Harald (1969). Abtönungspartikel. Bad Homburg: Gehlen. Weydt, Harald and Ehlers, Klaas-Hinrich (1987). Partikel-Bibliographie. Internationale Sprachenforschung zu Partikeln und Interjektionen. Frankfurt am Main: Peter Lang. White, Lydia (1996). “The tale of the ugly duckling (or the coming of age of Second Language Acquisition research)”. In A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski (Eds.), Proceedings of the 20th Annual Boston University Conference on Language Development (pp. 1–17). Somerville, MA: Cascadilla Press. Wichmann, Anne (2002). “Looking for attitudes in corpora”. In L. E. Breivik and A. Hasselgren (Eds.), From the COLT’s mouth ... and others’. Language Corpora Studies In Honour of AnnaBrita Stenström (pp. 247–260). Amsterdam: Rodopi. Wilkins, D. A. (1999). “Applied linguistics”. In B. Spolsky (Ed.), Concise Encyclopedia of Educational Linguistics (pp. 6–17). Amsterdam: Elsevier. Winter, Joanne (2002). “Discourse quotatives in Australian English: Adolescents performing voices”. Australian Journal of Linguistics, 22, 6–21. Yule, George (1996). Pragmatics. Oxford: Oxford University Press.
Appendix 1
Summary of the movie
The 24-minute silent Chaplin movie The Immigrant consists of two parts. The setting of the first part is an ocean liner transporting immigrants to the United States of America. On this ship, Chaplin is shown in various scenes: fishing, walking around the ship, playing cards and dice, eating, and falling in love with a young lady. The setting of second half is, for the most part, a restaurant somewhere in New York, supposedly. There he orders something to eat, meets the lady again, and desperately tries to find a solution for his problem: He has lost his money. In the end he is rescued by an artist who wants to engage him and the lady. The movie starts off with a shot of the ocean liner on the open sea. The next shot shows a bunch of immigrants, some lying on deck sleeping, some being sea-sick. Among the immigrants, a young lady caresses her mother, who seems to be ill. Chaplin is leaning over railing, apparently sea-sick, too. But when he turns around, he has got a fish on a hook, which wriggles out of his hand and falls into the bunch of sleeping immigrants. There, the fish happens to bite the nose of one of the sleepers. The camera then shows Chaplin walking up and down the deck, swaying from side to side since the ship is rolling heavily. Again, passengers are to be seen who are sea-sick. Chaplin comes to sit down at the bottom of the railing. A bearded man with a fur hat sits next to him and has the hiccups, which seems to “infect” Chaplin with an “impulse to throw up”. Then an inter-title (white letters on a black screen) says, “Dinner time”, and a member of the crew rings a bell. Everybody rushes to the dining room, except for the young lady and her mother. In the dining room, another slapstick scene is developed where a plump lady has fallen on the floor, due to the movement of the ship, Chaplin stumbles over her, and the two are rolling over each other. Finally, he gets up and finds a seat at the end of a bench at a long table. Food is given out in bowls, which also move back and forth on the table. Thus, Chaplin and the man sitting opposite him at first share a bowl, each taking a spoonful when the bowls arrives on their side. The young lady finally decides to leave her mother sleeping on deck and go to dinner. All seats are occupied, but Chaplin falls in love with the lady at first sight, offers her his seat, and leaves. Another inter-title says, “more rolling”, and then Chaplin is shown with some other men playing dice. Chaplin makes a big show of rolling the dice and wins. A huge bearded player seems to have lost all his money to Chaplin and walks away. When he finds himself next to the sleeping mother (with the daughter still gone) and discov-
Discourse markers in English discourse
ers money tied around her neck with a piece of string, he cuts the string with a knife, steals the money and walks back to the other players. This time, he forces Chaplin to play cards. Again, Chaplin makes a show of (apparently) shuffling the cards and keeps cheating throughout the game. In the process of the game, he exchanges money for the pistol of the bearded player. When he wins again, this player gets very angry, but Chaplin defends himself threatening the player with the newly acquired pistol and leaves. The bearded player in his rage destroys the wooden box that had served as a table for the game. During the gambling scene, short shots are shown of the young lady discovering the loss of the money and of the man with the fur hat still eating in the dining hall when all other passengers have already left. As Chaplin sits down to count his money, the man with the fur hat sits next to him and threatens to throw up into Chaplin’s hat. In an effort to save his hat, Chaplin moves on until he ends up sitting next to the young lady and her mother. Both are crying. On his inquiry the young lady tells him that (inter-title) “mother lost her money” and turns to her mother again. Spontaneously, Chaplin puts all the money he won into the young lady’s jacket pocket without her noticing. Then, on second thought, he takes it out again, takes out a few notes, and puts the rest back into the pocket. This happens twice. Finally he leaves, but a ship official, who watched him, stops him