Unity and Diversity of Languages
Unity and Diversity of Languages
Edited by
Piet van Sterkenburg University of Leiden, The Netherlands and Free University of Brussels, Belgium
John Benjamins Publishing Company Amsterdam / Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Unity and diversity of languages / edited by Piet van Sterkenburg. p. cm. Includes bibliographical references and index. 1. Linguistics. 2. Language and languages. P125 .U55 2008 410--dc22 isbn 978 90 272 3248 9 (Hb; alk. paper)
2008018020
© 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Table of contents
Preface
vii
Plenary sessions Sign languages East and West Susan D. Fisher
3
Lifting the veil: Uncovering language origin Jean Aitchison
17
Pragmatics and the lexicon Laurence R. Horn
29
Syntactic constraints Jane Grimshaw
43
Aspects of the neural representation of spoken language Grzegorz Dogil
59
From concepts to meaning: The role of lexical knowledge James Pustejovsky
73
Language rights, human development and linguistic diversity in a globalizing world Suzanne Romaine Formal semantics for interpreting temporal annotation Kiyong Lee
85 97
Parallel sessions Words, mind and brain Gary Libben
111
Information structure: Notional distinctions, ways of expression Caroline Féry & Manfred Krifka
123
Table of contents
Language policy: The first half-century Bernard Spolsky
137
Intercultural pragmatics, language and society Shoshana Blum-Kulka, Menahem Blondheim, Juliane House, Gabriele Kasper & Johannes Wagner
155
Historical linguistics in 2008: The state of the art Brian D. Joseph
175
Lexical semantics Sebastian Löbner
189
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report Johan van der Auwera & Hana Filip
201
Syntax: The state of the art Farrell Ackerman & James P. Blevins
215
Index of terms
231
Preface The Permanent International Committee of Linguists (Comité International Permanent des Linguistes, CIPL) is organizing the 18th Congress of Linguists to be held in Seoul on July 21–26, 2008, in close collaboration with the Linguistic Society of Korea. CIPL was founded in 1928 during the first International Congress of Linguists, which took place in The Hague (The Netherlands). The Congress constituted an important landmark in the study of linguistics since it was the first time that linguistics presented itself as an autonomous science to the world. The Permanent Committee included such leading linguists as Charles Bally, Franz Boas, Otto Jespersen, Daniel Jones, Antoine Meillet. Subsequent congresses were held in Geneva, Rome, Copenhagen, and after the war, among other places, in Cambridge (Mass.), Tokyo, and Quebec reflecting the idea that the sites of the congresses should not be restricted to Europe. The linguistic congresses have thus gradually developed into truly international gatherings and have constituted the only major event in linguistics until the drastic expansion of the field in the seventies of last century. The topics discussed at the congresses, too, reflected the rapid growth of linguistics as a science and its manifold interfaces with psychology, sociology, anthropology, computer science, philosophy and many other disciplines. At the same time more and more international linguistic conferences have been organized in the various subfields of linguistics. The International Congress of Linguists remains an important event because it fulfills the function of a general congress in linguistics and it is unique in this respect. The invited talks in this book address hot topics in various subdisciplines presented by outstanding and internationally well known experts. In addition, the state-of-the-art papers provide an overview of the most important research areas of contemporary linguistics. In her lecture Sign Languages East and West Susan Fischer is showing us sign languages are natural languages whose study can tell us a great deal about the phenomenon of human language, and indeed can contribute to linguistic theories. Although the modern linguistic study of sign languages goes back about 40 years, the systematic linguistic study of Asian sign languages is much more recent. There are surprising and interesting differences between Eastern and Western sign languages that make us rethink our ideas about sign languages just as typological studies in spoken language make us rethink our ideas about language in general.
Preface
Jean Aitchison outlines in Lifting the veil: Uncovering Language Origin some basic assumptions, ideas which are accepted by most serious scholars in the field. She also points out areas in which scholars have recently made useful advances in knowledge. Finally she pays special attention to some topics which are currently attracting widespread interest, especially grammaticalization, which has leapt to the forefront in recent discussions of language origin, as well as to the problem of universal conceptual metaphors. Laurence Horn focuses in his study Pragmatics and the Lexicon on the role of conversational implicature in the formation and distribution of lexical items and in meaning change, as well as the functioning of such contextually dependent notions as Roschian prototype effects and Aristotelian expectations. He has touched on a variety of lexical phenomena whose existence, form, distribution and history depend in crucial ways on principles of pragmatics. He proposes several such principles, including the two key components of a Manichaean model of implicature, whose interaction results in the expectation that the speaker will provide necessary specification for the interests of the hearer (Q) while avoiding unnecessary specification (R). While these principles reformulate observations about communication tracing back to Paul, Zipf, and Grice, he has also invoked Roschian prototypes and Aristotelian privation to account for the patterns exhibited by lexical clones and un-nouns. Without an understanding of the roles played by pragmatics, no treatment of lexical relations and meaning change can be fully explanatory. Jane Grimshaw is writing on Syntactic Constraints. He describes that harmonic bounding relations among syntactic candidates entail economy of structure. No phrase satisfies all syntactic constraints, hence every phrase has a cost: the violations it incurs. The more phrases a candidate contains, the more violations it incurs. Ranking is critical in selecting a candidate with more output structure, which can never harmonically bound a smaller candidate. In contrast, a smaller candidate can harmonically bound a larger, because its violations can be a proper subset of those incurred by the larger. This general result is illustrated for the distribution of complementizers in Korean and English. While optima containing multiple functional projections can be picked by ranked universal constraints, the number of functional projections is limited by harmonic bounding. In Aspects of the neural representation of spoken language Grzegorz Dogil is dealing with what has become explicit since the beginning of this century, that new imaging techniques can produce a deluge of information correlating linguistic and neural phenomena. Our understanding of the inter-relationship between data coming from linguistics and brain sciences has led to many fascinating discoveries in neurolinguistics, psycholinguistics and the cognitive studies of human language. He reports some major findings from experimental work investigating some aspects of spoken language processing in an unimpaired human brain. Besides he
Preface
presents findings of localization information provided by the hemodynamic fMRI based method and correlates them with high-quality temporal data generated by the electromagnetic MEG-technique. James Pustejovsky outlines in his contribution From Concepts to Meaning, a theory of compositionality for natural language utterances. He is aiming to extend the framework of Classic Generative Lexicon (GL) presented in Pustejovsky (1995) and Bouillon and Busa (2001) by characterizing the scope of the selectional mechanisms available to the grammar. The major claim of his paper is that coercion phenomena are ubiquitous in language, and must be accounted for within any theory of linguistic semantics. Coercion operations are subject to identifiable constraints, which can be stated as general conditions on the computation of semantic derivations. Linguists predict that 60 to 90% of the world’s 6,900 some languages may be at risk of extinction within the next 100 years, writes Susanne Romaine in her contribution Language Rights, Human Development and Linguistic Diversity in a Globalizing World. Although the fate of linguistic diversity has not figured centrally in the world debate on the consequences of globalization, addressing the question of why languages matter requires a new understanding of the critical role of language rights, and linguistic diversity in human development. Maintaining the world’s languages is part of a larger strategy of cultural survival. Policies that promote a community’s economic and cultural well-being will be likely to sustain linguistic diversity as well. Viewing language maintenance as integral to the promotion of sustainable, appropriate, empowering development means thinking about languages in the same way as other natural resources requiring careful planning and protection. Kiyong Lee aims at developing an OWL-Time ontology-based semantics to formally represent temporal and event-related information extracted from annotated text. Natural language text is annotated for such information in the XML format according to the guidelines specified by TimeML. The OWL-Time ontology, on the other hand, provides a formal language in which annotated information is translated into appropriate semantic representations. To this language, the notion of neighborhood is introduced to capture finer-grained aspectual and tense-related features in natural language, particularly, those features related to the past tense in Korean. It is claimed in his article Formal Semantics for Interpreting Temporal Annotation that OWL-Time and TimeML complement each other in processing relevant information and reasoning about time and events in natural language text. Gary Libben writes about Words, Mind and Brain. He argues that multimorphemic words offer valuable insight into the intersection of language, mind, and brain. As re-combinations of existing elements, multimorphemic words
Preface
have multidimensional links to other words. These links are fundamentally psychological. Thus, the study of lexical variables such as frequency, morphological family size, etc. can be seen as the study of mind-brain properties. He suggests that these mind-brain properties can be best studied through the use of experimental techniques that simultaneously correlate behavioural data and cerebral activity. It is expected that key advances over the next five years will involve the use of such measures to probe individual differences in lexical processing and the relation of lexical processing to the fundamentally social nature of human language. Manfred Krifka (et aliis) gives us a plausible explanation about Information Structure. He contends that information structure, the packaging of information to satisfy the immediate communicative needs, exerts a powerful force on all structural levels of language. He shows how this concept can be defined, he argues for focus, givenness, topic, frame setting and delimitation as important subconcepts, and he illustrates the wide variety in which these information structural functions are expressed in languages. “Language policy is now firmly established as a subfield of sociolinguistics”, says Bernard Spolsky in his article Language Policy: The First Half-century. It has not yet developed the firm relationship with political science that it needs but there are important signs of the beginning of cross-fertilization. Understandably, many scholars working with endangered languages become active in their support, lessening their objectivity but demonstrating their strong sense of social responsibility. Many others, convinced by the arguments for minority rights that they start by studying, tend to become advocates rather than analysts of specific language policies. The difficulties of actually managing languages lead from time to time to skepticism as to whether it is possible, and the complexity has delayed the development of any consensus about a theory. But the growing interest in the topic gives promise of interesting new developments. Shoshana Blum-Kulka (e.a.) discuss topics addressed in the session on Intercultural Pragmatics, Language and Society. The papers included represent four distinct themes: Globalization of Discourse, International Workplace Communication, Historical Pragmatics and Interculturality as Discursive Construction. These four themes represent key issues in the study of language and society, moving from the universal to the particular, from current concerns to historical underpinnings. Thus exploring the features of English as a Lingua Franca highlights universally shared properties of linguistic usage in a more and more globalized world, and work on communication at the workplace brings to the fore arguably universal principles of interaction used in bilingual communities of practice. On the other hand, the historical pragmatic study of written texts from periods others than the current one shares with all historical endeavors its emphasis on the unique and the particular, be it the use of a specific discourse marker or the norms
Preface
of politeness in given texts and periods. The recent shift in intercultural pragmatics towards constructionist approaches touches on themes of both sharedness and distinctiveness. Thus it allows for a double view on interculturality, conceptualizing interculturality as a discursive achievement, as an ongoing negotiation which may foreground (or suppress) cultural distinctiveness. Brian D. Joseph presents an overview of the current state of the field of historical linguistics, with an eye to identifying enduring questions and tested methodologies but also new opportunities and new methods. “Giving an account of the state of the art in a field as diverse and wide as Lexical Semantics is something one can only pretend to try”, writes Sebastian Löbner. His survey is therefore restricted to an attempt to describe the present state of the art as it is reflected in the contributions submitted for the Parallel Session on “Lexical Semantics” at this Seoul congress. Inevitably, the picture emerging from this approach will be biased, among others due to the geographical location of the conference. About one half of the contributions are from East Asian countries, from Korea, Taiwan, China, Japan, Thailand; approximately the same number of contributions have been submitted by European, Russian and North American linguists. His paper offers a brief survey of the talks presented, placing emphasis on theoretical orientation, methodology and major topics. In (Tense,) Aspect, Mood and Modality Johan van der Auwera & Hana Filip give an account of some of the very recent work on tense, aspect, mood and modality. The recency is taken very strictly: they focus on the work published in and after 2000. The imperfection of the survey is such that it cannot possibly be exhaustive, largely because of the obligatorily modest size of the paper. For this reason, the paper restricts the survey to those areas that they think have attracted most intellectual energy. In functional theories, they believe that it is mood and modality, more so than tense and aspect, that have fascinated linguists. For formal theories, it is their impression that it is aspect, rather than tense and mood and modality, that has been enjoying pride of place. Of course, these impresssions are influenced by the authors’ own research interests (functional and modal for van der Auwera, formal and aspectological for Filip). Farrell Ackerman (e.a.) argues in Syntax: The State of the Art that while the state of syntactic theory is at a crossroads, with favoured theoretical constructs, familiar methodologies, and consensus assumptions all in play, the prospects for syntactic research are arguably brighter than ever, providing data and problems for a new era of theory construction based on the detailed study of lesser studied languages and the utilization of quantitative, computational, and experimental techniques. We are deeply grateful to dr. Carole Tiberius for reading the drafts of the articles with a critical and discerning eye and for compiling the index of used
Preface
linguistic terms. She bears no responsibility, however, for what finally appears. Karin van Weerlee, secretary to the secretary-general of CIPL, gave a lot of support on the editorial side and offered many suggestions for improvement. We would like to thank the members of the Local Organizing Committee of CIL 18 in Seoul: Prof. Suk-Jin Chang, Prof. Hong Pin Im, Prof. Ik-Hwan Lee (Chair of the LOC), Prof. Chai-song Hong, Prof. Young-Se Kang, Prof. Yongsoon Kang, Prof. Seong-Kyu Kim, Prof. Chung-Mok Suh, Prof. Jaeyoung Han, Prof. Jong-Yurl Yoon, Prof. Jeonghan Han and Prof. Jin-ho Park for their invaluable help in the compilation of papers. We would also like to thank the members of John Benjamins’ staff for their help in preparing the present volume. The International Congress of Linguists remains an excellent and unique occasion to learn about linguistic research other than our own and to meet colleagues from other fields and to discuss with them linguistic questions of mutual interest.
Ferenc Kiefer, President of CIPL Piet van Sterkenburg, Secretary-General of CIPL
Plenary sessions
Sign languages East and West1 Susan D. Fischer
University of California, San Diego, USA
1. Introduction Deaf people cannot hear, yet they have an innate capacity for language. That capacity is, however, sorely tested for the majority of deaf children who are not exposed to a visually accessible language until, frequently, it is too late for them to fully master any language, signed or spoken. Those deaf children fortunate enough to have deaf parents, or to have hearing families who learn to sign when the children are first diagnosed, can develop a sign language that is equal in grammatical complexity and expressive ability to any spoken language. Deaf people have been denied access to sign languages due in part to misconceptions that even many linguists have about sign languages. In this paper, I am going to briefly debunk those misconceptions; I shall then talk about the kinds of insights that the examination of sign language structures can contribute to linguistics, as well as the other way around. Following that, I shall discuss the reasons for looking at a variety of sign languages, and then exemplify those reasons by showing some comparisons between Western and Asian sign languages.
2. Debunking misconceptions about sign languages When I tell acquaintances that I do research on sign languages, their first reaction is usually surprise that “sign language” isn’t universal. This misconception is based
. Acknowledgments: I am grateful to the countless Deaf consultants with whom I have worked in Japan, Taiwan, and the United States, as well as both Deaf and hearing colleagues and collaborators who have given feedback on this work. Some of the material in this chapter appears in different form in Fischer & Gong (in press). Jamie Gould served as the “talent” for the illustrations.
Susan D. Fischer
in turn on a number of erroneous assumptions; all of these myths are both false and easily falsifiable: 1. 2. 3. 4.
Sign languages are iconic, and [therefore?] bound to the concrete; Sign languages are just debased forms of spoken languages; Sign languages have no internal structure or grammar; Sign languages are not languages.
First, with regard to iconicity, it is true that many signs in sign languages are what we call motivated, i.e., one can understand the source once told what that source is. And the proportion of motivated words is undoubtedly higher in sign languages than in spoken languages.2 I have argued elsewhere (Fischer 1979) that sign languages have more motivated signs because they can; the visual channel is more conducive to motivated words than the auditory channel. Motivated signs become more arbitrary over time, similar to what has happened to Chinese characters (Frishberg 1975), so that the motivation is not there synchronically. Even motivated signs can express abstract concepts. For example, in ASL (American Sign Language), the sign SURPRISE, which can indicate both mental and physical surprise is made as in Figure 1.
Figure 1. (ASL) SURPRISE, OWE, JUSTICE.
Figure 1 also shows the ASL signs for “owe”, and “justice”; signs for countless other abstract concepts exist. Just because some signs are motivated does entail that all signs are motivated synchronically, and just because some signs are motivated, it does not follow that they are necessarily bound to the concrete. Humans are ingenious at making do with the linguistic materials they have at their disposal. Furthermore, although signs may look iconic, an iconic sign in one sign language is often different from an iconic sign for the same concept in another sign
. Even spoken languages vary with respect to the number of motivated words in the language; many Bantu languages have many idiophones, as does Japanese.
Sign languages East and West
Figure 2. ASL, JSL TREE, BASEBALL.
language, depending on which aspects of the referent are selected for representation. In the ASL sign for “tree”, for example, the arm represents the trunk of the tree and the fingers represent the branches and leaves; in the Japanese sign for “tree”, fingers of both hands outline the shape of the trunk. Another way in which motivated signs can differ from one another is perspective; in the ASL sign for “baseball”, the hands and arms swing an imaginary bat; in the Japanese and Taiwanese signs for the same concept, the hands represent the bat hitting the ball. See Figure 2. The perception that sign languages are iconic in turn feeds the myth that sign language is universal. Deaf people do seem to be able to communicate across national boundaries and with hearing persons who do not know how to sign. However, the way they do so is usually by going outside the linguistic system and using mime. They are also more used to having to use mime and are therefore
Susan D. Fischer
both more comfortable and more skilled at doing so. Many gestures from the surrounding culture will be grammaticized into a particular sign language; for example, the headshake that often accompanies negative sentences in English also marks negation in ASL, while the headnod that marks negation in Greek culture also marks negation in Greek sign language. The reason that people outside the deaf community sometimes get the impression that a sign language is just a debased form of the spoken language is due to the fact that users of the language will code-switch to make their signing approach the grammar of the surrounding spoken language. Real sign languages are therefore difficult for outsiders to observe. Contact between signed and spoken languages occurs through schools or families. Although sign language grammars are influenced by the structure of the spoken languages surrounding them, all sign languages have grammatical structures and features not to be found in the surrounding spoken languages. For example, ASL has object agreement where English does not, and JSL (Japanese Sign Language) has object and subject agreement as well as gender marking, which spoken Japanese does not. Note also that sign language families do not coincide with spoken language families: ASL and BSL (British Sign Language) are mutually unintelligible; TSL (Taiwan Sign Language) and CSL (Chinese Sign Language) are historically unrelated, though some CSL vocabulary and one grammatical process have recently found their way into TSL.3,4 ASL is closer to the sign languages of France, Russia, and Denmark; Users of JSL, TSL, and Korean Sign Language can interact easily even though Japanese and Mandarin are not related, and Japanese and Korean are related at best distantly. The impression that sign languages have no grammar may be engendered by the fact that transcription systems until quite recently have been woefully inadequate to represent the multilayered meaning contained in the sign; a gloss of a signed sentence makes it look impoverished, but the fault is in the gloss, not in the language. Another reason for the impression that sign language have no grammar may be that sign languages tend to depend more on inflections than on word order for the expression of grammatical relations; rich agreement is frequent and when present permits greater freedom of constituent order (Fischer 1974; Lillo-Martin 1991).
. Sasaki (2007) argues that CSL influence on TSL was earlier. . It should be noted here that signing is a channel of communication, so that while one can sign ASL or KSL, it is also possible to sign “in English” or “in Korean”. This is what deaf people tend to do when interacting with hearing people who do not sign, and may be one of the factors that leads many naïve people to believe that sign languages have no independent grammar of their own.
Sign languages East and West
The assertion that sign languages are not languages is in turn based on many of the myths we have now debunked. 3. Sign language and education Belief in the myths discussed above has had disastrous effects on the education of deaf children. If you think that sign languages are primitive and bound to the concrete, it is an easy step to think that they are easy and therefore should be denied to deaf people lest they fail to learn spoken language. And if you think that sign languages are nothing but pantomime or have no grammar, it’s an easy step to then say that it doesn’t matter when it is learned. Although sign languages have been accepted in the European Union as full-fledged minority languages, and are taught as foreign languages in American high schools and colleges, such is not the case in Asia. In terms of deaf education, East Asia is roughly in the position that the US was in 40 years ago: many deaf people go to special schools where sign language is forbidden until about age 12–13, when it is too late for most children to fully learn any language.5 Since at most 10% of deaf children come from deaf families, this means that relatively few deaf people become proficient signers before the critical period is over. Most deaf children learn to sign from their peers outside the classroom. While sign language is by no means a panacea for the deprivation of deaf children from knowledge of the world, early acquisition permits deaf children to get an idea of what language is like and what language is for; it also helps to alleviate deaf children’s social isolation. 4. Sign languages are languages When we look at other people’s criteria for calling something a language, for example the criteria implicit in Chomsky’s work, sign languages pass with flying colors. Sign languages can express anything; they are recursive, they have a notion of what is grammatical and what is not, and they conform to the universals that
. The reasons for denying sign language to deaf children are complex. Doctors and educators believe (wrongly) that language is the same as speech; they believe (wrongly) that if deaf children are allowed to sign, they will never learn to speak; they also believe (wrongly) many of the myths that we have elucidated above. They may therefore also believe (wrongly) that there is no critical period for sign language acquisition and that therefore no harm is being done if sign language acquisition is delayed. The results of such beliefs are evident all around us; not students who have failed in school, but educational systems that have failed generations of children.
Susan D. Fischer
have been posited within perspectives as diverse as Chomsky’s and Greenberg’s. Why this is true is not difficult to discern. Barring secondary disabilities, deaf people have the same capacity for language. If denied access to vocal language, they will find some other avenue to acquire language, even if they have to invent it. 5. Why study sign languages? The study of sign languages, their structure, processing, and acquisition, can answer fundamental questions about the nature of language and of the human capacity for language. It also permits us to examine the notion of language universals from a fresh perspective. Charles Hockett (1960) posited 13 by now well-known “design features” for language. One is the use of a vocal-auditory channel. Several others (rapid fading, arbitrariness of the signal, and complete feedback) are based on the assumption of a vocal-auditory channel. Obviously, if a vocal-auditory channel is a necessary condition for something to be called language, then sign languages fail to meet that condition. But is it fair to exclude sign languages because they do not meet that condition? If a signer feels her hands proprioceptively while the viewer sees them, thus constituting incomplete feedback, is that a good reason to exclude sign languages from the class of languages? I think not. Rather, let us use sign languages to determine which posited universals truly apply to all languages and which are determined instead by the channel of communication. It is true that sign languages are learned in a different context from most spoken language, but that context is not unique to the signing channel. One close analogy is early creoles (Fischer 1978). Early creole speakers are exposed to a pidgin; the majority of sign language users are not exposed to a full-fledged sign language and must therefore create it themselves to a large extent. One difference is that many signers are not exposed to any accessible language from birth; rather, they tend to learn it later in childhood or even as adults, and often from their peers rather than from their parents. If they are exposed to a sign language from birth, that language may have been imperfectly learned by the parents and hence in the case of both sign languages and creoles, children must invent a language. The fact that most deaf children learn language relatively late helps us address issues such as the critical period hypothesis for language. Indeed there do seem to be critical periods for sign languages as there are for spoken languages (Fischer 1998; Mayberry & Eichen 1991). 6. Relation of sign language to spoken language linguistics When sign linguistics began about 50 years ago, scholars were interested in how to apply linguistic theories to these newly-prominent languages. Thus, William
Sign languages East and West
Stokoe, the father of sign linguistics, applied Trager & Smith’s phonemic model to ASL and came up with important analyses of the subcomponents of signs (Stokoe, Casterline & Croneberg 1965). Liddell (1984) later was able to find important new insights into sign language phonology by applying more recent theories of nonlinear phonology to ASL. I and others have also attempted to apply syntactic theories to the study of various sign languages (see, for example, Fischer 1974, 1990; Lillo-Martin 1991; Sandler & Lillo-Martin 2006). In all of these cases, except for unavoidable differences due to the difference between sound and sight, current and prior linguistic theories apply as well to sign languages as they do to spoken languages. That said, providing insight is a two-way street. Sign languages can also help to adduce evidence for linguistic theories that spoken languages cannot. A good example of this is nonlinear phonology. We know from the earliest work in autosegmental phonology (Goldsmith 1976) that in order for a prosodic feature to appear, it must be attached to a segment. A number of prosodic features in sign languages do not have this requirement; such independence of prosody from segments is possible because of the nature of the gestural-visual channel in which sign language is communicated. One type of prosody in sign languages is nonmanual signals such as facial expression. Some signs in ASL have associated facial expressions which can appear by themselves without a change in meaning even if the segmental (manual) portion of the sign vanishes. One example is an ASL adverbial conjunction often glossed as “PAH!” because of the mouthing that occurs with it, though the meaning is something like “as a good result”. Versions of this sign with and without the manual portion count as repetitions. The point is that because facial expressions are truly independent of the hands, they need not attach to a segment in order to surface. Another example is given below.
Figure 3. YOU EAT WHAT.
A third type of evidence that sign languages can provide for linguistic theory is in the realm of abstract syntax and semantics. It is well documented in the sign
Susan D. Fischer
language literature that even for a wh-phrase or wh-word occurring in situ, it is possible to have a prosodic wh-marker that spreads throughout the scope of the wh-operator; the same is true of negative words and the scope of negation as marked by the prosodic negative marker. In a sentence meaning “What did you eat?” but glossed as YOU EAT WHAT, the wh-facial expression can extend through the entire sentence. Another way in which sign language has pushed linguistic theory is the importance of discourse; many processes central to sign language grammars, particularly those involved in anaphora rarely show up in isolated sentences; rather, one must look at connected discourse to observe them in their natural setting.
7. Comparisons of sign languages Linguists have gained valuable perspectives from comparisons among spoken languages. If we knew only about a narrow range of Indo-European languages, our view of what language is would be radically different; we might miss phenomena ranging from clicks to ergativity. We have seen that looking at a sign language can give us fresh perspectives on language itself, such as redefining what we mean by linguistic universals. By the same token, it is important to compare different sign languages to find out what the range of grammatical possibilities is within that channel, and, if you will, to find universals of sign languages. For the last 15 years or so I have focused my research on comparing Western and Asian sign languages, and have been surprised at how I have had to rethink my own assumptions about sign languages. 7.1 History The earliest work on sign languages was on Western sign languages. Work on Asian sign languages was largely limited to phonology, lexicography, or language planning (e.g., Kanda 1994; Yau 1978; Honna & Kato 1996; Ann 2005). This has been remedied in recent years by work such as that of Smith (1990), Kimura & Ichida (1995), Fischer (1996), Fischer & Osugi (1998, 2000), Ichida (2004), Gong (2005), Hong (2006), Morgan (2006), and Tang and her students, but the amount of literature on Asian sign languages is still small compared with the West. The grammars of sign languages always differ from those of the surrounding spoken languages; however, there is some interaction between signing and writing, and urban sign language grammars are frequently influenced by the grammar of the surrounding spoken languages, especially when some form of signing is used in education. For example, TSL branched off from JSL. Sixty years after the end of
Sign languages East and West
the Japanese occupation of Taiwan, there is some evidence that constituent order has shifted in TSL from Japanese-like to Chinese-like order. 7.2 Sign language characteristics After 50 years of research on Western sign languages, there is agreement on the commonalities among them. When we compare them with Asian sign languages, new insights come into play. I shall review here some features of Western sign languages and show how they contrast (or don’t) with Asian sign languages. 7.3 Verb classes All developed sign languages, Western and Asian, have three classes of verbs: those that do not change their form to show argument structure, those that change their form to show object and sometimes subject, and those that show source, goal, and location (Padden 1988; Meir 2002). Agreement is marked by a shift in the movement of the predicate, frequently accompanied by a shift in the facing of the palm of the hand. In addition to the traditional kind of agreement, in the JSL family, which includes Japanese, Korean, and Taiwan sign languages, agreement verbs permit and in some cases require gender marking in addition to person marking (Smith 1989; Fischer & Osugi 2000; Hong 2006). These languages also permit the systematic use of what Fischer & Osugi (2000) call indexical classifiers, which serve as agreement anchors and permit subject as well as object agreement to be so marked. 7.4 Constituent order There is a tendency for sign languages to be verb-final on the surface. This is due in large part to mechanism of agreement: in order for verbs to agree with subject/ object or source/goal, a referential locus for each argument needs to be established prior to the occurrence of the inflected verb. A predicate then moves between two loci or towards the locus [usually] of the object. It is important, however, to distinguish between frequent and underlying word order. ASL, while frequently head-final on the surface, largely shows characteristics of head-initial languages (see Koopman 1984 on languages whose surface word order differs from underlying order for principled grammatical reasons). JSL is underlyingly head-final. CSL is head-initial. TSL shows some signs of being mixed-headed, perhaps because it has influences from both JSL and spoken Chinese. Thus, for example, both TSL and JSL have auxiliaries that function like do-support in English, in that they serve to carry agreement when the verb does not. In TSL, these auxiliaries tend to come before the verb, but in JSL they tend to
Susan D. Fischer
come afterwards. Other auxiliaries or serial verbs in ASL vs. JSL follow the same pattern, occurring head-finally (Smith 1990; Fischer 1996). 7.5 Topicalization Western sign languages tend to be topic-prominent. They have both syntactic (movement to sentence-initial position) and prosodic (brow raise) markers for topic. This is one place where there is a contrast to Asian sign languages. While JSL has topicalization and prosodic as well as syntactic markers for topic, Chinese Sign Language has movement for topicalization, but does not seem to require a prosodic marker (Junhui Yang, p.c.).6 7.6 Simultaneity Perhaps to compensate for the relatively slow speed of transmission of signs vs. spoken words (Bellugi & Fischer 1972), Western sign languages use many simultaneous rather than linear grammatical processes. These include agreement (mentioned above), and aspect, and also predicate classifiers, which take the form of substitute handshapes that abstract away from features of the referent. Predicate classifiers usually require an antecedent and frequently encode theme or instrument. They occur in both Western and Asian sign languages, but are encoded differently; for example, the handshape representing persons in ASL is an extended index finger, while in CSL it is extended thumb and pinky. 7.7 Interface with written language Many Western sign languages borrow vocabulary from the written language into the sign language via letter-by-letter fingerspelling or initialization in which the handshape for the first letter of the intended written language word is substituted for the handshape of the base sign. TSL has no fingerspelling; CSL, KSL, and JSL have fingerspelling, but it has not integrated fully into those languages; many older signers do not fingerspell at all. All East Asian sign languages, however, use signs that either depict or trace Chinese characters, including KSL, which no longer uses Chinese characters but retains character signs for some place names. Character signs in East Asian sign languages seem to have the same function as fingerspelling in Western sign languages (see Ann 1998; Fischer & Gong in press).
. There is some evidence (Quinto-Pozos 1999), that the sign language of Mexico lacks topicalization.
Sign languages East and West
7.8 Prosodic marking of syntactic forms Sign languages also exploit simultaneity through the use of prosodic cues to show the scope of operators such as negation and interrogation. For negation in Western sign languages, a headshake spreads across the scope of the negative operator; for polar questions, a brow raise extends over the entire question, and for content, questions, the eyebrows are furrowed (see Fischer 2006). Although Asian sign languages use prosodic markers for syntactic purposes, there are some differences vis-à-vis Western sign languages. Polar questions appear to work the same way as in Western sign languages; however, in JSL, at least, one wh-marker may occur at the end of the sentence without being attached to any segmental portion of the utterance, or it may cliticize to the last element of the sentence regardless of what is being questioned (Fischer & Osugi 1998). A second difference is that in CSL and HKSL (Hong Kong, apparently a dialect of CSL) some prosodic markers do not need to extend over the scope of the negation or question operator the way they must in Western sign languages. 8. Conclusions We study a variety of spoken languages to learn about language universals. Studying sign languages similarly enriches and broadens our knowledge of language and grammar and gives us further insights into the human mind. I have been able only to touch on the topics that have interested me for several decades, but hope that this paper will convince readers that sign languages are worthy of linguistic study and spur them to investigate the sign languages of their own countries. Except for sight vs. sound, every grammatical phenomenon found in sign languages can also be found in some spoken language. A wise friend once said, “It isn’t deaf people’s ears but what is between them that matters”. Sign languages indeed show the resilience of the human capacity for language. References Ann, Jean. 1998. Contact between a sign language and a written language: character signs in Taiwan Sign Language. In Ceil Lucas (ed.), Pinky extension and eye gaze: language use in deaf communities. Sociolinguistics in deaf communities 4: 59–99. Washington DC: Gallaudet University Press. Ann, Jean. 2005. A functional explanation of Taiwan Sign Language handshape frequency. Language and Linguistics 6: 217–246. Bellugi, Ursula & Susan Fischer. 1972. A comparison of sign languages and spoken languages: rate and grammatical mechanisms. Cognition 1: 173–200.
Susan D. Fischer Fischer, Susan. 1974. Sign language and linguistic universals. In Christian Rohrer & Nicolas Ruwet (eds), Actes du Colloque Franco-Allemand de Grammaire Transformationelle, Band II: Études de Sémantique et Autres, 187–204. Tübingen: Niemeyer. Fischer, Susan. 1978. Sign language and creoles. In Patricia Siple (ed.), Understanding language through sign language research, 309–331. New York: Academic Press. Fischer, Susan. 1979. Many a slip ‘twixt the hand and the lip: Applying linguistic theory to non-oral language. In Robert Herbert (ed.), Metatheory III: Application of Linguistics in the Human Sciences, 45–75. East Lansing: MSU Press. Fischer, Susan. 1990. The head parameter in ASL. In W. Edmondson & F. Karlsson (eds), Proceedings of the fourth international symposium on sign language research, 75–85. Hamburg: Signum Press. Fischer, Susan. 1996. The role of agreement and auxiliaries in sign language. Lingua 98: 103–120. Fischer, Susan. 1998. Critical periods for language acquisition: Consequences for deaf education. In Amatzia Weisel (ed.), Issues unresolved: New perspectives on language and deaf education, 9–26. Washington DC: Gallaudet University Press. Fischer, Susan. 2006. Questions and negation in American Sign Language. In Ulrike Zeshan (ed.), Interrogative and Negative Constructions in Sign Languages. Sign Language Typology Series No. 1, 165–197. Nijmegen: Ishara Press. Fischer, Susan & Qunhu Gong. in press. East Asian sign languages. In Diane Brentari (ed.), Sign Languages. Cambridge: Cambridge University Press. Fischer, Susan, & Yutaka Osugi. 1998. Feature movement in wh-questions: evidence from sign languages. Theoretical issues in sign language research 6, Washington, DC. Fischer, Susan, & Yutaka Osugi. 2000. Thumbs up vs. giving the finger: indexical classifiers in NS and ASL. Theoretical issues in sign language research 7, Amsterdam. Frishberg, Nancy. 1975. Arbitrariness and iconicity: Historical change in American Sign Language. Language 51: 696–719. Goldsmith, John. 1976. Autosegmental phonology. Cambridge, Massachusetts: MIT Dissertation. Gong, Qunhu. 2005. Shouyu wenti jianghua (About Chinese Sign Language). In Yulin Shen et al. (eds), Shuangyu longjiaoyu de lilun yu shijian (Theories and practice of Deaf bilingual education), 39–60. Beijing: Huaxia. Hockett, Charles. 1960. The origin of speech. Scientific American 203: 88–89. Hong, Sung-Eun. 2006. Agreement verbs in Korean Sign Language (KSL). Theoretical issues in sign language research 9, Santa Catarina, Brazil. Honna, Nobuyuki & Mihoko Kato. 1996. On the formation of Japanese new signs. In Proceedings of the XIIth congress of the world federation of the deaf, 512–515. Vienna: WFD. Ichida, Yasuhiko. 2004. Head movement and head position in Japanese Sign Language. Theoretical issues in sign language reserach 8, Barcelona. Kanda, Kazuyuki. 1994. A computer dictionary of Japanese Sign Language. In Inger Ahlgren, Brita Bergman, & Mary Brennan (eds), Perspectives on sign language usage: papers from the fifth international symposium on sign language research, vol. 2: 409–419. Durham: ISLA. Kimura, Harumi & Ichida. 1995. Hajimete no shuwa. Tokyo: Nihonbungeisha. Koopman, Hilda. 1984. The Syntax of Verbs: from verb movement rules in the Kru languages to universal grammar. Dordrecht: Foris. Liddell, Scott. 1984. THINK and BELIEVE: sequentiality in American Sign Language. Language 60: 372–399. Lillo-Martin, Diane. 1991. Universal grammar and American Sign Language: Setting the null argument parameters. Dordrecht: Kluwer.
Sign languages East and West
Mayberry, Rachel & Ellen Eichen. 1991. The long-lasting advantage of learning sign language in childhood: Another look at the critical period for language acquisition. Journal of Memory and Language 30: 486–512. Meir, Irit. 2002. A cross-modality perspective on verb agreement. Natural Language and Linguistic Theory 20: 413–450. Morgan, Michael W. 2006. Interrogatives and negatives in Japanese Sign Language. In Ulrike Zeshan (ed.), Interrogative and Negative Constructions in Sign Languages. Sign Language Typology Series No. 1: 91–127. Nijmegen: Ishara Press. Padden, Carol. 1988. Interaction of morphology and syntax in American Sign Language. New York: Garland. Quinto-Pozos, David. 1999. Word order in Mexican Sign Language (LSM: lenguage de signos mexicano). Unpublished paper, University of Texas. Sandler, Wendy & Diane Lillo-Martin. 2006. Sign language and linguistic universals. Cambridge: Cambridge University Press. Sasaki, Daisuke. 2007. Comparing the lexicons of Japanese Sign Language and Taiwan Sign Language: a preliminary study focusing on the difference in the handshape parameter. In David Quinto-Pozos (ed.), Sign language in contact: sociolinguistics in deaf communities 13: 123–150. Washington: Gallaudet University Press. Smith, Wayne. 1989. Morphological characteristics of verbs in Taiwan Sign Language. Bloomington, Indiana: Indiana University dissertation. Smith, Wayne. 1990. Evidence for auxiliaries in Taiwan Sign Language. In Susan Fischer & Patricia Siple (eds), Theoretical issues in sign language research, vol. 1: Linguistics, 211–228. Chicago: University of Chicago Press. Stokoe, William, Dorothy Casterline & Carl Croneberg. 1965. A Dictionary of American Sign Language on Linguistic Principles. Silver Spring, MD: Linstok Press. Tai, James H-Y. & Shiou-fen Su. 2006. Taiwan shouyu de huying gangshi [Agreement in Taiwan Sign Language]. Language and Linguistics Monograph Series Number W-5, 341–363. Yang, Junhui & Susan Fischer. 2002. The Expression of Negation in Chinese Sign Language. Sign Language & Linguistics 5: 167–202. Yau, Shun-Chiu. 1978. Les signes chinois: langage gestuel des sourds chinois. Paris: Editions langages croisés.
Lifting the veil Uncovering language origin Jean Aitchison
University of Oxford, UK
1. Introduction Der den Ursprung der Sprache hüllende Schleier ist gelüftet, nicht vollends aufgedeckt (Grimm 1851/1905: 196). “The veil covering the origin of speech is lifted, not fully raised” is a statement made by Jakob Grimm when he wrote an imaginative essay on language origin over a century ago. The same assessment is probably true today. Views on language origin have passed through several stages. Early imaginative guesswork was effectively ended towards the end of the nineteenth century, when the influential linguist William Dwight Whitney condemned conjecture on the topic as “mere windy talk”, which had given “the whole question a bad repute among sober-minded philologists” (Whitney 1873: 279). Language origin was therefore dismissed by most serious scholars for over a century. Chomsky, for example, has claimed that the whole question is “a complete waste of time”, speculating that language came about when “some mutation took place in the cells of prehuman organisms” (Chomsky 1988: 183). Eventually, the subject was rescuscitated. Two respected American academics, Steven Pinker and Paul Bloom, wrote a landmark article (1990), which was published in a well-known interdisciplinary journal, Behavioral and Brain Sciences, in which original papers are accompanied by peer commentaries. Pinker and Bloom argued that language came about by ordinary evolutionary processes, and that there is now a “wealth of respectable new scientific information relevant to the evolution of language that has never been properly synthesized” (Pinker & Bloom 1990: 707). Arguably, this article “demolished some intellectual roadblocks in progress in understanding the relation between evolution and language” (Hurford 1990: 736). Largely due to this key publication, the field has been transformed into one that is both respectable and lively. Books have proliferated (e.g., Aitchison 1996, Jackendoff 2002; Wray 2002; Botha 2003; Tallerman 2005; Dessalles 2007; Heine & Kuteva 2007; Kenneally 2007), and so have international conferences with accompanying collections of papers (e.g., Hurford et al. 1998; Knight et al. 2000; Trabant & Ward 2001).
Jean Aitchison
In this paper, I will do the following. First, I will outline some basic assumptions, ideas which are accepted by most serious scholars in the field. Second, I will point out areas in which scholars have recently concentrated their attention, and made useful advances in knowledge. Thirdly, I will home in on some topics which are currently attracting widespread interest, especially grammaticalization, which has leapt to the forefront in recent discussions of language origin, and also the ongoing interest in universal conceptual metaphors. 2. Basic assumptions I will start with some basic assumptions, points that are widely accepted. First, language is/was primarily spoken. It is of course well-known that human communication is multi-modal, and that gesture may support spoken speech. Sign language is these days recognized as “full” language, but it is unlikely that such gestures were ever developed into a fully-fledged system in the early decades of language. The vocal-auditory channel is therefore considered by most researchers to be primary. Second, with regard to location, Africa is widely believed to be the area of origin for human language, though the exact location, once thought to be the Rift Valley, is still debated, and some southern African sites are under investigation (e.g., Leakey & Walker 2003). The notion of polygenesis is now thought to be unlikely, and claims about other possible areas of origin, such as China, have mostly faded. Third, in relation to date, language is of relatively recent origin, certainly compared with the evolution of the human species. At one time, full language was dated fairly late, at around 50,000 BP. But the likely date has been creeping back. Somewhere between 100,000 and 50,000 is now a plausible guess as to the time of fully developed language, though increasingly, researchers are trying to reconstruct a possible proto-language (e.g., Bickerton 2000; Wray 2000; Jackendoff 2002; Givón & Malle 2002; Heine & Kuteva 2007). The proto-language possibly emerged in stages, the earliest of which might have been in existence for millennia. “Rough equivalents for “Heave ho!”, “Hi!”, “No!”, “Help!” and several other proto-words might have been in use for a long time” (Aitchison 1996: 103). Fourth, the relationship of human language to animal communication is becoming clearer. The extent to which it is continuous or discontinuous with animal communication has been widely discussed (e.g., Hauser 1996; Hauser; Chomsky & Fitch 2002; Anderson 2004). The various constituents of language need to be assessed separately, since they contain different degrees of overlap,
Lifting the veil: Uncovering language origin
with auditory mechanisms showing strong overlap, the brain medium overlap, and articulatory mechanisms displaying little overlap (Aitchison 1998). This separation allows for more directed research. 3. Some recent steps forward In a number of areas, useful advances have been made. On articulatory mechanisms, the descent of the larynx, once thought to be a phenomenon restricted to humans, is now known to be a primitive trait that we share with other mammals (Fitch 2002), though humans are the only species whose larynx is permanently descended. Early hominids, like other mammals, lowered their larynx when vocal izing, then, arguably, “as speech increasingly became a feature of human life, it became more efficient, faster, and more accurate to have the larynx adopt that position all the time” (Anderson 2004: 312). This and other topics relating to the mechanisms of speech raise a general “chicken and egg” question: how many of these mechanisms involved exploitation of systems which had arisen primarily for some other function, and were later used for language, and how many evolved specifically to serve language? As the descended larynx shows, the answer is often unclear. An old feature may be adopted for use in language, and then further adapted. The relationship of language to music has also been explored. Mammals have a range of vocalizations, especially gibbons, whose beguiling musical “singing” recalls Otto Jespersen’s suggestion that the first utterances of speech might be “like something between the nightly love-lyrics of puss upon the tiles and the melodious love-songs of the nightingale” (Jespersen 1922: 434) – though gibbons turn out to be something of a disappointment: their songs are tuneful but individual, rather than socially adopted (Mitani & Marler 1989). From the point of view of vocalization, birdsong is the animal communication system that is most similar to human language, as has long been pointed out (e.g., Marler 1970, 1998), though their systems developed independently. Many birds utter inherited calls alongside songs which are partially learned, just as human babies produce early inbuilt cries, and later acquire a full language system. Single bird notes, and single human sound segments are relatively meaningless, it is the sequence of sounds which is relevant. Young birds have a type of twittering subsong, while human babies babble as a precursor to real language. Both birdsong and human language are normally controlled by the left side of the brain, and some species of birds have recognizable dialects. “Song learning in birds and language learning in humans is fairly similar. Each comes into the world with the ability to learn patterns from within a particular range … Birds and babies are
Jean Aitchison
both capable in principle of learning any of the patterns typical for their species” (Anderson 2004: 150). With regard to mental ability, symbolization is sometimes assumed to be the key to language: “Symbolic thought does not come innately built in, but develops by internalizing the symbolic process that underlies language” suggests Terrence Deacon (1997: 22). Yet a weakness of this idea is the fact that other ape species show some ability to symbolize (e.g., Marler 1998). Symbolization cannot therefore be unique to humans in the way that language as a system is unique. Furthermore, symbolization is turning out to be more complex and multi-faceted than was once assumed (Marler, ib.), so a general claim of symbolic thought is not particularly helpful, although the final level could be significant. The final stage of symbolization may be “pure naming”, an ability to label a person or object even when the labeller is not requesting it. Human children acquire the so-called “naming insight” some time in their second year, although deaf children come to it late. Schaller (1991) found a deaf Mexican adult, Ildefonso, who showed that naming ability might be a growing awareness, rather than a sudden “insight”: Ildefonso named first numbers, then nouns, then verbs. The naming insight has been attributed to “mind-reading”. Mind-reading is the ability to put oneself into another person’s shoes, as it were, and to envisage their mental state. This mastery is assumed by many to be a key proficiency underlying human language. Mind-reading is an awareness that develops with age: three year olds are typically unable to achieve it, though four year olds mostly can. A recent talking point has been its possible connection to mirror neurons, which are found in both humans and apes. An Italian neuroscientist Giacomo Rizzolatti noticed that a section of the frontal lobe of a monkey’s brain (the ventral premotor area) fired when the monkey performed certain actions such as putting food in its mouth, or reaching for an object. Unexpectedly, the same neurons fired when the monkey watched another monkey performing the same task. Rizzolatti labelled these mirror neurons, and suggested that he may have identified the neurological basis of mind reading (Rizzolatti & Arbib 1998; Malle 2002; Stamenov & Gallese 2002; Arbib 2005) – though this topic remains controversial. All these areas show the liveliness of the field of language origin, its interdisciplinary nature, and the wide range of topics covered. One of the difficulties is how to assess findings in so broad a field, especially those outside one’s own comfort zone. In the remainder of this paper, I will restrict myself to topics which are recognizably human linguistic ones, rather than archaeological, ethological or neurological.
Lifting the veil: Uncovering language origin
4. The linguistic scenario Scenarios describing the possible phases of language development in humankind have been proposed by several authors. (e.g., Aitchison 1996; Jackendoff 2002; Bickerton 2000; Heine & Kuteva 2007). As noted earlier, a few proto-words could have been in existence for centuries before full language emerged, though exactly when and how these spilled over into “real language” is unclear. Arbib (2005) uses the helpful term “language readiness” in relation to properties of the brain that provide the capacity to acquire and use language. In the 1960s, the linguist Charles Hockett attempted to list the “design features” of human language, and to identify those which make it unique (e.g., Hockett 1960; Hockett & Altmann 1968), and others have followed his lead, though not all have agreed fully on the features Hockett proposed (e.g., Aitchison 2008). This approach can be useful, in that researchers can home in on features either very important or unique to language. Of these, the interwoven features of “arbitrariness”, “discreteness”, “transmission”, “duality of patterning” and “productivity” may be the most relevant. Arbitrary features are of course those in which there is no link between the sound uttered and the item symbolized. Discrete features are those which are individually identifiable, rather than on a continuum. Transmission indicates that the system is at least partially handed down from one generation to another. Duality of patterning is a key aspect of language, as it is with some birdsong: each individual sound unit is normally meaningless, but only becomes meaningful when sounds are combined into longer sequences. Language is “open” or productive when sound segments and morphemes can be re-organized in a patterned way to form an indefinite number of new utterances. The key to human language is largely duality of patterning or in linguistic terminology, phonology and syntax, which, as far as we know, are unique to humans. Anderson comments: “No other animal seems to employ anything like a discrete combinatorial system” (2004: 313). How this all came about is therefore of key importance. 5. Grammaticalization I will now discuss linguistic processes which are still in operation, and may well have been in use since the beginning of language. As Heine and Kuteva state it: “The emergence and development of human language is the result of a strategy whereby means that are readily available are used for novel purposes” (2007: 17).