and accuses him of pickpocketing (inter-title: “You pickpocket!”). Chaplin denies and apparently explains, so that the official calls the young lady (inter-title: “Miss, come here!”) and asks her to check her pockets. When she discovers the money, she concludes that it must have come from Chaplin and wants to give it back. He insists that she keep it and receives a warm thank you from the lady. Next an inter-title says, “The arrival in the land of liberty”. The passengers are rushing up and are all looking into one direction. The Statue of Liberty is shown. Then, all passengers are rudely pushed into a corner of the ship by a crew member, who fastens a rope before them; in front of them, two officials set up a chair and a desk at which the passengers, one by one, will receive their papers. One of the officials also treats the passengers rudely. When he struggles with the man with the fur hat, trying to push him behind the rope, Charlie kicks him in the back. The official turns around, opens the rope and chooses the first two – the young lady and her mother – to pass. The young lady and the mother say good-bye to Charlie, who kisses their hands. Another inter-title says “Good-bye”. The two leave and Charlie is the next to pass the rope. Since he is staring at the departing ladies, he is not reacting to the officials, one of whom then kicks Charlie in the back. He also receives his papers and leaves the ship. In the second half, we find Chaplin (inter-title) “Later ... broke and hungry” in the streets. In front of a restaurant, he finds a coin on the sideway and puts in his trouser pocket. The coin immediately falls on the street again, presumably through a hole in the pocket. Chaplin, however, doesn’t notice it and goes into the restaurant to eat. Some other customers are already sitting in the restaurant at different tables. Chaplin takes a seat next to an apparently well-off customer. A huge waiter with thick black
Appendix 1. Summary of the movie
eyebrows and a fierce look on his face comes to the table and waits for Chaplin to order. Chaplin picks up the menu, but then points to the dish of his neighbor, indicating that he wants the same. When the waiter tells him repeatedly, by making hand signs, to take off his hat, Chaplin misinterprets these signs; he assumes the waiter refers to the man sitting next to him. Then, the head waiter takes the hat from his head twice, but Chaplin, who still does not realize what’s going on, puts it on again. Finally, the waiter takes the hat off, turns Chaplin’s head around, puts the hat onto the table and crushes it. Chaplin now makes fun of the waiter, pretending to put the hat on again twice, while he actually keeps it in his hands; then he hangs it up on the wall. Again, Chaplin takes the menu card, turning it sideways; the waiter then grabs it and turns it upside down. Finally, Chaplin mimes playing a flute. The waiter now disappears and comes back with a plate of beans and a French loaf of bread, which he puts down very brusquely before Chaplin. Chaplin starts eating the beans, picking them up one by one with his fork. Then, suddenly, he shoves them onto his knife and into his mouth. His well-off neighbor looks at him disgustedly and finally moves away from him. In between, short shots are shown of the young lady from the boat sitting at another table and looking very sad, and of two musicians in the background, a piano player and a violin player. As Chaplin is still eating, he looks to his right and discovers that the young lady is sitting there. While he looks at her, the beans he was about to eat fall into his coffee. He calls over to her, they both stand up and take each others hands, being delighted to see each other again. The young lady has a black and white handkerchief in her hands; when Chaplin sees that and sends a questioning look to the lady, they both become sad. Supposedly, this means that the mother of the young lady died. Chaplin then invites the lady to sit with him and have some beans, too. She accepts, and he orders the beans. While they are eating, they witness a scene in front of them. A customer is supposed to pay, but there seem to be some problems with that, so that the waiter (which is not the same waiter as Chaplin’s but a small guy) complains to the head waiter or manager. The manager calls to the back room, from which four other waiters emerge, including Chaplin’s huge waiter. They start beating the customer up brutally, with the huge waiter doing the bulk of it, once accidentally hitting the manager, lifting up the customer from the floor, etc., until they finally throw him out the door. When the customer’s waiter passes Chaplin and the young lady, they ask him, (inter-title) “What’s the trouble, waiter?” His (inter-title) response is, “He was ten cents short.” At that, Chaplin starts to count on his fingers, below the table. Then he checks his pockets, does not find the coin, becomes nervous, and finally discovers the hole in his pant pocket. Knowing that the same treatment he just witnessed is probably awaiting him, too, he practices some boxing moves, which earns him some irritated glances from the lady. The waiter appears and gives Chaplin his bill, but Chaplin order (intertitle) “coffee for the lady” to stall for time. The young lady tries to protest, but Chaplin puts his hand on her mouth to quieten her. The coffee is served.