Jean Aitchison
This has been labelled grammaticalization, a term which has been increasing in scope in recent years (e.g., Hopper & Traugott 2003), but ultimately describes how words can be converted into grammar. The lexicon must have preceded grammar. No sentences could have been formed without the words used to compose them. Grammaticalization, or grammaticization, is a term originally coined by the French linguist, Antoine Meillet, who defined it as “the attribution of a grammatical character to a previously autonomous word” (Meillet 1912/1948: 331), though Meillet does not sufficiently emphasize that grammaticalization is an ongoing, never-ending process. A useful updating of Meillet’s definition is that of Heine and Kuteva (2002: 2): “Grammaticalization is defined as the development from lexical to grammatical forms and from grammatical to even more grammatical forms”, which emphasizes the ongoing nature of the process. Yet even this view is ultimately too narrow, because the autonomous word which becomes grammaticalized does not itself necessarily disappear. It typically remains in the language alongside its newer grammaticalized form and usage: as Paul Hopper noted almost twenty years ago: “New layers are continually emerging … the older layers may remain to coexist with and interact with the newer layers” (Hopper 1991: 22). Grammaticalization involves a number of interrelated linguistic devices. Heine and Kuteva (2002: 2) list four main mechanisms: a. desemanticization: loss of meaning content; b. decategorialization (or recategorization, my addition JA): loss or change of morphosyntactic properties; c. erosion: loss of phonetic content; d. extension: use in new contexts. The first three of these potentially involve loss, while the last is reminiscent of Hopper’s layering. At the time of true language origin, nouns are likely to have come first, since nouns in English and many other languages can stand on their own, as names for people, animals and objects. These have been called “first-order entities” (Lyons 1977: 442). They are time-stable, in that they change minimally over time. Humans give priority to people, secondarily to animals, and thirdly to objects, though it is unclear where natural forces such as winds and waves should be classified, as these can be personified: “O wild West Wind, thou breath of Autumn’s being” wrote the British poet Shelley in his “Ode to the West Wind” (1819). Verbs can often be derived from nouns, as in Felix bicycled home, but the reverse is less usual, and normally requires extra morphological complexity, as in Vanessa’s discovery amazed us all (Aitchison 1996). Recategorization or reanalysis
Lifting the veil: Uncovering language origin
is the most probable route for the development of verbs from nouns. (Reanalysis is a word intentionally avoided by Heine and Kuteva (2002), as it can be theorydependent, but almost any linguistic term could be avoided for this same reason). The recategorization/ re-analysis of nouns possibly came about via ambiguous phrases, as seen in some modern pidgins. Tok Pisin, spoken in Papua New Guinea, can show how this might have happened. Words such as danis “dance” or pait “fight” are ambiguous as to their part of speech. Mi danis could be interpreted either as “I was at a dance”, or “I was dancing”, and mi pait could either be “I was at a fight” or “I was fighting” (Aitchison 1996). The original speaker may have intended the first meaning, but the hearer could have interpreted it as the second. As Heine and Kuteva, among others, have pointed out, “language change is a by-product of communicative intentions [and communicative (mis)interpretations, my addition JA] not aimed at changing language” (2007: 17). Nouns and verbs are not totally distinct, they are on a continuum. As Givón (1979: 321) stated, the most time-stable percepts are lexicalized as nouns, the least time-stable are rapid events and actions, such as flash, jump, which typically emerge as verbs, and are the lexical items which are most likely to have been re-interpreted. Other parts of speech (adjectives, adverbs, pronouns, and so on) depend on the existence of nouns and verbs, and can all get reinterpreted (Heine and Kuteva 2002 provide numerous examples). An intriguing finding of grammaticalization studies is that the same changes recur in different languages, particularly those relating to word meaning: words meaning know, for example, typically change from “know” to “know how to” to “be skilled at” to “be accustomed to”, as has happened in Tok Pisin, spoken in Papua New Guinea. All these meanings may be used alongside one another. The writers of a toothpaste advertisement possibly exploited the ambiguity in order to impress a wide range of potential customers: Colgate i save strongim tit bilong yu can be interpreted in its earlier meaning as “Colgate knows how to strengthen your teeth”. This later came to mean “Colgate is skilled at strengthening your teeth”, and eventually “Colgate is accustomed to strengthen your teeth’, though in spoken speech, the meaning “be accustomed to” is usually shortened to sa. Mi sa kukim long paia “I customarily/usually cook it on the fire”, explained Janet, one of my Papua New Guinea informants, but she retained the long form when she used save to mean “know”. Sometimes both sa and save came into the same sentence, with different meanings: mipela sa mekim dispela kain tu, mi no save tumas long kukim, tasol mama bilong mi em save gut tru long kukim. “We customarily make this kind too, I don’t know very much about cooking, but my mother she knows a lot about cooking”. The English ability word can behaves similarly, coming from Old English cunnan “to know”, “know how to”.
Jean Aitchison
However, it is not just grammaticalization that reveals that languages today behave in very much the same way as they always have. Typological studies reveal a similar story. In the languages of the world, animate actors tend to come first, and verbs and objects are closely bonded (Hopper & Thompson 1984, Aitchison 1996). A sample of over 400 languages around the world showed that almost 90% of them had SOV (subject-object-verb) or (SVO) subject-verb-object word order (Tomlin 1986) (though exact figures are impossible to calculate, mainly because of the difficulties caused by including some ergative languages in this count.) So far, I have considered primarily syntactic processes of grammaticalization. Yet at an earlier stage, it seems likely that humans paid attention only to certain ontological categories, which formed the conceptual framework to which the syntax could apply. Humans pay attention to space before time, e.g., from tree to tree came before from day to day. They pay attention to themselves and their bodies, before they move outwards and inwards. The human head was named before the word was re-applied metaphorically to describe the top of a mountain, as in Tok Pisin het bilong maunten. They learned also to describe inner feelings via outer events, as in I see what you mean (Sweetser 1990). These naming processes have probably been in use for millennia. 6. Final comments Darwin (1859/1964, 1871/1998) was the major pioneer in our understanding of evolution, and we still follow in his footsteps, with minor additions and emendations. Some people have puzzled over the evolutionary question of how any new behaviour, such as speech, came about in the first place. Yet this has been understood for a long time. The process has been labelled Baldwinian evolution, after the American psychologist James Mark Baldwin, who suggested in 1902 that behavioural variation might provide the key. A new trait might emerge when a subgroup of a species moves into a niche slightly different from its ancestors. Behavioural flexibility and learning can then amplify and bias natural selection. From the point of view of language, say, those who had better memory skills for vocabulary might be better able to communicate and therefore have a better chance of survival. They might hand this advantage on to their offspring, who in turn could further extend their language skills. Perhaps we need to ask why humans, alone of mammals, acquired the ability to speak. The answer may be convergence. Humans, like other apes, began with the advantage of good hearing, combined with the habit of habitual vocalization. Then humans began to walk upright, acquiring an L-shaped vocal tract which
Lifting the veil: Uncovering language origin
allowed them to produce stable vowels and a clear output. This converged with an enlarged brain, which eventually led to a theory of mind, and the naming insight. These two together led to a “naming explosion”, a huge increase in the lexicon. A large lexicon led to word combinations. Word combinations revealed inbuilt human preferences, such as the “me first” preference, which in the longer run led to animate first. Another preference was action-object closeness. These preferences eventually hardened into linguistic “rules” of subject first, followed by object verb or verb object. Language was on its way. “Man has an instinctive tendency to speak …” stated Charles Darwin in 1871, asserting that language “has been slowly and unconsciously developed by many steps” (1871/1998: 89). This comment was made over a century ago. We are now finally, slowly but consciously finding out the linguistic steps taken by our ancestors.
References Aitchison, Jean. 1998. On discontinuing the continuity-discontinuity debate. In James R. Hurford et al. (1998), 17–29. Aitchison, Jean. 1996/2000. The seeds of speech: Language origin and evolution. Cambridge: Cambridge University Press 1996, enlarged Canto edition 2000. Aitchison, Jean. 2008. The articulate mammal: An introduction to psycholinguistics, 5th edn. London: Routledge. Anderson, Stephen R. 2004. Dr. Dolittle’s delusion: Animals and the uniqueness of human language. New Haven: Yale University Press. Arbib, Michael. 2005. The mirror system hypothesis: how did proto-language evolve? In Maggie Tallerman (2005), 21–47. Baldwin, James M. 1902. Development and evolution. NY: Macmillan. Bickerton, Derek. 2000. How protolanguage became language. In Chris Knight, Michael StuddertKennedy & James R. Hurford (2000), 264–284. Botha, Rudolf P. 2003. Unveiling the evolution of language. Amsterdam: Elsevier. Chomsky, Noam. 1988. Language and problems of knowledge: The Managua lectures. Cambridge, MA: MIT Press. Darwin, Charles. 1859/1964. On the origin of species. Cambridge, MA: Harvard University Press. Darwin, Charles. 1871/1998. The descent of man. NY: Prometheus Books. Deacon, Terrence. 1997. The symbolic species: The co-evolution of language and the human brain. London: Allen Lane, The Penguin Press. Dessalles, Jean-Louis. 2000/2007. Why we talk: The evolutionary origins of language. Translated by James Grieve. Oxford: Oxford University Press. Fitch, W. Tecumseh. 2002. Comparative vocal production and the evolution of speech: reinterpreting the descent of the larynx. In Alison Wray (2002), 21–45. Givón, Talmy. 1979. On understanding grammar. NY: Academic Press. Givón, Talmy & Bertram F. Malle. 2002. The evolution of language out of pre-language. Amsterdam: John Benjamins.
Jean Aitchison Grimm, Jakob. 1851/1905. Über den Ursprung der Sprache. Auswahl aus den kleinen Schriften, 162–246. Hamburg: Im Gutenberg-Verlag Dr. Ernst Schultze. Hauser, Marc D. 1996. The evolution of communication. Cambridge, MA: MIT Press. Hauser, Marc D., Noam Chomsky & W. Tecumseh Fitch. 2002. The faculty of language: What is it, who has it, and how did it evolve? Science 298: 1569–79. Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Heine, Bernd & Tania Kuteva. 2007. The genesis of grammar: A reconstruction. Oxford: Oxford University Press. Hockett, Charles F. 1960. The origin of speech. Scientific American 203, 88–96. Hockett, Charles F. & Stuart Altmann. 1968. A note on design features. In Thomas A. Sebeok (ed.), Animal communication, 61–72. Bloomington, IN: Indiana University Press. Hopper, Paul J. 1991. On some principles of grammaticalization. In Elizabeth Closs Traugott and Bernd Heine (1991), 17–36. Hopper, Paul J. & Sandra Thompson. 1984. The discourse basis for lexical categories in universal grammar. Language 60: 703–83. Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd edn. Cambridge: Cambridge University Press. Hurford, James R. 1990. Beyond the roadblock in evolution studies. Behavioral and Brain Sciences 23: 736–7. Hurford, James R., Michael Studdert-Kennedy & Chris Knight (eds). 1998. Approaches to the evolution of language. Cambridge: Cambridge University Press. Jackendoff, Ray. 2002. Foundations of language: Brain, meaning, grammar, evolution. Oxford: Oxford University Press. Jesperson, Otto. 1922. Language, its nature, development and origin. London: Allen and Unwin. Kenneally, Christine. 2007. The first word: The search for the origins of language. NY: Penguin. Knight, Chris, Michael Studdert-Kennedy & James R. Hurford. 2000. The evolutionary emergence of language. Cambridge: Cambridge University Press. Leakey, Meave & Alan Walker. 2003. Early hominid fossils from Africa. Scientific American 13(2): 14–19. Lyons, John. 1977. Semantics. 2 vols. Cambridge: Cambridge University Press. Malle, Bertram F. 2002. The relation between language and theory of mind in development and evolution. In Talmy Givón and Bertram Malle (eds), 265–284. Marler, Peter. 1970. Birdsong and speech development: could there be parallels? Scientific American 58: 669–673. Marler, Peter. 1998. Animal communication and human language. In Nina G. Jablonski & Leslie C. Aiello (eds), The origin and diversification of language, Memoirs of the California Academy of Sciences Number 24: 1–20. Meillet, Antoine. 1912/1948. Linguistique historique et linguistique générale. Paris: Champion. Mitani, John Cary & Peter Marler. 1989. A phonological analysis of male gibbon singing behavior. Behavior 106: 13–45. Pinker, Steven & Paul Bloom. 1990. Natural language and natural selection. Behavioral and Brain Sciences 13, 707–784. Rizzolatti, Giaccamo & Michael A. Arbib. 1998. Language within our grasp. Trends in Neuroscience 21: 188–194. Schaller, Susan. 1991. A man without words. Berkeley, CA: University of California Press.
Lifting the veil: Uncovering language origin
Stamenov, Maxim I. & Vittorio Gallese (eds). 2002. Mirror neurons and the evolution of brain and language. Amsterdam: John Benjamins. Sweetser, Eve Eliot. 1990. From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Tallerman, Maggie (ed.). 2005. Language origins: Perspectives on evolution. Oxford: Oxford University Press. Tomlin, Russell. 1986. Basic word order: Functional principles. London: Croom Helm. Trabant, Jürgen & Sean Ward (eds). 2001. New essays on the origin of language. Berlin: Mouton de Gruyter. Traugott, Elizabeth Closs & Bernd Heine. 1991. Approaches to Grammaticalization, vol. 1. Amsterdam: John Benjamins. Whitney, William Dwight. 1873. Oriental and Linguistic Studies vol. 1. NY: Charles Scribner’s Sons. Wray, Alison. 2000. Holistic utterances in protolanguage: The link from primates to humans. In Chris Knight, Michael Studdert-Kennedy & James R. Hurford (2000), 285–302. Wray, Alison (ed.). 2002. The transition to language. Oxford: Oxford University Press.
Pragmatics and the lexicon Laurence R. Horn Yale University, USA
1. Lexical pragmatics: Minding ones’s Q’s and R’s For students of the lexicon and lexical semantics, the need for an explicit domain of lexical pragmatics was not recognized until the publication thirty years ago of McCawley’s “Conversational implicature and the lexicon.” The term “lexical pragmatics” itself seems not to have predated Blutner (1998), who offers the following characterization: Lexical pragmatics is a research field that tries to give a systematic and explanatory account of pragmatic phenomena that are connective with the semantic underspecification of lexical items. Cases in point are the pragmatics of adjectives, systematic polysemy, the distribution of lexical and productive causatives, blocking phenomena, [and] the interpretation of compounds. (Blutner 1998: 115)
While these phenomena can be fruitfully explored through the lens of lexical pragmatics (as I have undertaken to do in earlier works: cf. Horn 1972, 1984, 1989, 1993, 2006, 2007, along with related work by McCawley 1978; Atlas & Levinson 1981; and Levinson 2000), the domain has overflowed these banks. (See for example the Relevance-theoretic characterization in Wilson & Carston 2007.) My focus in the present study will be on the role of conversational implicature in the formation and distribution of lexical items and in meaning change, as well as the functioning of such contextually dependent notions as Roschian prototype effects and Aristotelian expectations. Here and elsewhere, the particular notion of implicature I assume is based on the Manichaean model (Horn 2007) invoking countervailing Q and R principles. While derived from the program of Grice (1989) in its broad outlines, the inspiration for this dualist approach is as neo-Paulist as it is neo-Gricean. In his Prinzipien der Sprachgeschichte – whose English translation appeared in 1889, precisely a century before Grice’s posthumous collection saw the light of day – Paul surveys a range of phenomena whose form and distribution reflect the interplay of two functional principles, the tendency to reduce expression
Laurence R. Horn
and the contextually determined communicative requirements on sufficiency of information: The more economical or more abundant use of linguistic means of expressing a thought is determined by the need … Everywhere we find modes of expression forced into existence which contain only just so much as is requisite to their being understood. The amount of linguistic material employed varies in each case with the situation, with the previous conversation, with the relative approximation of the speakers to a common state of mind. (Paul 1889: 351)
The direct descendants of Paul’s dualism include the two opposed communicative economies of Zipf and Martinet and the interdefined halves of Grice’s Maxim of Quantity. While most linguists would free-associate G.K. Zipf ’s name with the (speakerbased) principle of least effort, the Zipfian framework (1935; 1949: 20ff.), actually distinguishes a speaker’s economy, which would tend toward “a vocabulary of one word which will refer to all the m distinct meanings,” from the countervailing auditor’s economy, tending toward “a vocabulary of m different words with one distinct meaning for each word.” The Speaker’s Economy places an upper bound on the form of the message, while the Hearer’s Economy places a lower bound on its informational content. By Zipf ’s law of abbreviation, the relative frequency of a word is inversely correlated with its length; the more frequent a word’s tokens, the shorter its form. Frequency, and its effect on utterance length and phonological reduction, is relativized to the speaker’s assumptions about the hearer and their shared common ground: High frequency is the cause of small magnitude … A longer word may be truncated if it enjoys a high relative frequency [either] throughout the entire speech community [or] if its use is frequent within any special group. (Zipf 1935: 31 – 32)
Zipf ’s two mutually constraining mirror-image forces are periodically invoked (or rediscovered) in the diachronic and psycholinguistic literature: The linguist must keep in mind two ever-present and antinomic factors: first, the requirements of communication, the need for the speaker to convey his message, and second, the principle of least effort, which makes him restrict his output of energy, both mental and physical, to the minimum compatible with achieving his ends. (Martinet 1962: 139) The speaker always tries to optimally minimize the surface complexity of his utterances while maximizing the amount of information he effectively communicates to the listener. (Carroll & Tanenhaus 1975: 51)
This minimax of effort or complexity on the one hand and informative content or distinctness on the other is directly reflected in the tension between articulatory economy and perceptual distinctness, as explained in the work of phoneticians and phonologists from Lindblom and MacNeilage to Hayes and Flemming; cf. Horn (2006) for references and discussion. It has also been clear for some time
Pragmatics and the lexicon
that the Zipfian parameter of familiarity is a major player on the phonological field; segmental and prosodic reduction and simplification mark (or unmark) familiar or frequent items, while unfamiliar or unpredictable words are assigned (or retain) extra stress or pitch. More apposite for our purposes is the recognition of the trade-off between brevity and clarity within classical rhetoric, as captured in Horace’s dictum Brevis esse laboro; obscurus fio (“I strive to be brief; I become obscure” – Ars Poetica, line 25). Similar sentiments appear elsewhere, often encapsulating the Golden Mean (or Goldilocks?) Principle: “If it is prolix, it will not be clear, nor if it is too brief. It is plain that the middle way is appropriate…, saying just enough to make the facts plain” (Aristotle, Rhetoric, 3.12–3.16). The speaker aims to maximize ease and brevity, correlates of least effort; well before Zipf, the “principle of economy”, “the principle of least effort”, or “laziness” figure prominently in reflections by Sweet (1874), Sayce (1880), Paul (1889), and Passy (1890) on sound change, synonymy avoidance, word meaning and meaning change. On the other hand, the hearer requires sufficiency of content and discriminability of form. Speaker and hearer are aware of their own and each other’s desiderata, a mutual awareness that generates a variety of effects based on what was said and what was not. It is this interaction that led me to fold the maxims of conversation (Grice 1989: 26ff.) into two general mirror-image principles, dubbed Q and R in deference to Grice’s (first) Quantity submaxim and Relation maxim respectively. On this program, implicatures may be generated by either the Q Principle (essentially “Say enough,” a generalization of Grice’s first submaxim of Quantity) or the R Principle (“Don’t say too much,” subsuming the second Quantity submaxim, Relation, and Brevity). The hearer-oriented Q Principle is a lower-bounding guarantee of the sufficiency of informative content; collecting the first Quantity submaxim along with the first two “clarity” submaxims of Manner, it is systematically exploited to generate upper-bounding (typically scalar) implicata. The R Principle, by contrast, is an upper-bounding correlate of Zipf ’s principle of least effort dictating minimization of form; it collects the Relation maxim, the second Quantity submaxim, and the last two submaxims of Manner, and is exploited to induce strengthening implicata. The application of Q-based upper-bounding scalar implicature allows for a systematic and economical treatment of sets of logical operators and ordinary nonproposition-embedding predicates that can be positioned on a scale as defined by unilateral entailment: (1) Q-scales: logical operators 〈all, most, many, some〉 〈always, usually, often, sometimes〉
Q-scales: “ordinary” values 〈hot, warm〉 〈cold, cool, lukewarm〉
Laurence R. Horn
〈and, or〉 〈certain, likely, possible〉 〈the, a〉
〈excellent, good, OK〉 〈adore, love, like〉 〈thumb, finger〉
In each case, given an informationally stronger value S and a weaker value W plotted on the same scale 〈S, W〉 as determined canonically by unilateral entailment, in asserting […W…] I implicate, ceteris paribus, that I was not in an epistemic position to have asserted […S…] salva veritate, i.e., that I don’t know that S, and hence, all things being equal, that I know that ¬[…S…] holds. We thus predict that in general scalar predications will be ascribed a one-sided (lower-bounded) linguistic meaning pragmatically enriched to a two-sided communication: (2)
a. b. c. d.
You ate some of the cake. It’s possible she’ll win. He’s a knave or a fool. It’s warm.
what is said → what is communicated “some if not all’’ “some but not all’’ “at least possible’’ “poss. but not certain’’ “…and perhaps both’’ “…but not both’’ “…at least warm’’ “…but not hot’’
This model allows us to provide a satisfactory ending to the story of *O, i.e., to predict the non-occurrence of values corresponding to the O or southeast vertex of the Square of Opposition (Horn 1972; to appear b):
(3) A: all F are G
I: some F are G
E: no F are G
O: some F aren’t G; not all F are G
Thus, alongside the possible quantificational determiners all, some, no, we never find (in any language) an O determiner *nall; corresponding to the quantificational adverbs always, sometimes, never, we have no *nalways (= “not always”, “sometimes not”). We may have equivalents for both (of them), one (of them), and neither (of them), but never for *noth (of them) (= “not both”, “at least one … not” = the Sheffer stroke); we find connectives corresponding to and, or, and sometimes nor (= “and not”), but never to *nand (= “or not”, “not … and”). This asymmetry extends to modal operators (e.g., can’t = ¬CAN; mustn’t = MUST¬) and other values that can be plotted on the Square, as well as to O→E drift, the tendency for sequences that might be expected to yield O meanings get E interpretations, as when Dutch nimmer (lit., ¬IMMER, i.e., “nalways”) can only be understood
Pragmatics and the lexicon
as “never” , or when French Il ne faut pas que tu meures (lit. = ¬[You must die]) = “You must [not die]” . On the neo-Gricean approach to this asymmetry, the systematic, cross-linguistically attested restriction on the lexicalization or direct expression of values mapping onto the O vertex is attributable to the mutual Q-based implicature relation obtaining between the two subcontraries I and O and to the marked status of negation dictating the preference for I over O forms (cf. Horn, to appear b for elaboration).
2. R in the lexicon: Narrowing and strengthening Unlike the upper-bounding associated with Q-based implicature, R-based effects involve articulatory reduction and pragmatic strengthening: the speaker makes her contribution relatively brief or uninformative and counts on the hearer to recognize the operation of least effort and fill in the missing material. Where Q enjoins the speaker to “Provide necessary specification”, R dictates “Avoid unnecessary specification.” One obvious place to look for R-based effects in the lexicon is in processes like acronymy, blending, and clipping, reducing relatively long or complex descriptors for frequently invoked referents. These processes, direct reflexes of Zipf ’s Law of Abbreviation, cannot operate unchecked: “The demands of the speech functions must set a limit to the economic tendency” (Stern 1931: 257). That is to say, Q constrains R. R-based effects are a robust factor in semantic change, but here it is largely information rather than articulatory complexity that is economized. In the wellknown process of sense narrowing, we can distinguish Q-based narrowing, which is linguistically motivated and results from the hearer-oriented tendency to avoid ambiguity, from R-based narrowing, the socially motivated restriction of a set-denoting term to its culturally salient subset or member. Q-based narrowing, where the existence of a specific hyponym of a general term licenses the use of the general term for the complement of that hyponym’s extension, is illustrated in (4). (4)
animal cat finger lion rectangle
(including or excluding humans, birds, fish) (including or excluding kittens) (including or excluding thumbs) (including or excluding lionesses) (including or excluding squares)
In R-based narrowing, a general term denoting a set narrows to pick out a culturally or socially salient subset, allowing the speaker to avoid overtly specifying the subdomain via the assumption that the hearer will fill in the intended
Laurence R. Horn
meaning. Unlike the cases in (4), this process does not depend on the hearer’s inference from what was not said, but counts instead on his recognition of what need not be said. The result may be a complete referential shift as in (5) or the development of autohyponyms (where the original broader sense persists alongside the narrowed meaning) as in (6): (5)
corn deer hound liquor poison wife
(6)
drink (in particular [+alcoholic]) friend (in the sense of “friend-plus’’ or “umfriend’’) man (orig. “human’’, now chiefly “male adult human’’) number (in particular “integer’’) smell (as intransitive, = “stink’’) temperature (in particular, one in the “fever’’ range) Ger. Frau, Fr. femme, Span. mujer (“woman’’, “wife’’)
(“wheat’’ [England], “oats’’ [Scotland], “maize’’ [U.S.]) (originally “(wild) animal’’, as in Ger. Tier) (originally “dog’’, as in Ger. Hund) (originally “liquid substance’’) (originally “potion, drink’’) (originally “woman’’, as in (6) below)
The locus classicus of socially motivated R-based narrowing is euphemism, ranging from sex (sleep with, make love; lover) to excretion (toilet, go to the bathroom) to death and illness (disease, accident, undertaker). The Manichaean model of pragmatics is criticized by Carston (2002: Chap. 3; 2005: 314–15) on the grounds that both Q-based and R-based implicature involve “a strengthening of communicated content”; I have defended the model (Horn 2006; to appear a) in part through invoking the distinction between informative and rhetorical strength. But note also that the dualistic model makes it possible to tease apart the two conflicting motivations for lexical narrowing, and to see why we can predict that it is only in the latter case, where the hearer has no access to the information required to reconstruct the shift, that regularly leads to conventionalized semantic shifts (as opposed to on-line meaning restriction). Related to R-based lexical narrowing, and again motivated by social considerations – in particular, those relating to what Erving Goffman describes as respect for negative face – is the strengthening of negative expression. As Bosanquet (1911: 281) puts it, “The essence of formal negation is to invest the contrary with the character of the contradictory.” Across a wide range of languages we find a tendency for the speaker to weaken the force of her intended negative judgments, counting on the hearer to fill in the intended stronger negative. In English, the resultant contrary negatives in contradictory clothing include affixal negation, the
Pragmatics and the lexicon
so-called “neg-raising” phenomenon, and simple litotes, as illustrated in (7a–c) respectively (cf. Horn 1989: Chapter 5). (7) R-based negative strengthening a. contrary readings for affixal negation He is unhappy (> ~[He is happy]) I disliked the movie (> ~[I liked the movie]) b. “neg-raising” effects across clause boundaries I don’t believe it’ll snow (= I believe it won’t) I don’t want you to go (= I want you not to go) c.
litotes (understatement) in simple denials He’s not happy with it (> ~[He’s happy with it]) You don’t like kimchee (> ~[You like kimchee])
In each case a general formally contradictory negation is strengthened to a specific, contrary understanding; where the constructions differ is in the degree of conventionalization of this strengthening. I say you don’t like kimchee precisely to avoid acknowledging your antipathy directly; at the same time, I count on your willingness to fill in the intended R-strengthened (contrary) interpretation rather than simply settling for the contradictory negation literally expressed. In an embedding environment, this same practice is responsible for the “neg-raising” effect seen in (7b), resulting in the understanding of a negation outside the scope of certain predicates of opinion, desire, or likelihood as if it had lower-clause scope. Here again, the contrary meaning (“x believes that not-p”) is sufficient but not logically necessary to establish the truth of the contradictory (“x does not believe that p”), yet it is treated as if it were necessary – not surprisingly, both because it represents the inductively salient case that makes the contradictory true and because there may be social and cultural constraints against the direct expression of the stronger contrary. (See Horn 2000 for a general account of the R-strengthening of sufficient to necessary-and-sufficient conditions, focused on so-called conditional perfection.) 3. Two case studies in lexical pragmatics: The clone and the un-word Within Relevance Theory (Sperber & Wilson 1986, Carston 2002), effort is conceived of as a property of the hearer’s processing tasks rather than the speaker’s production, but surely both are relevant in the shaping of linguistic form. In particular, Carston (2005: 314) challenges the status of arguments from Brevity (and the R principle subsuming it), whose role as a component of speaker’s effort she rejects. One area worth exploring in this light is that of repetition in discourse and reduplicative constructions in the lexicon and syntax.
Laurence R. Horn
There is a general, iconically motivated tendency for repetition, particularly of verbal or clausal units, to represent increased salience along one or more physical dimensions, as determined by aspectual considerations. When you say “We walked and walked and walked” or “We kissed and we kissed and we kissed” or “Tomorrow, and tomorrow, and tomorrow, creeps in this petty pace”, your iteration requires greater locutionary effort and commensurately implies a greater distance traversed, intensity achieved, or desperation felt. Our focus here, however, is on a lexical correlate of repetition instantiated in English and some other languages in the construction variously termed the double, contrastive focus reduplication (Ghomeshi et al. 2004), or the lexical clone. In many contexts, the reduplicant singles out an element or subset of the extension of the noun corresponding to a true, real, default, or (in the sense of Rosch 1978) prototype category member: a dog dog may be an actual canine (excluding hot dogs or unattractive people) or it may be a golden retriever or collie (excluding Chihuahuas and toy poodles), a salad salad (in Western culture) is based on lettuce, not tuna, potatoes, or squid, a DOCTOR doctor is an M.D., not a Ph.D. in linguistics, and so on. On the other hand, my request for a drink drink asks for the real, i.e., [+ alcoholic] thing: not a prototype potable but the culturally salient one, with the clone functioning as a quasi-euphemism. This is what we can call the value-added or intensifying use of the clone construction. This understanding is especially prevalent in the case of cloned adjectives: “It’s HOT hot” will tend to convey “very hot”, “hot-plus”, with the appropriate scalar intensification in whichever sense (temperature or spiciness) the context favors. Why should the clone exist, given its apparent communicative redundancy? As first observed in Dray (1987) (see also Horn 1993: 48–51), the lexical clone offers a natural laboratory for observing the interplay of the Q and R Principles. A clone XX is more effortful (to produce, and presumably to process) than the simple nominal X. On the other hand, the clone XX is less informative, and arguably less effortful (for speaker and/or hearer) than a phrase or compound YX, with a modifier Y distinct from the modified X. XX must be both necessary (as against X), given R, and sufficient (as against YX), given Q, to narrow the domain appropriately. (As Dray observes the speaker might find it harder to characterize the narrowed domain by spelling out the default overtly than by invoking it via the clone.) Dray also stresses the essential role of context in establishing and inferring the appropriate interpretation, as illustrated by the elegant minimal pair in (8): (8) a. Oh, we’re just `LIVing together living together. [` = fall contour] b. Oh, we’re not vliving together living together. [v = fall-rise]
Pragmatics and the lexicon
With the prototype clone in (8a), the speakers present themselves as just roommates, not romantically or sexually involved, while the negated clone in (8b) must be interpreted in the opposite (value-added) sense, with the result that both affirmative and negative sentences assert a platonic status. Three distinct functions can be isolated for the clone construction in English: (9) i. singling out prototype category members (esp. with nouns) ii. assigning a value-added or intensifying use (esp. with adjs.) iii. picking out a literal, as opposed to figurative, use
Interestingly, the same functions and contextual dependence turn up in certain non-clone constructions, as recognized by Austin (1963: 70): [A] definite sense attaches to the assertion that something is real, a real such-and-such, only in the light of a specific way in which it might be, or might have been, not real. “A real duck’’ differs from the simple “a duck’’ only in that it is used to exclude various ways of being not a real duck – but a dummy, a toy, a picture, a decoy, &c.; and moreover I don’t know just how to take the assertion that it’s a real duck unless I know just what on that particular occasion, the speaker has it in mind to exclude…[T]he function of “real’’ is not to contribute positively to the characterization of anything, but to exclude possible ways of being not real.
We might add that the real duck – perhaps imprinting on the DUCK duck – may also be a healthy rather than diseased or wounded specimen, or perhaps a normal duck rather than an albino or dwarf. The same range of functions is attested for clones cross-linguistically (see the survey in Ghomeshi et al. 2004: §2.2). In Chinese, as Chao Li (p.c.) notes, adjectives can sometimes be cloned in a syllable-by-syllable fashion: (10) A: Ta ba
fangjian dasao ganjing-le
ma?
he obj room sweep clean-perf quest
“Did he sweep the room clean?’’
B: Ganjing shi ganjing-le, dan hai bu shi ganganjingjing. clean be clean-perf but still not be very.clean “Yes, it’s clean, but not very clean’’ (= not CLEAN clean)
Other languages obtain some or all of the same pragmatic effects from a prefix rather than a clone or a dedicated modifer. Poser (1991) analyzes Japanese ma as a Roschian function: maX “restricts the denotation of the base form to [X’s] canonical point”, and with natural kind nouns to the core or prototype category members: (11) siro “white’’ kuro “black’’
massiro “snow white’’ makkuro “pitch black’’
Laurence R. Horn
kita “north’’ natu “summer’’ hadaka “naked’’ iruka “dolphin’’
makita “due north’’ manatu “midsummer’’ mappadaka “stark naked’’ mairuka “the common dolphin’’
In each case, Poser argues, the relevant salient exemplar is picked out, but salience is locally determined. Thus ma- represents a lexicalized counterpart of the deictic or “online” narrowing associated with the lexical clone construction in English. (Japanese ma- is represented by 真, the kanji character for Chinese zhēn (“real, really”), a lexical item that intensifies adjectives (zhen re = “really hot”) and adds the value “real, authentic” to some nouns, although not as productively as with ma.) The speaker can count on the hearer to be implicitly aware of the lexically and culturally prototype meaning normally picked out by the use of underspecified modifiers (clones, prefixes, adjectives), while at the same time recognizing that the hearer will be able to override the default interpretation if salient aspects of the context render another understanding more plausible. Given the role of general properties of meaning and context in deriving the narrowing signaled in these uses, we are dealing here with lexical pragmatics rather than lexical semantics, although conventionalization remains a possibility. English offers a constructional opposite of the prototype clone in the form of the un-noun. In accordance with the principles governing these productive formations (cf. Horn 2002, 2005), a class A un-noun of the form un-X is not an X, but is situated just outside a given category with whose members it shares a salient function – the un-cola is a carbonated soft drink just outside the cola domain, i.e., 7-Up, an unturkey is a large vegan treat designed for the holiday table, an un-publication is a scholarly article that was circulated but never published. Class A un-nouns are privatives in Aristotle’s sense. Privation for Aristotle is a fundamentally pragmatic notion of opposition, defined by disappointed expectation; privatives are marked exceptions lacking a property one would expect to find instantiated at the species or genus level. Just as not just any creature without teeth or eyesight cannot be called toothless or blind, as Aristotle notes (Categories 12a28–33), so too not just anything that isn’t a cola (e.g., a red wine, hamburger, or T-shirt) could qualify as an uncola. An A-class un-noun unX is Almost a member of the category X, but not quite. A B-class un-noun unX, on the other hand, is just Barely an X, i.e., a peripheral member of a given category. Thus, an uncollege (in fact, the U. S. Naval Academy) may lack fraternities, cheerleaders, and binge drinking, an unpolitician is someone seeking office who is not a traditional office-seeker – not a POLITICIAN politician (Bill Clinton, perhaps?); an unhotel is technically a hotel but one too homey or funky to qualify as a good exemplar of the class – not a HOTEL hotel. If the
Pragmatics and the lexicon
lexical clone marks the core of category membership, the B-class un-noun represents the periphery; an unmovie is the opposite of the movie-movie. But what counts as a default instance of the unX or the XX will be grounded inextricably in the speaker’s assumptions (and her assumptions about the hearer’s assumptions) concerning cultural norms and utterance context. 4. Concluding remarks In this study I have touched on a variety of lexical phenomena whose existence, form, distribution and history depends in crucial ways on principles of pragmatics. I have proposed several such principles, including the two key components of a Manichaean model of implicature, whose interaction results in the expectation that the speaker will provide necessary specification for the interests of the hearer (Q) while avoiding unnecessary specification (R). While these principles reformulate observations about communication tracing back to Paul, Zipf, and Grice, I have also invoked Roschian prototypes and Aristotelian privation to account for the patterns exhibited by lexical clones and un-nouns. Without an understanding of the roles played by pragmatics, no treatment of lexical relations and meaning change can be fully explanatory.
Acknowledgments Earlier versions of this material were presented at the LSA Linguistic Institute at Michigan State U. (Summer 2003), the LSA 2004 annual meeting (Boston), the 9th National Conference on Pragmatics (Shanghai, July 2005), and the Workshop on Word Meaning, Concepts, and Communication at Cumberland Lodge in Windsor (Sept. 2005). I am grateful to the participants at those events and to Barbara Abbott, Robyn Carston, Chao Li, and JL Speranza for comments and suggestions. Needless to say, …
References Atlas, Jay & Stephen Levinson. 1981. It-clefts, informativeness, and logical form. In Peter Cole (ed.), Radical Pragmatics 1–51. New York: Academic. Austin, John L. 1963. Sense and Sensibilia. Oxford: Clarendon Press. Blutner, Reinhard. 1998. Lexical pragmatics. Journal of Semantics 15: 115–162. Bosanquet, Bernard. 1911. Logic, Vol. 1, 2nd edn. Oxford: Clarendon.
Laurence R. Horn Carroll, John & Michael Tanenhaus. 1975. Prolegomena to a functional theory of word formation. Papers from the Parasession on Functionalism, 47–62. Chicago: Chicago Linguistic Society. Carston, Robyn. 2002. Thoughts and Utterances. Oxford: Blackwell. Carston, Robyn. 2005. Relevance theory, Grice, and the neo-Griceans: A response to Laurence Horn’s “Current issues in neo-Gricean pragmatics.” Intercultural Pragmatics 2: 303–20. Dray, Nancy. 1987. Doubles and modifiers in English. M.A. thesis, U. of Chicago. Ghomeshi, Jila, Ray Jackendoff, Nicole Rosen & Kevin Russell. 2004. Contrastive focus reduplication in English. Natural Language and Linguistic Theory 22: 307–57. Grice, Herbert Paul. 1989. Studies in the Way of Words. Cambridge: Harvard U. Press. Horn, Laurence. 1972. On the Semantic Properties of Logical Operators in English. UCLA dissertation. Distributed by Indiana University Linguistics Club, 1976. Horn, Laurence. 1984. Toward a new taxonomy for pragmatic inference: Q–based and R–based implicature. In D. Schiffrin (ed.), Meaning, Form, and Use in Context (GURT “84), 11–42. Washington: Georgetown University Press. Horn, Laurence. 1989. A Natural History of Negation. Chicago: U. of Chicago Press. Reissued Stanford: CSLI, 2001. Horn, Laurence. 1993. Economy and redundancy in a dualistic model of natural language. In Susanna Shore & Maria Vilkuna (eds), SKY 1993: 1993 Yearbook of the Linguistic Association of Finland, 33–72. Horn, Laurence. 2000. From if to iff: Conditional perfection as pragmatic strengthening. Journal of Pragmatics 32: 289–326. Horn, Laurence. 2002. Uncovering the un-word: A study in lexical pragmatics. Sophia Linguistica 49: 1–64. Horn, Laurence. 2005. An un-paper for the unsyntactician. In Salikoko S. Mufwene et al. (eds), Polymorphous Linguistics: Jim McCawley’s Legacy, 329–65. Cambridge: MIT Press. Horn, Laurence. 2006. Speaker and hearer in neo-Gricean pragmatics. Journal of Foreign Languages. 2–26. Horn, Laurence. 2007. Neo-Gricean pragmatics: a Manichaean manifesto. In Noel Burton-Roberts (ed.), Pragmatics, 158–83. Basingstoke: Palgrave. Horn, Laurence. to appear a. William James + 40: Issues in the investigation of implicature. In Klaus Petrus (ed.), Meaning and Analysis: New Essays on H. Paul Grice. Downloadable at http://www.yale.edu/linguist/faculty/horn_pub.html. Horn, Laurence. to appear b. Lexical pragmatics and the geometry of opposition: The mystery of *nall and *nand revisited. In Jean-Yves Béziau (ed.), Papers from the World Congress on the Square of Opposition. Downloadable at http://www.yale.edu/linguist/faculty/horn_pub.html. Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge: MIT Press. McCawley, James. 1978. Conversational implicature and the lexicon. In Peter Cole (ed.), Syntax and Semantics 9: Pragmatics, 245–59. New York: Academic Press. Martinet, Andre. 1962. A Functional View of Language. Oxford: Clarendon Press. Passy, Paul. 1890. Étude sur les changements phonétiques et leurs caractères généraux. Paris: Firmin-Didot. Paul, Hermann. 1889. Principles of the History of Language. Trans., H.A. Strong. London: Macmillan. Poser, William. 1991. Ma. In Carol Georgopolous & Roberta Ishihara (eds), Interdisciplinary Approaches to Language: Essays in Honor of S.-Y. Kuroda, 449–58. Dordrecht: Kluwer.
Pragmatics and the lexicon
Rosch, Elinor. 1978. Principles of categorization. In Eleanor Rosch & Barbara L. Lloyd (eds), Cognition and Categorization, 27–48. Hillsdale: Lawrence Erlbaum. Sayce, A.H. 1880. Introduction to the Science of Language. London: C. K Paul & Co. Sperber, Dan & Deirdre Wilson. 1986. Relevance: Communication and Cognition. Cambridge: Harvard U. Press. (2nd edn, 1995.) Stern, Gustaf. 1931. Meaning and Change of Meaning. Bloomington: Indiana U. Press. Sweet, Henry. 1874. History of English Sounds from the Earliest Period. London: English Dialect Society. Wilson, Deirdre & Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference, and ad hoc concepts. In Noel Burton-Roberts (ed.), Pragmatics, 230–59. Basingstoke: Palgrave. Zipf, George Kingsley. 1935. The Psycho-Biology of Language. New York: Houghton-Mifflin. Zipf, George Kingsley. 1949. Human Behavior and the Principle of Least Effort. Cambridge, MA: Addison-Wesley.
Syntactic constraints1 Jane Grimshaw
Rutgers University, USA
1. Introduction Economy of Structure is an important component in determining the set of structures and hence grammatical sentences allowed by a grammar. Economy of structure follows, I argue here, from Optimality Theory directly, not from auxiliary stipulations of economy. It follows from the logic of harmonic bounding, an architectural property of the theory, which results in a general preference for smaller structures over larger ones. The result holds given a particular set of universal constraints on structure. This paper exemplifies some of the ways in which economy of structure constrains the set of grammatical sentences using examples from English and Korean. I will argue that the distribution of certain complementizers in English and Korean can be explicated in terms of simple and general constraints on syntactic structure. English and Korean have distinct complementizer systems. Nevertheless there is no difference between them at a more abstract level. They are subject to the same constraints, and show the same harmonic bounding effects. Optimality Theory (Prince & Smolensky 2004) provides a principle which chooses an optimum among a set of competitors which do not violate the same constraints. Under this principle: “An optimal output form for a given input is selected from among the class of competitors in the following way: a form that, for every pairwise competition involving it, best satisfies the highest-ranking constraint on which the competitors conflict, is optimal.” (Grimshaw 1997). The interaction of the posited constraints guarantees that “larger candidates”, those with more structure, can harmonically bound “smaller” candidates, but not vice versa. Harmonic bounding is analyzed in Samek-Lodovici and Prince (2002). They state “Candidate sets are typically unbounded in size, but the set of distinct possible optima is inevitably finite. Almost all candidates are ‘‘losers’’, doomed
. This paper is part of a larger project which has been assisted by a plethora of colleagues, whose help I am extremely grateful for. I thank Daniel Altshuler, Seunghun Lee and Alan Prince for their contributions to this particular piece of the research.
Jane Grimshaw
Table 1. Harmonic bounding input i
C1
a
**
**
b
*
**
C2
C3
C4
C5
C6
*
C7
C8
*** *
never to surface because under any ranking there are superior alternatives. Such losers are said to be ‘harmonically bounded’.” In Table 1, C1-C8 are constraints, and a and b are the outputs paired with input i. These candidates stand in a special relation: candidate a is harmonically bounded by candidate b. The violations of b are a proper subset of the violations of a. Constraints C1, C5 and C7 prefer b. C2, C4, C6 and C8 do not distinguish between them. Hence no ranking of constraints can prefer a over b, and a is an inevitable loser, not optimal in any grammar. I show here that larger structures can be harmonically bounded by smaller ones, but not vice versa. This is because the set of universal constraints guarantees that any phrase will incur at least one violation. This is the “cost” of the phrase. As a result, a phrase is possible only where extra structure is preferred by at least one constraint (Grimshaw 2001, 2002). If there is no constraint which prefers the extra structure, a candidate without the phrase harmonically bounds the larger candidate. 2. The constraints The constraints which result in the key relationships of harmonic bounding are given in (1)–(5). The first three are alignment constraints (McCarthy & Prince 1993).
(1) HdLft: Every X-zero is at the left edge of an X-max. (2) specLft: Every specifier is at the left edge of an X-max. (3) CompLft: Every complement is at the left edge of an X-max.
The last two constraints I call “obligatory element constraints”. They require the presence of particular elements: the head and the specifier. ObHd played a central role in the analysis of Grimshaw (1997). ObSpec, which mandates that a phrase has a specifier, has an obvious similarity to Onset (Prince & Smolensky 2004). The constraint Subject and similar constraints posited in Grimshaw (1997), Grimshaw & Samek-Lodovici (1998), and Samek-Lodovici (2005), are special cases of ObSpec, which is responsible for the widespread obligatoriness of subjects, often characterized as an effect of the “EPP” (Chomsky 1981). (An ObComp constraint may also exist but would not play a role in the analysis here.)
Syntactic constraints
(4) ObHd: A phrase has a head. (5) ObSpec: A phrase has a specifier.
These constraints are not in any sense “economy constraints”; in fact, individually they can prefer larger, more complex structures over smaller, simpler ones. It is the collective effects of general constraints on structure that enforce economy. The constraints select the core phrase structures of English and Korean. In both grammars, the ranking of Speclft above HdLft and Complft eliminates candidates with right specifiers, regardless of the ranking of Hdlft relative to Complft. Specifiers are always at the left of a phrase in both languages, barring adjunctions and other perturbations of structure. Now the two grammars diverge. Optimal X-bars are left-headed in English, right-headed in Korean. Hdlft must dominate CompLft in English, while Complft dominates Hdlft in Korean (see (6)–(7)). In the English optimum, the head is as close as possible to the left edge, while in the right-headed optimum, it is the complement that is as close as possible to this edge. The English X-bar best satisfies Hdlft while the Korean X-bar best satisfies CompLft. (Note the key role played here by Complft, a constraint which has little effect in English, but chooses right headed structures when it dominates Hdlft.). Tables 2 and 3 each show the evaluation of two candidates, with a verb and two arguments as the input and alternative arrangements of a phrase as the output. Table 2. Orders for specifier, head and complement Verb (argument, argument)
Hdlft
CompLft
a W [Spec [Hd Comp]]
*
**
b [Spec [Comp Hd]]
**
*
CompLft
Hdlft
a [Spec [Hd Comp]]
**
*
b W [Spec [Comp Hd]]
*
**
specLft
Table 3. Orders for specifier, head and complement Verb (argument, argument)
specLft
(6) SpecLft >> Hdlft >> Complft English
(7) SpecLft >> Complft >> Hdlft Korean
The constraint ranking is different in English and Japanese but the candidates and the constraints are always the same. (I factor out lexical differences: Korean words in Korean candidates, English in English.) Thus the violations incurred by
Jane Grimshaw
each candidate on these constraints are the same for every grammar, as exemplified by Tables 2 and 3. The difference is the optimum selected, because this is decided by ranking, which is not constant across grammars. The constancy of violations under the logic of OT underlies the economy of structure, as we will see shortly. No order among a specifier, a head, and a complement in a single phrase can satisfy all three of the alignment constraints because only one of the elements can be at the left edge. Two of the three constraints must be violated in every phrase containing three elements. Moreover, since one of the three elements must be separated from the left edge by the other two, the rightmost element necessarily incurs two violations of the alignment constraint governing its position. (If the alignment constraints are “gradient”, (McCarthy & Prince 1993), one violation is incurred for every element separating an aligning item from the left edge of the phrase.) The total number of alignment violations in a phrase containing a head, a specifier and a complement must therefore be three, exactly as the tableaux show. By the same reasoning, every candidate containing two elements incurs one alignment violation, regardless of order. A phrase can escape the penalties imposed by the alignment constraints if it contains only one element (specifier, head or complement), or if it contains no elements. If the phrase contains one element, two alignment constraints are vacuously satisfied, and the sole element in the phrase is left aligned. If the phrase is entirely empty, all three alignment constraints are vacuously satisfied. However, each empty phrase violates both obligatory element constraints. Here and throughout I use “_” to indicate an empty position.
(8) Violations in one-element phrases and empty phrases. (English order.)
[_ _ Comp], [_ _ _] [_ Hd _] [Spec _ _]
violates ObSpec and ObHd violates ObSpec violates ObHd
In sum, any phrase necessarily violates at least one of these markedness constraints. Every phrase has a cost in the form of guaranteed constraint violation. No matter how phrases are combined, a structure incurs more violations of the alignment and/or obligatory element constraints than any structure it properly contains. This is the foundation of harmonic bounding, and thus of economy of structure. A parametric account (see, for example, Saito & Fukui 1998, and papers in Alexiadou & Hall 1997) has a totally different logic. It protects each language from the constraint governing the other. The difference between the two grammars and thus languages is the result of their being subject to different (versions of the) constraints. (9) Japanese: X final in X-bar, X-bar final in XP English: X initial in X-bar, X-bar final in XP
Syntactic constraints
Both the English and the Japanese outputs satisfy their respective constraints fully. English violates the Japanese version of the constraints, and Japanese the English version, but in each language every phrase is perfect: there is no cost to a phrase. Economy of structure cannot be derived from the nature of phrases, in a parametric theory.
3. Economy and complementizers A large body of recent research has established that there are several head positions and hence projections in the region of a clause where “complementizers” are located, normally above the subject. Heads in this region are thought to express clause type, subordination, and the discourse status of preposed material. Studies include: de Haan & Weerman (1985), Rögnvaldsson & Thráinsson (1990), Bhatt & Yoon (1992), Authier (1992), Iatridou & Kroch (1992), Vikner (1995), Rizzi (1997), Puskas (1997), Bayer (1999), Bhatt (1999), Ackema (2001), Poletto & Pollock (2004), Benincà & Poletto (2004), Westergaard & Vangsnes (2005), Julien (2007). Such structures often go by the names, “CP recursion”, or “Split CPs”, although only in a few cases are all the phrases projections of the same head. What determines the number of such projections that occurs in any given clause in any given language? Each layer of structure has a cost: the violations which it necessarily imports. A phrase is therefore possible when its benefits exceed its costs. The benefits can exceed the costs only by virtue of ranking. If a faithfulness or markedness constraint requiring the presence of a functional head dominates the structural constraints which otherwise limit the size of projections, an additional projection will appear in the optimum. A functional head such as a complementizer expresses or “parses” functional morpho-syntactic properties specified in the input. (Baković & Keer 2001 and Ackema 2001 investigate cases where complementizers are not faithful to input.) Bhatt & Yoon (1992) analyze complementizers in Korean in the following way. The particle -ko, which occurs only in subordinate clauses, and occurs regardless of whether the clause is interrogative or propositional, is a Subordinator. The particles -ni and -ta are Type markers. They occur in both subordinate and main clauses; -ta in propositions and -ni in interrogatives. (10) a. -ko [+ sub] b. -ta [+ dec] c. -ni [+ int] (11) a.
John-i wa-ss-ta John-nom come-past-dec “John came”
Jane Grimshaw
b. Bill-un [John-i wa-ss-ta-ko] sayngkakhanta Bill-top John-nom come-past-dec -sub thinks “Bill thinks that John came” c.
John-i wa-ss-ni? John-nom come-past-int “Did John come?”
d. Bill-un [John-i wa-ss-nya-ko] mwulessta. Bill-top John-nom come-past-int-sub asked “Bill asked if John came”
As illustrated in Table 4, faithfulness constraints referring to these features (FaithSub and FaithTyp) prefer candidates which contain the appropriate heads, and hence candidates with two complementizers over candidates with one. Table 4. Complementizers motivated by faithfulness think (Bill, P) P= come (John) + past + dec +sub
Faith Faith Ob Sub Typ Spec
W [_ [_ IP ta] ko] V a [_ IP ta] V
*W
b [_ IP ko] V
*W
c [_ [_ [_ IP ta] ko] ko] V
Ob Hd
Spec Comp Lft Lft
Hd Lft
**
**
*L
*L
*L
*L
***W
***W
Table 4 represents the assessment of the candidates in the form of a Comparative Tableau (Prince 2002). The winning candidate appears as “W”, and the columns represent the preferences of each constraint, in a comparison between each losing candidate and W. If the constraint prefers the losing candidate, “L” appears in the cell, if it prefers the winner, “W” appears. The structure of W is in Figure 1. > > > SubP Sub´ TypP
Sub Typ´
IP
ko Typ ta
Figure 1.
Syntactic constraints
Looking at Figure 1 and Table 4 we see that a complement-head phrase in Korean violates Hdlft once (a head-complement phrase in English violates Complft once). W contains two such phrases (apart from IP which is constant across candidates), one headed by ko and one by ta. The smaller candidates contain only one such phrase. Thus Hdlft prefers a and b to the winner. ObSpec makes the same choice, so the markedness constraints would choose a or b. However, W satisfies both FaithSub and FaithTyp, since -ko encodes +sub and -ta encodes +dec. The two smaller candidates, a and b, each violate one of these faithfulness constraints. Every constraint which favors a loser must be dominated by at least one constraint which prefers the winner (Prince 2002). The larger candidate W is therefore optimal in a ranking where both FaithSub and FaithTyp dominate ObSpec and Hdlft. Table 4 orders the constraints so as to be consistent with the known rankings; the one just established, and the previously known ranking in (7): SpecLft > > Complft > > Hdlft. This ranking chooses the larger structure over the smaller in a subordinate clause, but no ranking chooses a yet larger structure, such as c, with more than two complementizers. Candidate c is no more faithful than W, in fact it may be less so, if the constraint Unique (Grimshaw 2006a) is violated when a single input element has two output correspondents.2 This candidate also incurs one additional ObSpec violation and one additional Hdlft violation. No constraint prefers it to W, so it is harmonically bounded. Ranking chooses among structures which are not harmonically bounded. To put the point another way, the third CP layer imports one alignment violation and one obligatory element violation. We can think of this layer of structure informally as “extraneous”. Its presence in the output of candidate a does harm and does no good. This case shows that some extra structure is motivated by constraint ranking. Extraneous structure is ruled out by harmonic bounding. Rankings can force the presence of additional structure, but only minimally. The presence of structure above and beyond that which allows best-satisfaction of the constraints leads to harmonic-bounding. The main clause optimum, which has the structure in Figure 1, but without Sub and its projection, contains only a Typ head, because this candidate harmonically bounds a candidate with both Typ and Sub (as well as one which includes Sub alone). This is because FaithSub is vacuously satisfied in a main clause, so its ranking has no effect there.
. There is a knotty problem in the background here. Can the input for a single clause contain more than one specification of e.g., +dec, and if so, how does this affect the assessment of faithfulness?
Jane Grimshaw
Table 5. Complementizers in main clauses come (John) + past + dec −sub
Faith Faith Sub Typ
Ob Spec
W [_ IP ta] a [_ [_ IP ta] ko]
*W
b [_ IP ko]
*W
Ob Hd
Spec Lft
Comp Lft
Hd Lft
*
*
**W
**W
*
*
*W
This brings us to the analysis of English that. Observing that it marks subordination (so is not found in main clauses) and that it marks propositions (so is not found in interrogatives) Bhatt and Yoon suggest that it is both a subordinator and a type marker. The analysis I give here pertains to verbs including assert, respond, and argue which require that in their complements. (See Bolinger 1972 for discussion.3 Subordinate clauses without that are analyzed in Doherty 1997, Grimshaw 2006b, in prep.) (12) a. The minister responded that the bill would pass. b. *The minister responded the bill would pass. Table 6. A complementizer in a subordinate clause respond (minister, P) Faith Faith P= pass (bill) + fut + dec +sub Sub Typ W V [_ that IP] a V IP
*W
*W
Ob Spec
Ob Hd
Spec Lft
Hd Lft
Comp Lft
*
*
L
L
As in Table 4, I do not record IP-internal violations in Table 6. The constraints are ordered to reflect the previously known ranking SpecLft > > CompLft > > Hdlft from (6), and are consistent with the rankings yet to be determined. The CP layer (bolded in W) imports violations of ObSpec and Complft. Its specifier is empty and the complementizer separates IP from the edge of CP. However, the CP layer makes it possible to satisfy FaithSub and FaithTyp because that is associated with both +sub and +typ, if we follow Bhatt and Yoon. (The +fut specification is realized by would, its form determined by sequence of tense.) The ranking of either FaithSub or FaithTyp over ObSpec and Complft chooses the larger structure, with the faithfulness constraint(s) forcing the presence of a phrase despite the preference of the markedness constraints. The subordinate
. That also occurs in subjunctive and relative clauses, which are not analyzed here.
Syntactic constraints
clause therefore has the structure within CP-1(Long Adjunct not included) in Figure 2: (13) that [ +sub], [ +typ] (14) FaithSub or FaithTyp > > ObSpec, Complft > > > CP-2 C´-1
Spec
CP-1 C-1
C´-1
Spec
that
C-2
Long Adjunct
IP
that
Figure 2.
Table 7. Violations in a main clause with and without a complementizer pass (bill) +future +dec -sub
Faith Faith Sub Typ
W [Spec I VP] a [_ that [Spec I VP]]
Ob Spec
Ob Hd
*L *W
Spec Lft
Hd Lft
Comp Lft
* *W
*
*W
In contrast that is impossible in main clauses, where the smaller structure is optimal. (15) The bill will pass *That the bill will pass
As before, the candidates tie on SpecLft, Hdlft and ObHd. (I show IPinternal violations in Table 7, since they are relevant here.) The smaller candidate with no CP layer is preferred by Complft and ObSpec. The inclusion of that makes it possible to realize +dec, but at the cost of realizing +sub, which conflicts with the input specification. The CP candidate violates FaithSub, and the IP candidate violates FaithTyp. The IP candidate is chosen in the ranking of (16): (16) FaithSub or Complft or ObSpec > > FaithTyp
The main clause therefore has the structure within IP in Figure 2. In a subordinate clause, ranking prefers the larger structure over the smaller. Nevertheless, harmonic bounding imposes a limit on its size. It must be a CP and no bigger than a CP. (17) is ungrammatical. CP recursion is in principle possible, but not here.
Jane Grimshaw
Table 8. Harmonic bounding of the that that candidate respond (minister, P) P = pass (bill) + fut + dec + sub W V [_ that IP]] a V [_ that [_ that IP]]
Faith Faith Ob Sub Typ Spec
Ob Hd
Spec Lft
Hd Lft
Comp Lft
*
*
**W
**W
(17) *The minister responded that that the bill would pass
IP-internal violations are not shown here. The presence of that satisfies the faithfulness constraints. The violations in the candidate containing one that are a subset of the violations in a, which is therefore harmonically bounded. (Candidate a also violates the faithfulness constraint Unique mentioned above.) For this input the optimal candidate can never have the output structure which includes everything within CP-2 of Figure 2. The sequence that + that at the top of a subordinate clause is thus ruled out by virtue of harmonic bounding. No appeal need be made to some special grammatical mechanism such as the “complementizer haplology filter” of McCloskey 2006: 107. (At least such an appeal is unnecessary if the ungrammaticality of inversion with a complementizer, noted in McCloskey’s paper, follows from the line of reasoning governing that. Also see Ackema 2001 for another case.) English contrasts with Korean where two complementizer candidates are not harmonically bounded. The difference is that the faithfulness constraints prefer the more complex structure, because each functional head parses a different feature in the input. A single that parses both features in the input, so a second that cannot lead to improvement. Despite the analysis just explicated, a structure with two complementizers is not always defeated by ranking, nor is it always harmonically bounded, even when the two complementizers are identical and parse the same input specifications. If some constraint is better satisfied in the larger structure, and if the ranking allows the constraint to enforce its preference, the optimum will contain two complementizers. While the first sentence in (8) is clearly ungrammatical, speakers are uncertain about, and frequently produce, sentences like the second. See McCloskey 2006. When a lengthy adjunct intervenes between the two complementizers the double CP structure is much improved. (See Grimshaw 2006b for an analysis of the short adjunct cases.) (18) a. *The minister responded that soon that the bill would pass. b. ?The minister responded that at some time in the very near future that the bill would pass.
Syntactic constraints
If these long adjuncts are located in a specifier position, the constraints as ranked for English choose the candidate containing two occurrences of that, which is not harmonically bounded in this case. They choose the structure in Figure 2, which includes every element within CP-2. Table 9. Preposed constituent in specifier position respond (minister, P) Faith Faith Ob P = LA pass (bill) + fut + dec + sub Sub Typ Spec
Ob Hd
Spec Lft
Hd Lft
Comp Lft
W V [_ that [LA that IP]]
*
*
***
a V [_ that [LA _ IP]]
*
*W
L
**L
b V [_ _ [LA that IP]]
*
*W
*
**L
*
****W
c V [_ that [_ that [LA that IP]]]
**W
Table 9 shows the violations incurred by the two-that candidate, a candidate with three occurrences of that and two candidates with only one that.4 (Edges targeted for alignment are bolded, “LA” stands for a long adjunct in specifier position, and IP-internal violations are not shown.) The input is identical to the previous two cases, apart from the inclusion of a long adjunct, which I simply represent as LA in the input. The long adjunct is the specifier in all candidates under consideration. SpecLft and the faithfulness constraints (apart from Unique, alluded to above) are satisfied in all candidates. Hdlft and Complft both prefer a and b, which each have just one complementizer. However, ObHd prefers the desired winner. Thus, a ranking in which ObHd dominates Hdlft and Complft chooses the double that candidate. This ranking is independently established for English in Grimshaw 2001, 2002. It enforces the presence of a head even though a phrase with a head incurs more alignment violations than one with no head. Any candidate containing three complementizers, even one with this input, is harmonically bounded for familiar reasons (see candidate c). It contains an extraneous phrase, which accrues additional violations of ObSpec and Complft, and no constraint prefers it to the winner. In sum, two complementizers are better than one when a long adjunct is in specifier position, by virtue of ranking but three complementizers are worse than two, by harmonic bounding. . There is a further one-complementizer candidate, of the form [Spec that IP], where the long adjunct is the specifier. This candidate satisfies ObSpec and has one less Complft violation than W. Nonetheless, it is not optimal because of the existence and ranking of a constraint which assesses violations of alignment at the edge of subordinate clauses: Hdlftsub. This proposal is made in Grimshaw (2006b).
Jane Grimshaw
The rankings motivated here are (19) for English and (20) for Korean: (19) SpecLft > > HdLft > > Complft FaithSub or CompLft or ObSpec > > FaithTyp FaithSub or FaithTyp > > ObSpec, Complft (20) SpecLft > > CompLft > > HdLft FaithSub, FaithTyp > > ObSpec, HdLft
The rankings in (19) and (20) are both compatible with more than one total ranking. The phenomena examined here do not determine the complete rankings for Korean and English, hardly a surprise. To show that there is at least one consistent ranking for each language, I provide the ranking diagrams in Figures 3 and 4. SPECLFT
FAITHSUB
COMPLFT
FAITHTYP
OBSPEC
HDLFT Figure 3. Korean Rankings. FAITHSUB SPECLFT
OBSPEC
HDLFT COMPLFT FAITHTYP Figure 4. English Rankings.
4. Conclusion The distribution of complementizers, and hence the shape and size of clauses, is governed by faithfulness constraints which are satisfied via the realization in an output of features in the corresponding input. Their distribution is also governed by markedness constraints which assess the presence and position of elements in phrases.
Syntactic constraints
In the rankings governing Korean and English, these constraints determine differences in grammatical patterns for the two languages. One is head final, the other head medial. One has a two head complementizer system, and the other a single syncretic complementizer. English has no complementizer in a main clause, Korean has one of its two complementizers in these clauses. These choices are determined by the ranking of conflicting constraints, which differs from grammar to grammar. The choice of more complex structures over less complex is not general. It depends on ranking, and thus is language specific; it depends on the input and thus holds in some cases but not in others, and it depends on the comparison: it is found only in comparisons where there is conflict among the relevant constraints. The preference for simpler structures is found for all rankings and thus is universal, it is independent of the input, and thus holds in all cases, and it holds regardless of the interaction between the relevant constraints. Ranked constraints can only choose from among candidates that are not harmonically bounded. A candidate containing more phrases can never harmonically bound one containing fewer. It can only be preferred under a particular ranking. Since the larger candidate necessarily violates the constraints under discussion more than the smaller, the violations of the larger cannot be a proper subset of the violations of the smaller. The choice of less economical structures over more economical alternatives is contingent: it holds for some kinds of phrases under particular rankings. The choice of economical structures over less economical alternatives is general: it holds for all kinds of phrases under all rankings of the responsible constraints, i.e., for all grammars. Economy of structure is entailed.
References Ackema, Peter. 2001. Colliding complementizers in Dutch: Another syntactic OCP effect. Linguistic Inquiry 32: 717–727. Alexiadou, Artemis & T. Alan Hall (eds). 1997. Studies on universal grammar and typological variation. John Benjamins. Authier, J. -Marc. 1992. Iterated CPs and Embedded Topicalization. Linguistic Inquiry 23: 329–336. Baković, Eric & Edward Keer. 2001. Optionality and ineffability. In Géraldine Legendre et al. (eds), Optimality–Theoretic Syntax, 97–112. MIT Press. Rutgers Optimality Archive 384. Bayer, Josef. 1999. Final complementizers in hybrid languages. Journal of Linguistics 35: 233–271. Benincá, Paola & Cecilia Poletto. 2004. Topic, focus, and V2. In Luigi Rizzi (ed.), The structure of CP and IP, 52–75. Oxford University Press. Bhatt, Rakesh & James Yoon. 1992. On the composition of COMP and parameters of V2. In Dawn Bates (ed.), Proceedings of 10th annual west coast conference on formal linguistics, 41–52.
Jane Grimshaw Bhatt, Rakesh. 1999. Verb movement and the syntax of Kashmiri. Kluwer Academic. Bolinger, Dwight. 1972. That’s that. The Hague: Mouton. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. de Haan, Ger & Fred Weerman. 1985. Finiteness and verb fronting in Frisian. In Hubert Haider and Martin Prinzhorn (eds), Verb second phenomena in Germanic languages, 77–110. Dordrecht: Foris. Doherty, Cathal. 1997. Clauses without complementizers: finite IP complementation in English. The Linguistic Review 14: 197–220. Grimshaw, Jane. 1997. Projection heads and optimality. Linguistic Inquiry 28: 373–422. Grimshaw, Jane. 2001. Economy of structure in OT. Rutgers Optimality Archive 434. Grimshaw, Jane. 2002. Economy of Structure in OT. In Angela Carpenter, Andries Coetzee & Paul de Lacy (eds), Papers in Optimality Theory II. University of Massachusetts Occasional Papers 26: 81–120. Amherst, MA: GLSA Publications. Grimshaw, Jane. 2006a. Chains as unfaithful optima. In Eric Baković, Junko Ito & John McCarthy (eds), Wondering at the natural fecundity of things: Essays in honor of Alan Prince, Paper 6. http://repositories.cdlib.org/lrc/prince/. Also Rutgers Optimality Archive 892.2. Grimshaw, Jane. 2006b. Location specific constraints in matrix and subordinate clauses. Rutgers Optimality Archive 857. Grimshaw, Jane. In prep. The size and shape of clauses: Economy of structure as harmonic bounding. Grimshaw, Jane & Vieri Samek-Lodovici. 1998. Optimal subjects and subject universals. In Pilar Barbosa et al. (eds), Is the best good enough? MIT Press, Cambridge, MA. 193–219. Iatridou, Sabine & Anthony Kroch. 1992. The licensing of CP-recursion and its relevance to the Germanic verb-second phenomenon. In Working papers in Scandinavian linguistics, vol. 50: 1–24. Julien, Marit. 2007. Embedded V2 in Norwegian and Swedish. lingBuzz/000475. McCarthy, John & Alan Prince. 1993. Generalized alignment. In Geert Booij & Jaap van Marle (eds), Yearbook of morphology, 79–153. Boston: Kluwer. McCloskey, James. 2006. Questions and questioning in a local English. In Raffaella Zanuttini, Héctor Campos, Elena Herburger & Paul H. Portner (eds), Cross-linguistic research in syntax and semantics: Negation, tense and clausal architecture, 87–126. Georgetown University Press. Poletto, Cecilia & Jean-Yves Pollock. 2004. On the left periphery of some Romance wh-questions. In Luigi Rizzi (ed.), The structure of CP and IP, 251–296. Oxford University Press. Prince, Alan. 2002. Arguing optimality. In Angela Carpenter, Andries Coetzee & Paul de Lacy (eds), Papers in optimality theory II. 269–304. University of Massachusetts occasional papers 26. Amherst, MA: GLSA Publications. Rutgers Optimality Archive 562. Prince, Alan & Paul Smolensky. 2004. Optimality theory: Constraint interaction in generative grammar. Blackwell: Malden, MA; Oxford, UK; Carlton, Aus. Puskas, Genoveva. 1997. Focus and the CP domain. In Liliane Haegeman (ed.), The new comparative syntax, 145–164. Longman. Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In Liliane Haegeman (ed.), Elements of Grammar, 281–337. Kluwer Academic Publishers. Rögnvaldsson, Eiríkur & Höskuldur Thráinsson. 1990. On Icelandic word order once more. In Joan Maling & Annie Zaenen (eds), Modern Icelandic syntax. Syntax and Semantics 24: 3–40. San Diego: Academic Press.
Syntactic constraints
Saito, Mamoru & Naoki Fukui. 1998. Order in phrase structure and movement. Linguistic Inquiry 29: 439–474. Samek-Lodovici, Vieri. 2005. Prosody syntax interaction in the expression of focus. Natural Language and Linguistic Theory 23: 687–755. Samek-Lodovici, Vieri & Alan Prince. 2002. Fundamental Properties of Harmonic Bounding. Rutgers Optimality Archive. 785. Vikner, Sten 1995. Verb movement and expletive subjects in the Germanic languages. Oxford University Press. Westergaard, Marit & Østein Vangsnes. 2005. On dialects of wh and V2 in Norwegian. Journal of Comparative Germanic Linguistics 8: 117–158.
Aspects of the neural representation of spoken language Grzegorz Dogil
University of Stuttgart, Germany
1. Introduction It has become explicit since the beginning of this century, that new imaging techniques can produce a deluge of information correlating linguistic and neural phenomena. Our understanding of the inter-relationship between data coming from linguistics and brain sciences has led to many fascinating discoveries in neurolinguistics, psycholinguistics and the cognitive studies of human language (cf. Demonet et al. 2005). In this paper we report some major findings from the experimental work investigating some aspects of spoken language processing in an unimpaired human brain. We present findings of localization information provided by the hemodynamic based fMRI method and correlate them with high-quality temporal data generated by the electromagnetic MEG-technique.1 2. Matured imaging technology Neurolinguistics, psycholinguistics and cognitive approaches to language were confronted with a technological revolution at the end of the 20th century due to the general availability of two non-invasive brain imaging techniques, which allow a non-invasive investigation of the human brain at action. fMRI – functional magnetic resonance imaging – uses the blood oxygenation level dependent (BOLD) effect as an indirect marker of brain activation.
. We provide an illustrative graphic in the appendix with all the relevant anatomical and neurophysiological detail so that a reader can figure out the localizations discussed in this paper for herself.
Grzegorz Dogil
Local neuronal activity gives rise to a decline in blood oxygenation which, in turn, causes an increase of blood flow. The hemodynamic response outweighs the oxygen demand yielding accumulation of oxyhemoglobin in the respective brain region. Since the magnetic properties of oxyhemoglobin are different from those of deoxyhemoglobin, imaging sequences allow to detect the change of the magnetic resonance signal in the activated areas. Magnetic resonance was discovered by Edward Percell and Felix Block (Nobel Price in physics 1952). It was patented for medical purposes in 1977 and in 1992 it was used by Seiji Ogawa for visualization of the BOLD-Effect. By now it is a standard method for localization of any activity in the human brain (cf. Frackowiak, Friston and Frith 2003, for a standard introduction). MEG – magnetoencephalography – records the magnetic fields that surround the neural electrical currents in cortical pyramidal cells. Neuromagnetic signals are about 100 million times weaker than the earth´s magnetic field. To pick up such weak magnetic fields, the most sensitive detectors for magnetic flux – SQUIDS [Super Conducting Quantum Interference Devices] – are used (cf. Clarke & Braginski 2004, for technological details). Magnetic fields are not affected by hair, skin, skull or other tissue, thus, the signal generated in the brain (even the fetal brain) is not distorted by artifacts of conduction. MEG mainly picks up signals from radial sources and, thus, it is best suited to pick up signals from pyramidal neural cells that are located in the fissures of the human brain. MEG has made its main contributions in the recording of auditory evoked potentials because the auditory cortex mainly lies in the Sylvian fissure. The fact that MEG can separately record the activity in each hemisphere has made it particularly interesting for studying human speech and spoken language (cf. Hämäläinen et al. 1993, for a seminal introduction to Magnetoencephalography). MEG-based spoken language research has developed a standardized odd-ball study design which allows the evocation of two well-understood components of the auditory response of the human brain, N100 and mismatch negativity (MMN). Figure 1 below illustrates a typical result of an experiment using the standard MEG paradigm. The component N100 is elicited by an abrupt change in any of the parameters of the auditory input signal. MMN reflects a mismatch between the representation of a sound event in memory and the new event (cf. Näätänen 1995, 2001). MMN can only arise when a memory trace (a phonological representation) of a stimulus has been formed. Indeed, ongoing research has uncovered a variety of MMN responses to various types of linguistic phonetic contrasts (cf. Sharma & Dorman 2000; Phillips et al. 2000; Shestakova et al. 2002; Hertrich et al. 2002; Lipski 2006).
Aspects of the neural representation of spoken language A sequence of auditory stimuli containing frequent standard stimuli (black squares) and rare deviant stimuli (grey squares):
The stimuli elicit the N100 at about 100 ms after stimulus onset. The response to the deviant stimuli (grey line) diverges from the response to the standards (black line) around 150 ms. The difference (shaded area between standard and deviant curve) constitutes the MMN. +
stimulus onset N100
standard deviant MMN
Figure 1. Schematic illustration of the oddball paradigm and the resulting N100 and MMN components.
3. Some major findings 3.1 Prosody Prosody is the most characteristic trait of spoken language. Prosodic features, unlike other linguistic textual features, are often produced without conscious intention and are open to forms of interpretation which rely on emotional or non-cognitive processes. At the same time, the prosodic organization of human communication is highly correlated with the semantic, syntactic, morphological and segmental packaging in speech production. Such diversity of functions is expected to be represented by very complex neural systems. However, all these functions are implemented by three simple phonetic parameters: duration, intensity and pitch. The straightforward phonetics of prosody could be expressed by an equally simple neural correlate. The scope of prosodic functions and cues in language processing has led to various hypotheses concerning the neurolinguistic and neuroanatomical basis of prosody. At least four hypotheses have been particularly influential (Baum & Pell 1999). The right hemisphere hypothesis states that all aspects of prosody are independently processed by the right hemisphere and integrated with the linguistic information (which is processed by the left hemisphere) via interhemispheric connections of the corpus callosum (Blumstein & Cooper 1974).
Grzegorz Dogil
The functional lateralization hypothesis assumes that there is a continuum from linguistic to affective functions of prosody and that processing shifts from the left hemisphere, which is involved with the linguistically-based tasks, to the right hemisphere, which supports affectively-based tasks (Dapratto et al. 1999). Clinical (Gandour et al. 2001) and neuroimaging evidence (Meyer 2000) corroborating this hypothesis points out that linguistic prosody should possibly be further subdivided into phrase vs. syllable level prosody. The left hemisphere seems to be dominant for short frames (syllables) and the right hemisphere for long frames (intonational phrases). The subcortical processing hypothesis claims that prosodic functions are highly dependent on subcortical processes and are not lateralized to one or the other hemisphere. According to this hypothesis, prosody is a basic, pre-linguistic channel of communication, which is neuronally the same for humans and their animal ancestors (Cancelliere & Kertesz 1990). The acoustic cues hypothesis contends that duration, pitch (and possibly intensity) are supported by highly specific and independently lateralized neural components (Van Lancker & Sidtis 1992). All four hypotheses find support in the clinical observation of prosodically impaired subjects. There are methodological problems with the interpretation of data from patients if it is used in isolation. Such data may reflect neural reorganization or the development of compensatory strategies. Apart from that, the process of clinical observation is very complex in itself, and it can be hardly empirically traced and fractionalized into relevant cognitive components. Furthermore modern cognitive theories question the assumption of a simple correspondence between multivariate linguistic systems (like prosody) and highly complex neural systems (like whole brain hemispheres). In our research we assume a fractionalized and elaborated model of prosody generation (Dogil 2003) and test its individual components with experiments designed to reveal the functions of the active, healthy brain. The main interest in the experiment described below concerns the role of prosodic function (“intonational meaning”) and the influence of the size of address frames in prosody generation. 3.1.1 Methods, subjects and materials The study recruited ten healthy native German subjects (five females, five males, mean age 26.2 years, range 21–32 years). All participants were right-handed as determined by standardized inventory, and none of them had a history of neurological disorders. Informed consent had been obtained from each subject. Subjects were paid for the participation in the experiment.
Aspects of the neural representation of spoken language
The experiment was designed in accordance with the cognitive subtraction paradigm. This paradigm requires a pair of tasks which are similar except for the presence (“activation”) or absence (“control”) of the one cognitive process that is examined. Under activation conditions, subjects were asked to produce a sentence-like sequence consisting of five syllables [dadadadada] with various pitchaccent types and locations (the FOCUS condition), with various boundary tone types (the MODUS condition), and with various kinds of emotional state marking (the AFFECT condition). As a control condition [MONO condition] for the statistical analysis they were asked to produce the sequences [dadadadada, dididididi, dododododo, dududududu] in a monotonous voice (with a syllable frequency of ca. 5Hz). Reactions were stimulated visually. We used reiterant syllables in order to reduce the influence of the syntactic, semantic, morphological and segmental factors on prosody generation. The aspects of prosody that were controlled in this experiment were in accordance with the model of prosody generation, correlated only with different address frames (syllables, feet, intonational phrases) and parameter settings (duration, intensity and pitch). 3.1.2 Procedure and imaging Subjects lay supine in the MR scanner (1.5 Tesla whole body scanner, Siemens Vision), the heads being secured by means of foam rubber in order to minimize movement artifacts. The stimuli were presented visually every 15 sec. for a period of three seconds. The pauses between the stimuli were 12 sec. long. Subjects produced the recognized item immediately after stimulus presentation. Every 60 sec. there was a paradigm change, initiated by an acoustic instruction. Each stimulus was presented eight times. In four out of these presentations the “prosodic” reaction was required. In the other four cases the subjects rendered the item in a monotonous manner. 28 parallel axial slices (thickness = 4mm, gap = 1mm) were acquired across complete brain volume by means of multislice echoplanar imaging sequence T2*EPI (TE = 39ms, TR = 3s, α = 90°, FOV = 192mm, 64² matrix). Imaging data was processed using SPM 99 (Wellcome Institute of Cognitive Neuroscience, London). Functional images were movement corrected and coregistered with anatomical images. Spatial normalization procedures were performed. The normalized fMRI data were spatially smoothed (Gaussian filter, 6mm FWHM). The activation threshold was set at p < 0.05 (corrected). 3.1.3 Results We observed a rather similar activation pattern for each condition compared to rest. Significant foci of activation were located in the motor cortex extend i ng to premotor areas [MC], in the supplementary motor cortex [SMA], in the cerebellum [CER], and in occipital regions [OCC]. These findings reflect the
Grzegorz Dogil
involvement of speech motor control [MC, SMA, CER] and processing of the visual stimuli [OCC]. In the direct contrasts between the prosodic conditions and their respective control condition these “speech” networks disappeared. This indicates that all tasks were comparable concerning the required cognitive components. As intended, the only difference between activation and control conditions was the prosody processing part – active in the former and absent in the latter. The subtraction FOCUS-MONO revealed a pattern of activation exclusively lateralized to the left hemisphere. Foci of activation were located in the anterior part of the superior temporal gyrus (STG, BA22 (Brodmann Area)) and in the inferior frontal gyrus (IFG, BA47). Additional clusters of activation were found in the left MC and in the anterior cingulum. The contrast MODUS-MONO revealed a focal spot of activation located also in the left IFG (BA 47). Subtracting MONO from AFFECT we observed focal activation exclusively in the right IFG (BA 47/11). 3.1.4 Discussion – Prosody processing in the inferior frontal gyrus The results show that both cortical hemispheres subserve processing prosody during speech production. This is clear evidence against a view that the right hemisphere is dominant for all types of prosodic computation. Furthermore, our findings do not confirm a crucial role of subcortical structures in prosody processing since we did not observe any significant activation in that part of the brain. However, our results are consistent with the functional lateralization hypothesis. Generating linguistically controlled prosodic features on the syllable level (FOCUS) as well as on the phrase level (MODUS) revealed activation patterns exclusively in the left hemisphere, while generating different emotional expressions revealed exclusively right hemisphere activation. Considering the exact localization of activated foci, our production data show interesting similarities with recent fMRI studies on prosody perception. In another experiment by our group, Wildgruber et al. (2000: 341) compared the identification of prosodically expressed emotions (happiness, anger, fear, sadness, disgust) with a non-prosodic, phonological control task (identification of the vowel following the first [a] in a sentence). Subtraction of the control task from the emotive identification task elicited activation in the right STG and bilateral activation in the IFG (BA 47), with the right hemisphere showing overall greater activation in terms of BOLD magnitude and spatial extent. In a different experiment, Wildgruber et al. (2001: 627) compared the perception of affective prosody (rating of emotional expressiveness of a sentence) with the perception of linguistic FOCUS (rating of appropriateness of question-answer pairs with respect to accent placement, i.e., a classical focus test). Activation pat-
Aspects of the neural representation of spoken language
terns of both tasks again involved the IFG (BA 47/11). The linguistic task (focus identification) revealed additional activation in the left BA 44/45 (classical “Broca’s area”). Focal activation in the orbito-basal frontal cortex during perceptual prosody processing was also shown by Meyer (2000) in a series of fMRI studies using lowpass filtered speech, which preserves the prosodic “intonational” features and obliterates the segmental cues for speech. Analogous activations were found by Dapratto et al. (1999) in a prosody perception study which used meaningful sentences in a selective attention paradigm. Loevenbruck et al. (2005) found significant activation of the left inferior frontal gyrus (BA 45 and/or 47) in an fMRI experiment on the production of deixis, involving both syntactic and prosodic deixis (i.e., focus). These findings point to a crucial role of the deep frontal operculum (BA 47 and 11) of the inferior frontal gyrus in the processing of the basic phoneticphonological aspects of prosody. Besides adding evidence to the correlation between the function of prosody processing and the neuroanatomical region of the inferior frontal gyrus, our results show interesting similarities between speech production and perception: both modalities recruit identical anatomical structures when certain functions are executed. Based on the referenced studies and on our own data we can assume that the setting and interpretation of basic prosodic parameters is one such function. Only when these basic parameters have to be integrated with higher linguistic or paralinguistic information, other areas become additionally activated. In particular, integrating prosodic and linguistic focus activated the classical Broca’s area, while identification of emotions based on aprosodic information activated the right superior temporal gyrus (the right hemisphere analogon to classical Wernicke’s area). 3.2 The speech network In the research by our group on speech production (Riecker et al. 2000a,b; Dogil et al. 2002, 2003; Ackermann & Riecker 2004) we have identified a cortical area which appears to play a role in interfacing phonetic and phonological processing: the left anterior insula (cf. appendix for localization). In the functional imaging studies of speech production the insular cortex has often been implicated, sometimes against the expectations of the authors. Seminal PET investigations of language processing by Petersen et al. (1988, 1989) found insular activation in a cognitive subtraction experiment in which the loud repetition of nouns was subtracted from the passive listening to words. The other areas activated were the motor cortex (MC), the mesionfrontal cortex encompassing the supplementary motor area (SMA) as well as portions of the cerebellum (CER).
Grzegorz Dogil
In a controlled, speech-oriented PET study Wise et al. (1999) were able to attribute the pivotal function of speech motor control to the hemodynamic response in the anterior insular cortex with the significant effects found only at the left insula. The separation of phonological selection from active performance of speech motor movements was the subject of an early experiment by our group. Riecker et al. (2000a) provide details of the procedure and the results. Here we consider only those aspects of the experiment that are relevant for this overview. We investigated both overt and covert production of highly overlearned word strings such as the names of the months of the year. This simple task was considered selective for speech motor control given the relatively low demands placed on syntactic, semantic and prosodic processing. As a control condition subjects were asked to reproduce overtly and covertly a nonlyrical tune. The choice of Mozart’s serenade ‘‘Eine kleine Nachtmusik’’ served the purpose perfectly – everybody knew the tune, and nobody had an association of a propositional meaning that might go together with it. We found complementary cerebral networks subserving singing and speaking. Reproduction of a nonlyrical tune elicited activation in the right motor cortex, the right anterior insula, and the left cerebellum, whereas the opposite, mirror image response pattern emerged during a speech task. In contrast to hemodynamic responses in motor cortex and cerebellum, activation of the intrasylvian cortex turned out to be associated with overt task performance. Hence anterior insula does not support the prearticulatory processes, but, as our findings suggest, it supports the actual coordination of multiple cortical sources which in turn organize the coordination of vocal tract musculature during speech production. The neuronal connectivity of the intrasylvian cortex (besides its connections to motor cortex and limbic structures, intrasylvian cortex receives input both from primary/secondary somatosensory and auditory areas) designates it as a pivotal node in the speech processing network. Ackermann & Riecker (2004) argue that the intersylvian cortex is engaged in crossmodal processing of speech, rather than in elementary functions in the speech domain, such as muscle coordination. Systems of overlearned coordinated movement like speaking and singing appear to be represented in a fairly abstract way in the human brain, i.e., rather than by the summation of the basic motor programs they appear to be represented by a crossmodal system which seems to include the aspects of production and perception. As in the case of prosody discussed in section 2, speech and singing appear to have a nodal representation in the brain across modalities. Is insula designated to represent abstract features of online speech coordination or is it its connectivity with other brain areas which lets it emerge as a pivotal node? A whole generation of phoneticians has followed an argument that speech is
Aspects of the neural representation of spoken language
special because of the unique link between production and perception (Liberman 1996). It is therefore expected that there must be a strong connectivity supporting the speech production system described above and the speech perception system. 3.3 On arguing that natural speech is special for the human brain Benson et al. (2001), following central assumptions of Motor Theory, carried out a detailed fMRI study to account for the localization of processes for speech perception and non-speech perception. Cortical areas reacting to passive listening to speech of varying complexity (vowels, CV syllables, CVC syllables) and quality (natural, synthetic) were recorded. For non-speech sounds, piano tones, chords and chord progressions (both “natural” and synthetic) were used as stimuli. Benson et al. identified seven regions which were activated in speech processing: left and right posterior superior temporal gyrus (pSTG), left ventral occipitotemporal cortex (BA 21 and 37), left inferior superior temporal gyrus (SMG), left middle frontal gyrus (BA 6/7), right superior SMG, and anterior insula. Critically, Benson et al. found that bilateral posterior superior temporal gyrus was selectively involved in the processing of complexity in speech. Especially the left pSTG appeared to be highly and selectively activated when the brain processed the differences in the complexity of speech. Thus, the authors concluded that speech and nonspeech sounds were encoded differently. The theoretical prediction that linguistic representations (which underly speech processing) are effective at very early stages of perception appears to be validated by the experimental results. The result was further refined in a recent study by the same group (cf. Whalen et al. 2006). The central role played by the left posterior superior temporal gyrus (left pSTG) in mediating speech perception, speech production and linguistic representation, observed in Benson’s et al. and Whalen et al. studies is congruent with the only electrical stimulation study of this area (cf. Boatman et al. 1998), with other imaging studies (cf. Dehaene-Lambertz et al. 2005; Vouloumanos et al. 2001) as well as speech perception models supported by both clinical and imaging data (cf. Hickok & Poeppel 2004). Another interesting discovery of the Benson et al. study was the differential activation of the broad cortical areas caused by natural vs. synthetic speech stimuli. Listening to synthetic speech activates classical “auditory” and “linguistic” networks localized in the left and right perisylvian cortex, whereas listening to natural speech additionaly activates “emotional” and “attitudinal” areas in the prefrontal cortex (BA 9/10). These additional activations are apparently caused by postural speaker characteristics present in natural speech stimuli. Differential temporal course, localization and lateralization for synthetic and natural stimuli were also found in MEG studies. Hertrich et al. (2002) compared
Grzegorz Dogil
the auditory responses to synthetic and natural syllables in an oddball experiment ([ba] was used as the standard, [da] as deviant). The MMN (mismatch negativity) response to natural stimuli was stronger in the right hemisphere. It was suggested that this result could reflect the early response of the perceptual system to the non-linguistic speaker characteristics present in the natural stimuli and absent from the synthetic stimuli used in the experiment. The perfect temporal resolution of MEG helped in discovering the late onset of MMN in natural stimuli (300–400 ms long MMN was recorded for natural stimuli but not for synthetic stimuli). Hertrich et al. hypothesized that synthetic stimuli which lack the postural features of natural speech do not grasp the brain attention to a degree which would maintain a long and strong mismatch response. Moreover, it was found that the natural deviants [da] presented to the right ear evoked strong, left lateralized MMN’s. This effect was absent for synthetic stimuli under passive-listening condition. This result suggests that natural spoken language evokes linguisticphonetic processing automatically and leads to left hemispheric responses typical of language processing. 4. Conclusion The first general conclusion that we draw from this selective overview is that spoken language is special to the human brain. When compared to synthetic or other unnaturally generated speech, naturally spoken language evokes significantly more distributed areas (cf. Benson at al. 2001 fMRI experiment discussed above) and it grasps significantly more of the brain attention (cf. Hertrich et al. 2002 MEG results discussed at the end of the previous section). Spoken language has apparently developed a symbiotic relation to neural processing, which is able to resolve not only linguistic but also emotional and attitudinal features of language, but only if language is presented in its natural form. This result has implications for experimental procedures in classical psycholinguistics and for speech synthesis. The other important finding concerns the nodal areas in the brain which apparently specialize in processing various aspects of spoken language. We have presented evidence from production and perception that the anterior inferior frontal gyrus (BA 47/11) mediates the processing of the phonetic-phonological aspects of prosody. These parameters have to be integrated with higher linguistic or paralinguistic information, and this is when other areas become additionally activated. Similarly, two nodal areas of the phonetic-phonological interface in segmental speech representations have been identified. The first is the left anterior
Aspects of the neural representation of spoken language
insula, implicated in a number of clinical and neuroimaging studies on spoken language and its distortions (cf. Dronkers 1996). Again, it has been argued that the neural connectivity of the insular cortex might be a prerequisite for its pivotal role in interfacing phonetic and phonological representations (cf. Dogil & Mayer 1998; Ackermann & Riecker 2004; Hillis et al. 2004). An additional nodal area has been proposed by Hickok & Poeppel (2004) in their influential model of neural language processing. Hickok & Poeppel (2004) stipulate that a single area of the brain – a site at the boundary of the parietal and the temporal lobes deep inside the Sylvian fissue (area Spt) – is a critical interface between sensoryand motor-based representations in speech. The evidence, which initially came from a neuroimaging study by Buchsbaum, Hickok and Humphreys (2001) in which subjects listened to pseudo words and then covertly rehearsed them, has been validated in a number of recent fMRI and MEG studies (cf. Hickok & Poeppel 2007). The results presented in this overview have obvious implications for classical, behavioural-experimental psycholinguistcs & classical, clinical data based neurolinguistics. The correlation of the results of all fields is already being addressed in few studies (cf. modeling studies of Hickok & Poeppel 2004/2007; Guenther et al. 2006; Dogil et al. 2003). However, it can also be observed that neuroimaging studies are assigned over-proportionally high weight in the scientific community. It is easier to publish research results which are supported by neuroimaging techniques than results that are derived from clinical data or classical behavioural experiments. We can only hope that the most recent applications of mature neuroimaging technology, like for example bio-neurofeedback (cf. Sitaram et al. 2007), will lead to new forms of cooperation between classical psycholinguistics, clinical linguistics and neuroimaging dominated brain research. References Ackermann, H. & Riecker, A. 2004. The contribution of the insula to motor aspects of speech production: A review and hypothesis. Brain and Language 89: 320–328. Baum, S. & Pell, M. 1999. The neural bases of prosody. Aphasiology 13: 581–608. Benson, R.R., Whalen, D.H. et al. 2001. Parametrically dissociating speech and nonspeech perception in the brain using fMRI. Brain and Language 78: 364–396. Blumstein, S. & Cooper, W. 1974. Hemispheric processing of intonation contours. Cortex 10: 147–158. Boatman, D., Hart, J. et al. 1998. Right hemisphere speech perception revealed by amobarbidal injection and electrical interference. Neurology 51: 458–464. Buchsbaum, B., Hickok, G. & Humphries, C. 2001. Role of left posterior superior temporal gyrus in phonological processing for speech perception and production. Cognitive Science 25: 663–678.
Grzegorz Dogil Cancelliere, A. & Kertesz, A. 1990. Lesion localization in acquired deficits of emotional expression and comprehension. Brain and Cognition 13: 133–147. Clarke, J. & Braginski, A.I. (eds). 2004. The SQUID Handbook. Wiley-VCH. Dapratto, M., Hariri, A., Bialik, M. & Bookheimer, S. 1999. Cortical correlates of affective vs. linguistic prosody: an fMRI study. NeuroImage 9: 1054. Dehaene-Lambertz, G., Pallier, C. et al. 2005. Neural correlates of switching from auditory to speech perception. Neuroimage 24: 21–33. Demonet, J.F., Thierry, G. & Cardebat, D. 2005. Renewal of the neurophysiology of language: Functional neuroimaging. Physiological Reviews 85: 49–95. Dogil, G. & Mayer, J. 1998. Selective phonological impairment: a case of apraxia of speech. Phonology 15: 143–188. Dogil, G., Ackermann, H., Grodd, W., Haider, H., Kamp, H., Mayer, J., Riecker, A. & Wildgruber, D. 2002. The speaking brain: a tutorial introduction to fMRI experiments in the production of speech, prosody and syntax. Journal of Neurolinguistics 15: 59–90. Dogil, G., Ackermann, H., Mayer, J., Riecker, A. & Wildgruber, D. 2003. Das Sprechnetzwerk im menschlichen Gehirn: Evidenz aus der funktionalen Kernspintomographie und aus der Klinik. In H. Müller & G. Rickheit (eds), Neurokognition der Sprache. Tübingen: Narr. Dogil, G. 2003. Understanding Prosody. In G. Rickheit, Th. Hermann & W. Deutsch (eds), Psycholinguistics: An International Handbook, 544–566. Berlin: De Gruyter. Dronkers, N.F. 1996. A new brain region for coordinating speech articulation. Nature 384. 159–161. Frackowiak, R., Friston, K. & Frith, C. (eds). 2003. Human Brain Function. Elsevier: Oxford. Gandour, J., Ponglorpisit, S., Khunadorn, F., Dechongkit, S., Boongirt, P., Boonklam, R. & Potisuk, S. 2001. Lexical tones in Thai after unilateral brain damage. Brain and Language 43: 275–307. Guenther, F., Gosh, S.S. & Tourville, J.A. 2006. Neural Modelling and Imaging of the Cortical Interactions Underlying Syllable Production. Brain and Language 96: 280–301. Hämäläinen, M.S., Hari, R., Ilmoniemi, R.J., Knuutila, J. & Lounasmaa, O.V. 1993. Magnetoencephalography – theory, instrumentation, and applications to noninvasive studies of the working human brain. Reviews of Modern Physics 65: 413–497. Hertrich, I., Mathiak, K., Lutzenberger, W. & Ackermann, H. 2002. Hemispheric lateralization of the processing of consonant-vowel syllables (formant transitions): Effects of stimulus characteristics and attentional demands on evoked magnetic fields. Neuropsychologia 40: 1902–1917. Hickok, G. & Poeppel, D. 2004. Dorsal and ventral streams: A framework for understanding aspects of the functional anatomy of language. Cognition 92: 67–99. Hickok, G. & Poeppel, D. 2007. The cortical organization of speech processing. Nature Reviews, Neuroscience 8: 393–402. Hillis, A., Work, M., Barker, P., Jacobs, M., Breese, E. & Maurer, K. 2004. Re-examining the brain regions crucial for orchestrating speech articulation. Brain 127: 1479–1487. Liberman, A.M. 1996. Speech: A Special Code. Cambridge, MA: MIT Press. Lipski, S. 2006. Neural correlates of fricative contrasts across language boundaries. Stuttgart: University of Stuttgart dissertation. Loevenbruck, H., Baciu, M., Segebarth, Ch. & Abry, Ch. 2005. The left inferior frontal gyrus under focus: an fMRI study of the production of deixis via syntactic extraction and prosodic focus. Journal of Neurolinguistics 18. 237–258.
Aspects of the neural representation of spoken language
Meyer, M. 2000. Ereignisbezogene hämodynamische Korrelate der Verarbeitung von grammatischer und prosodischer Information. Doctoral Thesis, University of Leipzig, Germany. Näätänen, R. 1995. The mismatch negativity: A powerful tool for cognitive neuroscience. Ear and Hearing 16: 6–18. Näätänen, R. 2001. The perception of speech sounds by the human brain as reflected by the mismatch negativity MMN and its magnetic equivalent MMNm. Psychophysiology 38: 1–21. Petersen, S.E., Fox, P.T., Posner, M.I., Mintum, M. & Raichle, M.E. 1988. Positron emission tomographic studies of the cortical anatomy of single-word processing. Nature 331: 585–589. Petersen, S.E., Fox, P.T., Posner, M.I., Mintum, M. & Raichle, M.E. 1989. Positron emission tomographic studies of the processing of single words. Journal of Cognitive Neuroscience 1: 153–170. Phillips, C., Pellathy, T. et al. 2000. Auditory cortex accesses phonological categories: An MEG mismatch study. Journal of Cognitive Neuroscience 12: 1038–1055. Riecker, A., Ackermann, H., Wildgruber, D., Dogil, G. & Grodd, W. 2000a. Opposite hemispheric lateralization effects during speaking and singing at motor cortex, insula and cerebellum. NeuroReport 11: 1997–2000. Riecker, A., Ackermann, H., Dogil, G., Wildgruber, D., Mayer, J., Haider, H. & Grodd, W. 2000b. Articulatory/Phonetic Sequencing at the Level of the Anterior Perisylvian Cortex: A Functional Magnetic Resonance Imaging (fMRI) Study. Brain and Language 75: 259–276. Sharma, A. & Dorman, M. 2000. Neurophysiological correlates of cross-language phonetic perception. Journal of the Acoustical Society of America 107: 2697–2703. Shestakova, A., Brattico, E. et al. 2002. Abstract phoneme representations in the left temporal cortex: magnetic mismatch negativity study. Journal of Cognitive Neuroscience and Neuropsychology 13: 1813–1816. Sitaram, R., Caria, A., Veit, R., Gaber, T., Rota, G., Kuebler, A. & Birbaumer, N. 2007. fMRI Brain-Computer Intreface: A Tool for Neuroscientific Research and Treatment. Computational Intelligence and Neuroscience. 1–10. Van Lancker, D. & Sidtis, J. 1992. The identification of affective-prosodic stimuli by left- and right-hemisphere-damaged subjects: All errors are not created equal. Journal of Speech and Hearing Research 35: 963–970. Vouloumanos, A., Kiehl, K. Werker, J. & Liddle, P. 2001. Detection of sounds in the auditory stream: Event-related fMRI evidence for differential activation to speech and nonspeech. Journal of Cognitive Neuroscience 13: 994–1005. Whalen, D., Benson, R. Richardson, M. et al. 2006. Differentiation of speech and nonspeech processing within primary auditory cortex. Journal of the Acoustical Society of America 119: 575–581. Wildgruber, D., Ackermann, H., Klein, M., Riecker, A. & Grodd, W. 2000. Brain activation during identification of affective speech melody: influence of emotional valence and sex. NeuroImage 11: 341. Wildgruber, D., Hertrich, I., Ackermann, H., Riecker, A. & Grodd, W. 2001. Processing of linguistic and affective aspects of speech intonation evaluated by fMRI. NeuroImage 13: 627. Wise, R., Greene, J., Büchel, C. & Scott, S.K. 1999. Brain regions involved in articulation. Lancet 353: 1057–1061.