Discourse markers in English discourse
A moment later, a man in shabby clothes enters the restaurant, polishing a coin. He sits down at a table next to Chaplin and orders a coffee with the huge waiter. Chaplin tries to talk to the man but is interrupted repeatedly by the waiter. When the man pays his coffee, the huge waiter puts the coin into his pockets, but does not notice that it falls onto the floor. Chaplin tries to reach out for the coin, but every time he does so, the waiter turns around and looks at him sharply, wondering what Chaplin is up to. In one attempt, Chaplin steps onto the coin with a loud bang, which makes the two musicians at the entrance of the room behind stop playing and wondering where the noise came from. When the waiter returns to find that out, Chaplin pretends to tie his shoelaces until the waiter turns his back on him; then he grabs the coin. Again, the waiter comes with the bill, and Chaplin pays with this coin. The waiter, having become suspicious, tests the coin by biting it. It bends. The waiter then turns on Chaplin, who repeatedly let himself slip from the seat almost to the floor and grabbed by the waiter. Finally, he orders another coffee for himself. This scene has been witnessed by an apparently wealthy and fat customer at the other end of the restaurant who is making hand signs as if drawing. Chaplin and the lady see him staring at them and wonder at his behavior. Chaplin looks back to him using the salt and pepper shakers as “looking glass”. The rich customer then comes over to their table, repeatedly bows to them, sits down and talks to them. Chaplin, listening intently, rests his arms in his beans, brushes the sleeves off and shoves the plate aside, where the artist rest his arm in them. In the course of the conversation, he explains that (intertitle) “I’m an artist”. While the three are talking, the waiter comes again with Chaplin’s bill. The artist takes Chaplin’s bill and offers to pay for it, but Chaplin takes the bill back, indicating that there is no need for the artist to pay for him. The bill goes back and forth between the two until the artist makes a motion indicating that he gives up offering to pay, which Chaplin had not expected. The artist asks the waiter to (inter-title) “Bring my check also!” When the artist pays his bill, the waiter returns several coins to him on a little plate. The artist, who is deeply in conversation with the lady, leaves one of them as a tip and takes no further notice of it, so that Chaplin, after some consideration, puts his own bill on the plate with the coin and thus pays. He even gets a small coin back, which he generously offers as a tip for the waiter. The three of them (the artist, Chaplin, and the lady) get up to leave the restaurant. The waiter dances attendance on the artist in expectance of a tip; when the artist does not seem to know what he wants, he makes angry faces behind the artist’s back. Outside, it is raining. The artist says to Chaplin and the lady (inter-title), “I’ll engage both of you now, starting tomorrow”, and turns to leave, but Chaplin goes after him and asks for (inter-title) “a couple of dollars on account”, which he gets. The artist leaves and Chaplin and the lady turn around a corner. A door with a sign “marriage licenses” is shown; Chaplin points to the door and talks to the lady, but she protests, first looking serious, then with a smile on her face. Chaplin kisses her, but she averts her face and still, with a smile, shakes her head. He then knocks on the door, but she pretends to run away. Chaplin catches her with his cane; a man comes out of the door
Appendix 1. Summary of the movie
who seems to be the official and waits for them to decide. Finally, Chaplin carries the wriggling lady through the office door. The movie ends with the intonation of the wedding march.
Appendix 2.1
Questionnaire used in Long Beach
Group___ Partner A (saw half of movie)
Partner B (saw whole movie)
Demographic Questionnaire Sex: Male Female Age: ____ University level:
Freshman Sophomore Junior
Senior Grad
Preparing to enter university ALI class (if applicable) ____________ Relation with partner (check description that applies best): Friend Acquaintance Stranger What is your first language? English other (please specify) _____________ If English is not your first language, please turn over and continue with the questions. Otherwise you are done with this questionnaire.
Discourse markers in English discourse
How long have you lived in the US? All my life Since age ______ I’ve arrived within the last 12 months I’ve arrived within the last 4 weeks What is your primary means of communication? (select both English and other if both apply)
With family member(s) – of older generation – of my generation With friends At school
English
Other (please specify)
_____________ _____________ _____________ _____________
Where have you learned your English previous to coming to the US? In the following country: _______________ In the following settings (select all which apply): Family for ____ years Friends for ____ years School for ____ years University for ____ years Language Institute for ____ years Other (please specify) _________________ for ____ years
Appendix 2.2
Questionnaire used in Giessen
Group___ Partner A (saw half of movie)
Partner B (saw whole movie)
Demographic Questionnaire Sex:
Male Female
Age: ____ University level:
1st/2nd semester 3rd/4th semester “Hauptstudium” approaching final exams graduate student (“Doktorand/in”)
Relation with partner (check description that applies best): Friend Acquaintance Stranger What is your first language? English German other (please specify) _____________ If English is not your first language, please turn over and continue with the questions. Otherwise you are done with this questionnaire.
Discourse markers in English discourse
Have you ever lived in the US or any other English-speaking country? No Yes in ___________ for ____ years / ____ months, at age (approx.) _____ in ___________ for ____ years / ____ months, at age (approx.) _____ in ___________ for ____ years / ____ months, at age (approx.) _____ What is your primary means of communication? (select both English and other if both apply) With family member(s) – of older generation – of my generation With friends (over the last year) At school / university (over the last year)
German
English
Other (please specify)
_____________ _____________ _____________
_____________
Where have you learned your English? (please do not include times of living in an English-speaking country) In the following country: _______________ In the following settings (select all which apply): Family for ____ years Friends for ____ years School for ____ years University for ____ years Language Institute for ____ years Other (please specify) _________________ for ____ years
Appendix 3
Transcription symbols
As used in the GLBCC, based on DuBois (1991). Table A3. UNITS Intonation unit Truncated intonation unit Word Truncated word
{carriage return} -{space} -
SPEAKERS Speaker identity/turn start Speech overlap
A:, B:, C:, R: [ ], [[ ]], [[[ ]]] plus vertical alignment
TRANSITIONAL CONTINUITY Final Continuing Appeal
. , ?