Grzegorz Dogil
Appendix A schematic illustration of the relevant areas of the human brain depicting the gyri of the left hemisphere (the right panel), the intrasylvian cortex with the insula and the Heschl’s convolutions (the central picture) and the Brodman areas (the left panel). Left panel, adapted from “Brodmann’s ‘Localization in the Cerebral Cortex’” (1999). Translated and edited by Garey, Laurence Hardback. Imperial College Press. Right panel, adapted from http://library.med.utah.edu/ed/eduservices/handouts/index.php Creative Commons Attribution-NonCommercial-ShareAlike 2.5 License.
From concepts to meaning The role of lexical knowledge James Pustejovsky
Brandeis University, USA
1. Introduction Linguists typically speak as though the information associated with words and the phrases they construct is fixed and immutable.1 The data suggest otherwise. For example, what conditions if any does a predicate impose on its arguments, and how are these conditions communicated? Somewhere in the grammar is the information associated with how predicates and their arguments compose to make meaningful expressions. In this paper I examine the mechanisms by which an argument satisfies the conditions imposed by a predicate. In Pustejovsky (1995), Busa (1996), Bouillon (1997), and subsequent work in classic Generative Lexicon (Bouillon & Busa 2001), the question of lexical creativity in the context of models of semantic compositionality was addressed. These works focused on integrating existing analytic tools, such as type, argument, and event structures, into a unified semantic representation, while introducing a new data structure called qualia structure. One of the major goals of this work was a systematic explanation of the phenomena associated with sense shifting associated with polysemy, the logical or systematic relationship between interpretations of a word or phrase in syntax. To this end, semantic composition was enriched with type coercion operations, exploiting richer logical forms including qualia structure. Upon examination, we observe that these cases divide naturally into two general phenomena: inherent and selectional polysemy. Inherent polysemy is where multiple interpretations of an expression are available by virtue of the
1. This work is supported by a grant from NSF-CRI 0551615.
James Pustejovsky
semantics inherent in the expression itself; selectional polysemy, on the other hand, is where any novel interpretation of an expression is available due to contextual influences, namely, the type of the selecting expression. Interestingly, these polysemies are not always in complementary distribution, but rather can both be present in certain contexts or constructions. In the discussion that follows, we examine this distinction in the broader linguistic context of predicate argument selection, and how to limit the application of semantic coercions more systematically. 2. Challenges to strict compositionality In this section, we examine the relation between the type of an argument as selected by a predicate, and the role this argument subsequently plays in the computation of the sentence meaning. One of our major goals here is to account for the contextual modulations in meaning that occur in actual argument selection data. The notion of compositionality assumed in linguistic theory broadly, and in semantics, specifically, is the view inherited from Frege, paraphrased below: “The meaning of an expression is a function of the meanings of its parts and the way they are syntactically combined. (Partee 1984: 153)”. If one assumes a compositional approach to the study of meaning, then two things immediately follow: (1) one must specify the meaning of the basic elements of the language; (2) one must formulate the rules of combination for how these elements go together to make more complex expressions. The first aspect involves fixing the semantics of lexical items, while the second entails defining a calculus for how these elements compose to form larger expressions, i.e., function application and function composition. While linguistic constructions seem to conform to these principles in most cases, there are many well-known cases that do not fit into the conventional compositional paradigm (Partee 2004), from idioms (kick the bucket) and non-intersective modification (fake gun) to adjectival scope issues (occasional sailor). There is, however, another construction which poses a problem for the principle of compositionality in semantics, namely type mismatches. As has long been noted (Grimshaw 1979), predicates do not always get the types they ask for in the phrases they take as their arguments; that is, predicates often seem to appear with arguments of the “wrong type.” For example, in (1a), a countable individual entity is being “coerced” into the food associated with that animal, namely bits of chicken, while in (1b), the mass terms water and beer are being packaged into unit measures (Pelletier 1975). In (2), while aspectual verbs typically select for an event, the NPs in object position here have
From concepts to meaning: The role of lexical knowledge
taken on event denotations. Similarly, in (3), both object NPs are assumed to have propositional interpretations.2 (1) a. There’s chicken in the salad. b. We’ll have a water and two beers. (2) a. Roser finished her thesis. b. Mary began the novel. (3) a. Mary believes John’s story. b. Mary believes John.
These examples illustrate that semantics must accommodate specific type-shifting and coercing operations in the language in order to remain compositional. But it doesn’t stop there. Coercion, actually is all around, and we see it in the way some verbs are polysemous between agentive and non-agentive senses (cf. (4–5) below). (4) a. John killed the plant intentionally. b. The drought killed the plant. (5) a. The bottle rolled down the hill. b. Billy rolled down the hill as fast as he could.
(cf. Wechsler 2005)
It is manifested in the way “concealed question” NPs change their denotation, chameleon-like in certain contexts, as illustrated in (6). (6) a. Mary asked me the weirdest question today. b. Bill knows a good wine for such dinners. c. Jill forgot her friend’s name.
In a similar manner, factives and verbs of denial will subtly shift the meaning of their complements, in order to maintain the type being selected. (7) a. The Senator regretted her question. b. The MP denied reports he was absconding with Department funds.
The example in (7a) is interesting, since the object NP denoting a question is interpreted contextually as a factive. In (7b), what is worth noting is that the events of the reporting are not being denied, but rather it is the content of the reports being denied. In Pustejovsky and Boguraev (1993), it is argued that failing to explain such meaning correlations in language weakens the notion of compositionality. Listing multiple entries for verbs to select distinct selectional contexts merely pushes the
2. Cf. Pustejovsky (1995), Jackendoff (2002), Copestake and Briscoe (1995), and others for discussion of coercion phenomena and their treatment.
James Pustejovsky
problem on to another level of interpretation, keeping composition in place locally, but failing to explain it globally. In order to explain such cases, Pustejovsky (2007) presents a general theory of composition that distinguishes between four distinct modes of argument selection: (a) function application; (b) accommodation; (c) coercion by introduction; and (d) coercion by exploitation. We examine such an analysis below. 3. The building blocks for composition Generative Lexicon (henceforth GL) aims to provide a compositional semantics for language that accounts for the contextual modulations in meaning that occur in real linguistic usage. That is, it can be seen as focusing on the distributed nature of compositionality in natural language. One important aspect of this “context modulation” is systematic polysemy. Recently, there has emerged an appreciation of how complex this problem is (e.g., Egg 2004), as well as a new understanding of the parameters at play in the interpretation of polysemous expressions. Within GL, two factors have been identified as contributing to the interpretation of polysemous terms: the nature of the expression’s lexical semantic representation; and mechanisms for exploiting this information in context compositionally. In recent work, this distinction has been identified with inherent versus selectional poly semy (Pustejovsky 2008). Indeed, polysemy cannot truly be modeled without enriching the various compositional mechanisms available to the language. In particular, lexically driven operations of coercion and type selection provide for contextualized interpretations of expressions, which would otherwise not exhibit polysemy. This is in contrast with Cruse’s (2000) view that it is not possible to maintain a distinction between semantic and pragmatic ambiguity. Cruse suggests that polysemy is best viewed as a continuous scale of sense modulation. The view within GL is generally that a strong distinction between pragmatic and semantic modes of interpretation should be maintained if we wish to model the complexity and provenance of the contributing factors in compositionality. The notion of context enforcing a certain reading of a word, traditionally viewed as selecting for a particular word sense, is central both to lexicon design (the issue of breaking a word into word senses) and local composition of individual sense definitions. However, most lexical theories continue to reflect a static approach to dealing with this problem: the numbers of and distinctions between senses within an entry are typically frozen into a grammar’s lexicon. This sense enumerative approach has inherent problems, and fails on several accounts, both in terms of what information is made available in a lexicon for driving the disambigua tion process, and how a sense selection procedure makes use of this information (cf. Pustejovsky & Boguraev 1993 for discussion).
From concepts to meaning: The role of lexical knowledge
Classic GL (Pustejovsky 1995) proposes that a lexical item has available to it the following computational resources: Lexical Typing Structure; Argument Structure; Event Structure; and Qualia Structure. This model defines a language for making types, where qualia (Formal, Constitutive, Agentive, Telic) can be unified to create more complex concepts out of simple ones. Following Pustejovsky (2001, 2006), we will assume three levels of complexity in the type system, characterized as: natural, artifactual, and complex types. This characterization will hold for types in all domains, i.e., individuals, prop erties, and propositions. Most treatments of GL representations are framed in terms of typed feature structures (Copestake et al. 1993; Bouillon 1997; Pustejovsky & Boguraev 1993). Viewed as such, the first two classes can be stated in terms of qualia. For example, a natural object can be given a function (i.e., a Telic role), and transformed into an artificial type. Complex types are reifications of multiple types, bound by an implicit relation (Pustejovsky 1994). For example, the noun book is doubly referring, in that it denotes both the informational content and the physical manifestation of that content. In this note, we will follow the formalization of qualia structure in Asher and Pustejovsky (2006) using a Type Composition Logic. The set of types is defined in (8) below. e the general type of entities; t the type of truth values. (σ, τ range over all simple types, and subtypes of e.) b. If σ and τ are types, then so is σ → τ. c. If σ and τ are types, then so is σ ⊗R τ; R ranges over A or T. d. If σ and τ are types, then so is σ • τ.
(8) a.
In addition to the conventional operator creating functional types (→), we introduce a type constructor • (“dot”), which creates dot objects from any types σ and τ, deriving σ•τ. This is essentially identical with the construction of complex types in Pustejovsky (1995). We also introduce a type constructor ⊗ (“tensor”) which can be seen as introducing qualia relations to a base type. To illustrate how the type system here is a natural extension of that in Pustejovsky (1995), consider a classic GL type feature structure for a term α, ignoring CONST for now: α
(9)
QUALIA =
formal = β telic = τ agentive = σ
Natural Types In Pustejovsky (1995), the type specification for an expression α, (i.e., the formal qualia value β) is distinct from the other qualia values in the semantic representation for α. The qualia structure, on this view, is the entire
James Pustejovsky
feature structure associated with the expression. Assume that the formal role is always present in the qualia, and hence will be considered the head type of the assignment; that is, [formal = β] is simply written β. The additional qualia values can be seen as structural complementation to the head type. Each qualia value will be introduced by a tensor operator, ⊗. To differentiate the qualia roles, we will subscript the operator accordingly; e.g., [telic = τ] can be expressed as ⊗T τ, [agentive = σ] can be expressed as ⊗Aσ. Now the feature structure for the expression α from (9) can be represented as a single composite type, as β ⊗T τ ⊗A σ. We can see the expression of natural typing throughout the major linguistic categories in the language: (10) a. Nouns: rock, water, woman, tiger, tree b. Verbs: fall, walk, rain, put, have c. Adjectives: red, large, flat, big
These will be our atomic types, from which we will construct our ⊗-types and •-types (artifactual and complex types, respectively). We will assume that the natural entity types, N, are just those entities formed from the Formal qualia value i.e., atomic types. The natural types are formally structured as a joint semi-lattice (Pustejovsky 2001), 〈N, 〉. Predicates that select for these natural types will be called natural predicates. We are in a position to define the natural predicates and relations that correspond to these types. The creation of functions over the sub-domain of natural types follows conventional functional typing assumptions: for any type τ in the subdomain of natural types, τ ∈N, τ → t is a natural functional type. For example, the verbs die and touch select natural type arguments: die: λx:eN [die(x)]; touch, λy:eN λx: eN [touch(x, y)]. Artifactual Types Within the type system, natural entities and natural func tions are defined as atomic types. Artifactual objects, that is, entities with some function, purpose, or identified origin, can now be constructed from the tensor constructor and a specific value for the TELIC or AGENTIVE role. I will adopt the term artifact, in a broad sense, to refer to artifactually constructed objects, natural objects that have been assigned or assume some function or use, as well as to humans with an identified role or purpose, e.g., agentive nominals. Following the discussion above, then, composing a natural entity type, en, with a Telic value by use of the ⊗-constructor results in an artifactual type: beer: liquid ⊗T drink; knife: phys ⊗T cut. As with the naturals, the creation of functions over the sub-domain of artifactual types is straightforward: for any type τ in the sub-domain of artifactual entity types, τ ∈A, τ → t is a artifactual functional type. Below are some examples of such functional
From concepts to meaning: The role of lexical knowledge
types, expressed as λ-expressions with typed arguments: λx:eA[spoil(x)]; λy:eAλx:eN [ fix(x, y)] Complex Types The final type we examine is the •-type, essentially a product type over the other two types. For any two types, there must be a coherence relation identifiable in the model between these two types (cf. Asher & Pustejovsky 2006). Examples of complex types are presented in Pustejovsky & Jezek (2008). The noun book, for example, denotes both information and the physical manifestation of this information. As a result, we type the noun book as phys • info. Creating functions over the sub-domain of complex types proceeds as before: for any type τ in the sub-domain of complex entity types, τ ∈C, τ → t is a complex functional type. The verb read is a complex functional type, since it selects a complex type as its direct object: read: phys • info → (eN → t); λy:phys • info λx:eN [read(x, y)]. 4. Mechanisms of argument selection There are two innovations required to adequately enrich the theory of selection. The first is a richer lexical representation, presented above. The second is a stronger theory of selection. Here we make reference to three mechanisms at work in the selection of an argument by a predicative expression (Pustejovsky 2008). These are: (11) a. pure selection (Type Matching): the type a function requires is directly satisfied by the argument; b. accommodation: the type a function requires is inherited by the argument; c. type coercion: the type a function requires is imposed on the argument type. This is accomplished by either: i. exploitation: taking a part of the argument’s type to satisfy the function; ii. introduction: wrapping the argument with the type required by the function.
Given this three-way distinction, we can now ask when polysemy arises in grammar. We will argue that the ability to assign more than one interpretation to a lexical or phrasal expression is a result of type coercion. Lexical items that are inherently complex in their meaning, what have been termed complex types (or dot objects), will assume the interpretation of whatever selectional context it appears in (even if multiple contexts are available). This phenomenon will be referred to as inherent polysemy, as the potential for multiple interpretations is inherent to the object itself. Most other cases of polysemy we will analyze as selectional in nature.
James Pustejovsky
Now let us examine more closely the types in our language and the mechanisms at work in argument selection. From the point of view of their internal structure, Natural types (e.g., lion, rock, water) are atomic.3 Conversely, artifactual (or tensor) types (e.g., knife, beer, teacher) have an asymmetric internal structure consisting of a head type that defines the nature of the entity and a tail that defines the various generic explanatory causes of the entity of the head type. Head and tail are unified by a type constructor ⊗ (‘‘tensor’’) which introduces a qualia relation to the head type: so, for instance beer = liquid ⊗Telic drink. Finally, complex types (or dot objects) (e.g., school, book, lunch etc.) are obtained through a complex typeconstruction operation on natural and artifactual types, which reifies two elements into a new type. Dot objects are to be interpreted as objects with a complex type, not as complex objects. The constituents of a dot type pick up specific, distinct, even incompatible aspects of the object (for instance lunch picks up event • food, speech picks up event • info etc.). See Asher and Pustejovsky (2006). The selection mechanisms introduced in (11) allow for modulation of types during semantic composition. Matching or Pure Selection takes place when the type call of the verb is directly satisfied by the argument. In this case, no type adjustment occurs. Accommodation occurs when the selecting type is inherited through the type of the argument. Coercion takes place when there is a mismatch (type clash) between the type selected by the verb and the type of the argument. This clash may fail to generate an interpretation (as in the case of *The rock died.): if the verb is non-coercive, and the argument fails to pass the pretest imposed by the verb’s type, it will not be interpreted by the interpretation function (the so-called fail early selection strategy (Pustejovsky 2006). Alternatively, the type clash may trigger two kinds of coercion operations, through which the type required by the function is imposed on the argument type. In the first case, i.e., exploitation, a subcomponent of the argument’s type is accessed and exploited, whereas in the second case, i.e., introduction, the selecting type is richer than the argument type and this last is wrapped with the type required by the function. The reason why two coercion operations are proposed instead of one is that the information accessible to semantic composition can be differently embedded in a noun’s semantics. In both cases, however, coercion is interpreted as a typing adjustment. But where should the type adjustments take place, what sort of adjustments should be made and how pervasive is coercion? These are questions we address in the following sections.
3. The linguistic motivations for establishing a fundamental distinction between natural and non-natural types and the conceptual underpinning of naturals are discussed in detail in Pustejovsky (2006).
From concepts to meaning: The role of lexical knowledge
5. Coercion in different contexts In our investigation, we take as our point of departure previous research on compositional mechanisms in semantics and discourse (cf. Asher & Pustejovsky 2000 and 2006; Pustejovsky 2006), where a set of semantic typing adjustments and rules are developed in order to account for the mismatches between selecting and selected type. We will view the compositional process as the interaction of inherent and selectional meaning potentials. What this means is that, regardless of context, some expressions are more potentially polysemous than others; still, context can wash away this distinction in many cases, as we will see. We first focus on domain-preserving type shifting within the domain ‘‘entity’’. The analysis in Pustejovsky (2006) illustrates how verb-argument composition is accounted for when coercion is operative throughout the grammar. For example, pure argument selection occurs when types match. In the exam ples below, the verb is subscripted with the type selected in object position, while the argument is subscripted with its inherent type. (12) a. Natural, en: John threwen [the rock]en. b. Artifactual, ea: Mary drankea [a coffee]ea. c. Complex, ec: Alice readec [the book]ec.
Pure selection may occur on any level. Now consider coercion, where a natural predicate occurs with a non-matching type in its argument. We look first at exploitation of the inherent type of the argument. Consider the verb throw when composing with arguments that do not match the requested natural type. (13) a. Exploiting from Artifactual, [en ⊗ τ]ea : John threwen [a ball]ea. b. Exploiting from Complex, [en • t]ec : John threwen [the book]ec. c. Exploiting from Complex, [en • t]ec : John believedt [the book]ec.
In both the constructions in (13a–b), the natural component, eN of the type structure is exploited to satisfy the composition. In (13c) the propositional component is exploited from a dot type construction.
Table 1. Verb-Argument Composition Table Verb selects Argument is
Natural
Artifactual
Complex
Natural Artifactual Complex
Selection Qualia Exploit Dot Exploit
Qualia Intro Selection Dot Exploit
Dot Intro Dot Intro Selection
James Pustejovsky
Now consider the converse case, exploitation, where the selecting type “wraps” the argument with material to satisfy the requisite typing. In (14a), a natural is raised to assume the role of an artifactual type, while in (14b), an artifactual is raised to a dot object; in each case, the introduction is sensitive to the predicative context from the governing verb. (14) a. Introduction of Artifactual, [eN ⊗ τ]ea: [The water]en spoiledea. b. Introduction of Complex, [eN • β]ec : John readec [the wall]ea.
We next turn our attention briefly to domain-shifting coercion operations, where the argument expected by the selecting predicate is not just a different type level than the one present, but a different domain, as discussed above in Section 2, and illustrated below in (15). (15) a. Mary believedt [that John is sick]t. b. Mary believedt [the rumour]en. c. Mary believedt [the book]ec.
We have argued previously (Pustejovsky 1995, 2006), that the propositional content from entities such as rumour and book can be used to satisfy the typing of such predicates. Notice, however, that regardless of coercion potential, some verbs simply cannot shift the complement to the desired type: (16) a. Mary said a lie. b. *Mary said a book.
There are two possible explanations for such a failure of context modulation in this case. First, by typing both predicates above as selecting for propositions, it is possible that we haven’t adequately accounted for what distinguishes say and believe as predicates, and lie and book as potential proposition-denoting entities. That is, lies are not merely of type t or s → t, but are also the product of a speech act. This could be subsequently reflected in the type of the verb say, as well as in propositional attitude nominals such as lie. The second possible explanation is that the verb say simply does not permit coercion to apply to its complement, the analysis adopted in Pustejovsky (1998). While descriptively adequate, such a solution requires each predicate to be indexed according to its coercion potential. At this point, more data is required to choose between these analyses, Further investigation will hopefully shed light on the appropriate solution. 6. Conclusion In this note, we have articulated the mechanisms of composition in language that account for sense modulation phenomena involving type mismatches in argument
From concepts to meaning: The role of lexical knowledge
selection. We proposed a framework of argument selection involving lexically encoded type coercion potentials. It remains to be seen how generative such mechanisms are and what aspects of the grammar impose constraints on type shifting operations. Acknowlegements I would like to thank Jessica Moszkowicz for comments on this paper. I would also like to thank Anna Rumshisky, Nicholas Asher, Elisabetta Jezek, Chungmin Lee, and Kiyong Lee, for comments and discusion on earlier versions of this work. References Asher, Nicholas & James Pustejovsky. 2000. The Metaphysics of Words in Context. Journal of Logic, Language, and Information. Asher, Nicholas & James Pustejovsky. 2006. A Type Composition Logic for Generative Lexicon, Journal of Cognitive Science 6: 1–38. Bouillon, Pierrette. 1997. Polymorphie et sémantique lexicale: le cas des adjectifs, Lille: Presses Universitaires du Septentrion. Bouillon, Pierrette & Frederica Busa (eds), The Syntax of Word Meaning, Cambridge University Press, Cambridge. Busa, Frederica. 1996. Compositionality and the Semantics of Nominals, Ph.D. Dissertation, Brandeis University. Copestake, Ann & Ted Briscoe. 1995. Semi-productive Polysemy and Sense Extension. Journal of Semantics. 15–67. Copestake, Ann, Antonio Sanfilippo, Ted Briscoe & Valeria de Paiva. 1993. The ACQUILEX LKB: An Introduction. In Ted Briscoe, Valeria de Paiva, and Ann Copestake (eds), Inheritance, Defaults, and the Lexicon, Cambridge University Press, Cambridge. Cruse, Alan D. 2000. Aspects of the micro-structure of word meanings. In Yael Ravin & Claudia Leacock (eds), Polysemy: Theoretical and Computational Approaches, 30–51. Oxford University Press. Egg, Markus. 2004. Flexible semantics for reinterpretation phenomena. Stanford: CSLI Publications. Grimshaw, Jane. 1979. Complement Selection and the Lexicon. Linguistic Inquiry 10: 279–326. Partee, Barbara H. 1984. Compositionality. In Fred Landman & Frank Veltman (eds), Varieties of Formal Semantics, 281311. Dordrecht: Foris. Jackendoff, Ray. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford University Press. Partee, Barbara H. 2004. Compositionality in Formal Semantics: Selected Papers of Barbara Partee. Oxford: Blackwell Publishers. Pelletier, Francis J. 1975. Non-Singular Reference: Some Preliminaries. Philosophia 5. Pustejovsky, James. 1995. The Generative Lexicon. MIT Press, Cambridge, MA. Pustejovsky, James. 1998. The Semantics of Lexical Underspecification. Folia Linguistica.
James Pustejovsky Pustejovsky, James. 2001. Type Construction and the Logic of Concepts. In Pierrette Bouillon & Frederica Busa (eds), The Syntax of Word Meaning, Cambridge University Press, Cambridge. Pustejovsky, James. 2006. Type theory and lexical decomposition. Journal of Cognitive Science 6. 39–76. Pustejovsky, James. 2008. From Concepts to Meaning. MIT Press, Cambridge. Pustejovsky, James & Branimir Boguraev. 1993. Lexical Knowledge Representation and Natural Language Processing. In Artificial Intelligence 63. 193–223. James Pustejovsky & Elisabetta Jezek. 2008. Semantic Coercion in Language: Beyond Distributional Analysis. Special issue of the Italian Journal of Linguistics.
Language rights, human development and linguistic diversity in a globalizing world Suzanne Romaine
Merton College, University of Oxford, UK
1. Introduction UNESCO (United Nations Educational Scientific and Cultural Organization) proclaimed 2008 as the International Year of Languages. However, there may not be much left to celebrate if linguists are right in predicting that 60 to 90% of the world’s 6,900 some languages may be at risk of extinction within the next 100 years (Nettle & Romaine 2000). UNESCO’s launch of the special year with the slogan “languages matter” will no doubt lead many to wonder exactly why languages matter and whether we should be concerned with the possible disappearance of so much of the world’s linguistic diversity. The issue of human rights and of language rights more specifically comes into the picture when we confront the fact that people do not normally give up their languages, cultures or identities willingly, but continue to transmit them, albeit in changed form over time. Not coincidentally, the vast majority of today’s threatened languages and cultures are found among socially and politically marginalized and/or subordinated national and ethnic minority groups. Estimates of the number of such groups range from 5,000–8,000 and include among them the world’s indigenous peoples, who comprise about 4% of the world population but speak up to 60% of its languages (Nettle & Romaine 2000). There is no reason to expect people would willingly give up their languages if they enjoyed equality of status. As the bedrock of the current political world order, the nation-state is the most critical unit of analysis because it is policies pursued within national boundaries that give some languages (and their speakers) the status of majority and others that of minority. Over 80% (N = 5,561) of the world’s languages are found in just twenty nation-states. The disappearance of a language almost always forms part of a wider process of social, cultural and political displacement where national cultures and languages are in effect those of dominant ethnic groups. It is for this reason that Skutnabb-Kangas (2000) has referred to language shift occurring within the context of forced assimilation as
Suzanne Romaine
linguistic genocide. Where education does not support or actively prohibits the use of a group’s native language, the state in effect moves children from a minority to the dominant group. Although the fate of linguistic diversity has not figured centrally in the world debate on the consequences of globalization, addressing the question of why languages matter requires a new understanding of the critical role of language rights, and linguistic diversity in human development. 2. Linguistic human rights (LHR) and human rights UNESCO’s declaration of February 21st as International Mother Language Day also brings the issue of rights to the fore because they placed special emphasis on international standard setting instruments related to multilingualism. The notion of linguistic human rights (LHR) is an attempt to link the debate about language rights with the relatively well-defined international legal framework already in existence for human rights. All the major international treaties and other legal instruments see human rights as inalienable entitlements inherent to the person and belonging equally to all human beings. As embodied above all in the Universal Declaration of Human Rights (1948), these consist of both civil and political rights as well as economic, social and cultural rights. The second category includes, among others, labor rights, right to an adequate standard of living, social security, food, housing, clothing, education, and health. The inalienability of our common entitlement follows from the fact that we are all human, and therefore all the same, but paradoxically the need to guarantee rights in law arises from the fact that we are diversely different. The pervasive presence of some degree of multilingualism in practically every nation in the world, whether officially recognized or not, indicates a universal need for multilingual/multicultural policy and planning to ensure that members of different language groups within nations have access to and can participate in national affairs without discrimination. Despite the attempt to guarantee freedom of expression and non-discrimination on grounds of language as fundamental human rights, and implicit recognition of the intimate connections between language and forms of cultural expression, national policies are radically out of line with the realities of multilingualism. The majority of countries in the world actually operate either de facto or de jure as monolingual in recognizing only one language for use in education. Only a quarter of all nations recognize more than one language (Edwards 2007: 44). When a multilingual country uses one or more languages exclusively in public schools, and in the administration of government services and activities, it is making a distinction based on language. In showing a preference for some language(s), whether designated as official or national or not, the state’s decision benefits those for whom
Language rights, human development and linguistic diversity in a globalizing world
the chosen language(s) is a primary language, to the detriment or disadvantage of others who either have no or lower proficiency and are denied the benefit of using and identifying with their primary language (Romaine 2007). Efforts at developing international norms of minority rights at both global and regional levels have resulted in a post-1990 flood of international declarations, conventions and recommendations, including, for example, the United Nations Declaration on the Rights of Persons Belonging to National or Ethnic, Religious, and Linguistic Minorities (1992), the Framework Convention for the Protection of National Minorities (1995) and the Declaration on the Rights of Indigenous Peoples (2007). The need for specifically targeted minority rights such as LHR has likewise emerged as a response to the lack of clarity within international human rights about the extent of promotion rights. Generic minority rights to culture and language are too vague to be meaningful, with majorities and minorities usually wanting more than can typically be guaranteed in international law. There are important differences between “tolerance rights” and “promotion rights”. Most democracies provide for freedom of government interference in private language use, but many are reluctant to make legal provision for promotion in the public sector of languages other than the dominant language(s). Some still regard the concept of language rights as “regressive” because they see them as encouraging the persistence of ethnic differences leading to conflict and divided loyalties. Kymlicka and Patten (2003: 11) doubt whether international law will ever be able to specify more than the most minimal of standards. The question of whether and when language shift can be required or expected in deliberative democracies is still unresolved. Likewise, one can question whether it is legitimate for the state to insist that all children be schooled in the majority language of the state as the sole or main medium of instruction. The current framework combining generic rights for all minorities with targeted rights for indigenous peoples and national minorities, is riddled with gaps and inconsistencies. Despite insistence from some experts that there should be no distinction in law between the linguistic rights of autochthonous and allochthonous minorities in human rights treaties (de Varennes 1996), national ethnic minorities still have many more internationally and nationally coded rights than immigrants under current international conventions and laws. These deficiencies indicate how far we still are from a coherent account of the kinds of targeting appropriate in international law. Targeted norms have emerged in an ad hoc way and are often presented as unique exceptions to the rule of generic minority rights. Kymlicka (2007) argues that this sort of ad hoc “mono-targeting” where one particular type of group is singled out for distinctive legal rights, while according all other groups only generic minority rights, is unlikely to be stable. Similarly, Romaine (2007) contends that the legal approach to reconciling status differences in languages with equality in a world where majority rights are implicit,
Suzanne Romaine
and minority rights are seen as “special” and in need of justification, is fraught with difficulty. Because the adoption of a large number of western practices by many traditional peoples appears to suggest that they have assimilated, demands for what outsiders see as “special” rights in order to maintain their language and culture often generate resentment. In the absence of both clarity about defining which groups should be targeted as well as bodies empowered to enforce norms, rights to maintain distinctive cultural and linguistic identities are further undermined by an almost complete lack of recourse for individuals whose rights have been violated. Signing an agreement is not the same thing as implementing and complying with it. The non-binding nature of some declarations and recommendations permits states to claim that they are meeting the requirements, even if only in a minimalist fashion. The “name and shame” approach is of limited efficacy. Skutnabb-Kangas (2000: 494–5, Table 7.1) used UNESCO data to rank 193 countries in terms of how many of the 52 Universal Human Rights instruments they had signed. Norway emerged as the top country with 44, while the United States, which signed only 15, ranked the lowest of any western industrialized democratic government, in 161st place (along with Bahrain, Comoros, Monaco and Somalia). The Cook Islands, Marshall Islands, Nauru, and Niue ranked in last place with only 1. In addition to all these difficulties, many legal experts and political theorists see signs of an impending retreat from the international commitment to multiculturalism. Bell (2007: 602) for instance, refers to a standoff over how far it is possible to accommodate the marginalized without committing suicide as a state, which is countered by concerns of the marginalized not to let states define them. Many are not hopeful about achieving more than minimal tolerance rights plus minor accommodations within LHR. Although the LHR movement has focused on securing a universal right to mother tongue primary education in line with UNESCO’s (1953: 6) much cited axiom “that the best medium for teaching is the mother tongue of the pupil”, Kymlicka and Patten (2003: 36) contend that this sounds more like the conclusion of an argument than the argument itself. A more realistic way forward may reside not in trying to specify a particular policy, but to establish fair background conditions under which members of different language communities can survive.
lobal distribution of linguistic diversity and 3. G language endangerment In a rapidly globalizing world with a handful of very large languages and many thousands of small ones, maintenance of linguistic diversity is inextricably linked
Language rights, human development and linguistic diversity in a globalizing world
to the survival of small communities. Nine languages with more than 100 million speakers each are spoken by just over 41% of the world’s population. Nearly 80% speak a total of only 83 languages, each with 10 million or more speakers. Leaving aside the world’s largest languages, the remaining 6,800 some are spoken by only 20% of the population. These include the smallest languages spoken by a mere .2% of the world’s population. Most, if not all, of these may be at risk because small languages can disappear much faster than larger ones due to the vulnerability of small groups to external pressures in a rapidly changing world. Resource extraction, the spread of mechanized agriculture and development projects damage the environment at the same time as they displace and marginalize people from places they traditionally relied on for their food, shelter, cultural practices and spiritual well-being. Without adequate access to education, health care, and water, they are frequently not recognized by the governments of the states in which they reside and thus are deprived of the right to participate in or direct their own sustainable human development. Discontinuities in transmission of culture and language are frequently accompanied by large human and social costs manifested in poverty, poor health, drug and alcohol abuse, family violence, and suicide. The African continent (and subSaharan Africa in particular) is a case in point, with high linguistic diversity and poverty. Africa as a whole is home to around 960 million people speaking 2,092 (30.3%) of the world’s 6,912 languages. Most of these are concentrated in six of the twenty most linguistically diverse countries (Chad, Democratic Republic of the Congo, Tanzania, Nigeria, Cameroon, and Sudan). In 2006 more than a third of the world’s poor lived in Africa, where mean income has been falling since 1981. The situation of indigenous people is nothing short of dire. Health is systematically worse than among the non-indigenous population, particularly where loss of land and customary resource bases has rendered people unable to maintain their traditional livelihoods and cultural practices, including their languages. In the case of the 300,000 to 500,000 Pygmy people, who live in ten central African countries, those who are able to maintain their traditional forest-based life have better health than those who have lost access to the forest through logging and farming. Forests are a vital component of a Pygmy sense of physical and spiritual well-being. Pygmy communities living outside the forest in fixed settlements are unable to meet their food needs and experience higher rates of infectious diseases such as malaria, tuberculosis, HIV/AIDS and parasites. In the absence of traditional cultural practices reducing social tensions, domestic violence against women and alcohol abuse have increased (Ohenjo et al. 2006: 1939–1941). The San people, hunter-gatherers once referred to as Bushman, today number about 30,000 to 40,000, and are widely recognized as the most impoverished, disempowered, and stigmatized ethnic group in Southern Africa. The human development index for 32,000 some San living in Namibia, where the second largest
Suzanne Romaine
population resides, is not only the lowest, but they are also the only group whose human development index fell between 1996 and 1998. Health and welfare among San in resettlement areas have declined; alcohol consumption and violence against women have increased. Ohenjo et al. (2006: 1943) regard these trends ultimately as “a problem of poverty stemming from the loss of land and livelihoods without a viable alternative”. The circumstances of the Kxoe community of San people in Namibia, who number only about 4,000, have greatly deteriorated since the 1960s to the point where their prospects for survival are extremely poor. Once highly mobile huntergatherers, they were settled on agricultural schemes in the western part of the Caprivi Strip of northeastern Namibia after independence in 1990. An influx of workers involved in the construction of the trans-Caprivi highway linking Namibia to Zambia, Botswana and Zimbabwe has made the Kxoe (especially young girls) particularly vulnerable to HIV/AIDS, which occurs at a higher rate along this major transport corridor than anywhere else in sub-Saharan Africa. Since 1996 AIDS has been the major cause of death for all age groups, and the Caprivi region has the worst health indicators in the country (Brenzinger & Harms 2001). The largest population of San numbering some 47,000 live in Botswana, where they comprise about 4% of the population. Despite the fact that Botswana has the fourth highest per capita income in Africa, the poorest 10% of the population consists largely of San and related minority communities, who receive only 0.7% of the nation’s income (Ohenjo et al. 2006: 1942). The National Constitution makes no reference to the San among the main eight tribes of the country. The government has justified its relocation of San people from the Central Kalahari Game Reserve by claiming that the San deplete the natural resources of the reserve and that providing services to them on the reserve is too expensive. Since their displacement, the San have been unable to adapt to their new surroundings, where they have no means of subsistence, and are increasingly dependent on the government for food relief and cash-for-work programs. Although the conventional wisdom on which most development theories are premised often assumes that people and places are poor because they lack resources, poverty is clearly a complex phenomenon. Cause and effect are often confused in arguments that poverty is caused by lack of infrastructure, services and employment opportunities in the rural communities where most indigenous peoples live. Thus, prevailing policies have generally entailed assimilation and integration into the dominant society’s modes of production and employment, often requiring migration or forced relocation to urban areas. Such strategies are motivated by misperception and stereotyping of indigenous peoples (particularly hunter-gatherers), their lifeways and languages as primitive, backward, and obstacles to development. President Festus Mogae of Botswana, for instance, asked, “How
Language rights, human development and linguistic diversity in a globalizing world
can you have a stone-age creature continuing to exist in the time of computers? If the Bushmen want to survive, they must change, otherwise, like the dodo they will perish” (Ohenjo et al. 2006: 1942). However, the experience of the San, Pygmies and most other indigenous peoples shows that if people are deprived of traditional means of subsistence, but unprepared for employment in other than the lowest and poorest paid jobs, a declining quality of life will result from development projects and assimilation policies (Wilmer 1993: 133). Taylor (2007: 16), for instance, draws attention to what he sees as “a clear contradiction between the desire of many Indigenous people to live in remote areas in small dispersed communities on traditional lands, and the general thrust of government policy that is intent on securing Indigenous participation in the mainstream urban economy as the core means to enhance well-being.” Many indigenous people define themselves in terms of their close relationships to land and community, where a good life is associated with maintaining traditional hunting, gathering, and herding practices. Without access to land, indigenous peoples find it hard to maintain their ways of life and their cultural identity on which transmission of their languages depends. Mill (1859/1869) recognized diversity in life modes as essential for liberty and happiness when he wrote: If it were only that people have diversities of taste, that is reason enough for not attempting to shape them all after one model. But different persons also require different conditions for their spiritual development; and can no more exist healthily in the same moral, than all the variety of plants can exist in the same physical, atmosphere and climate. The same things which are helps to one person towards the cultivation of his higher nature, are hindrances to another… Unless there is a corresponding diversity in their life modes, they neither obtain their fair share of happiness, nor grow up to the mental, moral, and aesthetic stature of which their nature is capable.
Hallett et al. (2007: 394) conclude that “the generic association between cultural collapse and the rise of public health problems is so uniform and so exceptionless as to be beyond serious doubt.” By some estimates, suicide rates are as much as 40% higher among indigenous peoples than among other populations (Hunter & Harvey 2002). This is especially true for young people, whom some have characterized as the lost, broken or stolen generation. Many young people are weakly integrated into traditional culture, and disconnected from their elders, and family support networks. Suicide is a choice of last resort when things go so badly wrong with identity development that youth see no viable way of linking their past, present and future selves. By contrast, when communities are successful in promoting their cultural heritage, they are better positioned to claim ownership of their past and future (Lalonde 2006). The positive effects reverberate in a variety
Suzanne Romaine
of measures relating to youth health and welfare: suicide rates fall, along with rates for intentional injuries. Fewer children get taken into care and school completion rates rise. Now we have proof that language maintenance matters too. In British Columbia First Nations communities where fewer than 50% reported conversational language knowledge had more than six times (96.59 per 100,000) the number of suicides as communities with higher levels of language knowledge. Among the latter the suicide rate was 13.00 per 100,000, well below the provincial average for both Aboriginal and non-Aboriginal youth (Hallett et al. 2007: 396). In fact, reported language knowledge proved to have predictive power over and above that of the six other cultural continuity factors identified in previous research. Most importantly, these studies from Africa and Canada provide strong empirical evidence against the view that assimilation to dominant cultures is harmless or even beneficial to individuals and communities in the ways suggested by prevailing policies and ideologies.
olicy implications of LHR, linguistic diversity and 4. P human development Important policy implications follow from recognizing that “human development is essential for realizing human rights, and human rights are essential for full human development” (UNDP 2000: 2). Impediments to participation in and the right to exercise control over development processes affecting peoples such as the San are increasingly being recognized internationally as human rights violations. Former United Nations High Commissioner for Human Rights, Mary Robinson made explicit the linkage between poverty and the denial of human rights when she observed: “I am often asked what is the most serious form of human rights violations in the world today, and my reply is consistent: extreme poverty” (UNDP 2003: iv). The interdependence of human rights and development emerges nowhere more clearly than in the global debate about poverty reduction in the context of the eight Millennium Development Goals (MDGs) adopted by 147 nations of the UN General Assembly at the United Nations Millennium Summit in 2000. Despite the Human Development Report’s (2000: 8) characterization of poverty eradication as “a central challenge for human rights in the 21st century”, and UNDP’s (United Nations Development Program) forceful reiteration of the maxim that “poverty is a denial of human rights” (UNDP 2003: iv), these principles have taken backstage to economic growth in what has become the largest and arguably most ambitious initiative on the international development agenda. Indeed, the Human Development Report (2006: v) regards the MDGS as a normative framework for human development. However, the MDGs, like most other large development undertakings, were not initiated by poor countries, but were prompted primarily by the United States,
Language rights, human development and linguistic diversity in a globalizing world
Europe and Japan. MDGs assume that poverty will be reduced by growth in gross domestic product through top down infusion of aid. Targeting funding to economically disadvantaged areas and populations does not directly address the socioeconomic and cultural processes and policies that lead to inequalities and poverty in the first place. Despite the rights-conscious approach articulated in the Human Development Reports, which stress that human rights are not “optional extras”, but rather “binding obligations that reflect universal values and entail responsibilities on the part of governments” (UNDP 2006: 4), the discourse of free market economics is preempting and displacing that of human rights (Romaine in press). Such is the power exercised by the dominant discourse within international development circles that it would be hard to find anyone who disagrees with an agenda dedicated to economic growth. Who would not want to reduce poverty by half by 2015? Yet paradoxically, the goal of halving the number of people living on less than $1 a day has been criticized by some as too ambitious and by others as not ambitious enough. Setting the benchmark at only halving poverty rather than eliminating it altogether leaves open the possibility that minorities will constitute the majority of those persons still living in poverty in 2015. Acknowledging that “integrating human rights in poverty reduction strategies does not so much changes [sic] “what” is to be done as to “how” and “why” activities are undertaken (UNDP 2003: iv), it is worth weighing up the economic and human costs of continuing business as usual. Despite evidence of growing rather than decreasing diversity in many education systems, in some countries the trend has been not towards recognition of the need for policy and planning, but the imposition of ever more centralized provision and greater intolerance of diversity. Despite some encouraging developments in some countries, in most parts of the world schooling is still virtually synonymous with learning a second language. Education for minorities in many parts of the world still operates in ways that contradict best practices, with fewer than 10 percent of the world’s languages used in education (Skutnabb-Kangas 2000). The continuation of educational policies favoring international languages at the expense of local ones is part of the development fiasco. Use of local languages is inseparable from participatory development. In the first of a series of annual reports assessing the implementation of MDGs (Global Monitoring Report 2004), at no point is there any reference to a right to education, or to the empowering effects that such a designation might have, despite the fact that every single government in the world (except Somalia and the United States) has explicitly accepted that there is such a right (by virtue of ratifying the Convention on the Rights of the Child). UNESCO’s (2003) position paper on education in a multilingual world also endorsed many of the recommendations that have emerged from the debate about linguistic human rights as a means of reaching consensus on the rights
Suzanne Romaine
of linguistic minorities to ensure social justice. These include the rights of indigenous and minority groups to education in their own language, access to the language of the larger community, and that of the national education system, and international languages. Discussion of a Universal Declaration of Linguistic Rights is taking place under the auspices of UNESCO. Such legislation aims at guaranteeing at an individual level that everybody can identify with their mother tongue(s) and have this identification accepted and respected by others; learn the mother tongue(s) fully, orally (when physiologically possible) and in writing. In most cases, this requires education for indigenous and minority children through the medium of their mother tongue(s); use of mother tongue(s) in official situations (including schools); and that everybody whose mother tongue is not an official language in the country where they are resident, can become bilingual (or multilingual, if they have more than one mother tongue) in the mother tongue(s) and (one of) the official language(s) (according to their own choice). 5. Conclusion The neoliberal ideology underpinning the MDG approach to poverty reduction through prioritizing economic growth has propelled human rights and development down increasingly separate paths. The rhetoric portraying western models of government, development, and education as the only one for contemporary societies continues the colonizing agenda of the past. Rejecting that colonization entails rejecting the discourse as well. It is time for a reconceptualization of the MDGS and a new understanding of poverty and development and their connections to linguistic diversity. Although the adoption of ILO (International Labor Organization) Convention No. 169 in 1989 marked a break with the integrationist approach to indigenous peoples and instead “recognises the aspirations of indigenous peoples to exercise control over their own institutions, ways of life and economic development, which includes the maintenance and development of their identities, languages and religions”, as of 2007 it has been ratified by only nineteen countries. In September 2007 the 192 member nations of the United Nations General Assembly passed the UN Declaration of Rights for Indigenous Peoples with 143 votes in favor and 11 abstentions. Bolivia is the first country in the world to adopt the entirety of the declaration into national law after having elected the first indigenous president (Evo Morales) in the history of the country. Among the objectors were the United States, Canada, Australia and New Zealand, who claimed it gave excessive property and legal powers.
Language rights, human development and linguistic diversity in a globalizing world
Events such as these reveal a significant mismatch between the ambitions of indigenous peoples and the responses of states. States have been generally more willing to engage with socio-economic issues of equity and access than political issues of self-determination and difference that often matter more to indigenous peoples. Refusal to come to grips with indigenous demands for self-determination cripples efforts to overcome indigenous poverty. The two are profoundly connected, and public policy has to take this into account (Cornell 2006). Maintaining the world’s languages goes hand in hand with achieving and maintaining greater selfdetermination as part of a larger strategy of cultural survival. Policies that promote a community’s economic and cultural well-being will be likely to sustain linguistic diversity as well. Viewing language maintenance as integral to the promotion of sustainable, appropriate, empowering development means thinking about languages in the same way as other natural resources requiring careful planning and protection.
References Bell, Christine. 2007. “Liberal multiculturalism” and the limits of international law. Ethnopolitics 6(4): 599–602. Brenzinger, Matthias & Gundel Harms. 2001. HIV/AIDS threat and prevention in marginalized ethnic groups. Experiences among the Kxoe in Namibia. Eschborn: Deutsche Gesellschaft fűr Technische Zusammenarbeit (GTZ) GmbH. Cornell, Stephen. 2006. Indigenous peoples, poverty and self-determination in Australia, New Zealand, Canada and the United States. In Robyn Eversole, John-Andrew McNeish & Alberto D. Cimadamore (eds), Indigenous peoples and poverty: An international perspective. Zed Books in association with CROP International Studies in Poverty Research. Edwards, John. 2007. Societal multilingualism: reality, recognition and response. In Peter Auer & Li Wei (eds), Handbook of multilingualism and multilingual communication, Handbooks of Applied Linguistics [HAL] 5: 447–469. Berlin: Mouton de Gruyter. Global Monitoring Report 2004. Policies and actions for achieving the Millennium Development Goals and related outcomes. Washington, DC: The International Bank for Reconstruction and Development/The World Bank. Hallett, Darcy, Michael J. Chandler & Christopher E. Lalonde. 2007. Aboriginal language knowledge and youth suicide. Cognitive Development 22(3): 392–399. Hunter, Ernest & Desley Harvey. 2002. Indigenous suicide in Australia, New Zealand, Canada and the United States. Emergency Medicine 14(1): 14–23. Kymlicka, Will. 2007. Multicultural odysseys. Ethnopolitics 6(4): 585–597. Kymlicka, Will & Alan Patten. 2003. Language rights and political theory. Annual Review of Applied Linguistics 23: 3–21. Lalonde, Christopher E. 2006. Identity formation and cultural resilience in Aboriginal communities. In Robert J. Flynn, Peter Dudding & James G. Barber (eds), Promoting resilient development in young people receiving care: International perspectives on theory, research, practice and policy, 52–71. Ottawa: University of Ottawa Press.