PAUSES AND LENGTHENING Lengthening Long pause Medium pause Short pause Latching
= ...(N) {N=time in seconds} ... {approx. between 0.5 and 1.0 sec.} .. {less than 0.5 seconds} (0)
VOCAL NOISES Inhalation Exhalation Laughter
(H) (Hx) @
QUALITY Laugh quality Quotation quality Low voice / sotto voce Whisper
<@ @><SV SV> <WH WH>
TRANSCRIBER’S PERSPECTIVE Researcher’s comment Uncertain hearing Indecipherable syllable
(( )) <X X> X
Discourse markers in English discourse
Table A3. (continued) SPECIALIZED NOTATIONS Codeswitching Intonation unit continued Intonational emphasis
& {capital letters}
NON-WORD NOTATIONS Filled pause Agreement (backchannel) Negation
uh, um mhm, uh huh nhn
Author index
A Abelson, R. Abraham, W. Ahrens, R. Aijmer, K. , –, , , , Andersen, E. S. – Andersen, G. , , , –, –, , –, , , , , , , –, , , –, –, –, , , Anping, H. , , Ariel, M. , , –, Austin, J. L. B Baba, J. , Bardovi-Harlig, K. , , Barron, A. –, , Bazzanella, C. Beile, W. Bell, B. –, , , –, –, Bell, D. M. Bhatia, T. K. , Bialystok, E. , – Biber, D. , , , , , , –, , , , , , , , , – Blackwell, S. Blakemore, D. , –, , , , –, , , Blass, R. , , Blum-Kulka, S. , –, Blyth, C., Jr. –, , , , –, , , , Bolinger, D. , Breivik, L. E. Brinton, L. J. , –, , , Brown, G. Brown, P. , Brumfit, C. Bublitz, W. – Burkhardt, A.
Bürki-Cohen, J. Bussmann, H. , Butters, R. R. Byram, M. , C Cameron, D. , Canale, M. Caron, J. Caron-Prague, J. Carter, R. A. Catford, J. C. Cenoz, J. , Chafe, W. L. –, , Channell, J. Cherrington, R. Clyne, M. G. Cohen, A. Cook, G. Cook, V. J. Core, M. G. Cots, J. M. , Couper-Kuhlen, E. Crystal, D. , , , , , –, , , D Dahl, M. Dailey-O’Cain, J. , –, , –, , –, Davies, A. De Cock, S. – De Fina, A. de Haan, P. Degand, L. , , Drescher, M. , , Du Bois, J. W. , , , , , E Edelhoff, C. Edmondson, W. –, Ehlers, K.-H. Ehlich, K.
Ellis, R. , , , , Erman, B. , , –, , , , , , –, , , , –, , –, –, , –, –, , –, , Ervin-Tripp, S. , Escudero, P. F Ferrara, K. W. , –, , , –, –, Finell, A. , , Fischer, K. , , , , Flowerdew, J. Foster-Cohen, S. H. , , Fox Tree, J. E. , –, , , – Fox, B. – Fraser, B. , , –, , , , , –, , – Fuller, J. M. , , , , , G Gass, S. M. , , Golato, A. , , Goldberg, J. A. Goss, E. L. , Granger, S. , , Greasley, P. Greenbaum, S. Gregg, K. R. – H Hakuta, K. , Hartford, B. S. , Hasselgren, A. , , Hasund, K. , , Hays, P. R. , , , , –, , , , He, A. W. , , , , Helt, M. E. , ,
Discourse markers in English discourse Hölker, K. Holmes, J. , , , , –, , , , –, , –, , –, , , –, Hopper, P. I. , House, J. , –, Hudson, R. , , , , –, Hymes, D. J James, A. R. , Jaspersen, R. – Jaworski, A. Jefferson, G. Jespersen, O. Johansson, S. Johnson, H. , Johnson, K. , Jucker, A. H. , –, , , , –, , , –, –, , , , –, , , , , –, , –, , , –, , , , , , , –, , , – K Kachru, B. B. Kasper, G. , , –, , Kaufmann, A. , Keller, E. , – Kramsch, C. , Kryk, B. Kyratzis, A. L Lakoff, R. , , , –, , –, Lamiroy, B. , Lange, D. , , –, , , Lauerbach, G. E. Laver, J. Lee-Wong, S. M. , Lenk, U. , , , –, , , – Levinson, S. , Lindsey, B. , , , , Lyons, C. Lyons, J.