Suzanne Romaine Mill, John Stuart. 1859/1869. On liberty. London: Longman, Roberts & Green. 4th edn. Nettle, Daniel & Suzanne Romaine. 2000. Vanishing voices. The extinction of the world’s languages. New York: Oxford University Press. Ohenjo, Nyang’ori, Ruth Willis, Dorothy Jackson, Clive Nettleton, Kenneth Good & Mugarua Benon. 2006. Health of indigenous people in Africa. The Lancet 367(9526): 1937–1946. Romaine, Suzanne. 2007. The impact of language policy on endangered languages. In Matthias Koenig & Paul de Guchteneire (eds), Democracy and human Rights in multicultural societies, 217–236. Aldershot: Ashgate/UNESCO. Romaine, Suzanne. In press. Biodiversity, linguistic diversity, and poverty – some global patterns and missing links. In Wayne Harbert, Sally McConnell-Ginet, Amanda J. Miller & John Whitman (eds), Language and poverty. Clevedon: Multilingual Matters. Skutnabb-Kangas, Tove. 2000. Linguistic genocide in education – or worldwide diversity and human rights? Mahwah, NJ: Lawrence Erlbaum. Taylor, John. 2007. Indigenous peoples and indicators of well-being: Australian perspectives on United Nations global frameworks. Social Indicators Research 82: 1–16. UNDP (United Nations Development Program). 2000. Human Development Report 2000. Human rights and human development. New York: Oxford University Press. UNDP (United Nations Development Program). 2003. Poverty reduction and human rights: A practice note, June 2003. New York: United Nations Development Program. UNDP (United Nations Development Program. 2006. Human Development Report 2006. Beyond scarcity: Power, poverty and the global water crisis. New York: Oxford University Press. UNESCO. 1953. The use of vernacular languages in education. Monograph on Fundamental Education VIII. Paris: UNESCO. UNESCO. 2003. Education in a multilingual world. Position paper on education. Paris: UNESCO. de Varennes, Fernand. 1996. Language, minorities and human rights. Dordrecht: Martinus Nijhoff. Wilmer, Franke. 1993. The indigenous voice in world politics. Newbury Park: Sage.
Formal semantics for interpreting temporal annotation Kiyong Lee
Korea University, Korea
1. Introduction As was initiated by Hobbs and Pustejovsky (2003), the current work presented here continues to develop an owl-time ontology-based compositional semantics for formally representing information about time and events in annotated text. Besides various works related to the owl-time ontology, the proposed owl-time-based semantics relies on Katz (2007), Pratt-Hartman (2007), and Bunt (2007) for its design and construction, for each of them independently had also developed a formal semantics for timeml. The input to the proposed semantics is a text annotated by timeml. It is an xml-based markup language for annotating information about time and eventualities in a text, which was developed by Pustejovsky et al. (2005a, b) and now is being developed as an international standard by the iso-timeml joint working group (2007). The proposed semantics for temporal annotation is thus based on the timeml markup language as well as the owl-time ontology with the tacit claim that these two complement each other in extracting and annotating relevant temporal information and reasoning about time and events in natural language text. Triggered by some morpho-syntactic features, timeml focuses on temporal categories (e.g., yesterday, December 31, 2007, from Monday to Friday) and eventdenoting verbs (e.g., destroy) and nouns (e.g., destruction) and marks them up with appropriate attributes and their values in the xml format. owl-time, on the other hand, translates such information contained in the xml elements into finergrained logical forms or representations of their semantic content and consequent implications. As newly introduced to the proposed owl-time-based semantics, the notion of neighborhood plays an important role in subsequent discussions. To this notion, there are two technical subnotions: (1) the neightborhood function N which maps instants to (denumerable) sets of intervals and (2) the designated neighborhood
Kiyong Lee
interval N (t) of some instant t among those (denumerably many) neighborhood interval in one of the sets generated by N. The choice of this interval is constrained by some pragmatic factors such as the perception of a speaker about the structure of time involved in the occurrence of an event. Particularly applied to the notion of the present time, the notion of neighborhood is used to represent the present time interval as the neighborhood N (n) of the present moment of time n, thus differentiating these two temporal notions. It is also used to formulate two different notions of the past: the near past with before (T, n) and the remote past with before (T, N (n)). In section 5, this distinction is shown to play a key role in formally explaining the perfective interpretation of the past tense in Korean.
2 TimeML for temporal annotation timeml specifies a set of guidelines for annotating temporal and event-related information in natural language text. For convenience, it adopts the XML format. These guidelines are formulated for both human annotators and automatic tagging. The specification of timeml introduces three basic xml tags: <EVENT>, <TIMEX3>, <SIGNAL> and three link tags:
, <SLINK> , . Then for each tag, a list of attributes and their values is specified. The following illustrates how some of these tags are used with appropriate attributes and their values in annotating a simple example in English.1 (1) a. Mia wasn’t sick on Sunday b. <TimeML> Mia wasn’t
<EVENT eid = ”e3” eiid = ”ei3” eventType = ”state” pos =”adjective” tense = ”past” aspect = ”none” polarity = ”neg”> sick <SIGNAL sid = ”s1”> on <TIMEX3 tid = ”t4” type = ”date” value = ”2008–02–10” temporalFunction = ”true” anchorTimeID = ”t0”>
. Some minor changes are introduced into the current version of timeml. First, the convention of putting attribute-values in upper-case is removed. Second, the attribute name class is renamed eventType. Third, the attribute value name “OCCURRENCE” is subclassified into “process” and “transition”.
Formal semantics for interpreting temporal annotation Sunday
Note: created on t0 = 2008–02–12
Here, the adjectival part headed by sick is annotated as being of the type state. Its tense is “past”, its aspect “none”, and its polarity “neg”. The value of Sunday is calculated on the basis of the document creation time t0, namely February 12, 2008, and is marked up as such in the <TIMEX3> -tagged element. Finally, based on the <SIGNAL> -tagged element, the is also established as it is. The entire fragment of the annotated text as a whole negates Mia’s being sick on Sunday, stating that there was no such an interval on Sunday during which Mia was sick. All of the relevant event-time related information packed into this annotation need be extracted and represented at the level of the owl-time translation.
3. A sketch of the OWL-time ontology The owl-time ontology adopted here is laid out in Hobbs and Pan (2004) and some relevant parts are outlined here briefly. Instants and Intervals: The owl-time ontology admits both instants and intervals as temporal entities. tseq (T) introduces a sequence of temporal entities. There are two particular instants: the beginning and end points of an interval. They are expressed by the predicates, begins and ends, as a relation over instants and intervals: begins (t, T) and ends (t, T), where the lower-case t stands for an instant and the upper-case T stands for an interval by convention. The predicate inside is also a relation over an instant and an interval: if some instant t satisfies inside (t, T), then it is understood that neither begins (t, T) nor ends (t, T) holds. Interval Relations: The predicate intDuring (T, T') is a subinterval relation over the intervals T and T'. Its inverse is intContains (T', T ). Temporal Relations: In the owl-time ontology, temporal entities are partially ordered, thus allowing the notion of branching time. There are three predicates over pairs of temporal entities: before, after, and overlaps. The predicates before and after are relations over temporal entities, while the predicate overlaps is a relation over pairs of intervals. before (T, t) holds if and only if for some t' such that ends(t', T), the relation before (t', t) holds. after is the inverse relation of
Kiyong Lee
before. overlaps (T, T') holds if and only if there is some instant t such that both inside (t, T) and inside (t, T') hold. Linking Time to Eventualities: Time may be linked to states and events. There are four predicates, atTime (e, t), during (e, T), holds (e, tempEntity), and timeSpan (e, S), where t stands for an instant and T an interval, and also where tempEntity is either an instant or an interval and S is either a temporal entity or a sequence of temporal entities. timeSpan links eventualities to temporal entities or their sequences. atTime (e, t) is understood as stating that some (punctual) event e obtains or takes place at an instant t and during (e, T) as stating that some (extended) state or event e obtains during an interval. If an eventuality obtains during an interval, it obtains at every instant inside the interval and during every subinterval. In case both holds (e, T) and interval (T) hold, then during (e, T) also holds or obtains. Time and Duration Stamps: ISO 8601: 2004 represents the year, month, day, and other times by the use of hyphenation, as illustrated with 2007–12–30. OWLTime, on the other hand, introduces two predicates, timeOf and durationOf, to deal with time and duration stamps, that simply list these dates and times. For example, 2007–12–30 can be dealt with in OWL-Time in two different ways: either (1), as an instant, timeOf (t, 2007, 12, 30) or (2), as an interval, durationOf (T, 2007, 12, 30). The interval referring to the date December 30, 2007, as is specified in (2), for instance, is understood as satisfying the above property. Neighborhood Interval (new): There are two notions related to neighborhood: one is a function N that maps instants to (denumerable) sets of intervals and another is a designated neighborhood interval N (t) of some instant t. While the value of the neighborhood function N (t) for some instant t denotes a denumerable set of intervals around t, the notation N (t) is used to designate some interval in N (t). The choice of such an interval N (t) is pragmatically controlled and loosely called the neighborhood of t. 4. OWL-Time-based semantics Event Types: The proposed owl-time-based semantics assumes an event ontology laid out in Pustejovsky (1991). Here, eventualities are subtyped into state (e.g., be sick, love, know, live), process (e.g., walk, climb, hiccup), and transition (e.g., arrive, leave, reach to the top). A state is viewed as a single homogenous state of affairs, a process as a sequence of related events, and a transition as an event that terminates. Furthermore, in Moens and Steedman (1988), a transition, particularly related to the perfective aspect, is treated as a
Formal semantics for interpreting temporal annotation
culminated process, which is accompanied by a consequent state. These distinctions are reflected in the owl-time translation rules.2 Translation Rules: The owl-time-based semantics proposed here aims at con structing a compositionally applicable set of translation rules. This task is very much constrained by both the structure of timeml and the ontology of owl-time in theory and also by the range of kinds of annotated text in practice. The translation rules presented here are thus only a fragment for illustration. There are two types of rules: (i) dRules, denotation rules, each of which applies to an xml-tagged element annotating a part of text to derive an owl-time σ-form of some sort, and (ii) cRules, combinatorial rules, each of which combines two owl-time σ-forms into the f(a) format of functional application. As proposed by Harry Bunt (through personal communication), the form of each rule is a triplet <eii, tj\ej formula>, which is reflected in naming each rule. The first denotation dRule 0 is named σeiiTjProp and this should be understood formally as standing for <eii, Tj, Prop>, where Prop denotes the propositional content of an event instance identified as eii which is related to the time tj.
OWL-Time translation forms:
• Parameters: eii (event instance id), t (instant), T, Ti , Tj (interval), Rtp (binary, Type), Rta (binary, Tense and Aspect), Rtr (binary, TRel). • dRule 0 for TLinked Propositional Frame σ eiiTjProp: <TimeML> <EVENT eid = ”ei” eiid = ”eii” content =”word” eventType = ”type” tense = ”tense” aspect =”aspect” polarity = ”pol”> word … ⇒ 1. pol = pos ⇒ σ eiiTjProp := λRtpλRtaλRttr ∃ {eii, Ti , Tj}
[word'(eii) ^ Rtp(eii)(Ti) ^ Rta(eii)(Ti) ^ Rtr(Tj)(Tj)] 2. Pol = neg ⇒ σ eiiTjProp := λRtp λRta λRtr ∃{eii, Ti, Tj} [Word'(eii) ^ Rtp(eii)(Ti) ^ Rta(eii)(Ti) ^ Rtr(Tj)(Ti)]
. Following the recommendations of the iso-timeml Working Group, no distinction is made between punctual and extended eventualities, although the owl-time ontology is rich enough to accommodate such a distinction.
Kiyong Lee
•
dRule 1 for Propositional Frame σ eiiProp: <TimeML> <EVENT eid=”ei” eiid=”eii” content=”word´” eventType=”Type” tense=”tense” aspect=”aspect” polarity=”pol”> word ⇒ 1. pol = pos ⇒ σeiiProp : = λRtpλRta ∃{eii, Ti}
•
• •
•
[word'(eii) ^ Rtp(eii)(Ti) ^ Rta(eii)(Ti)] 2. pol = neg ⇒ σeii jProp : = λRtp λRta ∃{eii, Ti} [word'(eii) ^ Rtp(eii)(Ti) ^ Rta(eii)(Ti)] dRule 2 for Propositional σ type: 1. type=state ⇒ σ type : = λTλe[during (e, T)] 2. type=process ⇒ σ type : = λTλe[timeSpan (e, T) ^
2. a spects = progressive ⇒ σaspect : = λTλe [timeSpan(e, T)] 3. a spects = none ⇒ σaspect : = λTλe [T = T] • cRule 2 for Combining Tense and Aspect: • •
•
• •
σtenseAspect : = σtense(σaspect)
cRule 3 for Tense-Aspect Typed Propositions: σeii(Tj)PropTypeTA : = σeii(Tj)PropType(σtenseAspect)
dRule 5 for : word ⇒ σtimex : = λT [durationOf(T, y, m, d)] dRule 6 for : ⇒ σTRel : = λAλTλT' [intDuring(T, T' ) ∧ A(T' )] cRule 4 for Timex-Constrained Temporal Relation: σtRelTimex : = σtRel(σtimex)] : = λTλT' [intDuring(T, T' ) ∧ durationOf(T', y, m, d))
cRule 5 for Combining <event> with : σeiiTjPropTypeTAtRelTimex : = σeiiTjPropTypeTA(σtRelTimex)
Formal semantics for interpreting temporal annotation
Illustration: Here is an example that illustrates the step-by-step combinatorial process of obtaining the final representation of the annotated fragment just given:
(2) a. b. c. d.
Mia wasn’t sick on Sunday Parameters: {ei3, T3 (event time), T4 (Sunday:2008–02–10)}. The present time interval N(n) = (2008–02–15)} dRule 0: σei3T4Prop : = λRtpλRtaλRtr ¬∃ {ei3, T3, T4} [sick(ei3) ∧ Rtp(ei) (T3) ∧ Rta (ei3) (T3) ∧ Rtr (T4) (T3)] dRule 2: σType: = λTλe[during(e, T)]
e. cRule 1: σei3T PropType : = σei3T Prop (σType) 4 4 : = λRtaλRtr ∃ {ei3, T3, T4} [sick (ei3) ^ during(ei3, T3) ^ Rta(ei3)(T3) ^ Rtr(T4)(T3)] f. dRule 4: σtense : = λAλT [before (T, N(n)) ^ A(T)] g. dRule 3: σaspect : = λe [T = T] h. cRule 2: σTA : = λTλe [before (T, N(n))] i. cRule 3: σei3T PropTypeTA : = σei3T PropType(σtense Aspect) 4 4 : = λRtr ∃{ei3, T3, T4} [sick (ei3) ^ during(ei3, T3) ^ before (T3, N(n)) ^ Rtr (T4)(T3)] j. dRule 5: σtimex : = λT[duationOf (T, 2008, 02, 10)] k. dRule 6: σtRel : = λAλTλT' [intDuring (T, T' ) ^ A (T4)] l. cRule 4: σtRelTimex : = σtRel(σtimex) : = λTλT' [intDuring(T, T') ^ duationOf(T', 2008, 02, 10)] m. cRule 5: σei3T PropTypeTAtRelTimex := σei3T PropTypeT A(σetRelTimex) 4 4 : = ∃{ei3, T3, T4}[sick (ei3) ^ during (ei3, T ) ^ before(T , N(n)) 3 3 ^ intDuring(T3, T4) ^ durationOf (T4, 2008, 02, 10)]
Here the interval variable T3 is existentially bound within the scope of negation and also restricted to be a subinterval of the interval T4 dated February 10, 2008. Thus, this representation is now interpreted as stating that there was no interval in the past, namely on Sunday February 10, 2008, during which Mia was sick. 5. Representing and interpreting the past tense in korean The semantics of the past tense in Korean has some interesting issues. First, the anchoring of tense, either unmarked or past-marked, may depend on the tense of the sentence-final verb or the matrix sentence as in mia-ka wain-ul masi-ko [tense unmarked] ca-ss-ta “Mia drank wine and then slept” or mia-ka wain-ul imi masi-essta-ko [past-marked] hay-ss-ta “Mia said that she had already drunk wine”. Second, there are two suffixal forms of the past tense: the simple past -ess- and the double past -ess.ess-. They both refer to eventualities in the past, but carry different logical or pragmatic implications, as formally analyzed by C. Lee (1987: 436–440). The
Kiyong Lee
simple past often implies the continuation of a state itself or a consequent state resulting from an event of the transition type, like the present perfective in English, especially when it is used with so-called resultative verbs, as generally treated as exceptions by Chang (1996: 121–122), Lee and Ramsey (2000: 174–175), Sohn (1999: 363), and many other Korean linguists. This paper restricts its discussion to the second issue that has been much debated without any formal explanation. 5.1 Two forms of the past tense The simple past verbal ending -ess-, or its variant, and the double past verbal ending -ess.ess- are each suffixed to an adjective or verb stem to form an adjective or a verb. Here are simple examples: (3) a.
mia-ka yeypp-ess-eyo mia-subj pretty-past-sentenceEnding ‘Mia was pretty’
b. mia-ka yeypp-ess.ess-eyo mia-subj pretty-doublepast-sentenceEnding ‘Mia used to be pretty’
Both of these sentences talk about the state of Mia’s being pretty in the past. But, unlike the simple form, the double past form is used to denote an event that held in the past, but terminated at a certain point in the past. For this reason, the double past is often called the remote past, as in Lee & Ramsey (2000: 178), or treated as denoting “an event or a state remoter or more definitely completed than that of the past …” by Chang (1996: 122). These two differently denoted states can be represented in owl-time with two different forms of the past: before (T, n) and before (T, N(n)), as shown below: (4) a.
simple past: mia-ka yeypp-ess-eyo σei1SimplePast : = pretty(ei1) ∧ during (ei1, T) ∧ before (T, n)
b. double past: mia-ka yeypp-ess.ess-eyo σei2DoublePast : = pretty(ei2) ∧ during (ei2, T) ∧ before (T, N(n))
Furthermore, the double past carries an implication that the eventuality referred to terminates at the end of the interval during which it used to hold. This terminating property of the double past as the remote past is added to σei2DoublePast in the following:
(5) σremotePast : = λR[σei2DoublePast ∧ R(ei2)(T2)](λT λe[ends(t, T) ∧
during (e, N(t))])
These distinctions explain two facts in Korean: (1) the modification of the simple past with temporal adverbs like cikum ‘‘now’’ that denotes the present time and (2) the interpretation of the simple past like the present perfective in English.
Formal semantics for interpreting temporal annotation
5.2 The present perfective interpretation of the past tense In English, present-time denoting temporal adverbs like now may not modify past-tensed verbs, although they can modify present perfective verbs. In contrast, temporal adverbs denoting the past like yesterday may modify past-tensed verbs, but then fail to modify with present perfective verbs. If this is the case, then the past here should be treated as before (T, n) so that an overlap between a portion of the past and the neighborhood of the present moment of time becomes possible as in Korean. (6)
a. b. c. d.
Mia has gone to New York now *Mia went to New York now Mia went to New York yesterday *Mia has gone to New York yesterday
Simple past-tensed verbs in Korean, on the other hand, may be modified by temporal adverbs denoting the present time or its neighboring intervals like cikum, icey, pangkum, kumbang or sibang as well as those adverbs or nouns referring to the past like ecey “yesterday” or cajnyen “last year”, as was analyzed with formal descriptions in K. Lee (1976, 1998). Here are examples: (7) a.
ecey mia-ka pusan-ey ka-ss-ta yesterday Mia-subj Busan-goal go-past-decl ‘Mia went to Busan yesterday’ subj: subject, decl: declarative sentence ending
b. cikum mia-ka pusan-ey ka-ss-ta now Mia-subj Busan-goal go-past-decl ‘Mia has gone to Busan now’
Through the owl-time translation given below, the following two issues can be resolved: (i) how present-time denoting adverbs like cikum “now” can modify verbs with the past tense in Korean and also (ii) how such verbs are interpreted as the present perfective as in English.
(8) a.
cikum mia-ka pusan-ey ka-ss-ta
b. cikum
mia-ka pusan-ey <event eid=”e1” eiid=”ei1” content=”goToBusan” eventType=”transition” tense=”past” aspect=”none”> ka-ss-ta
c.
σcikum : = λTλT' [intDuring(T, T' ) ∧ T' = (N (n))]
Kiyong Lee
d. σei1T0PropTypeTA : = λRtr∃ ∧ {T0, T1} [goToBusan(ei1) ∧ holds(ei1, T1) ∧ ends(t, T1) ∧ ¬ holds(ei1, N(t)) ∧ before (T1, n) ∧ r = resultingState(ei1) ∧ during(r, N(t)) ∧ Rtr](T0)(T1))] e.
σei1T0PropTypeTAtRelTimes : = σei1T0PropTypeTA(σcikum) : = ∃ {T0 , T1} [goToBusan(ei1) ∧ holds(ei1, T1) ∧ ends(t, T1) ∧ ¬ holds(ei1, N(t)) ∧ before (T1, n) ∧ r = resultingState(ei1) ∧ during(r, N(t)) ∧ intDuring(T1, T0) ∧ T0= N(n)]
Here, the temporal adverb cikum “now” restricts the range of the interval T1 to the present time N(n) as its subinterval, namely intDuring(T1 , T0), where T0 = N(n). This does not block its cooccurrence with the simple past before (T1 , n), for it allows an overlap between T1 and N(n) unlike the remote past before (T, N(n)). Furthermore, since both ends(t, T1), and overlaps(T1, N(n)) hold, the following two also hold: inside(t, N(n)) and overlaps(N(t), N(n)) so that the resulting state r holds in N(n) as well as in N(t) and also that both the initiation and the termination points are inside, or within, N(n), although the event time T1 is still before n, and thus in the past. Note also that the event does not terminate on the end point of T1, but only after that point because the event still holds on that point. The translation above thus allows the interpretation that not only the main event of Mia’s going to Busan has taken place in the neighborhood of the present moment of time, but also its consequent state, possibly of being in Busan, holds at the present time. Such is the intepretation that is attributed to the present perfective in English. 6. Concluding Remarks In this paper, an owl-time-based temporal semantics has been proposed to devise a compositional way of translating annotated text with timeml into owltime σ -forms that stand for semantic representations. These forms are interpreted informally as carrying various sorts of fine-grained temporal and event-related semantic content and implications corresponding to the attributes and their values related to the event-type, tense, and aspect in text as annotated according to the timeml specifications. To this task, both timeml and owl-time make their unique contributions. The former extracts appropriate markables from a text and marks them up as simply as possible by assigning appropriate values to the attributes in each of the timeml tags. The latter, then, unpacks the marked-up information into owl-time σ -forms with a rich expressive power that are interpreted (informally) for their semantic content and implications. These two are thus found to complement each other in doing the task of temporal semantics.
Formal semantics for interpreting temporal annotation
The designated neighborhood interval N(t) of some instant t is shown to play an important role in applying the proposed owl-time-based temporal semantics to natural language, particularly to the interpretation of the past tense in Korean. By formulating two types of the past, before (T, n) and before (T, N(n)), it is now possible to formally describe the notions of the near past and the remote past.3 This then is shown to formally analyze the modification of past-tensed verbs with present-denoting adverbs and also the intepretation of the past tense as the present perfective, especially with verbs of the type transition or so-called resultative verbs in Korean. This paper has not done any model-theoretic work for temporal annotation semantics, as has been initiated by Katz (2007) and Pratt-Hartman (2007). This is one of the future tasks that need be done.
Acknowlegements I am very much indebted to Harry Bunt, Tyrone Cashman, Suk-Jin Chang, Jae-Woong Choe, Roland Hausser, Chungmin Lee, Hwan-Mook Lee, Chongwon Park, and James Pustejovsky for improving this paper.
References Bunt, Harry. 2007. The semantics of semantic annotations. Proceedings of the 21st pacific Asia conference on language, information and computation 13–28. The Korean Society for Language and Information. Chang, Suk-Jin. 1996. Korean. Amsterdam: John Benjamins. Hobbs, Jerry & Feng Pan. 2004. An ontology of time for the semantic web. acm Transactions on Asian language information processing 3(1): 66–85. Hobbs, Jerry & James Pustejovsky. 2003. Annotating and reasoning about time and events. Proceedings of aaai spring symposium on logical formalizations of commonsense reasoning(Stanford, CA, March 2003). Reprinted in Mani et al. (eds), (2005). 301–315. ISO-TimeML Working Group. 2007. ISO CD 24617–1: 2007 Language resource management – semantic annotation framework (SemAF)- part 1: time and events. ISO/TC 37/SC 4/ WG 2. Katz, Graham. 2007. Towards a denotational semantics for TimeML. In Schilder et al. (2007). 88–106.
. Chongwon Park (through personal communication) pointed out the possibility of applying these distinctions to the analysis of a sentence like Bridget went to the coffee machine now, which he claims that many native speakers accept as well-formed.
Kiyong Lee Lee, Chungmin. 1987. Temporal expressions in Korean. In Verschueren, Jeff & Marcella Bertuccelli-Papi (eds), The pragmatic perspective: selected papers from 1985 international conference on pragmatics, 435–453. Amsterdam: John Benjamins. Lee, Iksop & S. Robert Ramsey. 2000. The Korean language. Albany, NY: State University of New York Press. Lee, Kiyong. 1976. sikanlon: cikum-ui uimi (A theory of time: the meaning of now). Language reserach (Seoul National University), 12(2): 161–174. Reprinted in K. Lee (1998): 21–43. Lee, Kiyong. 1998. Sicey-wa yangsang: kanung seykey uimilon (Tense and modality: the possible worlds semantics). Seoul: Taehak-sa. Mani, Inderjeet, James Pustejovsky & Rob Gaizauskas (eds). 2005. The language of time: a reader. Oxford: Oxford University Press. Moens, Marc & Mark Steedman. 1998. Temporal ontology and temporal reference. Computational linguistics 14(2). 15–28. Reprinted in Mani et al. (eds). (2005). 93–114. Pratt-Hartman, Ian. 2007. From TimeML to interval temporal logic. Proceedings of the seventh international workshop on computational semantics (IWCS-7) 166–180. Pustejovsky, James (1991). The syntax of event structure. Cognition 41. 47–81. Reprinted in Mani et al. (eds). (2005). 33–60. Pustejovsky, James, Robert Knippen, Jessica Littman & Roser Saurí. 2005a. Temporal and event information in natural language text. Language resources and evaluation 39(2–3). 123–164. Pustejovsky, James, Robert Ingria, Roser Saurí, José Castanõ, Jessica Littman, Robert Gaizauskas, Andrea Setzer, Graham Katz & Inderjeet Mani. 2005b. TimeML: Robust specification of event and temporal expressions in text. In Mani et al. (eds) (2005). 545–558. Schilder, Frank, Graham Katz & James Pustejovsky (eds). 2007. Annotating, extracting and reasoning about time and events. Berlin: Springer. Sohn, Ho-Min. 1999. The Korean language. Cambridge: Cambridge University Press.
Parallel sessions
Words, mind, and brain Gary Libben
University of Alberta, Canada
1. Words such as language, mind, and brain My goal in this paper is to discuss how a consideration of the processing of words in general, and the processing of multimorphemic words in particular, may inform our conceptions of the relation of the constructs (and the words) language, mind and brain. Words are the building blocks of language use. More than any other aspect of language, they appear to be autonomous elements that are acquired as units, sometimes implicitly and sometimes explicitly. Words imbue us with a sense of control, as though the ability to name things would give us power over them. Words bind persons to one another, for, to fully function as words, they must be shared in form and function among individuals. Perhaps most importantly, the words that we hear, see, say, or write shape our perception of reality by both creating complex units and distinguishing among them. We all have had the experience in which the acquisition of a technical vocabulary in some domain, (e.g., types of automobiles) enhances our perception of differences among members of that domain. We have probably also all had the experience in which acquiring a word for an abstract superordinate concept (e.g., globalization) creates a rubric under which formerly disunited elements are now bound together. As an example of this phenomenon, consider Heraclitus’ famous observation that a man can never step into the same river twice. Surely, this observation would lose a considerable amount of its force if we did not have a word for “river” that binds together its ever-changing waters, or proper names for particular rivers that bind the ever-changing waters to a particular location and course. Here, as in many other cases, the words play a large role in creating the reality. Now, what about the words “language”, “mind” and “brain”? How much of our research reality have they created? In my view, the use of these distinct words has created a great deal of our research reality, with both positive and negative consequences. On the positive side, the distinct words have allowed us to make intellectual progress in the understanding of human language ability by dividing the problem in such a manner that it can be dealt with by scholarly specialists.
Gary Libben
The distinct words have allowed us to formulate questions such as “How is language processed in the mind? or “Where is language represented in the brain?”. The negative consequences of the use of these distinct terms are, of course, closely related to their advantages. Dividing the problem of lexical processing into the distinct domains of linguistic analysis, psychological modeling, and neuronal instantiation makes it perhaps too easy to relegate the problem of the intersection of language, mind and brain to the domain of future research. It also enables us to be less thorough in the study of this intersection than we might be forced to be in the constituent scholarly domains of linguistics, psychology, or neuroscience. Worst of all, it might allow us to form possibly nonsensical questions, such as the two posed above. The reason why questions such as “How is language processed in the mind? or “Where is language represented in the brain?” are quite likely nonsensical is that they take as their unstated premise that language is fundamentally something “out there” and distinct from the internal constructs of “mind” and “brain”, which serve to enable language processing and/or storage. While it is uncontroversially true that human language has manifestations “out there” such as in speech or writing, the fundamental and central nature of language is that it is internal to human beings. Taking the domain of words as the focus of the discussion, I will argue that all properties of words are psychological properties and that these properties are intrinsically tied to properties of the brain and its operation, which we call mind. 2. Words such as languages, mindful, and brainless At the outset of this paper, I claimed that words are the building blocks of language. This seems to suggest that words are the basic simple elements from which more complex language events are built. There are very good reasons to believe that this is not the correct way to look at the role of lexical processing in language communication. More, likely, it seems to me, is that the exquisite complexity of language is evident at all levels, and perhaps most evident at the level of lexis. The source of this complexity is that most words of most languages are multimorphemic. Even in English, a language well known for its impoverished morphology, most uninflected words are composed of more than one morpheme (Libben 2007). Thus, when we consider the manner in which the study of words can inform our understanding of the intersection of language, mind and brain, it is important to consider that multimorphemic words represent the normal rather than the exceptional cases. The centrality of multimorphemic words both within and across languages is what makes them particularly revealing of cognitive functioning. Because
Words, mind, and brain
multimorphemic words are, by definition, composed of smaller meaningful units, they present to us two faces of the mental operation. On the one hand, (and I will restrict myself here to derived and compound forms) words need to be considered as part of memory. They must be acquired and maintained in memory; they must be associated in complex ways. Yet, on the other hand, words also have a computational side that is, at least conceptually, distinct from memory representation and associative links. Such a computational face must be posited if we are to account for the human ability to create and interpret novel words such as foamstick, foamstair, and foamable. The mechanisms that underlie this computational ability are not yet clear, and it may also be the case that not all types of novel multimorphemic words can be interpreted using the same types of processes. In the examples above, foamstick could conceivably be understood in terms of analogy with other compounds that end with stick (e.g., lipstick, nightstick, breadstick). It will be noted that these all have different semantic relations between head and modifier (i.e., a stick for lips, a stick used at night, a stick made of bread). Research by Gagné and colleagues has suggested that such semantic relations among compound elements influence how compound words are related to one another (Gagné 2001, 2002). The novel form foamstair might require different computational processes for its interpretation. In this case, analogy based on the constituent elements will not get us very far, because neither the left constituent foam nor the right constituent stair participate in other English compound words with a frequency that could serve as the basis for analogy. Finally, the interpretation of the novel derived form foamable, seems to be able to invoke the “pre-compiled” computation that constitutes the function and meaning of the suffix –able. The relation noted above between the manner in which novel multimorphemic forms are processed and the properties that their constituents have in existing words of the language offers us an important cue to how lexical processing can be best modeled and understood. Leaving aside the chicken and egg question for the moment, we can imagine that existing words in the language create the patterns which enable the interpretation of new complex words. These patterns can involve abstract morphological configurations (say, that the prefix re- can only attach to verbs), semantic patterns (say, that the prefix re-, when added to a stem, will acquire meanings associated with “back” or “again”), and also patterns that are exemplar-based. Although it is not entirely clear to what extent all putative abstract patterns are in fact generalizations of exemplar-based computation (Baayen 2007), the result for language processing is very clear: Once a word such as broadcast is part of our minds, related new forms such as sportscast and podcast are more easily incorporated and interpreted. A recent example of this is the coinage, Godcast, modeled on podcast. This word seems to be easily understood correctly, as a form of internet religious audio programming and
Gary Libben
already has a Google hit frequency of over 43,000. We can expect that within a very short period of time it will no longer be regarded as a novel form. But, now, an interesting question to think about is: When the (no longer novel) word Godcast is recognized, what role will the related words broadcast and podcast play in the recognition process? Recent findings suggest that they will continue to play a very large role. It is not only the case that there is a spread of association from the target word (in this case Godcast) to related words, but also that the number of morphological relatives that a word has influences how it will be activated in the first place (Moscoso del Prado Martin, Bertram, Schreuder & Baayen 2004). Findings such as these seem to point to a perspective in which words are perhaps less naturally modeled as distinct and encapsulated representations in the mind/ brain. Rather, they might be best seen as patterns of subactivation that can be isolated under specific circumstances (i.e., when a participant is required to identify one and only one word). Under such circumstances, we expect that two processes play an important role: one that results in a relative increase in the activation of the subnetwork, and one that results in a relative decrease in the activation of other potentially related subnetworks. This interplay between activation and deactiviation in the processing of words in general, and the processing of multimorphemic words in particular, underlines the fact that the identification of a word is not likely to be a discrete mental event, but rather the spotlight we shine on a particular aspect of cognitive brain activity at a particular moment in time. Recent work in both psycholinguistics and neurolinguistics (Libben, Buchanan & Colangelo 2004) has highlighted how important the role of deactivation is in normal lexical processing. Perhaps for this reason, multimorphemic words constitute, for the language user, a double-edged sword in online language processing. On the one hand, they have the advantage of being linked to a network of associations to their pre-existing constituents and to the ways in which those constituents have been encountered and used in other multimorphemic constructions. On the other hand, the richer the network of related but not identical forms to which they are associated, the greater the need to engage in the deactivation of those forms in order to uniquely isolate a word. It is important to note that framing issues of the processing of multimorphemic words in this manner suggests that the properties of words are, at bottom, the properties of their mental instantiation. In my view, this is a critically important feature – one that may serve to help us better understand the underlying nature of the most robust variables in lexical processing. These variables include lexical frequency, constituent frequency, abstractness, morphological productivity, and morphological family size. In the next section I make a case for their being fundamentally psychological in nature.
Words, mind, and brain
3. All lexical variables are psychological One of the greatest dangers associated with seeing words as entities that are “out there” and of dissociating the terms language, mind and brain, is that it leads us to conceptions in which the properties of words such as lexical frequency are seen as stimulus properties that are external to mental processing such that they can influence that processing as independent variables in language processing experiments. In my view, measures such as lexical frequency do not affect processing, they are processing. Of course, values associated with lexical frequency seem to be external because we can look them up in external sources, for example in a word frequency list, a corpus, or a similar lexical database. But it is important to remember that these frequency values only have predictive utility if we can use them as estimates of how often an individual participant has encountered or used a particular word form. As such, the variable is not really a property of a word, but a property of the cognitive system as it related to that word. Consider, as a very close analogy, what it means when we think about mental processes that might differ between seeing someone who is familiar to us walking down the street and someone who is less familiar. First, it would be very obvious that this familiarity variable is much more a property of the individual perceiver, than of the individual who is being perceived. Second, we might imagine that a closer inspection of the underlying nature of familiarity in such a context will reveal to us that the familiarity effect is but a shorthand for differences in the manner of processing, rather than the properties of any fixed representations in the mind/brain. I would claim that all the lexical variables that we typically employ and measure as though they were independent variables in language processing experiments are fundamentally of this sort. That is, they are not independent variables at all, but part of the participant’s response. We might consider as an example of the consequences of this view, the variable morphological family size. One way to define this variable is as the total number of derived words that have this target word as a constituent (Schreuder & Baayen 1997). A variant of this that seems to be particularly well linked to the processing of multimorphemic words is the variable “positional morphological family size”. In this case, we count not simply how many words a constituent can be found in, but specifically how many words have that constituent in a particular location. So, for example, the positional family size of the constituent –cast in broadcast will be the total number of words that have cast as the final constituent of a biconstituent compound (in this case, 24, if only simple compounds are counted). The fact that this variable has been found to be very predictive of the recognition
Gary Libben
latencies of members of the morphological family suggests (at least to me) that this lexical variable is in fact not an independent variable that influences how words will be processed in the mind/brain but, in fact, is an estimate of properties of the response. It seems reasonable to assume that positional family size effects reflect the co-activation of the morphological neighbours of a word and that the total co-activation is facilitatory for certain tasks such as lexical decision (see below for a discussion of this task). Interestingly, positional family size effects are enhanced when opaque members of the morphological family are removed from the set (de Jong, Feldman, Schreuder, Pastizzo & Baayen 2002). Thus, recognition of a target such as broadcast will be facilitated by the activation of cablecast, faxcast, telecast, etc., but not by the presence of typecast in the language users’ vocabulary set (assuming that the language user interprets the words typecast and broadcast correctly as involving a different sense of –cast). Now, I am hoping that I have not introduced any claims thus far that are particularly counter-intuitive. Upon reflection, we would all agree that words are fundamentally internal constructs that correspond to patterns of representation and/or activation in the brain. We would probably agree that the meanings of words for each individual boil down to how that individual interprets them and expects that others will interpret them. Finally, I expect we would agree that multimorphemic words, by definition, require a rich set of inter-lexical links in order to function as they do in human natural language. I hope that I have also been able to highlight that, despite the fact that we would consider these claims to be true, our first-pass conceptualization of lexical processing involves thinking about words and their properties as external stimuli which affect changes to the individual. Most often, the goal of language processing experimentation, using this first-pass view, is to measure the effect that such stimuli have on properties of some response. In the following section, I wish to argue that this is not enough. All true lexical variables are internal and what we need in psycholinguistic methodology is an enhanced means by which we can tap into the processes that mediate between an external stimulus and an observable response. It seems to me, that such techniques are currently available and can be deployed in a systematic and coordinated manner. 4. Methodology: The state of the art and the shape of the future Until quite recently, the psycholinguistic study of lexical processing was clearly dominated by a single experimental paradigm – the lexical decision task. In a typical use of this task, words are presented individually on a computer screen and participants as asked to decide, as quickly as possible, whether the presented word is an existing word of the language. The reasoning behind the use of the
Words, mind, and brain
lexical decision paradigm is that it provides a relatively pure measure of the ease of lexical access (the first contact with a stored lexical representation). The latency of a “yes” response in the task is thus interpreted as a measure of how easily the mental representation of the word is accessed. The latency and accuracy for nonwords is taken to reflect the extent to which they are word-like. The lexical decision paradigm can be modified by distorting the stimuli (e.g., by presenting part of a word in upper case letters and the other part in lower-case letters). Much more common, however, are manipulations in which the target stimulus is preceded by another prime stimulus. This prime can be presented for very brief durations, so that participants are not consciously aware of having seen it and, therefore, cannot develop conscious strategies to predict the target stimulus from the prime. This frees the experimenter to develop prime stimuli that can be related to the target along a variety of linguistic dimensions (e.g., phonologically, as in great → mate, semantically, as in friend → mate, morphologically, as in roommate → mate, or simple in terms of form, as in cremate → mate). A clear trajectory in psycholinguistic research in this domain is to ameliorate the basic single-word lexical decision paradigm along two dimensions. The first dimension involves relating the results of behavioural paradigms such as lexical decision to the neural correlates of language processing. Thus a number of studies have joined a lexical decision task to the measurement of event-related potentials (ERPs) so that the electrical activity of the brain, measured at various locations on the scalp, can be correlated to prime presentation, target presentation, and response. ERP measurements do not have very good spatial resolution, so they do not provide the best measures of which areas of the brain are most active at any particular time. However, they do have excellent temporal resolution, so it is possible to synchronize the ERP analysis and the stimulus presentation with accuracy in the millisecond domain. Similar approaches have been taken to relate behavioural measures of lexical processing to brain activities using fMRI and MEG. Of these, the MEG procedure, despite the fact that it is currently very costly to employ, seems to offer the most reliable measures of localized activity in the brain as well as the precise time at which this activity is seen. The second dimension in which recent research has sought to ameliorate standard paradigms such as lexical decision is in the development of paradigms in which the words are presented in more ecologically valid contexts. By this, we normally mean sentence context as opposed to single word presentation. For a long while, now, eye-tracking research such as that conducted by Pollatsek and colleagues (Pollatsek & Hyönä 2006) has very effectively probed lexical and morphological processing by having participants simply read sentences silently. In this research, there is no overt decision that the participant needs to perform.
Gary Libben
Rather the dependent variables in the experiments are properties of the eyemovements themselves (e.g., total reading time, eye fixations at particular locations in the sentence, backward eye movements to particular locations). It seems to me, that in the very near future, we will see substantial development along both these dimensions, so that the measurement of brain activity in psycholinguistic research will be as commonplace as the measurement of response latency is now. I would also expect that any relatively complete study of lexical processing and morphological effects will obligatorily contain some evidence of how such lexical items are processing in sentence (or preferably, discourse) context. Perhaps most, importantly, I would see as the next important development, the use of psycholinguistic research to simultaneously probe a participants’ processing using a variety of measures which include, for example, behavioural response latency, eye-movements, and ERP response. The reason why this is a key methodological advance is that is significantly cuts down the length of the inferential chain that needs to be employed in experimental paradigms such as simple lexical decision. It also moves us away from the conceptualization of lexical processing as a behavioural response to an external lexical stimulus. As I have claimed above, the important lexical stimuli are internal to the cognitive system. Under this view, the primary function of the psycholinguistic experiment is to set the context, triggers, and response criteria within which lexical processing can be investigated. This processing unfolds over time and it is quite conceivable that many important aspects of the process do not remain available in the final behavioural response, such as the lexical decision keypress. The advantage of simultaneously measuring eye-movements, ERP, and response latency is that we gain from the first two measures critical information on what is happening before the lexical decision response takes place and to what extent differences in response latency are related to data that can be obtained from earlier processing. I would expect that this emphasis on the process unfolding over time will also lead us to behavioural measures that are less punctual in nature. Eye-tracking, for example, naturally offers an ongoing record of processing (to the extent that processing is correlated with eye-movements). In our laboratory, we have been using the tracking of hand movements during writing to obtain similar measures of ongoing processing activity. One would also expect that recent technical explosions of mass storage capacity will allow us to maintain and analyze much more intricate raw ERP patterns that may reveal processes that have been thus far “hiding behind” the averaging procedure. I think that it is safe to say that the direction in which we are moving methodologically is toward analyses that concentrate less on the stimulus component of the experiment and less on the ultimate response. Instead, what we really want is a methodology that will uncover, as fully as possible, the activity of language by
Words, mind, and brain
particular individuals under particular triggering circumstances. If I were to hazard a guess at the next important advances in the understanding of language processing in general and lexical processing in particular, it would be in the understanding of individual variation in language processing and what that individual variation can tell us about the extent to which mental lexical organization is shaped by experience and constrained by properties of human cognition and brain operation. In the final section below, I offer some concluding thoughts on how individual experience could shape the mental organization of the types of words that have been the focus of our discussion. 5. Individual differences In Section 2 above, we considered the manner in which the processing of multimorphemic words necessarily involves complex associations to other cognitive representations and processes. We took as an example of this, the manner in which words such as Godcast are interpretable because of the semantically and morphologically related recent coinage podcast. We also considered the manner in which recognition of even relatively well entrenched words such as broadcast (a 20th century lexical extension of an agricultural term dating back to the 18th century) obligatorily involves the activation of morphological relatives such as telecast, but perhaps less so, an item such as typecast, which is less related semantically. Now if we consider why this might be the case, we would have to conclude that it is the experience of the language user that creates patterns of associations whose strength and shape are influenced by the interaction of form and meaning. Most speakers of English will have associations between broadcast and telecast, and we might expect that fewer will have strong associations between broadcast, forecast, and typecast. But these predictions can be at best approximate, for a great deal will depend on the nature of individual mental associations and, moreover, the depth and variety of meanings that an individual speaker will have associated with the root cast. One might imagine, the more differentiated and entrenched the root meaning is, the more differentiated will be the associations among the compound forms. In the absence of this differentiated knowledge, in this case the knowledge that the free morpheme cast most likely means “to throw”, the relations of all words with cast as the compound head, will be likely much more homogenous. This homogeneity will reflect the fact that what they all have equally in common (in the absence of any semantic differentiation) is the shared element -cast. For such individuals, these compounds would be like English words containing a variant of the Greek root ballein (which happens also to mean “to throw”) such as hyperbole, symbol, and metabolism.
Gary Libben
These conjectures, if on the right track, suggest that the nature of native speaker lexical ability might be much more variable than we might have suspected. Individual lexical experience varies greatly with exposure to particular semantic domains and education. Some native speakers of English know that the root bol in symbol means to cast. Others do not. Speakers who are proficient in languages other than English will have associations that differ from those of monolinguals in significant and measurable ways. Thus, once again, it is not the words that have the meanings, or the morphological structure, or the semantic associations. These are all properties of individual human minds which will share some lexical characteristics with other members of a speech community but not others. There is a great deal in this line of thinking that remains undeveloped and uncertain. Yet, recent proposals (e.g., Feldman 2006; Pulvermüller 2004) seem to offer very promising advances in proposing how we can think about words of the mind/ brain. However, a survey of the neurolinguistic and psycholinguistic literature in the domain of lexical and morphological processing will still reveal a pressing need for better metaphors for how words are products of the mind/brain. At present, we are still encumbered by metaphors in which words, as external entities, are acquired by human brains, stored in human brains, and retrieved from human brains. It is my hope that in this paper, I have been able to provide an alternative view of lexical processing – one which is centered in the psychological functioning of mind/brain. I hope I have also been able to point out that if we ignore the investigation of individual differences in this domain, we might likely miss an opportunity to understand the bounds of human lexical capacity and the biological and environmental factors influence the realization of that capacity. References Baayen, R. Harald. 2007. Storage and computation in the mental lexicon. In Gonia Jarema & Gary Libben (eds), The Mental Lexicon: Core Perspectives, 81–104. Oxford: Elsevier. Feldman, Jerome A. 2006. From Molecule to Metaphor. Cambridge, MA: MIT Press. Gagné, Christina L. 2001. Relation and lexical priming during the interpretation of noun-noun combinations. Journal of Experimental Psychology: Learning, Memory and Cognition 27: 236–254. Gagné, Christina L. 2002. Lexical and relational influences on the processing of novel compounds. Brain and Language 81: 723–735. Jong, Nivia de, Laurie B. Feldman, Robert Schreuder, Matthew Pastizzo & R. Harald Baayen. 2002. The Processing and Representation of Dutch and English Compounds: Peripheral Morphological and Central Orthographic Effects. Brain and Language 81: 555–567. Libben, Gary. 2007. Reading Complex Morphological Structures. In Sally Andrews (ed.), From inkmarks to ideas: Current issues in lexical processing, 192–215. Hove: Psychology Press. Libben, Gary, Lori Buchanan & Annette Colangelo. 2004. Morphology, Semantics, and the Mental Lexicon: The failure of deactivation hypothesis. Logos and Language 4: 45–53.
Words, mind, and brain
Moscoso del Prado Martin, Fermin, Raymond Bertram, Robert Schreuder & R. Harald Baayen 2004. Morphological family size in a morphologically rich language: The case of Finnish compared to Dutch and Hebrew. Journal of Experimental Psychology: Learning, Memory, and Cognition 30: 1271–1278. Pollatsek, Alexander & Jukka Hyönä. 2006. Processing of morphologically complex words in context: What can be learned from eye movements. In Sally Andrews (ed.), From inkmarks to ideas: Current issues in lexical processing, 275–298. Hove: Psychology Press. Pulvermüller, Friedemann. 2004. Lexical access as a brain mechanism. Behavioral and Brain Sciences 27: 297–298. Schreuder, Robert & R. Harald Baayen 1997. How complex simplex words can be. Journal of Memory and Language 37: 118–139.
Information structure Notional distinctions, ways of expression Caroline Féry
Universität Potsdam, Germany
Manfred Krifka
Humboldt-Universität zu Berlin, Germany
1. Introduction Information Structure, the packaging of information to satisfy the immediate communicative needs, exerts a powerful force on all structural levels of language. We show how this concept can be defined, we argue for focus, givenness, topic, frame setting and delimitation as important subconcepts, and we illustrate the wide variety in which these information structural functions are expressed in languages. 2. What is Information structure? The phenomena we subsume under the notion of information structure (IS, for short) have enjoyed the attention of linguists for a long time. They have been identified since the medieval Arab grammatical tradition by different linguistic schools in a number of ways. To mention the perhaps most influential one, the Prague School initiated by Mathesius has argued that the identification of given material (the theme) and the highlighting of new material (the rheme) exerts a powerful force on language structure. Today, the effects of IS are recognized in every theoretic framework that strives for a comprehensive view of linguistic structure, and they are investigated in a wide variety of distinct languages – witness the contributions to the Parallel Session on Information Structure at CIL 18. But what is IS? Following Chafe (1976), we understand it to refer to the packaging of information that meets the immediate communicative needs of the interlocutors, i.e., the techniques that optimize the form of the message with the goal that it be well understood by the addressee in the current attentional state.
Caroline Féry & Manfred Krifka
One such feature, for example, is the highlighting of constituents, called focus. In (1), a question creates a particular attentional state, which is recognized by the focus in the answer, expressed by pitch accent on tiger (cf. 1a). Pitch accent on road, as in (1b), would lead to an infelicitous answer, even though the truth conditions of (1a) and (1b) are the same, as it does not fit to the context question. (1) {What did you see on the road?} a. We saw a tiger on the road. b. #We saw a tiger on the road.