M Macaulay, R. –, –, – Magnan, S. S. Maschler, Y. , Matras, Y. , McCarthy, M. , McDonough, J. Meehan, T. –, , –, , , Miller, J. , –, –, , , , Mosegaard Hansen, M.-B. , , Moxey, L. M. Mukherjee, J. Müller, S. N Nayar, P. B. Nelson, C. L. Nikula, T. , , , – Nunan, D. O Odlin, T. Olshtain, E. Olynak, M. Östman, J.-O. , , , –, , , , , , –, –, , –, , P Paikeday, T. M. Phillipson, R. Piepho, H.-E. Piller, I. Poulisse, N. Q Quirk, R. R Rampton, M. B. H. Rathmayr, R. Redeker, G. –, , –, , –, –, , , , , , , , , , –, , , Richards, J. C. Rieger, C. L. , , Risselada, R. , ,
Ritchie, W. C. , Romaine, S. , , , –, , , Romero Trillo, J. , , , , –, Rösler, D. Rouchota, V.
S Salmons, J. C. , Sanders, T. , Sanford, A. J. Sankoff, G. –, , , , Savignon, S. J. Schank, R. Schiffrin, D. , –, , , , , , , , –, , –, –, , –, –, , , –, , , –, –, –, , , , , , –, , Schmidt, R. W. Schourup, L. –, –, , –, , , –, , , , , , –, , , Schrock, J. C. , –, , Schubert, L. K. Schwarz, H. Selinker, L. – Sharwood Smith, M. Shumin, K. Simon-Vandenberge, A.-M. Sinclair, J. M. , Smith, S. W. , –, , –, , , , , , , –, , –, , , –, , , , –, , , –, , Souto Silva, R. Sperber, D. –, Spolsky, B. , , , Spooren, W. , , Stenström, A.-B. , Stubbe, M. , , , , –, , , – Svartvik, J. , –, , –, –, , –, –, , –, –, , , , , Svennevig, J. Swain, M.
Author index
T Tagliamonte, S. , , , , –, Takahara, P. O. Tannen, D. , Tarone, E. – Tauroza, S. , –, , , – Tognini-Bonelli, E. , Tottie, G. Traugott, E. C. , Trim, J. L. M. Trommer, S. Trosborg, A. ,
U Underhill, R. –, , , , – Unger, C. J. , , ,
V Valencia, J. F. , Varonis, E. M. Vine, B.
W Walz, J. Watts, R. J. , , , , –, –, ,
Weinert, R. , –, –, , , , Wennerstrom, A. Weydt, H. White, L. Wichmann, A. Wilkins, D. A. Wilson, D. –, Winter, J. , , , Y Yule, G. , Z Ziv, Y. , –, , –
Subject index
A abroad , , –, , , , , –, –, –, acknowledge that the speaker is right –, acquaintance , , , , , acquisition of English see formal vs. informal context Advanced Learners’ Dictionary see Oxford or Macmillan adverb , –, , –, –, –, , , age –, , , , –, , –, , , , , , , –, –, , , American English / influence see British English / influence and Santa Barbara Corpus of Spoken American English amplification – answer see continuing, contributing, direct, indirect appeal for understanding , –, , approximate number/quantity –, , –, –, – approximation –, , , authentic , –, B backchannel –, , –, , , , –, , , basic meaning , , , Bergen Corpus of London Teenage Language see COLT bilingual , –, – bottom-up approach ,
boundary , –, , , , , –, , British National Corpus British data , , , , , – influence , , , , , –, , speakers , , –, , ,
C Cambridge International Dictionary of English see CIDE children, child language , , , , Chinese , , , CIDE , –, , – classroom , , , , , , (see also teaching) Cobuild (Collins Cobuild English Language Dictionary; Collins Cobuild English Dictionary) , , –, , , , –, , , , , , , , , COED , –, , , , , coherence , –, , , –, , , –, cohesion Collins Cobuild see Cobuild Collins PONS Globalwörterbuch Englisch-Deutsch , COLT –, –, , , , –, , , , , common ground , , –, , (see also shared knowledge) comparable data , , , ,
competence , , , –, , –, , , , conceptual , , , –, –, , Concise Oxford English Dictionary see COED conclusive well –, , , , conjunction , , –, , , –, , –, consequence see result content search , –, , , , , –, context – formal vs. informal see formal context of recording (restricted, unrestricted) – continuing an opinion/answer –, , contributing an opinion , –, core meaning see basic meaning corpus size – corpus-based corpus-driven , , correcting –, Cross-Cultural Speech Act Realization Project culture, cultural values , , , , , , D dictionary see Cobuild, Collins, COED, CIDE, Longman, Macmillan, OED, OALD, Webster’s direct answer –, –, , , direct speech –, , , –, , , –, , , –, , , ,, (see also quotative) direct translation –, ,
Discourse markers in English discourse directness – discourse completion test , , distance , , Dutch – E error , , , –, , ethnicity evaluating , –, example giving an example (so) –, , , introducing an example (exemplifying like) , –, , , , , –, , , , exemplification (you know) –, –, , expectations see response to self-raised expectations explanation well , , –, you know , –, , , – like , –, –, –, –, , – F face threatening act , –, , – false start , , –, , , , filled pause , –, , –, , –, filler , , film see movie Finnish , first language –, –, , , , , , , , , , fixed expressions , , –, , , , , , – fluency focus (lexical focus), focuser –, –, –, –, –, , –, , – formal vs. informal context of recording –