Chafe’s metaphor of packaging suggests that IS never affects truth conditions. However, one should be aware of the fact that the markings of IS can have truthconditional effects, for example with focus-sensitive particles like only. In (2a), the speaker may have seen other animals on the road, but the only place where a tiger was spotted was on the road. In (2b), no other animal was seen there. Depending on the placement of the pitch accent, this English sentence is true in different contexts. (2) a. We only saw a tiger on the road. b. We only saw a tiger on the road.
One and the same linguistic device, sentence accent, can be used for packaging as well as for constructing the truth-conditional content. There are two ways of dealing with this: One is to assume that the two uses of the same feature are essentially unrelated, just as the uses of accent in English to express focus and to distinguish words such as record and record. The other is to assume that the feature is to be interpreted in a particular way that makes sense for the purposes of information packaging and of building information content. The second alternative is more attractive, as we should not assume multiple meanings if possible. We will see that focus indeed can be interpreted in this way. We will first provide definitions of the notions of IS, and then examine some of the linguistic means used for the realization of IS. The grammatical devices for focusing, defocusing or topicalizing will turn out to be parts of a set of reflexes existing independently in the language under consideration. We wish to point the readers to Féry, Fanselow & Krifka (eds), (2006) for a more comprehensive exposition of some of the points discussed here. 3. The notions of information structure If we want to talk about communication as transfer of information and its optimization relative to the momentary needs of interlocutors, it is useful to adopt
Information structure: Notional distinctions, ways of expression
a model of information exchange rooted in the notion of Common Ground. The original notion of CG (cf. Stalnaker 1974) saw it as a way to model the information that is mutually known to be shared, which is continuously modified in the course of communication. This allowed for modeling the distinction between presuppositions, as requirements for the input CG, and assertions or the proffered content, as the proposed change in the output CG. This distinction is relevant for information packaging, as the CG changes continuously, and information has to be packaged corresponding to the CG at the point at which it is uttered. CG does not only consist of a set of propositions that is presumed to be mutually accepted, but also of a set of entities that have been introduced into the CG before, the discourse referents. They can be taken up by pronouns or by definite NPs, which express requirements to the input CG. The choice of anaphoric expression depends on the recency of the antecedent, again a notion that falls squarely within Chafe’s notion of packaging. The properties of CG mentioned so far have to do with the truth-conditional information in the CG, so we can subsume them under the heading of CG content. But any notion of CG that can be applied to the analysis of real communication must also contain information about the manifest communicative interests and goals of the participants. For example, questions typically do not add factual information to the common ground, but indicate informational needs on the side of one participant that should be satisfied by a conversational move of the other. We call this dimension of the common ground CG management, as it is concerned with the way the CG content is supposed to develop. Just as CG content, the tasks of CG management is shared by the interlocutors, with the understanding that the responsibility for it may be asymmetrically distributed among participants. Turning now to definitions of the IS categories that we consider crucial, we propose first a three-way distinction between focus, givenness and topic. A general definition of focus, making use of a central insight of Alternative Semantics (Rooth 1992), appears in (3). (3) Definition of focus: Focus indicates the presence of alternatives that are relevant for the interpretation of linguistic expressions.
One prominent use of focus is the identification of context questions in answers, as in (1). The idea is that the meaning of a question identifies a set of alternative propositions, the answer picks out one of these, and the focus within the answer signals the alternative propositions inherent in the question. The alternative denotations indicated by focus have to be comparable to the denotation of the expression in focus, i.e., they have to be of the same type, and often also of the same ontological sort (e.g., persons or times). They also can be
Caroline Féry & Manfred Krifka
more narrowly restricted by the context of utterance. The complement of focus is “background.” The second notion to be defined is givenness: (4) Definition of givenness: A feature X of an expression α is a Givenness feature iff X indicates whether the denotation of α is present in the CG or not, and/or indicates the degree to which it is present in the immediate CG.
Schwarzschild (1999) develops a refined theory of interaction between givenness and focus, which checks givenness recursively and states that constituents not in focus must be given, and that focus has to be applied only when necessary, i.e., to prevent that a constituent is given. But while focus is restricted in Schwarzschild’s theory, it cannot be eliminated totally. In fact focus/background and given/newness cannot be reduced to just one opposition, as these pairs of notions are only partially overlapping. For example, given expressions, like pronouns, can be focused. The notion of “topic” comes with a complementary part called “comment.” Reinhart (1982) integrates it into a theory of communication that makes use of the notion of CG. According to her, new information is not just added to the CG content in form of unstructured propositions, but is rather associated with entities, just like information in a file card system is associated with file cards that bear a particular heading. A definition based on this insight appears in (5). (5) Definition of topic: The topic constituent identifies the entity or set of entities under which the information expressed in the comment constituent should be stored in the CG content.
For example, while (6a,b) express the same proposition, they structure it differently insofar as (a) should be stored as information about Aristotle Onassis, whereas (b) should be stored as information about Jacqueline Kennedy. (6) a. [Aristotle Onassis] T [married Jacqueline Kennedy]C b. [Jacqueline Kennedy]T [married Aristotle Onassis]C
Theories of IS often introduce additional notions beyond focus, givenness and topic. One such notion is contrast, which is distinguished from pure information focus that shows up, e.g., in the answer to constituent questions. With contrast, the alternatives have to be given explicitly, and usually it is also assumed that only one of the contrasted alternatives is acceptable. For example, answers to alternative questions would qualify as having contrastive focus: (7) A: Do you want tea or coffee? B: I want tea.
Information structure: Notional distinctions, ways of expression
There is a plausible argument that we do not need contrastive focus as a separate basic notion, as we already have introduced givenness; hence contrastive focus can be defined as that subtype of focus in which the alternatives are given. The uniqueness assumption may follow if we assume that apparent non-uniqueness arises because alternatives can be combined (e.g., we saw a tiger and a baboon on the road), but that the explicit enumeration of alternatives that does not include a combination (e.g., or both in (7.A)) suggests that such combinations should be disregarded. Another important notion is contrastive topic, as in the following example: (8) A: What are your sisters playing? B: My younger sister plays the violin, and my older sister, the flute.
The phrase my younger sister is a contrastive topic. Rising accent indicates that at the current position of discourse, other topics could have been chosen (e.g., my older sister). Again, we do not need contrastive topic as a basic notion. If focus, in general, indicates the presence of alternatives, then focus within a topic would indicate the presence of alternative topics at the current point in discourse. In (8), the choice of my younger sister as topic indicates that there are other possible topics that can be described by my X sister; and indeed, the second clause, my older sister identifies such an alternative topic. Contrastive topics are to be expected whenever a question is too complex to be answered on the basis of one single topic; they indicate a particular answering strategy by introducing subtopics (cf. Büring 2003). There is a phenomenon that is somewhat related to topichood, which has been called frame setting (cf. Jacobs 2001). (9) A: How is John? B: {Healthwise/As for his health}, he is FINE.
This is a statement about John, but it is restricted to those aspects that concern John’s health (in contrast, e.g., to his financial sitation). We call phrases like healthwise “frame setters”. Clearly, focus plays a role with frame setters, just as with contrastive topics, as they express a certain restriction of the ensuing predication to some perspective that is not clearly identified by the context already – if the health perspective were already established, there would be no need to express it explicitly. Frame setters restrict context-sensitive expressions, like be fine, to the specified dimension, or delimit the predications that can be made. For example, As for his health, he had a serious flu recently is fine, but As for his financial situation, he has a serious flu recently is not. There is an obvious similarity between contrastive topics and frame setters that is reflected in the way how these information-structural functions are marked
Caroline Féry & Manfred Krifka
(e.g., by a B-accent in English, or by the postposition nun in Korean). Both express that the predication is restricted in some way – e.g., (8b) restricts the predication plays the violin to the younger sister (where the expected value is the sisters in general), and (9) restricts the predication to aspects concerning John’s health. It is useful to introduce a new term for this function: delimitation. It is a genuine phenomenon of IS, as it responds to the current informational need of the addressee: It is indicated that the issue at hand is broader, and that the ensuing speech act concerns only a part of this more general issue. Hence, delimitation can be defined as follows: (10) Definition of Delimitation A delimiter α in within an expression […α….] always comes with a focus within α indicating alternatives α′, α″ etc. It indicates that the current informational need is not totally satisfied by […α…] but would be satisfied by additional expressions [---α′---], [α″---] etc.
We do not claim that the notions of focus, givenness, topic, frame setting and delimitation exhaust what there is to say about IS. For example, in an argumentative discourse, the current informational need might dictate the selection and ordering of arguments to gain support for a particular conclusion. But such effects go beyond the limit of the sentence, and relate it to discourse structure. Here we will stay within the confines of the sentence (in a particular context), and we will try to illustrate some of the ways in which the IS notions specified above are expressed in languages.
4. The expression of information structure How do languages mark the various IS distinctions? While there is considerable variety in the strategies that we find in different languages, they always have a relationship to prosody: focus tends to be prosodically prominent, and givenness tends to be prosodically non-prominent, while topic tends to form a separate prosodic phrase, and is thus also prominent (the same holds for frame setters and delimiters in general). But this prosodic connection is achieved by different grammatical correlates in different languages, depending on the languages’ general properties. And languages differ in the obligatoriness of expressing IS distinctions; for example, it has been shown that in Northern Sotho (Bantu) and Hausa do not express the focus of answers as rigidly as English (cf. Zerbian 2006; Hartmann & Zimmermann 2007). In English, focus and topic correlate with pitch accents, and givenness is often expressed by deaccenting, see (1) and (2). But in a number of Asian
Information structure: Notional distinctions, ways of expression
and African languages, pitch accents only play a minor role, if at all, and morphological and syntactic means are prevalent. In tone languages, phrasing can replace the pitch accents of intonation languages, and particles can play the role of boundary tones. An extreme case of prosodic marking of IS is ellipsis, where only the focused part of a sentence is pronounced, and the given part is just deleted. Following Jackendoff (1972), we assume that IS roles are identified at the surface syntactic structure by features, in the way shown in (11) to (13). (11) a. We only saw a tiger [on the road]F. b. We only saw [a tiger]F on the road. (12) {What did you see on the road?} [We saw]G [a tiger]F [on the road]G. (13) [As for tigers,]T [we saw one on the road]F.
We examine syntactic, phonological and morphological reflexes of IS in the next subsection, and show in each case how they relate to prosody. 4.1 Sentence position First, IS roles are often associated with sentence positions. Halliday (1967–8) holds that the initial position is a necessary condition for a “theme” (a topic in our terminology). This preferred place for a topic is easily explained: since it is the address at which the information of the sentence is supposed to be stored, it makes sense to introduce it right at the beginning. However, topics are not necessarily located sentence-initially. In the following Korean sentence, the topic dezaato-wa “dessert” is placed after a quantifier phrase and is thus not initial.1 A subscript P shows a prosodic phrase (p-phrase), and a subscript I a larger intonation phrase (i-phrase). (14) ((Nwukwuna-ka)P ([dessert-nun]T)P (ice cream-ul mek-ess-ta)P)I everyone-nom dessert-top ice cream-acc eat-past-dec ‘As for dessert, everyone ate ice cream.’
Sentence-final topics, sometimes called “anti-topics” are also possible, as illustrated in (15) for Cantonese and (16) for French. (15) ((Go loupo)P (nei gin-gwo gaa)P, ([ni go namjan ge]T)P)I. clf wife 2.sg see-exp dsp this clf man dsp ‘The wife you have seen, of this man.’
. Thanks to Shin-Sook Kim for providing this sentence.
Caroline Féry & Manfred Krifka
(16) ((Pierre l’ a mangée)P, ([la pomme]T)P)I. Peter it-acc has eaten, the apple ‘Peter has eaten the apple.’
The common property of topics is their separation from the remainder of the sentence. They tend to form their own i-phrase (intonation phrase), which allows for a clear intonational separation. In languages like Japanese or Cantonese, particles not only signal the role of the constituent as a topic, but also add place for a boundary tone. This allows the topic in (16) to be inserted in a non-initial position. Focus has also been associated with special positions in certain languages. Hungarian has been described as a language which obligatorily places an exhaustive focus preverbally (É. Kiss 1998), and Italian as a language with clause-initial (Rizzi 1997) or clause-final (Samek-Lodovici 2006) foci. An alternative explanation, which accounts for the Hungarian facts without forcing an association between focus and preverbal position, assumes that Hungarian is phonologically a left-headed language, both for prosodic words and prosodic phrases. Focus wants to be prominent and the preferred stress position is at the beginning of the main i-phrase, directly after the i-phrase of the topic. The initial position is occupied by the narrow focus, as often as possible, and happens to be the verb in all other cases (see Szendrői 2003). But focus may also be located postverbally: In (17), both the VP and the dative object are focused and the accusative object is given, but the dative object is postverbal. In such cases, focus is indicated by pitch accent only. (17) ((Tegnap este)P)I ((bemutattam Pétert)P (marinak)P)I. yesterday evening prt-introduced-I Peter.acc Mary.dat ‘Yesterday evening, I introduced Peter to Mary.’
In Italian, given elements may be moved out of the matrix clause, and typically, it is this movement which causes finality of focus. In (18), adapted from SamekLodovici (2006), Parigi is the focus, and the following constituents are rightdislocated as they are given. Italian is a language with final stress, both at the level of the p-word, where it is trochaic, and at the level of the p-phrase, and syntactic reorganization helps prosody in moving narrow foci to the furthest possible rightward position. Thus, both in Hungarian and in Italian the peripheral position of focus is not a special feature of focus, but reflects the general preference for prominence. (18) ((L’ho incontrato [a parigi]F)P, (Luigi)P, (ieri)P)I. (I)him have-met in Paris, Luigi, yesterday ‘I met Luigi in Paris yesterday.’
4.2 Accents There have been numerous attempts in the literature to relate specific information roles to the form of pitch accents. Bolinger (1958) introduces a distinction between
Information structure: Notional distinctions, ways of expression
accent A, a falling accent, and accent B, a fall-rise accent, for English, and Jackendoff (1972) and Liberman & Pierrehumbert (1984) relates the former to focus and the latter to topic, as in (19). Manny has accent B, and Anna accent A. (20) {What about Manny? Who did he come with?} (([manny]T)P (came with [anna]F)P)I.
Büring (2003), for German, and Steedman (2000), for English, establish an obligatory relationship between contours and roles by having pitch accent contours participate in the definition of topics and foci. Attempts to relate forms of accents to specific IS roles are found for other languages as well. For instance, Frota (2000) claims that narrow foci in Portuguese are always associated with a certain kind of accent. An alternative explanation is possible which only indirectly relates IS to the forms of accents. The preference for associating specific contours with IS roles can be explained by general properties of the language. As far as topics are concerned, the preference for sentence-initiality is paired with a preference for rising tones. The rising tone is just a reflex of the non-finality of this accent. And the falling contour often found on focus may be related to the late position of a focus in a sentence. Do languages with pitch accents necessarily use them for topics and foci? The question bears on the necessity of accents (and of deaccenting) in general in relation to focus/topic/givenness. Jackendoff formulates a rule which directly relates a focus with an accent. “If a phrase P is chosen as the focus of a sentence S, the highest stress in S will be on the syllable of P that is assigned highest stress by the regular stress rules” (1972: 247). Nearly all models relating focus with phonology rely on a direct correspondence between semantics and phonetics and require an accent signaling the presence of a focused constituent (see for instance Cinque 1993; Rooth 1992; Selkirk 1995; Schwarzschild 1999; and many others). But in fact, there are examples in which the association between focus and accent seems to be cancelled. One such case is the so-called Second Occurrence Focus (cf. Partee 1999; Beaver et al. 2007; Féry & Ishihara 2005), which combines elements of association with focus and givenness. Only vegetables in (21b) is associated with the focus operator only, and is thus a focus, but it is also given, because it is repeated from (21a). (21) a. {Everyone already knew that Mary only eats [vegetables]F}. b. If even [Paul]F knew that Mary only eats [vegetables]SOF , then he should have suggested a different restaurant.
There are only weak correlates of accent, and no pitch excursions in the postnuclear position, although Féry & Ishihara (2005) show that a pitch accent is indeed present in the prenuclear position.
Caroline Féry & Manfred Krifka
Other cases of absence of accent on a focus are a consequence of avoidance of stress-clash and the consequent deaccenting. In (22a), herself is a so-called intensifier which is claimed to be obligatorily accented in the literature. But in the presence of an adjacent narrow focus, accent on herself disappears. The same is true of the association with focus adjacent to a parallel focus in (22b), cf. Rooth (1992). In (22c), the answer to the question is completely deaccented. Instead the additive particle also carries the stress. (22d), from Reis & Rosengren (1997), shows that a contrastive topic (Peter in Krifka’s 1999 analysis) can also be realized without excursion if another, more prominent topic (Gauguin) is adjacent. (22) a. Marie-Luise even grows rice herself. b. People who grow rice only eat rice. c. {John said that Mark is coming, but what did Sue say?} She also said that Mark is coming. d. {Boy, Paul possesses a Gauguin.} Einen gauguin besitzt Peter auch ‘Peter also owns a Gauguin’
We see that there is no strict association between focus and accent or topic and accent. Accent is a preferred option but it is not obligatory. It is only present if the phonological structure of the sentence allows it. To sum up, the preference for associating some specific contours with IS roles, or just pitch accents can be explained by general properties of the language.2 4.3 Morphological markers Morphological markers are compatible with the general claim of this section that the marking of focus and topic is always prosodically prominent if can be shown that the presence of a particle changes the prosody of the sentence it appears in. Examples confirming this claim appeared in (11) for Japanese and in (15) for Cantonese topic markers. A number of tone languages have been studied as for their focus realizations which do not seem to have other correlates of IS than optional presence of morphological markers. Examples involving morphological markers for focus appear in the Gur languages Buli in (23) and Ditammari in (24), both from Fiedler et al. (to appear). In Buli, the focus marker kà precedes the focused constituent. But when the focused túé is sentence-initial, the marker kà is not obligatory. As for Ditammari, the focus marker nyā follows the focused constituent.
. Many languages do not use pitch accents to highlight a focused element, but rather raise the pitch register of a focused phrase as a whole. This happens for instance in Mandarin Chinese (Xu 1999), in Korean, in Georgian and in Hindi.
Information structure: Notional distinctions, ways of expression
(23) Q: A:
What did the woman eat? ò ŋòb kà túé. 3.sg eat fm beans ‘She ate beans.’
(24) Q: A:
What did the woman eat? ò dī yātũrà nyā. 3.sg eat beans fm ‘She ate beans.’
These markers have a delimiting function in creating a prosodic boundary. We thus propose that the prosodic connection of the focus and topic markers is to be found in the phrasing properties of a constituent delimited by such a marker. Even if not enough is known about the exact behavior of particles, it seems to be a valid generalization that they always appear at the periphery of the constituent they mark. In languages without special markers for IS, that is in languages which do not have pitch accent and which have only optional morphological markers, the answer to wh-question typically involves ellipsis of the given material. Only the constituent in focus is realized, a strategy which we propose to analyze in prosodic terms. 5. To conclude This paper has proposed definitions of IS concepts in a model of information exchange that makes use of the notion of Common Ground. We also argued – in the limited place available – that there are universal principles in the way how IS roles are coded. One tendency, the tonal prominence of focused and the de-prominence of given expressions, can be explained by the effort code (Gussenhoven 2004): Higher or lower expenditure in prosodic explicitness reflects the importance of subconstituents for communication. Others, like the tendency of topics to be prosodically separated from and to precede the rest, can be traced back to the optimal flow of information transmission. In spite of such commonalities, we have pointed out drastic differences in the ways how they are realized in the grammars of individual languages.
References Beaver, David, Brady Zack Clark, Edward Flemming, T. Florian Jäger & Maria Wolters. 2007. When semantics meets phonetics: Acoustical studies of second occurrence focus. Language 83: 245–276. Bolinger, Dwight. 1958. A theory of pitch accent in English. Word 14: 109–149. Büring, Daniel. 2003. On D-trees, beans, and B-accents. Linguistics and Philosophy 26: 511–545.
Caroline Féry & Manfred Krifka Chafe, Wallace L. 1976. Givenness, contrastiveness, definiteness, subjects, topics and point of view. In Charles N. Li (ed.), Subject and Topic. 27–55. New York: Academic Press. Cinque, Guglielmo. 1993. A null theory of phrase and compound stress. Linguistic Inquiry 24: 239–297. Féry, Caroline, Gisbert Fanselow & Manfred Krifka (eds). 2006. The notions of Information Structure. Working Papers of the SFB632, Interdisciplinary Studies on Information Structure (ISIS) 6. Potsdam: Universitätsverlag Potsdam. http://www.sfb632.uni-potsdam.de/isg. html. Féry, Caroline & Shinichiro Ishihara. 2005. Interpreting second occurrence focus. Ms. University of Potsdam. Fiedler, Ines, Katharina Hartmann, Brigitte Reineke, Anne Schwarz & Malte Zimmermann. To appear. Subject focus in West African languages. In Malte Zimmermann & Caroline Féry (eds), Information Structure in Different Perspectives. Frota, Sónia. 2000. Prosody and Focus in European Portuguese. New York: Garland Publishing. Gussenhoven, Carlos. 2004. The phonology of tone and intonation, Cambridge University Press, Cambridge. Halliday, Michael A.K. 1967–68 Notes on transitivity and theme in English. Journal of Linguistics 3 & 4. Hartmann, Katharina & Malte Zimmermann. 2007. In place – out of place: Focus in Hausa. In Kerstin Schwabe & Susanne Winkler, On information structure, meaning and form, 365–403. Amsterdam: John Benjamins. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. Jacobs, Joachim. 2001. The Dimensions of Topic-Comment. Linguistics 39: 641–681. Kiss, É., Katalin. 1998, Identificational focus versus information focus, Language 74: 245–273. Krifka, Manfred. 1999. Additive particles under stress. Proceedings of SALT 8. Cornell, CLC Publications. 111–128. Liberman, Mark & Janet Pierrehumbert. 1984. Intonational invariance under changes in pitch range and length. In Mark Aronoff & Richard T. Oehrle (eds), Language Sound Structure, 157–233. Cambridge, MA: MIT Press. Partee, Barbara H. 1999. Focus, quantification, and semantics-pragmatics issues. In Peter Bosch & Rob van der Sandt (eds), Focus: Linguistic, Cognitive, and Computational Perspectives, 213–231. Cambridge: Cambridge University Press. Reinhart, Tanya. 1982. Pragmatics and linguistics. An analysis of sentence topics. Philosophica 27: 53–94. Reis, Marga & Inger Rosengren. 1997. A modular approach to the grammar of additive particles: The case of German auch. Journal of Semantics 14: 237–309. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of Grammar: Handbook in Generative Syntax, 281–337. Dordrecht: Kluwer Academic Publishers. Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1: 75–116. Samek-Lodovici, Vieri. 2006. When right dislocation meets the left-periphery: A unified analysis of Italian non-final focus. Lingua 116: 836–873. Schwarzschild, Roger. 1999. GIVENness, AvoidF and other constraints on the placement of accent. Natural Language Semantics 7: 141–177. Selkirk, Elisabeth O. 1995. Sentence prosody: Intonation, stress and phrasing. In John Goldsmith (ed.), Handbook of Phonological Theory, 550–569. Cambridge, MA: Blackwell.
Information structure: Notional distinctions, ways of expression
Stalnaker, Robert. 1974. Pragmatic presuppositions. in Milton K. Munitz & Peter K. Unger (eds), Semantics and Philosophy. 197–214. New York: New York University Press. Steedman, Mark. 2000. Information Structure and the Syntax-Phonology Interface. Linguistic Inquiry 31: 649–689. Szendrői, Kriszta. 2003. A stress-based approach to the syntax of Hungarian focus. The Linguistic Review 20 (1): 37–78. Xu, Yi. 1999. Effects of tone and focus on the formation and alignment of F0 contours. Journal of Phonetics 27: 55–105. Zerbian, Sabine. 2006. Expression of Information Structure in the Bantu Language Northern Sotho, Doctoral diss., Humboldt University, Berlin.
Language policy The first half-century Bernard Spolsky
Bar-Ilan University, Israel
1. Birth of the field The academic study of language policy is not yet fifty years old if we take Haugen (1959) as the first major paper, but it has already matured into a vibrant and growing sub-field of sociolinguistics with established journals, international meetings, university courses, and textbooks designed for their students. The field of sociolinguistics originated as a blend of problem-oriented and theory-driven research, according to the recollections of those who participated in its development (Paulston & Tucker 1997). Grimshaw (1997) recalls that some early scholars were driven to an interest in the sociology of language by “ameliorative concerns about unfairness in the delivery of health services, provision of educational opportunity, or in participation in legal institutions.” While some attempted to build theoretical models, no basic theory emerged, and one of the founders, Ferguson (1997) remained pessimistic about the possibility. This was in part because of the number of disciplinary fields – anthropology especially ethnography, descriptive linguistics, dialectology, political science, sociology, and social psychology – which have all made their contributions to the study of language in its social context. It was the problem-oriented approach to a number of language-related issues that encouraged the growth of studies of language policy and language planning. 2. Language planning for linguistic utopia In the optimistic mood that followed the end of the Second World War, planning was seen as the method of solving social, economic, and even political problems. It seemed appropriate to take a similar approach to language-related issues. The collapse of British, French, Belgian, Dutch, and Portuguese empires was starting to produce newly independent nations faced with complex linguistic problems that resulted from the lack of fit between political and linguistic boundaries. India with its 387 living languages (according to Grimes 2000) and Africa with its many new multilingual nations posed conspicuous challenges. One approach was to look back
Bernard Spolsky
to how similar problems had been tackled in the late 19th century during the period of creation of new nations. The classic pioneering work on language planning investigated the way that newly independent Norway set out to create a language distinct from the Danish that had previously been dominant: studies by Haugen (1959, 1961, 1966) described the political compromise that produced two standard varieties of Norwegian and the requirement that all school pupils learn both. In India, a series of Rockefeller foundation summer schools taught by American linguists starting in 1954 had laid the basis for a renewed interest in linguistics which came to focus on language planning with the establishment in 1969 of the Central Institute of Indian Languages. In the United States, the establishment in 1959 of the Centre for Applied Linguistics, supported by the Ford Foundation, grew out of the realization of “the potential contributions of (Applied) linguistics to education and national development (Fox 2007).” The Ford Foundation later funded a pioneering survey of language use and language teaching in East Africa (Ford Foundation 1975), a model unfortunately not emulated as witness the current shortage of census and survey data on language use in most of the world. The wide range of nations involved in these early discussions and the breadth of academic interest were admirably represented in the collected papers from the 1966 Airlie House conference (Fishman, Ferguson & Das Gupta 1968). Individual papers dealt with sub-Saharan Africa (specifically, Sudan, Nigeria, South Africa, West Africa, the Congo, East Africa, and Nigeria) and North Africa, India, Israel, Scandinavia, Haiti, Papua-New Guinea, Peru, Paraguay, and the West Indies, concentrating on language status in developing multilingual nations but including concern for language education policy. Shortly after, a group of American sociolinguists met to plan an international study of language planning processes: their deliberations were published in Rubin & Jernudd (1971) although the results of their study did not appear until Rubin et al. (1977); again, this pioneering initiative was not repeated, and there are few if any studies of the effectiveness of corpus planning. In the meantime, the extraordinary productivity of the new students of language policy warranted another collection (Fishman 1974) which added several important theoretical studies and a number of additional countries and regions: Turkey, Australia, Arab Countries, Latin America, Indonesia and Malaysia, Philippines, French Polynesia, China, Canada and Ireland. At the same time, in the United States, the developing civil rights initiative started to produce activity and research into language education policy (Fishman 1971b; Fishman & Lovas 1970, 1972; Mackey 1970). Originally conceived as a more effective method of teaching English (Bilingual education 1967; Bilingual Education Programs 1967), bilingual education was seized on as a method of empowering Spanish speakers and other linguistic minorities, and benefited from US government support for 34 years (Crawford 2002). Cooper (1989) recognized language
Language policy: The first half-century
education as one of the three main activities of language management, alongside status and corpus planning proposed by Kloss (1969). Given the wide gap in many parts of the world between home and school languages (Spolsky 1974), it has grown into a major field in its own right. 3. The view from Prague While other countries and nationals were increasingly involved, much early work in language planning was US inspired or conducted. One important exception was the interest of Prague School scholars in sociolinguistic issues, particularly the studies of Vilém Mathesius and Bohuslav Havránek of language cultivation, essentially the definition and demarcation of Standard Czech. Concerned specifically with corpus planning, the Prague School linguists established principles for the standardization of languages with a strong literary tradition (Neustupný 1968). Prague-trained but by 1966 living in Australia, Neustupný (Neustupný 1997) presented at the Airlie House conference some of these notions calling for the development of a general theory of language problems and language policy. The Prague School position was set out in two papers given at the 1973 Georgetown Round Table (Garvin 1973; Prague School 1973). A decade later, Jernudd & Neustupný (1987) published the outline of a theoretical model of language policy and management. 4. From planning to policy The last thirty years of the 20th century produced a large number of detailed studies of specific cases of language planning and of various aspects of language policy. Much of this is associated with the scholarly research and editorial energy of Joshua A. Fishman. Starting in 1973, Mouton published under his editorship the International Journal of the Sociology of Language and a parallel book series Contributions to the Sociology of Language. There have so far been 190 issues of the journal, most concentrating on the sociolinguistics of a specific country, region, or language, and a significant proportion describing and evaluating language policy, and 94 monographs in the Contributions series. Fishman himself has tackled in authored or edited volumes a wide range of the most important topics: the structure of a bilingual community (Fishman, Cooper & Ma 1971), nationalism (Fishman 1973), bilingual education (Fishman 1976), the development of writing systems (Fishman 1977), the spread and globalization of English (Fishman, Cooper & Conrad 1977), language and ethnicity (Fishman 1989), reversing language shift
Bernard Spolsky
(Fishman 1991), the first language congresses (Fishman 1993), post-imperial English (Fishman, Rubal-Lopez & Conrad 1996), ethnic identity (Fishman 1999), endangered languages (Fishman 2001), and language purism (Fishman 2006). Other important research was published in two sociolinguistics journals: Language in Society and Journal of Multilingual and Multicultural Development. Three journals are now devoted specifically to study of language planning and language policy: Language Problems and Language Planning which began in the 1980s, Current Issues in Language Planning (since 2000) and Language Policy (since 2002). As a result, there are now scholarly studies of language policy for virtually every nation. The majority focus on policies developed by national governments. Not unnaturally, they deal with locally salient issues – the policies adopted by newly independent African states and their general failure to replace colonial languages with indigenous ones, the development of autonomy in post-Franco Spain and the consequent efforts to re-establish Catalan and Basque, the linguistic consequences of the movement for Quebec’s independence, the major language policy changes that followed the breakup of the Soviet Union and the similar breakup of Yugoslavia, the various efforts to develop a language policy for the European Union, the struggle in supranational organizations to add linguistic to human rights, the late recognition of linguistic minorities in Latin America, and the consequences of territorial solutions and autonomy in Belgium and the United Kingdom. Continuing a research project that began with the study of the linguistic consequences of colonialism (Calvet 1974), Calvet (1987) (published a decade later in English as Calvet 1998) saw multilingual societies as the source of conflict and “language wars,” arguing that unless interfered with, linguistic communication in these situations was self regulating, often resulting in the development of lingua francas. National governments do however commonly interfere, attempting to impose national languages that will supersede regional varieties. In one of the first books attempting to relate language planning to other aspects of social planning, Cooper (1989) starts with four defining examples, the foundation of the Académie française, the promotion of Hebrew in Palestine, the US feminist-inspired campaign against linguistic chauvinism, and the Ethiopian mass literacy campaign. Each of these, he shows, had nonlinguistic goals – the establishment of centralized political power, the building of a new nation, women’s rights, and the recruitment of student power to the new revolutionary regime. Against this background, he proposes and discards a number of possible definitions for language planning, offering finally his own: “Language planning refers to the deliberate efforts to influence the behavior of others with respect to the acquisition, structural, or functional allocation of their language codes (page 45, italics in original).” In this definition, Cooper adds to the original two activities of corpus and status planning proposed by Kloss (1966) a third, acquisition planning, thus including the
Language policy: The first half-century
whole field of language education within language policy. He concludes without a theory of language planning not just because of its complexity but because it serves so many different goals including “economic modernization, national integration, national liberation, imperial hegemony, racial, sexual, and economic equality, the maintenance of elites, and their replacement by new elites” (1989: 182). 5. Language policy as linguistic imperialism Although it was not written as a monograph on language policy, Phillipson’s doctoral dissertation (Phillipson 1990) quickly published as Phillipson (1992) played a pioneering role in the investigation of colonial and postcolonial language policy. There have been disagreements with its methods and applicability (Fishman et al. 1996), but Phillipson showed the value of analyzing documentary evidence of language planning, illustrating Cooper’s insistence on the need to identify the agent of planning, and raised fears about the power of governments and others to control society through language. The major weakness in his approach and that of a number of other scholars who fret over the relation between political power and language policy, such as Shohamy (2006), is that they assume much more focused and successful effort on the part of bureaucrats and politicians than the evidence warrants, resulting perhaps from the gap between studies of language policy and political science. With some exceptions (France and Québec are obvious cases), there is commonly inconsistency, disagreement, confusion, and lack of implementation rather than the kind of planned control and management that they assume. The evidence they submit for their interpretations are selected from the many published and unpublished reports, recommendations, opinion pieces, and proposals that virtually litter the field, making it quite easy to find support for differing arguments. How complex the reality can be is well illustrated by Evans (2002), which traced the debate in colonial India over language education, showing the existence of two major conflicting schools of thought, those who supported the French and Portuguese belief that all education must be in the Western metropolitan language, and the Orientalists who argued for initial education at least in the vernacular. The gap between rhetoric and practice becomes clearer in the work of Kaplan and Baldauf, who regularly draw attention to what they call “policy without planning,” which perhaps might be described as ineffective management. Kaplan (1994) sets out fundamental issues in language planning. Kaplan & Baldauf (1997) define language planning as “a body of ideas, laws and regulations (language policy), change rules, beliefs, and practices intended to achieve a plan to change (or to stop change from happening) in the language use in one or more communities.” Their aim in writing the book was to urge that the “underlying paradigm” needs to be
Bernard Spolsky
significantly revised (1997: 320). Their main emphasis is on macro level planning, by which they mean developments at the level of the national government, but they also insist on the need for planning at the micro level. They see status and corpus planning as closely interrelated and argue that language-in-education planning needs to be integrated with national planning. They reject the “one-language/ one-nation myth” arguing that every language fits in to an eco-system; language planning, they conclude is highly complex and therefore requires “the attention of a wide range of academic specialists as well as of the communities of speakers of all the languages and varieties involved (1997: 321).” Their position still reflects the optimism of the utopian post-Second World War economic planners who believed that social planning was possible, given the right experts and cooperative societies. Their book is a thorough review of earlier work in the field of language policy, looking at a very broad range of macro and micro policies in various countries at various levels: the relevant issues are carefully described and analyzed in detail and critically. They conclude with a very brief summary of national contexts; one of the principal tasks undertaken in the journal that they edit, Current Issues in Language Planning has been to commission long articles providing detailed accounts of national language planning in many countries. 6. But perhaps it is cultural? Schiffman (1996) takes a different approach, considering language policy as centrally located not in a political but a linguistic cultural context. Any language policy is bound by “linguistic culture”, defined as “the set of behaviours, assumptions, cultural forms, prejudices, folk belief systems, attitudes, stereotypes, ways of thinking about language, and religio-historical circumstances associated with a particular language (1996: 5).” He denies cultural determinism, but points out that culture is transmitted largely through language. He is doubtful about the existence of a multilingual citizenry, but rather assumes extended diglossia as a characteristic of the linguistic group (1996: 13). Language policy, he believes, is dichotomized into overt and covert aspects, so that the absence of an official language policy in the United States does not mean that there is not an actual implicit policy (1996: 15). In Switzerland, the declaration in the Federal Constitution of four national and three official languages disguises the reality of the educational policy that is actually established at the level of the Canton (1996: 18). He sets out a complex typology of multilingualism and of language policy recognizing the distribution of linguistic repertoires across linguistic registers. Official language policies are unlikely to recognize this ecological complexity. Breaking out of the contemporary Western secular scholarly mode, Schiffman insists on the fact that in cultures where sacred texts are revered, language and religion become virtually identical.
Language policy: The first half-century
The bulk of his book analyses five specific linguistic cultures. In France, language and linguistic culture is a core value; he traces in detail the development over the centuries of the hegemony of the French language. India, he briefly argues, made the post-Independence mistake of assuming that they could apply a Soviet model to language policy rather than accepting that Indian linguistic culture is diverse and multilingual. He analyses the fanatical quality of Tamil devotion to their language. In his discussion of United States policy, he traces the development of an American linguistic culture which did not accept language rights as central and in which pressure for multiculturalism is still firmly resisted. This he illustrates in more detail by showing that California has from its earliest days worked towards an English-only policy. All of this suggests to him that a nation’s linguistic culture dominates any effort to apply “many of the common ideals and courtesies supposedly cherished by Americans of goodwill (1996: 275).” 7. Models of language policy In an early article, Spolsky (1977) proposed a model of language education policy and argued that it needed to be looked at in a wider linguistic, sociological, economic, political, psychological, cultural, and religious context. Twenty years later, Spolsky & Shohamy (1999) elaborated this model in an investigation of the sociolinguistic ecology of Israel and the formulation of an appropriate language education policy. The model distinguishes between language policy, ideology, and practices. By practices was meant the implicit policy (as Schiffman put it), the unplanned changes that Kaplan and Bauldauf had noted, the full ethnography of speaking that Hymes (1974) had proposed, the ecology of language that Voegelin, Voegelin, & Schutz (1967) and Haugen (1972) had first named. By ideology was meant not just the views of language that Silverstein (1979) investigated, but also the beliefs about language and bilingualism and the values assigned to specific language varieties. Policy meant what Kaplan and Bauldauf called planning, and what Spolsky (2004) labeled language management. In another important recent book, Ricento (2006b) collected a number of essays on various aspects of language policy theory, methodology, and issues. In his overview and in the introduction which follows, Ricento (2006a) starts by noticing the difficulty of defining “language” in language policy, because of the flexibility of the concept and argues that an adequate theory must deal with ideology, ecology, and agency. Language policy debates deal with more than language matters. The field is multidisciplinary and interdisciplinary. He notes the emergence of a critical approach which sees language shift as the “manifestation of asymmetrical power relations (2006a: 15).” He mentioned also that language policies exist “in all societal domains (2006a: 19).”
Bernard Spolsky
8. Language politics A growing concern for language rights finally attracted a number of political theorists to the field of language policy. Kymlicka & Patten (2003b) present a baker’s dozen of articles originally presented at a conference in Ontario in 2001. In the introduction, Kymlicka & Patten (2003a) claim that this is the first attempt to look at the issue from the point of view of “normative political theory (2003a: 1).” The new interest follows, they say, from a number of political contexts in which language conflict has been appearing: Eastern Europe after the break-up of the Soviet empire, various Western concerns for regional minorities (such as Québec and Spain), Western problems of immigrant absorption, the language conflicts associated with the development of the European Union, and awareness of the threat to endangered indigenous languages. Internally, political theorists had already been concerned with multicultural models of citizenship, leading to the development of theories that would allow for liberal democracy. In addition, they are becoming aware of the need for the study of democracy to take into account not just voting but also deliberation and opinionformation, something in which language is obviously important, including the inclusion of linguistic minorities. Seeing the selection of a single language as exclusionary, linguistic diversity becomes important. Kymlicka & Patten (2003a) outline some of the main issues: the problem of language policy for internal communication in governments and businesses, language policy for public services, language policy for courts and legislatures, educational language policy, private language use, immigration and naturalization, and the language of official declarations. They see language rights as organized on four dimensions: tolerance versus promotion, norm-and-accommodation versus official-languages, personal versus territorial, and individual versus collective. The papers in the volume present a number of approaches to these questions. One approach is benign neglect, but this ignores that public institutions must make decisions. A second is the attempt to establish “linguistic human rights” but this does not recognize that human rights declarations have little to say about language and that nowhere are all languages equal. What is needed, they conclude is a method of evaluating various possible policies. Much language policy is driven by a convergence nation-building approach, which of course leaves the problem of minorities. A second approach puts main emphasis on the preservation of language diversity, but this can lead to interfering with individual language choice. They summarize some suggested alternatives, including that public institutions should deal with individual-identity claims even-handedly. Continuing this concern for contextualizing language policy, Wright (2004) stresses its political, social, and ethical dimensions. She deals with language learning controlled by economic or political factors, and also with three language choices selected for social mobility, economic advance, or identity. The first part
Language policy: The first half-century
concentrates on the role of standard languages, including detailed studies of Europe and of postcolonial developments. The second part deals with linguistic accommodation as a result of contact, including the development of lingua francas and of English as a global language. The third discusses language rights and the way that policy affects minority groups and small nations. 9. Language policy and management Spolsky (2004) set out a more elaborated model of language policy. He distinguished between practices, beliefs, and management. Language policy can apply not just to named language varieties, but also to linguistic items. This attempt to correct the confusion caused by dealing only with named languages is supported by a number of scholars, among whom Blommaert (2001), Calvet (1999), and Shohamy (2006) are particularly important. Named languages are the creation of ideology and management; in practice, individual variants rather than varieties turn out to be salient in the plurilingual code-switching reality of the modern multilingual ecology. Language management operates within a speech community, whatever its size. It is best studied within specified sociolinguistic domains (Fishman 1972), ranging from the family to the nation-state. In practice, there are regularly conflicts as one level attempts to influence or ignore another: for example, schools regularly try to influence home language policy, nations commonly ignore religious policies, and supranational organizations establish policy not only for their own internal working but also try to impose policies on nations. With the exception perhaps of Iceland, most nations in the world are multilingual, but for ideological or pragmatic reasons, the majority of them proclaim monolingual policies. Only when there is clear economic motivation (such as the global pressure for English proficiency) or when there are more than one accepted Great Tradition (Fishman 1971a) (such as Belgium) or under internal or external pressure (as in Canada or New Zealand) are nations likely to include in their constitution or laws some recognition of minority language groups. In these cases, one popular solution is territorial, granting some variety of autonomy to regions or communities where numbers warrant (as in Canada or Belgium or India). Four main sets of conditions co-occur with various kinds of language policy. The first is the sociolinguistic situation or ecology, or rather the perception of it:1 the
. Smalley (1994) showed the gap between the perception of Thailand as a monolingual nation and the actual complexity of varieties that make up its sociolinguistic ecology.
Bernard Spolsky
number and kinds of languages, the number and kinds of speakers of each, the usefulness of each language inside and outside the community. The second is the importance of national or ethnic or other identity within the community. In nation states, the national language takes pride of place: schools teach in it, citizens are expected to be able to use it. A counterforce may be values associated with ethnic or religious subgroups wishing to have their language recognized alongside the national language. Once in power, a former minority regularly asserts the primacy of its own language. The third set of conditions has to do with globalization, and the consequent tidal wave of English that is sweeping into almost every national sociolinguistic repertoire. A fourth is the slowly growing acknowledgement that language choice is a component of human and civil rights, expressed in international covenants and charters. Associated with this is increasing recognition of the problems faced by minorities and immigrants because of their limited proficiency in the national official language. In a series of papers and a monograph, Spolsky (2006b, 2006c, 2007a, 2007c, forthcoming; in press) develops further the notion of language management, agreeing with Nekvapil (2006) to keep the term language planning for the utopian efforts of the 1960s to develop workable policies for newly independent states. He agrees with Calvet (1987) that it is useful to start at the family level, or even, as Jernudd & Neustupný (1987) suggest, with the accommodation they recognize as the simple language management of an individual adjusting to the needs of successful communication. He argues that language management can usefully be studied not just in nations, but in various speech communities at other levels of society or in the linguistic domains that Fishman (1972) showed were essential for the complete description of sociolinguistic ecology. In various domains – family, religion, workplace, neighborhood, schools, health and legal services, the military, local and national government – some participants believe it appropriate to manage the language practices and beliefs of others, producing a complex blend of conflicting currents or occasional a consensus on a hegemonic view. Beyond this, supranational organizations, apart from dealing with their own need to handle complex linguistic demands of their common multilingualism, have shown an increasing tendency to suggest broader principles translated into charters of human and civil rights intended to cater to the needs of language minorities. 10. Salvaging and regenerating endangered languages One focus of this concern for minorities is the growing call for correcting the social, political, educational and linguistic harm inflicted on marginalized minorities, especially tiny indigenous groups not actually wiped out during previous
Language policy: The first half-century
conquests. Thus, a special branch of language policy deals with indigenous peoples (however defined) and with their endangered languages; some typical studies are Francis & Reyhner (2002), Hornberger (1997), King (1997), McCarty (2003), and Spolsky (2007b). Fishman (1991, 2001) has defined a sub-field of language policy which he has labeled “reversing language shift.” It covers a wide range of cases, from the efforts of national governments to guarantee the continued use of their national language, to the attempt to teach a dying or even dead language to members of the heritage community. It occurs in an assortment of situations, ranging from languages where the only speakers are a handful of elderly people who possibly no longer use it, to a situation where a language is widely spoken and even officially used but its supporters are afraid that it is being threatened by a world language (generally English) or by new immigrants (as the supporters of English Only in the US feel they are being threatened by Spanish). At the upper end of the continuum, the actions taken are usually attempts to pass laws; at the lower level, there is first a major task of basic descriptive linguistic work to reconstruct the language from limited sources, followed by a combination of language teaching and language cultivation. 11. Language education policy While the scholarly work in the 1960s tended to concentrate on decisions concerning language status, it turned out that the most important decisions were probably those concerning schools and education. Following Cooper’s addition of language acquisition policy to status and corpus planning, it is now realized that decisions on language or languages of instruction and on the teaching of additional languages are crucial. Much schooling began as part of religious education, so that it was normal practice to teach the sacred language (whether Latin or Greek or Hebrew or Arabic or Sanskrit) to children. With secularization, Western education continued for some centuries to consider its main role to be teaching classical Latin and Greek. Teaching in the vernacular carried on similar principles in teaching what was considered the purest form of a standard language. During the 19th century, schooling regularly concentrated on the standard variety; an educated person was one who no longer used a dialect. Colonial empires, to the extent that they provided any education for their subjects, followed one of two main strategies – the French and the Portuguese in Africa and elsewhere and the Spanish in America took it for granted that all teaching of a local elite must be in the metropolitan language. The British, after severe controversy in India, and the Belgians allowed for an initial period of a few years
Bernard Spolsky
of primary education in the vernacular before switching to the metropolitan language for the minority permitted to continue. Protestant missionaries commonly selected one local dialect of the local languages to serve for Bible translation and schooling; commonly, this dialect later became standard (as in Fiji). The language planners of the 1960s tended towards the British approach, accepting a notion of initial literacy teaching in the mother tongue, however defined. The argument over language of instruction remains a live one in contemporary language policy. Walter (2003, 2008) summarizes the evidence from empirical research showing the advantage of using the mother tongue, but in much of the world actual practice favors an international or national official language (Shin & Kubota 2008; Cha & Ham 2008). The alternative of using the vernacular (Reaser & Adger 2008) or developing varieties of dual medium instruction (Baker 2001; Reyes & Moll 2008), while finding considerable scholarly support, tends to occur only when there is a general acceptance of pluralism and the importance of minority rights. The choice of additional languages also tends to be political or economic. In some Romance-language nations, a second Romance language is preferred but in much of the world English is the first foreign language. In officially bilingual countries, the other language is usually the first additional language: French or German in Switzerland but Stotz (2006) has shown that it is not always successful and that some cantons now prefer English; Dutch or French in Belgium but with many problems (Deprez 2000); French or English in Canada but unevenly (Landry & Forgues 2007); and in Israel, English rather than official Arabic (Amara 2002; Saban & Amara 2002). 12. Conclusion Language policy is now firmly established as a subfield of sociolinguistics. It has not yet developed the firm relationship with political science that it needs but there are important signs of the beginning of cross-fertilization (Williams 2007; Wright 2004). Understandably, many scholars working with endangered languages become active in their support, lessening their objectivity but demonstrating their strong sense of social responsibility. Many others, convinced by the arguments for minority rights that they start by studying, tend to become advocates rather than analysts of specific language policies. The difficulties of actually managing languages (MacNamara 1971; Ó Murchú 1977; Spolsky 2006a, 2006d) lead from time to time to skepticism as to whether it is possible, and the complexity has delayed the development of any consensus about a theory. But the growing interest in the topic gives promise of interesting new developments.