of the usage of English / acquisition of English , , , –, –, , , –, –, , , fossilization frame, framing , , –, , , , , , , , – framework , –, , –, –, –, , –, , , , French , – friends –, , , –, –, –, –, –, , –, , , , , , , –, , –, , , – G gender , –, –, , –, –, –, , , , –, , , –, –, , generalization , – German expression –, , , grammaticalization , , Great Britain , , , , –, , (see also British influence / data / speakers) H hedge , , , , , , hesitation , , –, , , , , , , – I ICE ICLE ideational , , , –, , , –, , , , , , , , imagine the scene –, , –, , , , implication see see the implication implied result –, , , , , impolite ,
indirect answer –, –, , , , – indirectness informal see formal initial position –, , , , , , interactional , –, –, , , , , , , , , , , , –, , , –, , , , –, , , – interference , , International Corpus of English see ICE International Corpus of Learner English see ICLE interview –, , , , , –, , , , , invented material J Japanese , , , , , L L1 , , L2 –, lab –, , – large corpora – learner English – learning English see formal lexical focus see focus lexical search –, –, , , , –, LINDSEI LLC , –, , , London Lund Corpus of Spoken English see LLC London Teenage Language see COLT Longman Dictionary of Contemporary English Louvain International Database of Spoken English Interlanguage see LINDSEI M Macmillan Dictionary for Advanced Learners , , , , main idea unit , , –, , , main story , , , –, , , ,
Subject index map task –, modal particles , , movie narratives (silent movies) – multifunctionality –, –, –, , N narratives , –, , –, –, , , , , , , , , , – native speaker contact , , –, –, –, , , , native speaker – naturalistic negative face New Zealand , –, next scene , –, , , , , – O OALD , –, , , –, , , , , –, OED , , , , , , opinion see continuing an opinion, contributing an opinion, speech act marker – opinion optionality –, , , (see also syntactic features, syntactic position) orality –, , –, , , , overgeneralization overlap overuse , Oxford Advanced Learners’ Dictionary see OALD Oxford English Dictionary see OED P pause, filled see filled pause PERL , phonological features / criteria , , , , phonological reduction – pitch , , , plane , –, , –, ,
positive face pragmatic competence see competence pragmatic expression , , , , –, preposition , , –, , – procedural , , , – proposition , –, –, –, , , –, , –, , , , , , , , , psycholinguistics , , psychological distance see distance purpose (expressing purpose) Q question – questionnaire , –, , , , –, , , , , , , –, – quotation quality quotative well –, , , you know –, , like , –, , , , –, , –, –, , R real , , , –, , –, , , recording (restricted vs. unrestricted context) – reference to shared knowledge see shared knowledge relationship between partners , –, , , , , , , , , – relevance , , , , , , –, , , –, , –, –, , , repair , , , –, –, –, , –, , , , , rephrasing –, request – response to self-raised expectations –, ,
result or consequence –, , (see also implied result) rewording –, role play , , , , role –, –, –, , –, –, Russian S Santa Barbara Corpus of Spoken American Language , , scene see imagine the scene, next scene scope , , , , , search see content search, lexical search searching for the appropriate expression , –, , , , , , searching for the right phrase , –, , , , , see the implication –, , , –, self-raised expectations see response semantic content –, –, sequential , , , , –, , –, , shared knowledge , –, , –, , , , small corpora – social class –, –, sociocultural competence see competence Spanish , , speech act marker –, , , , speech act , –, , , spoken , –, –, , , , , , , , , , , , , , , , , , –, –, , , , , , , , , spontaneous speech , , , , , stranger , –, –, , , , , , , , , , , strategic competence see competence substitute
Discourse markers in English discourse summarizing – syntactic features , , , , , – (see also optionality) syntactic position –, , –, , T teaching –, , , , –, , , teenage language / teenager –, –, –, , –, , , textbook – textual , , , –, –, –, , –, , –, –, , , , , , , –, , , , –, , , , –, –, –, –, , , , –,
tone group , tone unit topic –, –, –, , , , , –, –, –, , –, –, , –, , , –, , , , –, transfer , –, –, , transition relevance place –, , , , translation , , , –, , , , , – truncation, truncated intonation unit, truncated word , , –, –, , –, , , –, , , –, , , turn , –, –, , , , , –, –, , , –, –, –, , , –, ,
U usage of English , , (see also formal)
V Valley Girl Talk –, voice quality , –, –, , , –
W waffle phenomenon Webster’s New World Dictionary of American English , –, , –, , Wellington Corpus of Spoken New Zealand English word class –, , written –, , , –, –, , , , , , ,
In the Pragmatics & Beyond New Series the following titles have been published thus far or are scheduled for publication: 144 BÜHRIG, Kristin and Jan D. ten THIJE (eds.): Beyond Misunderstanding. Linguistic analyses of intercultural communication. Expected April 2006 143 BAKER, Carolyn, Michael EMMISON and Alan FIRTH (eds.): Calling for Help. Language and social interaction in telephone helplines. 2005. xviii, 352 pp. 142 SIDNELL, Jack: Talk and Practical Epistemology. The social life of knowledge in a Caribbean community. 2005. xvi, 255 pp. 141 ZHU, Yunxia: Written Communication across Cultures. A sociocognitive perspective on business genres. 2005. xviii, 216 pp. 140 BUTLER, Christopher S., María de los Ángeles GÓMEZ-GONZÁLEZ and Susana M. DOVAL-SUÁREZ (eds.): The Dynamics of Language Use. Functional and contrastive perspectives. 2005. xvi, 413 pp. 139 LAKOFF, Robin T. and Sachiko IDE (eds.): Broadening the Horizon of Linguistic Politeness. 2005. xii, 342 pp. 138 MÜLLER, Simone: Discourse Markers in Native and Non-native English Discourse. 2005. xviii, 290 pp. 137 MORITA, Emi: Negotiation of Contingent Talk. The Japanese interactional particles ne and sa. 2005. xvi, 240 pp. 136 SASSEN, Claudia: Linguistic Dimensions of Crisis Talk. Formalising structures in a controlled language. 2005. ix, 230 pp. 135 ARCHER, Dawn: Questions and Answers in the English Courtroom (1640–1760). A sociopragmatic analysis. 2005. xiv, 374 pp. 134 SKAFFARI, Janne, Matti PEIKOLA, Ruth CARROLL, Risto HILTUNEN and Brita WÅRVIK (eds.): Opening Windows on Texts and Discourses of the Past. 2005. x, 418 pp. 133 MARNETTE, Sophie: Speech and Thought Presentation in French. Concepts and strategies. 2005. xiv, 379 pp. 132 ONODERA, Noriko O.: Japanese Discourse Markers. Synchronic and diachronic discourse analysis. 2004. xiv, 253 pp. 131 JANOSCHKA, Anja: Web Advertising. New forms of communication on the Internet. 2004. xiv, 230 pp. 130 HALMARI, Helena and Tuija VIRTANEN (eds.): Persuasion Across Genres. A linguistic approach. 2005. x, 257 pp. 129 TABOADA, María Teresa: Building Coherence and Cohesion. Task-oriented dialogue in English and Spanish. 2004. xvii, 264 pp. 128 CORDELLA, Marisa: The Dynamic Consultation. A discourse analytical study of doctor–patient communication. 2004. xvi, 254 pp. 127 BRISARD, Frank, Michael MEEUWIS and Bart VANDENABEELE (eds.): Seduction, Community, Speech. A Festschrift for Herman Parret. 2004. vi, 202 pp. 126 WU, Yi’an: Spatial Demonstratives in English and Chinese. Text and Cognition. 2004. xviii, 236 pp. 125 LERNER, Gene H. (ed.): Conversation Analysis. Studies from the first generation. 2004. x, 302 pp. 124 VINE, Bernadette: Getting Things Done at Work. The discourse of power in workplace interaction. 2004. x, 278 pp. 123 MÁRQUEZ REITER, Rosina and María Elena PLACENCIA (eds.): Current Trends in the Pragmatics of Spanish. 2004. xvi, 383 pp. 122 GONZÁLEZ, Montserrat: Pragmatic Markers in Oral Narrative. The case of English and Catalan. 2004. xvi, 410 pp. 121 FETZER, Anita: Recontextualizing Context. Grammaticality meets appropriateness. 2004. x, 272 pp. 120 AIJMER, Karin and Anna-Brita STENSTRÖM (eds.): Discourse Patterns in Spoken and Written Corpora. 2004. viii, 279 pp. 119 HILTUNEN, Risto and Janne SKAFFARI (eds.): Discourse Perspectives on English. Medieval to modern. 2003. viii, 243 pp. 118 CHENG, Winnie: Intercultural Conversation. 2003. xii, 279 pp. 117 WU, Ruey-Jiuan Regina: Stance in Talk. A conversation analysis of Mandarin final particles. 2004. xvi, 260 pp. 116 GRANT, Colin B. (ed.): Rethinking Communicative Interaction. New interdisciplinary horizons. 2003. viii, 330 pp. 115 KÄRKKÄINEN, Elise: Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. 2003. xii, 213 pp. 114 KÜHNLEIN, Peter, Hannes RIESER and Henk ZEEVAT (eds.): Perspectives on Dialogue in the New Millennium. 2003. xii, 400 pp. 113 PANTHER, Klaus-Uwe and Linda L. THORNBURG (eds.): Metonymy and Pragmatic Inferencing. 2003. xii, 285 pp.