Language policy: The first half-century
References Amara, Muhammad. 2002. The place of Arabic in Israel. International Journal of the Sociology of Language 158: 53–68. Baker, Colin. 2001. Foundations of bilingual education and bilingualism (3rd edn). Clevedon England & Buffalo USA: Multilingual Matters Ltd. Bilingual Education, United States Senate, 90th Congress, First Sess. 681 1967. Bilingual Education Programs, House of Representatives, 90th Congress, First Sess. 584 1967. Blommaert, Jan. 2001. The Asmara Declaration as a sociolinguistic problem: Reflections on scholarship and linguistic rights. Journal of Sociolinguistics 5(1): 131-142. Calvet, Louis-Jean. 1974. Linguistique et colonialisme: petit traité de glottophagie. Paris: Payot. Calvet, Louis-Jean. 1987. La guerre des langues: et les politiques linguistiques. Paris: Payot. Calvet, Louis-Jean. 1998. Language Wars and Linguistic Politics (Michel Petheram, Trans.). Oxford: Oxford University Press. Calvet, Louis-Jean. 1999. Pour une écologie des langues du monde. Paris: Plon. Cha, Yun-Kyung & Seung-Hwan Ham. 2008. The impact of English on the school curriculum. In Bernard Spolsky & Francis M. Hult (eds), Handbook of educational linguistics, 313–327. Malden MA and Oxford England: Blackwell Publishing. Cooper, Robert L. 1989. Language planning and social change. Cambridge: Cambridge University Press. Crawford, James. 2002. Obituary: The Bilingual Education Act: 1968–2002. http://ourworld. compuserve.com/homepages/JWCrawford/T7obit.htm (17 January 2008). Deprez, Kas. 2000. Belgium: from a unitary to a federalist state. In Kas Deprez & Theo Du Plessis (eds), Multilingualism and government: Belgium, Luxembourg, Switzerland, Former Yugoslavia, and South Africa, 17–29. Pretoria, South Africa: Van Schaik Publishers. Evans, Stephen. 2002. Macaulay’s Minute revisited: Colonial language policy in nineteenth-century India. Journal of Multilingual and Multicultural Development 23(4): 260–281. Ferguson, Charles A. 1997. History of sociolinguistics. In Christina Bratt Paulston & G. Richard Tucker (eds), The early days of sociolinguistics: memories and reflections, 77–95. Dallas, TX: The Summer Institute of Linguistics. Fishman, Joshua A. 1971a. National languages and languages of wider communication. In W.H. Whitely (ed.), Language use and social change, 25–56. London: Oxford University Press for the International African Institute. Fishman, Joshua A. 1971b. The politics of bilingual education. In James E. Alatis (ed.), Report of the Twenty-first Annual Georgetown Round Table Meeting of Languages and Linguistics, 47–58. Washington, DC: Georgetown University Press. Fishman, Joshua A. 1972. Domains and the relationship between micro- and macrosociolinguistics. In John J. Gumperz & Dell Hymes (eds), Directions in sociolinguistics, 435–453. New York: Holt˛Rinehart and Winston. Fishman, Joshua A. 1973. Language and nationalism: two integrative essays. Rowley, MA: Newbury House Publishers. Fishman, Joshua A. 1976. Bilingual education: An international sociological perspective. Rowley, MA: Newbury House Publishers. Fishman, Joshua A. 1989. Language and ethnicity in minority sociolinguistic perspective. Clevedon, Avon: Multilingual Matters Ltd.
Bernard Spolsky Fishman, Joshua A. 1991. Reversing language shift: theoretical and empirical foundations of assistance to threatened languages. Clevedon, England: Multilingual Matters Ltd. Fishman, Joshua A. 2006. Do not leave your language alone: the hidden status agendas within corpus planning in language policy. Mahwah NJ: Lawrence Erlbaum Associates Publishers. Fishman, Joshua A. (ed.). 1974. Advances in Language Planning. The Hague: Mouton. Fishman, Joshua A. (ed.). 1977. Advances in the creation and revision of writing systems. The Hague & Paris: Mouton. Fishman, Joshua A. (ed.). 1993. The Earliest stage of Language Planning: The “First Congress” Phenomenon. Berlin: Mouton de Gruyter. Fishman, Joshua A. (ed.). 1999. Handbook of language and ethnic identity. New York and Oxford: Oxford University Press. Fishman, Joshua A. (ed.). 2001. Can threatened languages be saved? Reversing language shift, Revisited: A 21st century perspective. Clevedon, Avon: Multilingual Matters Ltd. Fishman, Joshua A., Robert L. Cooper & A.W. Conrad. 1977. The spread of English: the sociology of English as an additional language. Rowley, Mass.: Newbury House. Fishman, Joshua A., Robert L. Cooper & Roxana Ma. 1971. Bilingualism in the barrio. Bloomington, IN: Research Center for the Language Sciences, Indiana University. Fishman, Joshua A., Charles A. Ferguson & Jyotirindra Das Gupta. 1968. Language Problems of Developing Nations. New York: Wiley. Fishman, Joshua A. & John Lovas. 1970. Bilingual education in a sociolinguistic perspective. TESOL Quarterly 4: 215–222. Fishman, Joshua A. & John Lovas. 1972. Bilingual education in a sociolinguistic perspective. In Bernard Spolsky (ed.), The language education of minority children: selected readings, 83–93. Rowley MA: Newbury House Publishers. Fishman, Joshua A., Alma Rubal-Lopez, & Andrew W. Conrad (eds). 1996. Post-Imperial English. Berlin: Mouton de Gruyter. Ford Foundation. 1975. Language and development: a retrospective survey of Ford Foundation language projects, 1952–1974. New York: Ford Foundation. Fox, Melvin J. 2007. Ford Foundation: Personal reflection. In Christina Bratt Paulston & G. Richard Tucker (eds), The early days of sociolinguistics: memories and reflections, 271–272. Dallas, TX: The Summer Institute of Linguistics. Francis, Norbert & Jon Reyhner. 2002. Language and literacy teaching for indigenous education: A bilingual approach. Clevedon England: Multilingual Matters Ltd. Garvin, Paul. 1973. Some comments on language planning. In Joan Rubin & Roger Shuy (eds), Language planning: Current issues and research, 24–73. Washington DC: Georgetown University Press. Grimes, Barbara A. (ed.). 2000. Ethnologue: Languages of the world (14th edn). Dallas, Texas: SIL International. Grimshaw, Allen. 1997. Origins and milestones. In Christina Bratt Paulston & G. Richard Tucker (eds), The early days of sociolinguistics: memories and reflections, 101–112. Dallas, TX: The Summer Institute of Linguistics. Haugen, Einar. 1959. Planning for a standard language in Norway. Anthropological Linguistics 1(3): 8–21. Haugen, Einar. 1961. Language planning in modern Norway. Scandinavian Studies 33: 68–81. Haugen, Einar. 1966. Language conflict and language planning: the case of Modern Norwegian. Cambridge, MA.: Harvard University Press. Haugen, Einar. 1972. The ecology of language: Essays by Einar Haugen. Stanford CA: Stanford University Press.
Language policy: The first half-century
Hornberger, Nancy H. (ed.). 1997. Indigenous literacies in the Americas: Language planning from the bottom up. Berlin: Mouton de Gruyter. Hymes, Dell. 1974. Foundations in sociolinguistics: an ethnographic approach. Philadelphia: University of Pennsylvania Press. Jernudd, Bjoern & Jiří V. Neustupný. 1987. Language planning: for whom? In L. LaForge (ed.), Proceedings of the international colloquium on Language planning, 69-84. Quebec: Presses de l’Universite Laval. Kaplan, Robert B. 1994. Language policy and planning: fundamental issues. Annual Review of Applied Linguistics 14: 3–19. Kaplan, Robert B. & Richard B. Baldauf. 1997. Language planning from practice to theory. Clevedon, Avon: Multilingual Matters Ltd. King, Kendall A. 1997. Indigenous politics and native language literacies: Recent shifts in bilingual education policy and practice in Ecuador. In Nancy H. Hornberger (ed.), Indigenous literacies in the Americas: Language planning from the bottom up, 267–283. Berlin: Mouton de Gruyter. Kloss, Heinz. 1966. German-American language maintenance efforts. In Joshua Fishman (ed.), Language loyalty in the United States, 206–252. The Hague: Mouton. Kloss, Heinz. 1969. Research possibilities on group bilingualism: a report. Quebec: International Center for Research on Bilingualism. Kymlicka, Will & Alan Patten. 2003a. Introduction: Language rights and political theory: Context, issues and approaches. In Will Kymlicka & Alan Patten (eds), Language rights and political theory, 1–51. Oxford UK: Oxford University press. Kymlicka, Will & Alan Patten (eds). 2003b. Language rights and political theory. Oxford UK: Oxford University Press. Landry, Rodrigue & Eric Forgues. 2007. Official language minorities in Canada: and introduction. International Journal of the Sociology of language 185. 1–10. Mackey, William. 1970. A typology of bilingual education. Foreign Language Annals 3(4): 596–608. MacNamara, John. 1971. Successes and failures in the movement for the restoration of Irish. In Joan Rubin & Bjoern Jernudd (eds), Can language be planned? Honolulu HI: University Press of Hawaii. McCarty, Teresa L. 2003. Revitalising indigenous languages in homogenising times. Comparative Education 29(2): 147–163. Nekvapil, Jiří. 2006. From language planning to language management. Sociolinguistica 20: 92–104. Neustupný, Jiří V. 1968. Some general aspects of “language” problems and “language” policy in developing societies. In Joshua A. Fishman, Charles A. Ferguson & Jyotirindra Das Gupta (eds), Language problems of developing nations, 285–294. New York: John Wiley and Sons. Neustupný, Jiří V. 1997. Sociolinguistics: Some other traditions. In Christina Bratt Paulston & G. Richard Tucker (eds), The early days of sociolinguistics: memories and reflections, 201–209. Dallas, TX: The Summer Institute of Linguistics. Ó Murchú, Mairtin. 1977. Successes and failures in the modernization of Irish spelling. In Joshua A. Fishman (ed.), Advances in the creation and revision of writing systems, 267–289. The Hague & Paris: Mouton. Paulston, Christina Bratt & G. Richard Tucker (eds). 1997. The early days of sociolinguistics: memories and reflections. Dallas, TX: The Summer Institute of Linguistics. Phillipson, Robert. 1990. English language teaching and imperialism. Unpublished Doctorate, University of Amsterdam, Amsterdam.
Bernard Spolsky Phillipson, Robert. 1992. Linguistic imperialism. Oxford: Oxford University Press. Prague School. 1973. General principles for the cultivation of good language (Paul L. Garvin, Trans.). In Joan Rubin & Roger Shuy (eds), Language planning: Current issues and research, 102–111. Washington DC: Georgetown University Press. Reaser, Jeffrey & Carolyn Temple Adger. 2008. Vernacular language varieties in educational settings. In Bernard Spolsky & Francis M. Hult (eds), Handbook of educational linguistics, 161–173. Malden MA and Oxford England: Blackwell Publishing. Reyes, Luis & Luis C. Moll. 2008. Bilingual and biliterate practices at home and school. In Bernard Spolsky & Francis M. Hult (eds), Handbook of educational linguistics, 147–160. Malden MA and Oxford England: Blackwell Publishing. Ricento, Thomas. 2006a. Theoretical perspectives in language policy: an overview. In Thomas Ricento (ed.), An introduction to language policy: Theory and method, 3–23. Malden MA and Oxford UK: Blackwell publishing. Ricento, Thomas (ed.). 2006b. An introduction to language policy: Theory and method. Malden MA and Oxford UK: Blackwell publishing. Rubin, Joan & Bjoern Jernudd (eds). 1971. Can language be planned? Honolulu: University Press of Hawaii. Rubin, Joan, Bjoern Jernudd, Jyotirindra Das Gupta, Joshua A. Fishman & Charles A. Ferguson. 1977. Language planning processes. The Hague: Mouton Publishers. Saban, Ilan & Muhammad Amara. 2002. The status of Arabic in Israel: Reflections on the power of law to produce social change. Israel Law Review 36(2): 5–39. Schiffman, Harold E. 1996. Linguistic culture and language policy. London and New York: Routledge. Shin, Hyunjung & Ryoko Kubota. 2008. Post-colonialism and globalisation in language education. In Bernard Spolsky & Francis M. Hult (eds), Handbook of educational linguistics, 206–219. Malden MA and Oxford England: Blackwell Publishing. Shohamy, Elana. 2006. Language policy: Hidden agendas and new approaches New York: Routledge. Silverstein, Michael. 1979. Language structure and linguistic ideology. In Paul R. Clyne (ed.), The elements: a parasession on linguistic units and levels, 193–247. Chicago: Chicago Linguistic Society. Smalley, William A. 1994. Linguistic diversity and national unity: Language ecology in Thailand. Chicago IL: University of Chicago Press. Spolsky, Bernard. 1974. Linguistics and the language barrier to education. In Thomas A. Sebeok, Arthur S. Abramson, Dell Hymes, Herbert Rubenstein, Edward Stankiewicz & Bernard Spolsky (eds), Current Trends in Linguistics: Linguistics and adjacent arts and sciences, vol. 12, 2027–2038. The Hague: Mouton. Spolsky, Bernard. 1977. The establishment of language education policy in multilingual societies. In Bernard Spolsky & Robert L. Cooper (eds), Frontiers of bilingual education, 1–21. Rowley, MA.: Newbury House Publishers. Spolsky, Bernard. 2004. Language Policy. Cambridge: Cambridge University Press. Spolsky, Bernard. 2006a. Fallas en la politica del lenguaje (Failures of language policy). In Roland Terborg & Laura Garcia Landa (eds), Los retos de la planificatión del lenguaje en el siglo XXI (The challenges of language policy in the XXI century), vol. 1, 91–111. Mexico: Universidad Nacional autónoma de México. Spolsky, Bernard. 2006b. Family language management: some preliminaries. In Anat Stavans & Irit Kupferberg (eds), Studies in Language and Language Education: Essays in Honor of Elite Olshtain. Jerusalem: Magnes Press.
Language policy: The first half-century
Spolsky, Bernard. 2006c. Language policy and management in the context of an ecological model of language diversity, Conference on Language, Diversity and Integration in the Enlarged EU: Challenges and Opportunities. Kaunas, Lithuania. Spolsky, Bernard. 2006d. Language policy failures – why won’t they listen? In Martin Pütz, Joshua A. Fishman & JoAnne Neff- van Aertselaer (eds), ‘Along the Routes to Power’: Explorations of Empowerment through Language, 87–106. Berlin & New York: Mouton de Gruyter. Spolsky, Bernard. 2007a. An ecological model for language policy and management, Conference on Languages and Cultures World Language Centre in Iceland, The Vigdís Finnbogadóttir Institute of Foreign Languages, University of Iceland, Reykjavík. Spolsky, Bernard. 2007b. Riding the tiger. In Nancy Hornberger (ed.), Can schools save indigenous languages? 152–160. Basingstoke UK: Palgrave Macmillan. Spolsky, Bernard. 2007c. Tolerance, liberty and language management: towards a liberal theory of language accommodation, LED 2007: Second International Conference on Language, Education and Diversity. The University of Waikato. Spolsky, Bernard. forthcoming. Managing language and languages. Cambridge England: Cambridge University Press. Spolsky, Bernard. in press. Religious language management. In Li Wei & Vivian Cook (eds), Contemporary Applied Linguistics (Vol. 2 Linguistics for the Real World ): Continuum. Spolsky, Bernard & Elana Shohamy. 1999. The languages of Israel: policy, ideology and practice. Clevedon: Multilingual Matters. Stotz, Daniel. 2006. Breaching the peace: struggles around multilingualism in Switzerland. Language Policy 5(3): 247–265. Voegelin, Carl F., Florence M. Voegelin, & Noel M.Jr. Schutz. 1967. The language situation in Arizona as part of the South West cultural area. In Dell Hymes & W.E. Bittle (eds), Studies in Southwestern ethnolinguistics, 403–451. The Hague: Mouton. Walter, Stephen L. 2003. Does language of instruction matter in education? In Mary Ruth Wise, Thomas N. Headland & Ruth M. Brend (eds), Language and life: essays in memory of Kenneth L. Pike, 611–635. Dallas TX: SIL International and the University of Texas at Arlington. Walter, Stephen L. 2008. The language of instruction issue: Framing an empirical perspective. In Bernard Spolsky & Francis M. Hult (eds), Handbook of educational linguistics, 129–146. Malden MA and Oxford England: Blackwell Publishing. Williams, Colin H. 2007. Linguistic minorities in democratic context. Basingstoke UK and New York: Palgrave Macmillan. Wright, Sue. 2004. Language Policy and Language Planning: from nationalism to globalisation. London, Basingstoke: Palgrave Macmillan.
Intercultural pragmatics, language and society Shoshana Blum-Kulka, Menahem Blondheim Hebrew University, Israel
Juliane House
Universität Hamburg, Germany
Gabriele Kasper
University of Hawai‘i at Mãnoa
Johannes Wagner
University of Southern Denmark
In the following we discuss topics addressed in the session on Intercultural Pragmatics, Language and Society at the 18th International Congress in Linguistics, Seoul, July 2008, organized by the first author. The papers presented were grouped into four subpanels with distinct themes, with each subpanel coordinated and summarized by one of the authors: Globalization of Discourse (Juliane House), International Workplace Communication (Johannes Wagner), Historical Pragmatics (Menahem Blondheim), and Interculturality as Discursive Construction (Gabriele Kasper). These four themes represent key issues in the study of language and society, moving from the universal to the particular, from current concerns to historical underpinnings. Thus exploring the features of English as a Lingua Franca highlights universally shared properties of linguistic usage in a more and more globalized world, and work on communication at the workplace brings to the fore arguably universal principles of interaction used in bilingual communities of practice. On the other hand, the historical pragmatic study of written texts from periods others than the current one shares with all historical endeavors its emphasis on the unique and the particular, be it the use of a specific discourse marker or the norms of politeness in given texts and periods. The recent shift in intercultural pragmatics towards constructionist approaches touches on themes of both sharedness and distinctiveness. Thus it allows for a double view on interculturality, conceptualizing interculturality as a discursive achievement, as an ongoing negotiation which may foreground (or suppress) cultural distinctiveness.
Shoshana Blum-Kulka et al.
1. Globalization of discourse The following topics were found to be of particular importance for capturing the complex phenomenon of the globalization of discourse. 0. Theoretical foundation In trying to establish a (tentative) theoretical foundation for the notion “globalization of discourse” a number of relevant concepts such as “English as a global lingua franca” (House 2003), “Community of practice” (Wenger 1998), “Accommodation in communication” (Giles & Ogay 2006); “Drift” (Sapir 1921), and “Levelling and Simplification” (Trudgill 1986) need to be considered. These concepts might be taken to underly an integrative, transdisciplinary characterization of gobalized discourse. Inside such a theoretical framework, linguistic, computer-mediated and pedagogic aspects will be shown to play a particularly important role in globalized discourse. 1. Particular linguistic aspects of globalized discourse at various linguistic levels 1a. The lexical level: Globalized discourse has been characterized as featuring a large amount of “internationalisms” (Braun et al. 2003) . Much has been made in the literature about the massive incursion of Anglo-American lexical items, so-called “Anglicisms”, into many of the world’s languages. Such borrowings have mostly been condemned as damaging local languages in their functional and expressive potential. However, there are also voices claiming that such borrowing in fact facilitates comprehension as it adds to the transparency and overall smoothness of interlingual and intercultural communication processes (cf. here Werner’s (2008) examples of Anglicisms imported into the German and Japanese languages). 1b. The semantic level: Globalized trends in the semantic development of certain formulae or illocutionary indicating devices such as please, thank you or sorry have recently been investigated in the literature. It has been shown that the semantic versatility of such seemingly passe-partout items as the above comes at a cost to the expression potential of these items in that they often turn out to be inappropriate when it comes to expressing “sincere” feelings. Analyses of the current substantial borrowings into many languages (from Cypriot Greek to Polish, from Dutch to Irish Gaelic) of Anglo- American terms that express politeness or other functions show that these borrowed terms are put to the service of pre-existing functions in the receiving languages while at the same time developing various additional indexing functions that result from shifts of emphases from their original primary functions in English. (cf. the paper by Terkourafi (2008) and cf. also Thomason (2001) and Heine & Kuteva (2005), who make an essentially similar point in connection with language variation and change).
Intercultural pragmatics, language and society
1c. The pragmatic text/discourse level: In interactions in English as a global lingua franca between two non-native speakers of English in the workplace that are characterized by asymmetric role relationships, certain discoursal features have been identified (regarding turn length, interruption, topic shift, floor management or the incidence of small talk). These features place this discourse type in the vicinity of native-non-native talk as it has been described abundantly in the Interlanguage literature. In both constellations, one of the interactants is dominant, and this dominance is indexed by the absence of small talk and/or mitigation. (cf. Yun Jung Kim 2008 and see for similar results House and Lévy 2008). In the absence of asymmetric role relationships in ELF talk, such features do not seem to characterize ELF discourse. Rather, the use of ELF in symmetrical conversational every day talk can be described as overall co-operative and markedly consensus-, solidarity- and harmony-oriented (House 2008 & House 2002). In discussions of the nature of English as a lingua franca (ELF) discourse, ELF has been likened to a species of “foreigner talk”. It is important, however, not to confound ELF with the idea of an Interlanguage or a learner language. ELF speakers explicitly do not aim to achieve English native competence, they are not “learners” but rather “users”, and they engage in intercultural communication (cf. Chen 2008). Globalized norms of written discourse in various genres have been shown to “drift” towards universal English-based rhetorical patterns – as was the case with Classical Latin’s and Classical Chinese’s influence in ways other than simple lexical borrowing. This drift might be explained with reference to exemplar-based models of natural language processing (cf. Eggington’s 2008). Standardized, homogenized “English-based” forms of rhetorical patterning have been observed in various academic, scientific, bureaucratic and economic global discourse types all over the world. However, while processes of homogenization are most visible in academic texts (particularly in the sciences), written business communication tends to be subject to (deliberate) local variation. In the field of “intercultural rhetoric”, recent studies of globalized and/or localized texts focus on contextualized text analysis, viewing underlying cultures as in principle hybrid and dynamic, and text production as subject to negotiation and accommodation processes (Canagarajah 2007). 1d. The level of socio-semiotics: Globalized discourse at this level of description is described as a reservoir of “globalized linguistic signs” through which globalized multilingual linguistic landscapes are created. Linguistic landscapes show characteristic uses of the written language in the public sphere making language visible in a specific new way, particularly so in urban spaces. Examples discussed in the literature are the linguistic
Shoshana Blum-Kulka et al.
landscapes in East Asian Cities (Beijing, Seoul, Shanghai, Tokyo, Singapore). Their representation scheme of discourse can be taken as relevant illustrations of linguistic usage in a globalized world (cf. Gorter 2006; Backhaus 2006 & Kim Sung-Do 2008). 2. Globalized discourse in computer mediated communication The role of advanced technology has been much discussed in connection with the globalization of discourse. Given the fact that the English language can no longer be said to be “owned” by its inner circle native speakers, the influx of English words into for instance Chinese blogs can be taken as proof of the fact that “English is going local”. This is particularly true of cases where not only the usual brand names, Hollywood film titles and other tokens of Western pop culture are used, but also perfectly “ordinary English words” (such as house, car etc.) – which invariably have perfectly suitable Chinese equivalents - are employed to achieve certain pragmatic functions such as being rebellious, showing off one’s global identity, or obfuscating everyday meanings of words in computer mediated communication. The English language is thus instrumentalized as a resource for one’s native tongue (cf. Lee 2006; Wengao Gong & Huaqing Hong 2008) . Opinion is divided as to whether linguistic inequality increases through monolithic use of English in the Internet across the non-English speaking world, or whether it is in fact through the Internet as a new “global language” that equality is actively being promoted. The rapidly spreading availability of the Internet’s “virtual universe” is seen by some as creating an egalitarian “ubiquitous society” without political, social or linguistic borders. Sooner or later, then, the digital divide will be overcome by the ubiquitous use of English as a lingua franca as a new cyber dialect (cf. Fusa Katada 2008) open to everybody. Recent studies have also looked at computer-mediated communication and at Internet domains as communities of practice. They examine language variation and change via, for instance, grammaticalization of discourse markers which index politeness and mitigation (cf. Cheshire in press; Kazuko Tanabe 2008). 3. Pedagogic Aspects of the Globalization of Discourse The growing impact of globalisation on the nature of teaching English as a foreign language (EFL) leading to Westernization and Americanization in methods and materials used in classrooms has polarized the profession in many Asian countries. Such a polarization is particularly noticeable in Korea, which has been under a huge economic, political and cultural influence from the USA for the past 50 years. While some Korean scholars still welcome Americanization for its promise of participating in Western style democracy and capitalist economic success, others extol nationalist processes and openly resist US style hegemony which, they claim, makes their compatriots see
Intercultural pragmatics, language and society
everything through American eyes. The resulting conflicting discourses are for instance reflected in EFL textbooks in Korea (cf. Leejin Choi 2008 ). 2. International workplace communication Empirical research on international workplace communication is related to at least three different major research areas: – Organizational Communication: There is a long tradition of researching organizational (workplace) communication. This has been done mainly by quantitative methods of social sciences (reference), to a lesser degree by qualitative studies (Boden 1994), as e.g. field work in an anthropological tradition. A subfield here is the study of cultures and subcultures in an organization (Martin 1992). – There is as well a long tradition to study social behavior in institutional environments. Following Drew & Heritage (1992), a large number of studies have investigated institutional interactions in fine granularity. While some institutions are exhaustively investigated, e.g. medical interaction (Maynard 2003; Heritage & Maynard 2006) , others are rather underresearched. – There is work on international communication, often placed inside a larger EFL or LSP environment (Bargiela-Chiapini & Harris 1997a, 1997b, Firth 1995; Ehlich & Wagner 1995) The intersection of these three fields is a rather sparsely populated field of enquiry which studies naturally occurring interactions in international workplaces. These studies are interested in the linguistic situation of professionals who, because of their professional skills, are engaged in work with different linguistic backgrounds. There are good reasons for why international workplaces have not been researched very much. There are purely methodological obstacles, i.e., data collection in companies is difficult. Further, the populations are ‘transient’ and often move rather swiftly in and out of the researchers reach. 3. Main issues in this field 1. The dominance of English as a lingua franca, (Firth 2008; Lüdi & Heiniger, 2005). 2. Cultural differences. Work concerning cultural differences is perhaps best represented in a volume edited by Bargiela-Chiapini and Harris (1997a). 3. The robustness of transnational interactions, (Firth 2008; Rasmussen 2000; Rasmussen & Wagner 2002). Firth’s work (1991; 1995; 1996) has been very influential in this area.1 Firth (1996) analyzes English as a lingua franca within
. For a different perspective on lingua franca discourse, see Knapp & Meierkord (2002).
Shoshana Blum-Kulka et al.
international commercial workplaces. He maintains that, in spite of the “nonnative”, “marked” or “abnormal” appearance of work-related talk, interlocutors routinely rely on general, arguably universal, principles of interaction in order to “imbue talk with orderly and “normal” characteristics”, and that interlocutors adapt to each other’s language use regardless of its fit with the norms of the relevant cultural groups (i.e., English native speakers). A similar perspective on lingua franca discourse as well as native/non-native interaction is taken, for example, in Meierkord 2000; Rasmussen & Wagner 2002; Wagner 1995a, 1995b. Firth (2008) focusses on “normalizing” behaviour in ELF interactions. This involves participants doing various kinds of interactional “work” that serves to render marked (socio)linguistic behaviour “normal” for the purposes immediately at hand. 4. Lately, a new focus has emerged on the interplay between building social relations and doing professional work. How professionally motivated social encounters tend to build social reations. Day, Jensen & Millar (2008), argue on the basis of detailed studies of business interaction that a central organizing feature of these data is their relational instrumentality; while global communication is very goal oriented – where getting the job done takes precedence over surface level features of linguistic correctness and politeness – there may nonetheless be an (re)establishment of important social relationships between participants. Moreover, these relationships are part and parcel of “getting the job done”. Wagner (2008) shows in a longitudinal study how “professional friendships” are build over time and how the participants in stressful situations draw on these social relations. Both papers challenge notions of instrumentality in communication as “impersonal” suggesting instead that by observing naturally occurring communication and grounding these observations in an in-depth understanding of their contexts, complex interplays of work communication and social networking can be uncovered. 4. Outlook The close analysis of the communicative life of multilingual workplaces is just beginning. There is much work to be done in exploring the relationships between communicative activity and the interlocutors’ use of bilingual resources. Future work in this area should enable us to better understand what bilingual communicative competence entails for professionals, and should provide the basis for more informed decisions concerning company language training and language policy.
Intercultural pragmatics, language and society
5. Historical pragmatics 5.1 Historical pragmatics: Distance, depth, and dynamism “A scholar who is quoted, his lips speak in his grave” – thus speak the lips of a scholar quoted in the Babylonian Talmud (ca. 650 CE, Yevamot, 97a). He was saying, in effect, that the words of the present can, should, and do dialogue with those of the past. Historical pragmatics (HP) is in many ways, an attempt to accomplish this kind of miracle and operationalize the notion it carries. It suggests that the history of language and speaking can be gleaned, to an extent, from written records – the graveyards of words, which originally are ever oral. This not at all trivial premise serves as a foundation for the study of historical pragmatics. But why would sociolinguists want to engage in it? Two opposite rationales present themselves as a general answer, one stressing distance, the other proximity. The first approach would liken the study of historical pragmatics to linguistic anthropology or cross-cultural pragmatics (Blum-Kulka, House & Kasper 1989), in that it allows us to extend our understanding of the complex links between language, society, and culture well beyond the matrix familiar to us. While linguistic anthropology stretches our knowledge and imagination in space, historical pragmatics does it in time, yielding an acquaintance with the distant, exotic, discursive worlds of the past. This extension of the familiar allows us to validate, modify, or even build theory on a broad and varied base of experience – accumulated over time rather than across space – in the quest for establishing universally valid insights. The alternative rationale for studying historical pragmatics highlights familiarity rather than strangeness, continuity rather than contrast. Since our present linguistic practices grew out of earlier ones, the understanding of the latter can help us interpret the former. Such an approach bellies another major potential of historical pragmatics: it can orient us to process of linguistic change in itself. It can help uncover the dynamics of shifts in linguistic conventions, in social relations and in what pragmaticians are particularly interested in: the interface between the two. In an age of rapid transformation in communicative affordances bred of new technologies, as well as of radical shifts in social organization, tracing the dynamic dimension of pragmatics is a major prospect. In sum, a historical approach can offer pragmaticians the proceeds of difference, depth, and dynamism. 5.2 Major Themes: The linguistic-pragmatic tension Echoing similar tensions in the fields of discourse analysis and sociolinguistics, studies in HP vary in focus between the poles of a pragmatic and a linguistic
Shoshana Blum-Kulka et al.
orientation. Thus one finds works launching their inquiry from specific linguistic forms such as the study of a distinct discourse marker or the pragmatic functions of a grammatical form (e.g., Visconti 2005; Alonso-Almeida in press). Other studies start with function such as politeness or speech acts phenomea, in historical texts (e.g., Kohnen 2000; Kopytko 1995). As it appears, the intensifying interest in the interface of pragmatics and linguistics, coupled with the opening, through the influence of pragmatic interests, of a grand new spectrum of texts for analysis, has been luring scholars and tilting the field towards the linguistic pole. One can notice this kind of shift by scanning the titles of papers published in the field’s by now established outlet: The Journal of Historical Pragmatics, launched by Andreas Jucker, the founding father of the field. The pragmatic pole of the HP project has yielded over the years a kaleidoscopic range of significant issues in pragmatics, like politeness and impoliteness (Kryk-Kastovsky 2006), argumentativeness (Blum-Kulka, Blondheim & Hacohen 2002), the realization modes of specific speech acts like promises (Arnovick 1994), address terms (Kohnen 2008), or threats (Rudanko 2004). 5.3 The methodological quandary of HP The study of language today is informed by a rich array of context and culture sensitive approaches, to include pragmatics, sociolinguistics, discourse analysis, and communications. Access to oral language is a key to many of these approaches and modern information and communication technologies, since the mid 20th century have made it possible to capture “fresh” and “naturally occurring” speech for close scrutiny and analysis. These efforts were aided by innovative, theoretically informed, techniques for transcribing speech and adding contextual nuance through multimedia representation. Historical pragmatics inevitably has to live without these key techniques. In response to this challenge HP has developed an array of remedies, that will, however, always remain partial; for as Harold Innis averred, the oral can be only “revealed darkly through the written or the printed word.” HP has responded to the challenge by seeking corpora that would appear to closely echo the oral world, such as court proceedings, political debates, public sermons, even cartoons (Danet & Bogoch 1992; Koch 1999; Kahlas-Tarkka 2007; Collins 2006). But of course, all of these can be no more than an approximation, as are literary or theatrical dialogues, further distanced from the lifeworld of speech by their nature as fiction. However, modern theories of language have no time constraints; therefore, we can apply them freely to historical texts. And indeed, a prominent theme running through works of historical pragmatics is the application of models such as speech acts or relevance theory to fictional and
Intercultural pragmatics, language and society
non-fictional historical texts (Sell 2001). Another limited but fruitful solution to the dilemma has been to probe the meta-communications on speech and conversation so as to glean from them insight into conversational norms and practices (Jacobs & Jucker 1995: 9–10; Watts 1995). 5.4 Scope of the field: Languages, eras, contexts A remarkable but not surprising aspect in the development of HP is its extremely broad language scope. Since pragmatics reflects social and communicative universals, it is luring to trace their manifestations in diverse linguistic and textual environments. Thus, while Jucker, Fritz and Lebsnaft (1999) have surveyed traditions of study which evolved in the Romance, German, and English worlds, extensive new work is being done on Japanese (e.g., Onodera & Suzuki 2007), Russian (Collins 2006), Hebrew (Blondheim & Blum-Kulka 2001), and other languages. The historical pragmatic subpanel in CIL 18 illustrate this inclusiveness, featuring texts in Greek (Mendels 2008), Aramaic (Blondheim & Blum-Kulka 2008) and Japanese (Shimomia 2008). Further, HP has been expanding its historical scope beyond early modern and modern history and into the medieval and ancient. In fact, to illustrate from the same conference, papers on HP are dedicated mainly to pre-modern texts, to include dialogical texts from the Old Testament and the Koran, (Liebes 2008), the Gospels (Mendels 2008), and liturgical drama from the high middle ages (Argenter 2008). The expanding scope of HP study has also underscored the great range of social situations and contexts covered in the field, ranging from courtrooms to courtship, seduction to sanctity. Across these geographic, temporal, and situational divides, HP demonstrates its potential to help uncover new data for analysis, to historically validate linguistic theorems, and to suggest new ideas and insights. Its fast accumulating wisdom is relevant to historians too, pointing to the prospect of adding the history the discourse of cultures to veteran fields of historical inquiry such as economic, social military and gender history. Such a project would serve the quest of reconstructing human development, in all its richness. 6. Interculturality as discursive construction Communication between members of different speech communities has been a long standing research topic in sociolinguistics, intercultural pragmatics, and intercultural communication (e.g., Kiesling & Bratt Paulston 2005; SpencerOatey, in press for recent collections). An unexamined tenet in much of this research is that participants’ membership in different cultural, ethnic, or language
Shoshana Blum-Kulka et al.
groups defines an interaction as “intercultural” ipso facto. Membership in different linguo-cultural groups is seen to determine participants’ understanding and behavior by default and to constitute a common source of “intercultural miscommunication”. As an alternative to socio-cultural and rationalist views, interculturality can be conceptualized as a contingent interactional accomplishment. From a broadly discursive-constructionist perspective, one may ask in what ways ostensibly “intercultural” encounters are treated as such by the participants. When is intercultural distinctiveness relevant for them and how do they orient to it through the organization of their interaction? Drawing on ethnomethodology, conversation analysis, membership categorization analysis, and recent versions of interactional sociolinguistics, a growing literature explores interculturality as a participant concern (e.g., Day 1994, 1998, 2006; Higgins 2007; Mori 2003; Nishizaka 1999). A key question addressed in this research is how participants co-construct themselves and each other as members of particular cultural groups, and how they problematize or reject such categorization. Specifically, the adopted discursiveconstructionist approaches enable investigators to examine in detail how participants invoke, reaffirm, and resist cultural ideologies, how, for the participants, cultural membership articulates with incumbency in other social categories, how they make and contest knowledge claims on cultural matters including language, and display and attribute cultural and language expertise. Methodologically, this research agenda requires that interactional data be analyzed in emic perspective, that is to show how the participants’ actions, orientations, understandings, and identities are publicly manifest in the details of their interactional conduct (Markee & Kasper 2004). Table 1 summarizes some important contrasts between sociostructural and rationalist conceptualizations of interculturality on the one hand and discursiveconstructionist approaches on the other. A recurrent theme in the literature are the devices through which participants assume or ascribe cultural membership. For instance, it has been shown how cultural objects such a films (Mori 2003) and “typical” food items such as kimchi (Korean fermented vegetables) and tsukemono (Japanese pickles) (Zimmerman 2007) are recruited as categorization devices and organize participation frameworks. In Mori’s and Zimmerman’s studies, the talk about such objects accomplishes social affiliation, whereas Fukuda (2008) shows rather the opposite. In her study of dinner table conversations between L1 and L2 speakers of Japanese, the L1 speakers attach familiarity with and enjoyment of “typical” Japanese foods to the category of Japanese ethnicity, thereby treating mainstays of Japanese cuisine as “racially exclusive possession of culture” (Yoshino 1992). As Fukuda proposes,
Intercultural pragmatics, language and society
Table 1. Contrasting perspectives on interculturality Sociostructural/rationalist
Discursive-constructionist
Unitary, static
Diverse, dynamic, resource & construction
– Locus
Personal trait
– Duration
Permanent, context-independent Dominant
Situated interactional accomplishment Emergent, contingent, contextual, contestable Variably relevant, other identities may be more salient Resource to accomplish actions & organize participation frameworks Construct (cultural) orientations & identities Interactional resource
Culture Cultural identity
– Relation to
other identities
– Relation to actions &
Determines actions & participation frameworks
Discourse practices and resources Role in interaction
Culturally determined
participation
Foregrounded cultural distinctiveness Research perspective
Hazardous, source of miscommunication Disaffiliative
Etic (interculturality as presupposed, based on interaction-exogenous theory)
Potentially affiliative, disaffiliative, or relationally inconsequential Emic (interculturality as a participants’ concern)
by laying exclusive claim to these objects, the L1 speakers reflexively construct the non-Japanese co-participants as cultural outsiders (“gaijinization”, Iino 1999) while auto-exoticizing their own cultural group. Here, then, the L1 participants orient less to “inter”-culturality than to firm ethnic boundaries. In contrast, other studies show how orientations to intercultural distinctiveness articulate with interactional projects that emphasize shared identities. Analyses of peer discussions between Japanese and international graduate students at a university in Japan (Ikeda 2008) and conversations between two adolescent ESL speakers and an L1 English speaking graduate student (Kim 2008) reveal that the participants contingently orient to intercultural matters without assuming primary membership in exclusive cultural groups. Instead, repeated “intercultural moments” (Ikeda) enable the graduate students to pursue “transcultural” social purposes, specifically to achieve collegial relations with their peers and membership in an academic community. In Kim’s study, shifts between foregrounding and backgrounding
Shoshana Blum-Kulka et al.
intercultural distinctiveness are interrelated with interactional matters such as topic development, claims to authority, and shifting participation frameworks. For the most part, studies have given focus to the resources and interactional practices through which participants orient to membership in different cultural communities. Much less attention has been paid to “intracultural” diversity as a participant concern. Yet Omori (2008) observes such concurrent orientations to national and local identities in his study of conversation activities in an ESL program. In discussing cultural practices in “American” holiday celebrations, the North American conversation partners instantiate cultural membership as “Americans” and as members of local communities with diverse practices. Simultaneously, they accomplish the institutional goal of providing to the ESL students opportunities for learning “American” culture. Participants’ orientation to cultural membership is often associated with claims to, of demonstrations of, expertise in matters of culture and language. Here the question arises just how such knowledge claims are interactionally achieved. Particularly relevant in this context is research on the linguistic resources through which speakers index evidentiality and epistemic stance (Chafe & Nichols 1986; Ochs 1996). Cook (2008) examines how one such resource, the Japanese epistemic stance marker desho(o), is deployed in conversations between foreign students on homestay in Japan and their Japanese hosts. In the host family members’ contributions, desho(o) implicitly conveys their expectations as to what the students should (not) know about Japanese culture and language. Likewise, the students effectively deploy the marker in making their own knowledge claims and constructing themselves as knowledgeable co-participants. One particular domain of expertise that may become relevant for participants in multilingual settings is language. Second language research in particular is replete with partitionings of people into “native” and “nonnative” speakers, proposing that membership in these categories is consistently relevant to the research purpose and correlated with linguistic knowledge and behavior. Against the prevalent etic view, ethnomethodological and constructionist perspectives require researchers to treat “native” and “nonnativeness” like any other social category, i.e., a participant concern. Responding to the many problems inherent in the notion of “nativeness”, and the argument that “nativeness” bears no necessary relation to either mastery of a language nor affiliation with it, several researchers have adopted Rampton’s (1990) suggestion to replace the categorical notion of “nativeness” with the continuous attribute of “(differential language) expertise” (Hosoda 2006; Kasper 2004). In either case, it falls on the researcher to demonstrate that and how participants visibly orient to native or nonnative identities, or to differential language expertise, at specific interactional moments, and to show what gets accomplished through it. Lee (2008) examines how students
Intercultural pragmatics, language and society
and teachers contingently make their native/nonnative speaker identities relevant as they engage in pedagogical tasks in ESL classrooms. However, language as an interactional topic and its association with expertise are not confined to language classrooms. Metalingual sequences generated in ordinary conversation may accomplish different practical purposes, such as invoking co-membership as multilingual speakers (Zimmerman 2008) or asserting linguistic authority over the co-participant (Takigawa 2008). Just as other cultural objects, language as a topic is a malleable resource to accomplish affiliation and disaffiliation. In addition to examining the construction of interculturality as discursive practice in adult and adolescent talk, a productive research direction explores how children and other novices gain membership in multilingual and multicultural settings over time. A long and diverse research tradition has examined the joint development of cultural, interactional, and literacy competencies from the perspectives of sociocultural theory (Vygotsky 1978; Wertsch 1998), language socialization (Garrett & Baquedano-Lopéz 2002; Kulick & Schieffelin 2004; Ochs 1996), and situated learning (Lave & Wenger 1991; Rogoff 2003). Predominantly though not exclusively, in the examined settings novices interact with more experienced “expert” co-participants who enable increasing participation through assisted performance. However, mounting evidence from a variety of learning environment shows that expertise is distributed between collaborating learners (Blum-Kulka & Snow 2004; Donato 2004; Lantolf & Thorne 2006; Ohta 2001). The contributions by Blum-Kulka and Gorbat (2008) and Cekaite and Aronsson (2008) provide further support for this key finding. In both cases, immigrant children (to Israel, Blum-Kulka; to Sweden, Cekaite & Aronsson) transform their initially peripheral participation through persistent engagement in peer activities, such as language play and impromptu “drills”. As an upshot of their study, Cekaite and Aronsson conclude that theories of sociocultural development as participation have to be extended to incorporate children’s constructions of and orientations to institutional hierarchies and control. The longitudinal studies of immigrant child L2 learners in specific educational contexts are complemented by research on ethnic minority children’s orientation to intercultural distinctiveness in several multilingual communities (Olshtain 2008).
References Alonso-Almeida, Francisco. In press. The pragmatics of and-conjunctives in Middle English medical recipes: a Relevance theory description. Journal of Historical Pragmatics 8(2). Argenter, Joan A. 2008. Liturgical drama in medieval Catalonia and its Latinate origin: the pragmatics of liturgy, verbal genre, verbal art, and vernacular performance. Paper presented at the 18th International Congress of Linguistics, July, Seoul.
Shoshana Blum-Kulka et al. Arnovick, Leslie K. 1994. The expanding discourse of promises in present-day English: A case study in historical pragmatics Folia Linguistica Historica 15(1–2): 175–191. Babylonian Talmud. Tractate Yevamot. 97a. Backhaus, Peter. 2006. Multilingualism in Tokyo: A look into the linguistic landscape. International Journal of Multilingualism 3: 52–66. Bargiela-Chiapini, Francesca & Sandra Harris (eds). 1997a. The Languages of Business: An international perspective. Edinburgh: Edinburgh University Press. Bargiela-Chiapini, Francesca & Sandra Harris. 1997b. Managing language: the discourse of corporate meetings. Philadelphia: Benjamins. Blondheim, Menahem & Shoshana Blum-Kulka. 2001. Literacy, orality, television: Mediation and authenticity in Jewish conversational arguing, 1–2000 AD. Communication Review 5: 499–527. Blondheim, Menahem & Shoshana Blum-Kulka. 2008. Historical continuity of conversational patterns: The evidence from the Talmud. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Blum-Kulka, Shoshana, Julian House & Gabriele Kasper (eds). 1989. Cross-cultural pragmatics: Requests and apologies. Norwood, NJ: Ablex. Blum-Kulka, Shoshana, Menahem Blondheim & Gonen Hacohen. 2002. Traditions of dispute: From negotiating talmudic texts to the arena of political discourse in the Media. Journal of Pragmatics 34(10–11): 1569–1594. Blum-Kulka, Shoshana & Catherine Snow. 2004. The potential of peer talk. Discourse Studies 6: 291–307. Blum-Kulka, Shoshana & Naomi Gorbat. 2008. “How do you say toilet in Russian?” The social grammar of young Hebrew L2 novices’ interaction with their Hebrew speaking peers. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Boden, Deirdre. 1994. The business of talk: organizations in action. Cambridge: Polity Press. Braun, Peter et al. 2003. Internationalismen. Tübingen: Niemeyer. Canagajarah, Suresh. 2007. Lingua franca English, Multilingual Communities and Language Acquisition. The Modern Language Journal 91: 923–939. Cekaite, Asta, & Karin Aronsson. 2008. Participation and classroom hierarchies: On L2 learning and social order in an immersion classroom. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Chafe, Wallace & Johanna Nichols (eds). 1986. Evidentiality: The linguistic coding of epistemology. Norwood, New Jersey: Ablex. Chen, Xinren. 2008. Foreigner Talk: An intercultural pragmatic perspective. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Cheshire, Jenny. in press. “Discourse variation, grammaticalisation and stuff like that”. Journal of Sociolinguistics. Collins, Daniel E. 2006. Speech reporting and the suppression of orality in seventeenth-century Russian trial dossiers. Journal of Historical Pragmatics 7(2): 265–292. Cook, Haruko. 2008. Construction of a knowledgeable person through desho(o): A case of JFL learners and their Japanese host families. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Danet, Brenda & Bryna Bogoch. 1992. From oral ceremony to written document: The transitional language of Anglo-Saxon wills. Language & Communication 12(2): 95–122. Day, Dennis. 1994. Tang’s dilemma and other problems: Ethnification processes at some multilingual workplaces. Pragmatics 4: 315–336.
Intercultural pragmatics, language and society
Day, Dennis. 1998. Being ascribed, and resisting, membership of an ethnic group. In Charles Antaki & Sue Widdicombe (eds), Identities in talk, 151–170. London: Sage. Day, Dennis. 2006. Ethnic and social groups and their linguistic categorization. In Kristin Bührig and Jan D. ten Thije (eds), Beyond misunderstanding. Linguistic analyses of intercultural discourse, 217–244. The Hague, Netherlands: Benjamins. Day Dennis, Astrid Jensen & Sharon Millar. 2008. Making communication work: instrumental and relational aspects of intercultural communication in the globalised workplace. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Donato, Richard. 2004. Aspects of collaboration in pedagogical discourse. Annual Review of Applied Linguistics 24: 284–302. Drew, Paul & John Heritage (eds). 1992. Talk at work: interaction in institutional settings. Cambridge: Cambridge University Press. Eggington, William. 2008. Is there a ‘drift’ towards universal English-based rhetorical patterns? Paper presented at the 18th International Congress of Linguistics, July, Seoul. Ehlich, Konrad & Johannes Wagner (eds). 1995. The Discourse of Business Negotiation. Berlin: Mouton de Gruyter. Firth, Alan 1991. Discourse at Work. Negotiating by telex, fax and phone. Unpublished Ph.D. dissertation, Aalborg University, Denmark. Firth, Alan (ed.). 1995. The Discourse of Negotiation. Oxford: Pergamon. Firth, Alan. 1996. The discursive accomplishment of normality: On “lingua franca” English and conversation analysis. In Johannes Wagner (ed.) Special issue of The Journal of Pragmatics 26: 237–258. Firth, Alan. 2008. Talk international: ‘Normalizing’ behaviour in work-based ‘lingua franca’ English encounters. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Fukuda, Chie. 2008. The deployment of interculturality in NS-NNS conversations: Covert alienation of NNS from Japanese culture. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Garrett, Paul B. & Patricia Baquedano-López. 2002. Language socialization: Reproduction and continuity, transformation and change. Annual Review of Anthropology 31: 339–361. Giles, Howard. & Tania Ogay. 2006. Communication Accommodation Theory. In Brian B. Whaley & Wendy Santer (eds), Explaining Communication: Contemporary theories and exemplars, 293–310. Mahwah, NJ: Erlbaum. Gorter, Durk. 2006. The study of the linguistic landscape as a new approach to multilingualism. International Journal of Multilingualism 3: 1–6. Heine, Bernd & Tanja Kuteva. 2005. Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Heritage, John & Douglas W. Maynard (eds). 2006. Communication in Medical Care: Interaction Between Primary Care Physicians and Patients. Cambridge University Press. Higgins, Christina (ed.). 2007. A closer look at cultural difference: ‘Interculturality’ in talk-ininteraction. Special Issue, Pragmatics 17(1). Hosoda, Yuri. 2006. Repair and relevance of differential language expertise in second language conversations. Applied Linguistics 27: 25–50. House, Juliane. 2002. Communicating in English as a lingua franca In Susan Foster-Cohen (ed.), EUROSLA Yearbook 2, 243–261. Amsterdam: Benjamins. House, Juliane. 2003. English as a lingua franca: A threat to multilingualism? Journal of Sociolinguistics 7: 556–578.