112 LENZ, Friedrich (ed.): Deictic Conceptualisation of Space, Time and Person. 2003. xiv, 279 pp. 111 ENSINK, Titus and Christoph SAUER (eds.): Framing and Perspectivising in Discourse. 2003. viii, 227 pp. 110 ANDROUTSOPOULOS, Jannis K. and Alexandra GEORGAKOPOULOU (eds.): Discourse Constructions of Youth Identities. 2003. viii, 343 pp. 109 MAYES, Patricia: Language, Social Structure, and Culture. A genre analysis of cooking classes in Japan and America. 2003. xiv, 228 pp. 108 BARRON, Anne: Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study abroad context. 2003. xviii, 403 pp. 107 TAAVITSAINEN, Irma and Andreas H. JUCKER (eds.): Diachronic Perspectives on Address Term Systems. 2003. viii, 446 pp. 106 BUSSE, Ulrich: Linguistic Variation in the Shakespeare Corpus. Morpho-syntactic variability of second person pronouns. 2002. xiv, 344 pp. 105 BLACKWELL, Sarah E.: Implicatures in Discourse. The case of Spanish NP anaphora. 2003. xvi, 303 pp. 104 BEECHING, Kate: Gender, Politeness and Pragmatic Particles in French. 2002. x, 251 pp. 103 FETZER, Anita and Christiane MEIERKORD (eds.): Rethinking Sequentiality. Linguistics meets conversational interaction. 2002. vi, 300 pp. 102 LEAFGREN, John: Degrees of Explicitness. Information structure and the packaging of Bulgarian subjects and objects. 2002. xii, 252 pp. 101 LUKE, Kang Kwong and Theodossia-Soula PAVLIDOU (eds.): Telephone Calls. Unity and diversity in conversational structure across languages and cultures. 2002. x, 295 pp. 100 JASZCZOLT, Katarzyna M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 2. 2003. viii, 496 pp. 99 JASZCZOLT, Katarzyna M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 1. 2003. xii, 388 pp. 98 DUSZAK, Anna (ed.): Us and Others. Social identities across languages, discourses and cultures. 2002. viii, 522 pp. 97 MAYNARD, Senko K.: Linguistic Emotivity. Centrality of place, the topic-comment dynamic, and an ideology of pathos in Japanese discourse. 2002. xiv, 481 pp. 96 HAVERKATE, Henk: The Syntax, Semantics and Pragmatics of Spanish Mood. 2002. vi, 241 pp. 95 FITZMAURICE, Susan M.: The Familiar Letter in Early Modern English. A pragmatic approach. 2002. viii, 263 pp. 94 McILVENNY, Paul (ed.): Talking Gender and Sexuality. 2002. x, 332 pp. 93 BARON, Bettina and Helga KOTTHOFF (eds.): Gender in Interaction. Perspectives on femininity and masculinity in ethnography and discourse. 2002. xxiv, 357 pp. 92 GARDNER, Rod: When Listeners Talk. Response tokens and listener stance. 2001. xxii, 281 pp. 91 GROSS, Joan: Speaking in Other Voices. An ethnography of Walloon puppet theaters. 2001. xxviii, 341 pp. 90 KENESEI, István and Robert M. HARNISH (eds.): Perspectives on Semantics, Pragmatics, and Discourse. A Festschrift for Ferenc Kiefer. 2001. xxii, 352 pp. 89 ITAKURA, Hiroko: Conversational Dominance and Gender. A study of Japanese speakers in first and second language contexts. 2001. xviii, 231 pp. 88 BAYRAKTAROĞLU, Arın and Maria SIFIANOU (eds.): Linguistic Politeness Across Boundaries. The case of Greek and Turkish. 2001. xiv, 439 pp. 87 MUSHIN, Ilana: Evidentiality and Epistemological Stance. Narrative Retelling. 2001. xviii, 244 pp. 86 IFANTIDOU, Elly: Evidentials and Relevance. 2001. xii, 225 pp. 85 COLLINS, Daniel E.: Reanimated Voices. Speech reporting in a historical-pragmatic perspective. 2001. xx, 384 pp. 84 ANDERSEN, Gisle: Pragmatic Markers and Sociolinguistic Variation. A relevance-theoretic approach to the language of adolescents. 2001. ix, 352 pp. 83 MÁRQUEZ REITER, Rosina: Linguistic Politeness in Britain and Uruguay. A contrastive study of requests and apologies. 2000. xviii, 225 pp. 82 KHALIL, Esam N.: Grounding in English and Arabic News Discourse. 2000. x, 274 pp. 81 DI LUZIO, Aldo, Susanne GÜNTHNER and Franca ORLETTI (eds.): Culture in Communication. Analyses of intercultural situations. 2001. xvi, 341 pp. 80 UNGERER, Friedrich (ed.): English Media Texts – Past and Present. Language and textual structure. 2000. xiv, 286 pp. 79 ANDERSEN, Gisle and Thorstein FRETHEIM (eds.): Pragmatic Markers and Propositional Attitude. 2000. viii, 273 pp.
A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com