Shoshana Blum-Kulka et al. House, Juliane. 2008. English as a lingua franca in globalized discourse. Paper presented at the 18th International Congress of Linguistics, July, Seoul. House, Juliane. & Magdalène Lévy. In press. Academic office hours as intercultural contact situations. Arbeiten zur Mehrsprachigkeit. Iino, Masakazu. 1999. Language use and identity in contact situations. In Lawrence F. Bouton (ed.), Pragmatics and language learning, vol. 9, 129–162. Urbana, IL: Division of English as an International Language, University of Illinois at Urbana-Champaign. Ikeda, Keiko. 2008. Reinterpreting intercultural moments: A case of peer discussions among foreign graduate students in Japan. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Jacobs, Andreas & Andreas H. Jucker. 1995. The historical perspective in pragmatics. In Andreas H. Jucker (ed.), Historical pragmatics: Pragmatic development in the history of English, 3–33. Amsterdam & Philadelphia: John Benjamins. Jucker, Andreas H., Gerd Fritz, & Franz Lebsnaft (eds). 1999. Historical dialogue analysis. Amsterdam & Philadelphia: John Benjamins. Kahlas-Tarkka, Leena & Matti Rissanen. 2007. The sullen and the talkative: Discourse strategies in the Salem examinations. Journal of Historical Pragmatics 8(1): 1–24. Kasper, Gabriele. 2004. Participant orientations in conversation-for-learning. The Modern Language Journal 88: 551–567. Katada, Fusa. 2008. What the emergence of ‘autolinguals’ say about the state of ‘EAGLE’ (English as a global lingua franca) in the ubiquitous age. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Kiesling, Scott F. & Christina Bratt Paulston (eds). 2005. Intercultural discourse and communication. Oxford: Blackwell. Kim, Younhee. 2008. ‘We are different, but not too different’: Interculturality construction as interactional business. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Kim, Sung-Do. 2008. The linguistic landscape of East Asian Cities from the perspective of intercultural pragmatics. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Kim, Yun Jung. 2008. Quantitative, topical and interactional dominance in EFL writing tutorial discourse: Interaction between NNS tutors and NNS tutees. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Koch, Peter. 1999. Court records and cartoons: Reflections of spontaneous dialogue in early romance texts. In Andreas H. Jucker, Gerd Fritz & Franz Lebsanft (eds), Historical dialogue analysis, 329–399. Amsterdam & Philadelphia: John Benjamins. Kohnen, Thomas. 2000. Explicit performatives in Old English: A corpus-based study of directives. Journal of Pragmatics 1(2): 301–321. Kohnen, Thomas. 2008. Linguistic politeness in Anglo-Saxon England? A study of Old English address terms. Journal of Historical Pragmatics 9(1): 140–158. Kopytko, Roman. 1995. Linguistic politeness strategies in Shakespeare’s plays. In Andreas H. Jucker (ed.), Historical pragmatics: Pragmatic development in the history of English, 515–540. Amsterdam & Philadelphia: John Benjamins. Knapp, Karlfried & Christiane Meierkord (eds). 2002. Lingua Franca Communication. New York: Peter Lang. Kryk-Kastovsky, Barbara (ed.). 2006. Historical Courtroom discourse, Special Issue of Journal of Historical Pragmatics 7(2).
Intercultural pragmatics, language and society
Kulick, Don & Bambi Schieffelin. 2004. Language socialization. In Alessandro Duranti (ed.), A companion to linguistic anthropology, 349–368. Malden, MA: Blackwell. Lantolf, James P. & Steven L. Thorne. 2006. Sociocultural theory and the genesis of second language development. Oxford: Oxford University Press. Lave, Jean & Etienne Wenger. 1991. Situated learning: Legitimate peripheral participation. Cambridge University Press. Manchester: St Jerome Press. Lee, Yo-An. 2008. Nonnative identity in classroom interaction: What constitutes pedagogical interaction? Paper presented at the 18th International Congress of Linguistics, July, Seoul. Lee, Jamie Shinhee. 2006. Linguistic constructions of modernity: English mixing in Korean television commercials. Language in Society 35: 59–91. Leejin, Choi. 2008. The impact of globalization on the English teaching in Korea: Representation in Korea EFL textbooks. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Liebes, Tamar. 2008. The uses of classic religious texts in making war and peace. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Lüdi, Georges & Monika S. Heiniger. 2005. L’organisation de la communication au sein d’une banque régionale bilingue. Sociolinguistica 19: 82–96. Markee, Numa & Gabriele Kasper. 2004. Classroom talks: An introduction. The Modern Language Journal 88: 491–509. Martin, Joanne. 1992. Cultures in Org: Three Perspectives. OUP. Maynard, Douglas W. 2003. Bad News, Good News: Conversational Order in Everyday Talk and Clinical Settings. Chicago: University of Chicago Press. Meierkord, Christiane. 2000. Interpreting successful lingua franca interaction. An analysis of non-native-/non-native small talk conversations in English. Linguistik online 5, 1/00. http://www.linguistik-online.de/1_00/MEIERKOR.HTM Mendels, Doron. 2008. The orality and the literacy of the synoptic gospels: A historical pragmatics perspective. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Mori, Junko. 2003. The construction of interculturality: A study of initial encounters between Japanese and American students. Research on Language and Social Interaction 36(2): 143–184. Nishizaka, Aug. 1999. Doing interpreting within interaction: The interactive accomplishments of a “Henna Gaijin” or “Strange Foreigner.” Human Studies 22: 235–251. Ochs, Elinor. 1996. Linguistic resources for socializing humanity. In John J. Gumperz & Stephen Levinson (eds), Rethinking linguistic relativity, 407–437. Cambridge: Cambridge University Press. Ohta, Amy S. 2001. Second language acquisition processes in the classroom: Learning Japanese. Mahwah, NJ: Lawrence Erlbaum. Olshtain, Elite. 2008. How do children perceive “intercultural distinctiveness” in contexts of minority-majority relations? Paper presented at the 18th International Congress of Linguistics, July, Seoul. Omori, Makoto. 2008. Construction of interculturality in an English conversational exchange program. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Onodera, Noriko O. & Ryoko Suzuki. 2007. Historical Changes in Japanese: Subjectivity and Intersubjectivity. Special issue of Journal of Historical Pragmatics 8(2). Rampton, Ben. 1990. Displacing the ‘native speaker’: expertise, affiliation and inheritance. ELT Journal 44: 97–101.
Shoshana Blum-Kulka et al. Rasmussen, Gitte. 2000. Zur Bedeutung kultureller Unterschiede in interlingualen und interkulturellen Gesprächen. München: Iudicium. Rasmussen, Gitte & Johannes Wagner. 2002. Language choice in international telephone conversations. In Kang Kwong Luke & Theodossia-Soula Pavlidou (eds), Telephone Calls. Unity and diversity in Conversational Structure across languages and cultures, 111–131. Amsterdam: John Benjamins. Rogoff, Barbara. 2003. The cultural nature of human development. Oxford: Oxford University Press. Rudanko, Juhani. 2004. ‘I wol sterve’: Negotiating the issue of a lady’s consent in Chaucer’s poetry. Journal of Historical Pragmatics 5(1): 137–158. Sapir, Edward. 1921. Language. New York: Harcourt, Brace and World. Sell, Roger D. 2001. A historical but non-determinist pragmatics of literary communication Journal of Pragmatics 2(1): 1–32. Shimomia, Tadao. 2008. Haiku and linguistics. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Spencer-Oatey, Helen (ed.). In press. Culturally speaking, 2nd edition. London: Continuum. Tanabe, Kazuko. 2008 The creation of polite mitigation in cyber-dialect in light of the theory of communities of practice. The case of young people’s computer mediated communication. Paper given at CIL 18, July, Seoul. Takigawa, Yuzuru. 2008. Problematic couple talk: American husbands and Japanese wives interacting in Japanese. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Terkourafi, Maria. 2008. Globalizing trends in the semantics of please, thank you and sorry. Paper given at CIL 18, July, Seoul. Thomason, Sara. 2001. Language Contact. An Introduction. Edinburgh: Edinburgh University Press. Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell. Visconti, Jacqueline. 2005. On the origins of scalar practices in Italian Journal of Historical Pragmatics 6(2): 237–261. Vygotsky, Lev S. 1978. Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Wagner, Johannes. 1995a. What makes a discourse a negotiation? In Ehlich Konrad & Johannes Wagner (eds), The discourse of business negotiation, 9–36. Mouton de Gruyter. Wagner, Johannes. 1995b. Negotiating activity in technical problem solving. In Alan Firth (ed.), The discourse of negotiation: Studies of the language in the workplace, 223–246. Pergamon. Wagner, Johannes. 2008. International workplace communication: business, professional friendship and language. Paper presented at the 18th International Congress of Linguistics, July, Seoul. Watts, Richard J. 1995. Justifying grammars: A Socio-pragmatics foray into the discourse community of early English Grammarians. In Andreas H. Jucker (ed.), Historical pragmatics: Pragmatic development in the history of English, 145–186. Amsterdam & Philadelphia: John Benjamins. Wengao Gong & Huaqing Hong. 2008. The globalization and glocalization of the English language: Insights form the use of English words in Chinese Blogs. Paper presented at the 18th International Congress of Linguistics, July, Seoul.
Intercultural pragmatics, language and society
Wenger, Etienne. 1998. Communities of Practice. Cambridge University Press. Werner, Angelika. 2008. Internationalismen in globalen Diskursen am Beispiel des Deutschen und Japanischen. Paper given at CIL 18, July, Seoul. Wertsch, James V. 1998. Mind as action. Oxford: Oxford University Press. Yoshino, Kosaku. 1992. Cultural nationalism in contemporary Japan. London: Routledge. Zimmerman, Erica. 2007. Constructing Korean and Japanese interculturality in talk: Ethnic membership categorization among users of Japanese. Pragmatics 17: 71–94. Zimmerman, Erica. 2008. Crafting multilingual and intercultural identities in talk: I know a little about your language. Paper presented at the 18th International Congress of Linguistics, July, Seoul.
Historical linguistics in 2008 The state of the art Brian D. Joseph
The Ohio State University, USA
1. Introduction Historical linguistics can be characterized as both the study of language change and the study of language history. These are two distinct but related enterprises: we learn about language change by studying particular events in the history of individual languages but the history of these languages also includes more than just the study of change; there are also elements to be recognized that have stayed relatively stable and unchanged over time. Moreover, historical linguistics is interested in determining and exploring relationships that languages show with other languages, whether genetic or diffusionary in nature; the former sort of relationship can provide indirect evidence of change, if related languages, sprung from the same source, nonetheless show, as they typically do, some differences between one another, whereas the latter type of relationship allows for the determination of instances of language change through language contact. Thus there is a vast amount of material that provides grist for the historical linguistic mill; in a real sense, every language has a history and thus is of potential interest to the historical linguist, if even just at the level of description, with the recording of the details of a given language’s historical development. Within the history of linguistics, there have been times when the venerable field of historical linguistics was linguistics, period, with little else of concern except the historical. During such times, it was almost impossible to engage in linguistics without being well-versed in historical methodology and without caring about the historical dimension to any description or account. So too, however, there have been times when historical linguistics has been virtually absent from the main stream of linguistic thought and practice, especially in the United States. The former period can be identified with the 19th century and into the early to middle of the 20th century, and the latter period with much of the second half of the 20th century. As Watkins (1989: 784) noted, reflecting on the field from the late 1960s on into the 1980s, “it is possible to get a Ph.D. degree in linguistics at
Brian D. Joseph
a number of fine and distinguished American universities without ever taking a course in historical linguistics, and there are good linguists teaching in my own department [at Harvard University] who have never had such a course”. This period coincides with the emergence of intense concern for scientific methodology and the development of models that might be called “linguistics as cognitive science” or “linguistics as a branch of biological science”,1 and for many practitioners, it was not clear where the study of language change and language history fit into those scientific conceptualizations. There is of course an irony in this admittedly somewhat caricatured view of the field, since historical linguistics in some sense can be said to have put linguistics on a scientific footing in the 19th century with its development of predictive means, via the recognition of the regularity of sound change, of accounting for certain aspects of language change. 2. State of the art: Infrastructural indicators To move to the current state of the art, the present day is a mix. If one looks just to external signs, then there are several indicators of health: – the existence of a well-attended biennial international scientific meeting devoted to historical linguistics, namely the International Conference on Historical Linguistics (ICHL) soon to occur in its 19th instantiation (2009, in Nijmegen) – a related book series, publishing, in John Benjamins’ Current Issues in Linguistic Theory series, selected papers from every ICHL since ICHL 3 (1977) and occasional special thematic volumes from workshops at (ICHL gatherings) – a flourishing journal dedicated to publishing the very best papers on a wide range of topics in historical linguistics, namely Diachronica (published by Benjamins), which is soon to expand from two to three issues per year – the emergence of new book series with historical linguistics as their focus, including the University of Edinburgh Press historical textbook series and the series Brill has launched with a purely Indo-European focus (in a sense taking historical linguistics back to its own roots in 19th century Indo-European comparative philology) – the maintenance of a lively electronic listserv, Histling ([email protected]. edu), maintained by Claire Bowern of Rice University, that serves numerous . This characterization of the field most famously belongs to Noam Chomsky; as Battistella (1996: 130) notes, “Chomsky views linguistics as a branch of psychology (in turn a branch of biology)”.
Historical linguistics in 2008: The state of the art
subscribers with announcements and opportunities for discussion of relevant issues – relatedly, the recognition of a field of inquiry that has come to be known as “contact linguistics”; while language contact has long been of considerable interest to linguists, it is now coming of age as a subfield, with the publication in recent years of several textbooks and surveys (e.g., Thomason 2001; Winford 2003). It is true that contact linguistics is not just focused on studies of language change and particular histories, but is equally attentive to on-going multilingualism on a social and individual scale; nonetheless, the importance of understanding language contact for understanding language change is now widely recognized – rapid expansion in the past 20 years of interest in “grammaticalization”, the study of the origins of grammatical forms and of developments with grammatical forms, is noteworthy too (and external signs of this keen interest can be cited, such as a now-biennial conference (New Reflections on Grammaticalization – the fifth one is coming up in summer 2008), various research materials including a compendium of grammatical changes (Heine & Kuteva 2002), a dictionary of grammaticalization terminology (Lessau 1994), and numerous volumes of articles from conferences and the like, too many to list here). As a glance at any of these many works on grammaticalization shows, there is more to grammaticalization nowadays than just historical issues, as those working within this framework tackle matters of emergent grammar, usage-based grammar, the location of language within general communicative and cognitive strategies, and so on. But the original impetus behind studies of grammaticalization was the historical side of the development of grammar and thus grammatical change, as two landmark works demonstrate: Meillet 1912, which provided a conceptual basis, as well as relevant terminology (he spoke of “grammaticalisation”, after all) for the grammaticalization “movement”, and Givón 1971, who offered a modern nudge, with his slogan “Today’s morphology is yesterday’s syntax”, in the direction that Meillet pointed towards. These various indicators represent what might be called the “infrastructure” of historical linguistics.2 With all of them taken into consideration, it certainly
. Another infrastructural issue that might be considered is institutional in nature. That is, the question of jobs in historical linguistics within the academy must be considered. Here, however, there is not the same array of positive indicators for the health of historical linguistics since jobs that focus just on historical linguistics or even which mention it as a desired secondary specialization are few and far between. I reckon that there have been no more than a dozen in my 30 or so years in the field.
Brian D. Joseph
seems that historical linguistics is on a solid footing and healthy once again. It is fair to say that there is good reason for these signs of health, since they rest in part on a recognition by those in the linguistic mainstream that historical linguistics really is a part of linguistics and has much to offer the field in general. Such a pronouncement may seem obvious to some, and it even has a ring of being selfserving (in that it is coming from a historical linguist). Nonetheless, as observed above, it is true that there have been times in the history of linguistics when historical analyses were viewed as niceties at best that could serve as distractions to making real progress with understanding language. There is thus within mainstream linguistics now a greater interest, more so than in the era surrounding the 1960s and 1970s perhaps (see the Watkins quote above), in addressing language change and learning about language in general from the examination of how it changes. It is therefore reasonable to explore somewhat the conceptual underpinnings of the relation between historical linguistics and linguistics proper, since such an exercise not only offers an important additional dimension on the state of the field but it also allows for some consideration of new trends and developments: what the key issues – both old and new – are, what remains to be done, what challenges lie ahead, and what opportunities there are to seize upon. 3. State of the art: Conceptual bases Ki parsky (1968: 174) described language change as “a window on the form of linguistic competence”. Since the goal of linguistic theory is to characterize the substance of the human linguistic abilities, i.e., competence, Kiparsky’s pronouncement opened the door to a possible and potentially quite fruitful marriage of theoretical and historical linguistics, and to be sure, this sentiment has been taken to heart from time to time by theoreticians and diachronicians alike.3 I often tell the students in my historical linguistics classes that in my view, to be a good historical linguist, one has to be a good linguist. Moreover, I stress to them the importance of recognizing that there are applications outside of the classroom for what we teach – and learn – in classes on language change and that one such venue for applications is specifically in classes on other areas of linguistics. Such recognition,
. Kiparsky maintains this position in later works; for instance Kiparsky (1988: 405) writes “It is this interplay of mutually constraining factors which gives historical linguistics its focal role in the study of language”.
Historical linguistics in 2008: The state of the art
I say, is key to their learning to be well-informed theoreticians, sensitive to the value of diachronic evidence for the testing of theoretical claims that are more synchronic in scope and focus. It also means that advances in theory or in our understanding of any aspect of language – advances that are often made on the synchronic front – can have an impact on our view of the diachrony of a given element or construct in a given language. For instance, before the middle of the 20th century, before the work of Alfred Tarski and Richard Montague in particular, there really was no such subfield within linguistics as formal semantics. And, huge advances have been made in our understanding of phonetics with the advent of acoustic phonetics as a well-developed science with its own methodologies and related technology. Consequently, in order to understand now just how, for instance the scope of negation or marking for definiteness might develop or change through time or how sounds are altered diachronically, the insights of, respectively, formal semantics and acoustic phonetics cannot be ignored.4 To do so would be folly, but it would also mean that one was not being the best historical linguist possible, since one was not being the best linguist possible. Being a linguist first and foremost is crucial to understanding how linguistic systems can change. Nonetheless, there is a downside, and on-going controversy, associated with an interest in looking to diachronic evidence for models of synchronic linguistic competence. In particular, Kiparsky’s treatment, in that same 1968 paper, of various linguistic developments in several languages engendered a notion that language change was simply equivalent to change in the grammatical apparatus employed in theoretical descriptions. And since the theoretical constructs one worked with were themselves subject to change due to shifts in the acceptance of particular linguistic theories and general analytic frameworks, the “location” of language change in properties of the grammar has depended on the shape and form of that assumed grammar. Thus while Kiparsky 1968 and 1971 embraced such classical generative phonological notions as rule ordering, which translated into treating rule reordering as a mechanism of phonological change, more recent theoreticians, working within a constraint-based optimality theory (OT) framework, look to rerankings of the relative strength of constraints as a primary mechanism of change. Other related questions come up below in sections 4 and 7.
. On the role of advances in acoustic phonetics to the understanding of sound change see especially the work of John Ohala, as summarized nicely in Ohala (2003). Deo (2006) is a nice example of the value of formal semantics to historical analysis.
Brian D. Joseph
4. S tate of the art: Old but persistent questions and old but useful methods In a now-classic paper, Weinreich, Herzog, and Labov (1968) articulated a number of basic issues that face anyone interested in examining language change; briefly stated, they are:5
(1) a. the “constraints” problem: What are the general constraints on change, if any, that determine possible and impossible changes and directions of change? b. the “transition” problem: By what route does language change? c. the “embedding” problem: How is a given language change embedded in the surrounding system of linguistic and social relations? d. the “evaluation” problem: How do members of a speech community evaluate a given change, and what is the effect of this evaluation on the change? e. the “actuation” problem: Why did a given linguistic change occur at the particular time and place that it did?
Although progress is being made on solutions to all these problems – in particular, see below section 5, regarding variationist methodology – they still remain as a driving force behind most research into the mechanisms of language change. And, just as such key questions remain from the advent of variationist approaches to language change, there are some key principles of even greater age, dating from the 19th century, that remain useful and almost essential. For instance, the Neogrammarian formulation of sound change as inherently regular to this day not only drives decisions about historical reconstruction, but it also provides a basis for determining relative chronology of changes, for separating out borrowings from inherited items, and for recognizing analogical change as opposed to sound change. Although questions are routinely raised about the validity of the Neogrammarian position on sound change, it seems well justified to recognize a type of change event, to adapt the phraseology of Hale (2003), that involves systematic (i.e., regular) and phonetically driven alteration of sounds, what may be called “sound change proper”, “sound change in the strict sense”, or simply “Neogrammarian sound change”, to introduce terms that I have employed in my teaching over the years; see Labov (1981) and Hale (2003) for discussion.
. Joseph (2001a) offers some discussion of each problem. Some of these problems are reminiscent of statements of problems in studying language change articulated by Coseriu (1958), namely the “rational” problem (why are languages not invariant? Why are they always changing?), the “general” problem (what are the conditions that lead to language change?), and the “historical” problem (why does any particular change occur when it does?).
Historical linguistics in 2008: The state of the art
Similarly, careful philological attention to data, arising out of their training in the Classics, allowed the Neogrammarians to have a solid empirical basis for their historical work, and that can be seen as the precursor to a modern interest in building theories on accurate descriptive studies and experimental results. And, old methods remain useful and in fact indispensable for progress along certain fronts. One in particular, the comparative method, despite challenges to its validity (cf. some of the discussion in Durie & Ross 1996), still is the most reliable and powerful method for certain types of historical questions, especially those involving relatedness among languages. Also, however, identifying across different languages cognate forms that are “congruent” historically in some way but are not identical means that at least one of those languages has undergone at least one change.6 In that way, the comparative method also provides indirect evidence of language change and therefore, depending on how much one believes in one’s reconstructions, for particular changes as well. The use of the comparative method thus not only allows for the recovery of aspects of language history but it also feeds into our understanding of language change. To return to the generative reinterpretation of language change as grammar change mentioned at the end of the previous section, one key question it raised (see Andersen 1973) but left unaddressed, namely why the rules should change, is a question that persists now. Inasmuch as that reinterpretation continues today in a reincarnated form in OT, a parallel question of why constraints should ever be reranked needs to be asked. One answer to both the earlier instantiation of that question and the current one is that changes in the grammar are not the mechanism of change but only the modeling of the results of completed changes, but to the extent that the grammar change view continues to hold appeal, this key question must continue to be asked. 5. State of the art: New methods As noted in section 3, new advances in linguistic theory have immediate consequences for any diachronically oriented linguist: determining how the view of language change is altered in the new theory and asking whether the new theory provides new insight into, or just new robe, as it were, for particular changes. In the
. Maybe both have changed and maybe more than one change has caused the difference between the outcomes of what was once the same element in the proto-language linking the two offspring languages in question. But minimally, such a situation indicates that at least one change has occurred.
Brian D. Joseph
case of new methods, the question is simply how to put them to best use in analyzing and accounting for change. There are three relatively new types of methodologies that add to tried and true methods mentioned in section 4 for studying language change. These are computationally based applications, variationist methods, and attitudinal/ideological approaches. Under computational applications, I include the modelling of change (as in Polinsky & van Everbroeck 2003), the mathematical testing of claims about frequency effects, relatedness, the role of chance, and the validity of tree phylogenies (as well represented in the many papers in Forster & Renfrew 2006 (including studies by Brett Kessler, April McMahon, Johanna Nichols, Donald Ringe, and Tandy Warnow, among others), and by such works as McMahon & McMahon 2006, again among others), the use of statistical tests more generally (including applications as employed by variationist sociolinguists – see below), and corpusbased studies (such as those emerging from the study of the Helsinki Corpus of English or any of the other large annotated or tagged corpora that offer material on different stages of well-documented languages). Under variationist methods, I have in mind the studies in the paradigm of William Labov and related approaches, as exemplified best in Labov (1994, 2001, and more recently 2007). In this line of inquiry, the fine details of inter-speaker and intra-speaker variation, especially with regard to phonology, are subjected to careful instrumental phonetic analysis and to various statistical tests using such programs as VARBRULE (Sankoff 1988) or GOLDVARB (Rand & Sankoff 1991), and the sometimes subtle but nonetheless significant trends shown by the variation are correlated with on-going changes progressing towards completion or with changes whose effects have settled into a stable variation. In addition, variationists these days are attending more to matters of style as reflected in usage, and to the role of identity in determining language use; these add up, for the historical linguist, to a crucial concern for external issues that affect language choices speakers make and thereby alter their realization of forms and their selection among variants. If, as Ohala (2003, and elsewhere) reminds us, the seeds of change are to be found in synchronic variation, then any factors which contribute to or promote certain choices out of a range of possible variants will necessarily have an impact on language change. Finally, in what might be thought of as an extension of the variationist interest in identity formation, the role of language ideologies (see Silverstein 1979 on this notion) in guiding speaker choices has come to be recognized as a potent force shaping the direction of change. In as yet unpublished work (Joseph 2006, 2007) on language contact in the Balkans involving phonology and the reactions of speakers to such contact, I have drawn (fruitfully, I believe) on the notion of
Historical linguistics in 2008: The state of the art
language ideologies, especially pertaining to a basic ideologically driven decision speakers make every day (perhaps falling under the rubric of the “evaluation question” stated above in (4)) of where to draw boundaries between one’s own language and the speech form of others. Further compelling applications await only the development of a sufficient level of interest on the part of scholars in this area. 6. State of the art: New opportunities The current state of the field of linguistics presents several novel opportunities for historical linguistics. I signal here three such avenues for new investigation. First, our understanding of language relationships and individual language histories can only be enriched and extended as our knowledge of under-documented and under-studied languages grows, with greater attention and resources being directed toward endangered and generally threatened languages. Of course, not all endangered languages are under-documented but many, perhaps most, are; nor are all under-documented languages endangered, but again, many, perhaps most, are. Thus the heightened awareness of the loss of languages expected in the coming few decades and the increased documentation (and additional funding for such basic research) that it has engendered mean that more will be learned about how various languages are related – or not – to one another. The boost to comparative linguistics from such efforts will mean too that reasonable inferences about the histories of particular languages will be able to be drawn, thus increasing the storehouse of information on possible language changes. Second, the populations that were the basis for variationist studies done now over a generation or more ago, such as the ground-breaking Labov (1963), have aged to a point where it is now fruitful to do follow-up studies, with the same speakers (or a comparable sample). This type of study, e.g., Pope, Meyerhoff & Ladd (2007) or Sankoff & Blondeau (2007) or the work of the Danish LANCHART project,7 allows for the tracing of language change in individuals throughout their lifetimes, a phenomenon not widely recognized as possible, and for the drawing of inferences about the consequences of such lifespan change for language change in a speech community at large. Finally, there are productive cross-disciplinary lines of inquiry to pursue. For instance, striking parallels between biology and linguistics, and more particularly, evolutionary biology and historical linguistics, have long been noted,8 and some . See http://lanchart.hum.ku.dk/; the project studies language change in real time. . See Atkinson & Gray (2005) and Ben Hamed (2004) for some discussion of these parallels.
Brian D. Joseph
of the computational methods noted above in section 5 exploit these parallels, especially with regard to rate of change and working out phylogenies. Still, one has to wonder if these parallels provide a good biological metaphor for linguistic work or if they have substantive consequences; one can note that there is no direct analogue in biological evolution to the common linguistic act of borrowing; in language contact, borrowing can be seen in choices individuals make about which words (for instance) to use, but in biological development, contact with others does not affect individual organisms but can only promote the selection by large populations of particular features.9 And, there are other potentially fruitful crossdisciplinary marriages that may yield insights into language change, e.g., studies on the psychology of bilingualism and what that might tell us about the sort of language contact that takes places within an individual’s mind. 7. Conclusion: A key remaining (set of related) question(s) Several questions have been posed here, but only few have been answered. I see that as reasonable, since the purpose of this presentation is to signal where we are in historical linguistics and where we need to be in the years to come. And while other questions could well be explored – such as the relation of language change to language acquisition by children, what creole languages can tell us about language change, how many distinct processes of language change there are,10 and so forth – I close here with another basic issue that maybe has not yet been answered, and discuss its ramifications:
(2) At what point do we consider a “change” to have occurred?
In particular, the issue here is what “counts” as an event of language change: is it the first introduction of some innovative feature alone (arguably revealing a change in some individual’s language and thus of relevance to the language taken
. For instance, flowers in one area might ultimately develop in the direction, i.e., select for, the color of other flowers in their area, especially if there is some evolutionary advantage to be had. But an individual flower would not change within its lifetime, unlike humans who can change aspects of their language within their lifetimes by contact with other speakers. . More particularly, is grammaticalization a separate process of change (or processes – see Janda 2001 & Joseph 2001b), or just a label for the result(s) of other processes of language change (e.g., analogy, metaphorical extension, sound change, etc.)? Similarly, is lexical diffusion something different from analogy (as, e.g., Phillips 2006 thinks) or just subsumed under analogy (as Kiparsky 1995 argues)? Does reanalysis have a status as a separate mechanism of change? And so on.
Historical linguistics in 2008: The state of the art
as a whole and viewed as the union of the speech of all its speakers) or is it instead the spread of the innovation to other individuals (and arguably within a speaker’s own usage, as to whether the innovation even persists and recurs in that one speaker)? Opinions are mixed on this point, and perhaps it will never be fully resolved. As a spin-off from that question, it is reasonable to ask how abstract language change is, as that may offer a handle on the question of when change is said to occur. More to the point (to return to the position of Kiparsky (1968, 1971) – see above section 3), which comes first, change in the surface manifestation of a language or change in the grammar underlying those surface forms? For instance, in a study involving formal approaches to grammaticalization phenomena, Amritavalli (2004) examined how functional heads grow out of lexical heads and become separate projections, but it is fair to ask what the cause and effect relationship is here regarding the shift in the location of the head: is the shift in the grammar the result of a shift in the way a once-lexical head is used or does the once-lexical head take on a new surface role once its status in the grammar is altered? And similar questions can be asked about any formal account of a change: does the change occur first in the surface and then the grammar “catches up”, so to speak, with the new reality of how forms have been redeployed by speakers, or are the speakers at the mercy, as it were, of restructurings of the grammar? Questions and more questions – that is what the spirit of intellectual inquiry is all about, and thus the robustness of the questions asked here about language change and its study can be taken as an index of the robustness of the field of historical linguistics early in the 21st century.
References Amritavalli, R. 2004. Some developments in the functional architecture of the Kannada clause. In Veneeta Dayal & Anoop Mahajan (eds), Clause structure in South Asian languages (Studies in Natural Language and Linguistic Theory 61). Dordrecht: Kluwer. Andersen, Henning. 1973. Abductive and deductive change. Language 49(4): 765–793. Atkinson, Quentin & Russell D. Gray. 2005. Curious parallels and curious connections: Phylogenetic thinking in biology and historical linguistics. Systematic Biology 54(4): 513–526. Battistella, Edwin. 1996. The Logic of Markedness. New York: Oxford University Press. Ben Hamed, Mahé. 2004. Evaluating congruence between morphological, genetic and linguistic evolution: models and methods. Ph.D. dissertation, University Pierre et Marie Curie (Paris VI), France. Coseriu, Eugenio. 1958. Sincronía, diacronía e historia. El problema del cambio lingüístico. Madrid: Gredos. Deo, Ashwini. 2006. Diachronic Change and Synchronic Typology: Tense and Aspect in Modern Indo-Aryan Languages. Ph.D. Dissertation, Stanford University.
Brian D. Joseph Durie, Mark & Malcolm Ross (eds). 1996. The Comparative Method Reviewed: Regularity and Irregularity in Language Change. Oxford: Oxford University Press. Forster, Peter & Colin Renfrew. 2006. Phylogenetic methods and the prehistory of languages. Cambridge: McDonald Institute for Archaeological Research. Givón, Talmy. 1971. Historical syntax and synchronic morphology: An archaeologist’s field trip. CLS 7: 394–415. Hale, Mark. 2003. Neogrammarian sound change. In Joseph & Janda (eds), 343–368. Heine, Bernd & Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Janda, Richard D. 2001. Beyond “pathways” and “unidirectionality”: On the discontinuity of language transmission and the counterability of grammaticalization. Language Sciences (Special Issue – Grammaticalization: A Critical Assessment, ed. by Lyle Campbell) 23(2–3): 265–340. Joseph, Brian D. 2001a. Historical linguistics. In Mark Aronoff & Janie Rees-Miller (eds), The handbook of linguistics, 105–129. Oxford: Blackwell Publishers. Joseph, Brian D. 2001b. Is there Such a Thing as “Grammaticalization”? Language Sciences (Special Issue – Grammaticalization: A Critical Assessment, ed. by Lyle Campbell) 23(2–3): 163–186. Joseph, Brian D. 2006. Analyzing Variable Outcomes in Contact Phonology: Structure, Ideology, and Bilingualism in Aromanian Borrowings from Greek. Poster presented at NWAV 35 (New Ways of Analyzing Variation), Columbus, OH, 10 November 2006. Joseph, Brian D. 2007. Broad vs. Localistic Dialectology, Standard vs. Dialect: The Case of the Balkans and the Drawing of Boundaries. Paper presented at ICLAVE 4 (4th International Conference on Language Variation in Europe), Nicosia, Cyprus, 19 June 2007. Joseph, Brian D. & Richard D. Janda (eds). 2003. The Handbook of Historical Linguistics. Oxford: Blackwell. Kiparsky, Paul. 1968. Linguistic universals and linguistic change. In Emmon Bach & Robert T. Harms (eds), Universals in linguistic theory, 171–202. New York: Holt, Rinehart and Winston. Kiparsky, Paul 1971. Historical linguistics. In William O. Dingwall (ed.), A survey of linguistic science, 576–642. College Park: University of Maryland Linguistics Program. Kiparsky, Paul. 1988. Phonological change. In Frederick J. Newmeyer (ed.), Linguistics: The Cambridge Survey. Vol. I: Linguistic theory: Foundations, 363–415. Cambridge: Cambridge University Press. Kiparsky, Paul. 1995. The phonological basis of sound change. In John A. Goldsmith (ed.), The handbook of phonological theory, 640–670. Oxford: Basil Blackwell. Labov, William. 1963. The social motivation of sound change. Word 19(3): 273–309. Labov, William. 1981. Resolving the Neogrammarian controversy. Language 57(2): 267–309. Labov, William. 1994. Principles of linguistic change: Internal factors. Oxford: Blackwell. Labov, William. 2001. Principles of linguistic change: Social factors. Oxford: Blackwell. Labov, William. 2007. Transmission and diffusion. Language 83(2): 344–387. Lessau, Donald A. 1994. A dictionary of grammaticalization (3 volumes). Bochum: Universitätsverlag Dr. N. Brockmeyer. McMahon, April & Robert McMahon. 2006. Language Classification by Numbers. Oxford: Oxford University Press. Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia (Rivista di Scienze), 12, no. 26, 6.
Historical linguistics in 2008: The state of the art
Ohala, John. 2003. Phonetics and historical phonology. In Joseph & Janda (eds), 669–686. Phillips, Betty. 2006. Word frequency and lexical diffusion. London: Palgrave. Polinsky, Maria & Ezra van Everbroeck. 2003. Development of gender classifications: Modeling the historical change from Latin to French. Language 79(2): 356–390. Pope, Jennifer, Miriam Meyerhoff, & D. Robert Ladd. 2007. Forty years of language change on Martha’s Vineyard. Language 83(3): 615–627. Rand, David & David Sankoff. 1991. GoldVarb: A variable rule application for the Macintosh (version 2.1). Montreal: Centre de recherches mathématiques, Université de Montréal. Sankoff, David. 1988. Variable rules. In Ulrich Ammon, Norbert Dittmar & Klaus J. Mattheier (eds), Sociolinguistics: An international handbook of the science of language and society, vol. 2, 984–997. Berlin: de Gruyter. Sankoff, Gillian & Hélène Blondeau. 2007. Language change across the lifespan: /r/ in Montreal French. Language 83(3): 560–588. Silverstein, Michael. 1979. Language Structure and Linguistic Ideology. In Paul Cline, William Hanks & Carol Hofbauer (eds), The Elements: A Parasession on Linguistic Units and Levels, 193–247. Chicago: Chicago Linguistic Society. Thomason, Sarah G. 2001. Language Contact: An Introduction. Washington, DC: Georgetown University. Watkins, Calvert W. 1989. New parameters in historical linguistics, philology, and culture history. Language 65(4): 783–799. Weinreich, Uriel, William Labov & Marvin I. Herzog. 1968. Empirical foundations for a theory of language change. In Winifred P. Lehmann & Yakov Malkiel (eds), Directions for Historical Linguistics, 95–188. Austin: University of Texas Press. Winford, Donald. 2003. An Introduction to Contact Linguistics. Oxford: Blackwell.
Lexical semantics Sebastian Löbner
Heinrich Heine University Düsseldorf, Germany
Giving an account of the state of the art in a field as diverse and wide as Lexical Semantics is something one can only pretend to try. The following survey is therefore restricted to an attempt to describe the present state of the art as it is reflected in the contributions submitted for the Parallel Session on “Lexical Semantics” at the 18th International Congress of Linguists in Seoul in 2008.1 Inevitably, the picture emerging from this approach will be biased, among others due to the geographical location of the conference. About one half of the contributions are from East Asian countries, from Korea, Taiwan, China, Japan, Thailand; approximately the same number of contributions have been submitted by European, Russian and North American linguists. Nevertheless, the semantic research presented at the conference exhibits a remarkable degree of uniformity in terms of topics, research methods and theoretical orientations. The following paper offers a brief survey of the talks presented, placing emphasis on theoretical orientation, methodology and major topics. I should begin by apologizing for the unbalanced treatment of contributions; the fact that some have been commented on in more detail than others does not in any way imply that they are more and the others less important. 1. Theoretical orientations Research in lexical semantics is guided by diverse scientific orientations, but the picture emerging from the conference is not as varied as one might expect.2 Some of the work, primarily from the former Soviet Union, follows a structuralist approach to lexical meaning, contributing research about the lexicon as a whole or
. For a recent general overview of the field see Pustejovsky (2006). The article also offers a brief description of Pustejovsky’s qualia approach to be mentioned below. . This is not due to selectional reviewing: theoretical orientation did not play a role in selecting the talks from the applications.
Sebastian Löbner
about greater parts of it. [Pavlov]3 is concerned with principles of sub-divisions of word classes in the lexicon; [Kotorova] tries to determine the overall degree of polysemy in Russian as compared to German; [Neðpore] explores the large field of motion verbs in Latvian by applying a classical structuralist approach based on syntagmatic and paradigmatic relations. A number of contributions reflect the influence of cognitive linguistics in one way or the other. Among the many studies on polysemy, some adopt the model of radial categories developed in Lakoff (1987), or more advanced models based on his, such as the theory of Principled Polysemy (Evans 2005). The notion of metaphor and metonymy as basic cognitive operations appears to be widely accepted. Other authors apply particular cognitive models to the analysis of lexical meaning [Matsumoto]. Work on the decomposition of lexical meanings either adopts predicate logic representations as in Dowty (1979), frame-like models such as Pustejovsky’s qualia theory (Pustejovsky 1995) or a quasi-syntactic approach (Ramchand 2008). 2. Methodology Three remarkable trends are characteristic of the research presented in the Session on Lexical Semantics: (1) in-depth case studies, (2) the massive use of corpuslinguistic data work, and (3) comparative work on two or more languages. Unlike syntax, morphology or phonology, lexical semantics is a discipline that faces an almost unlimited set of explananda, even in a single language – the lexica of natural languages being the vast structures that they are. In view of this colossal task, several strategies can be employed. One involves restricting the area of research to carefully chosen small sections of a given lexicon, sometimes even single lexical items. Although as a rule these studies do not plow up much ground in the field, they do unearth important facets of a more general nature. One possibility for handling the wide-ranging aspects of an analysis is to compare similar phenomena in two or more languages. These types of studies are demanding, since they involve intensive data work and require competence in more than one language or closely coordinated research with others. The session on Lexical Semantics exhibits a remarkable number of cross-linguistic studies. The other extreme of working in the field comprises attempts to cover as much ground as possible. Until recently, the only work that tried to account for the complete
. Names in square brackets refer to talks announced for the Parallel Session on Lexical Semantics at the conference; see the list of talks at the end of the article.
Lexical semantics
lexicon of a language was carried out by lexicographers. Needless to say these efforts, although eminently valuable for the linguist, do not afford much depth of analysis. Moreover, lexicographers compile their dictionaries for the general user rather than for the linguist. Scientific insights in lexical semantics and grammar is only modestly reflected in lexicography. In the last twenty years, however, first attempts have been undertaken by linguists to provide a cartography of complete natural language lexica, e.g., the WordNet enterprise. These network approaches differ from traditional dictionaries in that they present the single-item accounts in a global network of semantic relations. This kind of work represents a significant step toward the scientific exploration of complete natural language lexica. What made this work technically possible is the internet and advanced linguistic software. Today’s lexical semantics owes a great deal to corpus linguistics and this, in turn, has profited from the work of computational linguistics. After decades of theoretical and practical work in the field of language processing, linguistics now benefits from the availability of electronic corpora and elaborate software for the retrieval of relevant data. The use of these resources is becoming more and more standard in the field of lexical semantics. It allows working semanticists to verify their intuitions on the wide range of uses of lexical items and to gather statistical data on the frequency of uses or the co-occurrence with other items. These resources will ensure a dramatic improvement of data work, not only in terms of the quantity of data that can be dealt with but also in terms of the reliability of the results. Quite a few studies presented at the session make essential use of corpuslinguistic methods. They have been applied to investigations of meaning differences between near-synonyms or to the disambiguation of polysemous expressions and subtle differences in the use of particular lexical items, which can only be detected and analyzed by scrutinizing large amounts of text. Equally important, corpuslinguistic tools also allow the application of statistical methods to the data. The lexical semanticist is now in a much better position to make reliable statements about the relative frequency of different uses of a lexical item. 3. Major topics Perhaps the time will come when evidence on lexical meaning can be collected by neurosemantic experimentation. However, for the time being, the only evidence for the lexical meaning of a given linguistic item is the manner it is consistently used in actual speech. Since the vast majority of lexical items are used in the context of sentences, and sentences occur in some user context, the meanings of lexical items can only be reconstructed from indirect evidence. Lexical items in an
Sebastian Löbner
utterance are chosen with respect to their lexical meaning; they are inserted into a sentence, in which their meanings are combined with others in various ways to form more complex meanings and ultimately propositions. When the sentence is used in some actual context, it interacts with external contextual information. References are fixed for the referring expressions in the sentence, and in the course of this process the semantic information of the referring expressions is merged, or unified, with contextual information concerning the respective referents. The process of determining the references may either fail because the semantic information conveyed by the sentence contradicts available context information or because the semantic information would be irrelevant in the given context. In this case, the meanings of lexical items may be shifted, in order to make them fit the context. These modifications of meaning triggered by the given context include well known mechanisms such as metaphor and metonymy. The shifts are due to a general principle, which I have dubbed the Principle of Consistent Interpretation: 4 A complex expression is almost always interpreted such that its parts fit together and that the whole fits the context. Finally, the utterance is interpreted as a communicative act in a given social interaction. Hence, the lexical meaning of an item occurring in an actual utterance in context is merged with non-linguistic knowledge; it may even be shifted to refer to things different from those originally provided for in its lexical meaning; and its interpretation may further be constrained and modified for pragmatic reasons. Lexical semanticists find themselves in the position of utterance recipients, because the only evidence they can base their work on is the way in which lexical items are or can be used in actual contexts. (Even made-up examples of uses must be interpreted as though they were uttered in some concrete context.) Accordingly, lexical semanticists have to reconstruct lexical meanings by trying to undo all the modifications that lexical meanings may have undergone in concrete usages. They have to strip utterance meaning from implicatures and other inferences, recognize the results of meaning shifts and reconstruct their original input, isolate lexical information within the merger of linguistic and world knowledge that results from using expressions for references to objects within the given contexts. They then have to determine the exact meaning the item under scrutiny contributes to the meaning of the sentence. In this sense, lexical semantics presupposes sentence semantics and is based on the assumption of the Principle of Compositionality: The meaning of a complex expression is determined by the meanings of its components and the way in which the components are combined. Conversely, the lexical meaning of a lexical item can be defined as its contribution
. See Löbner (2002: 52f) for an introduction of the principle.
Lexical semantics
to the meanings of the sentences in which it occurs. Only after following these steps of re-construction can the lexical semanticist arrive at a hypothesis about the lexical meaning of expressions. The result does not, however, represent the end of the work to be done. Retrieved from singular instances of use, it will also have to be checked for consistency with other uses. And this, again, is not a trivial task, since so many lexical items are polysemous, which, above all, constitutes an inconsistency of use. Furthermore, as everyone knows who has done corpus-linguistic data mining, actual use of items is to a certain degree inconsistent owing to nonregular language usages, errors in electronic data, or lack of contextual knowledge necessary for interpreting the data. If a lexical item turns out to be lexically ambiguous, we have to ask: Do these meanings form a consistent whole? If so, how are the meaning variants interrelated? Is there a primary or central meaning variant from which the other variants derive? The working lexical semanticist is thus faced with a task of seemingly neverending complexity. It is no surprise then that, in one way or the other, most of the research on lexical semantics is addressed to the variability of meaning. 3.1 Polysemy 3.1.1 Models of polysemy: case studies The notion of polysemy itself is trivial: a lexical item is polysemous if it has more than one meaning. Understanding polysemy, however, is far from trivial; in fact, it constitutes one of the major challenges to the discipline. Given that lexical meaning is a constitutive and distinctive aspect of lexical items, a lexical item should strictly not have more than one meaning. Obviously, there are cases in which one would not admit that they represent one item with several meanings but rather different lexical items, which happen to have the same form; these cases are usually set apart from polysemy and considered as instances of homonymy. If we assume that a given case of lexical ambiguity is an instance of polysemy, i.e., several meanings belonging to the same lexical item, that decision has to be grounded on the fact that these are variants connected by recognizable relations, such as, once again, metonymy and metaphor. Cognitive semantics, in particular the work of Lakoff (1987) and others, has contributed substantially to understanding polysemy. According to the cognitivist view, the meaning variants of polysemous words are related by basic cognitive processes, such as metonymy, metaphor or narrowing. Often, though not necessarily, there is a basic meaning variant, from which the other meanings are derived in one or more steps. The meaning variants of a polysemous expression thus form a network of concepts interconnected by basic cognitive relations. One influential model proposed by Lakoff is the notion of a “radial category”: a network of
Sebastian Löbner
meaning variants has a central member, from which all others are derived by modifications of different kinds, possibly in several steps. The general idea that the meaning variants of polysemous expressions must be related in a principled way, was developed by, among others, Tyler and Evans in an approach they entitle Principled Polysemy. Most in-depth case studies presented in the session are concerned with determining the total meaning variants of some polysemous expression and the structure of the complex of meanings. Two studies apply the Principled Polysemy approach: a detailed study of Taiwanese ma “also” [Huang] and an investigation into the polysemy of Chinese xin “mind, heart” [Lin]. [Mari] suggests a model based on wittgensteinian family resemblance for the range of uses of French avec “with”. Another attempt to analyze the totality of uses of a given item in a principled way is the analysis of the German verb riskieren “to risk” in [Zaefferer]. In German it is possible to say both “risk one’s life” and “risk one’s death” for the same kind of situation5; one might conclude therefore that, given the opposition between life and death, the verb cannot mean the same in both cases. However, Zaefferer argues that the verb has the same meaning in both cases. The two possibilities of use are due to different ways in which the object argument of the verb can be specified; and it is these arguments that are systematically interrelated by metonymic relations. This appears to be a promising line of analysis. A great number of verbs exhibit these kinds of variation in argument specification6 and we are only beginning to understand the underlying principles and constraints. [Mueanjai’s] study offers a detailed analysis of the verbs for “give” in Thai and Vietnamese, hây and cho respectively. Both verbs have lexical as well as grammaticalized uses (e.g., benefactive marking). The study gives a full picture of the respective sets of meaning variants and their internal structure, as well as a semantic comparison of the two verbs. Cross-linguistic studies like this illustrate a general point of polysemy; although the variation of meaning to be observed with polysemous items is principled and motivated, it is still in need of lexicalization. Any two items in different languages, even if they have very similar basic meaning, cannot be expected to exhibit the same range of lexicalized meaning variants. [Chancharu] presents a similar investigation of the meanings and grammaticalized functions of two cognate verbs in Thai and Zhuang,7 both meaning “touch”. The polysemy of five Mandarin taste terms (suan “sour”, tian “sweet”, ku “bitter”, la “hot” and xian “salty”) is systematically investigated in [Chen]. . The same holds for the English verb risk. . For another example see the uses of the verb open mentioned in Löbner (2002: 115). . Another language of the Tai family, spoken in Southern China.
Lexical semantics
3.1.2 Polysemy vs. post-lexical meaning variation Apart from being polysemous, a given lexical item may exhibit further meaning variation when put to actual use. It is widely agreed upon that these additional readings are not on a par with lexicalized meaning variants, but result from postlexical processes of meaning modification. The phenomenon is known under different labels. Bierwisch (1982), among the first to mention it, calls the mechanisms “conceptual shifts”; therefore, he distinguishes metaphors, metonymies and what he calls conceptual differentiation (narrowing). More recently, the same phenomenon was labeled “selectional polysemy” in work by Pustejovsky (see [Jezek] for references). Croft and Cruse (2004) discuss the phenomenon as “contextual modulation”.8 Post-lexical shifts differ from polysemy in that they are not, and need not be, lexicalized; rather, they are triggered by the context in which an expression is used, or more precisely by applying the Principle of Consistent Interpretation to this instance of usage.9 Very often they are the result of obeying selectional restrictions of predicate terms.10 For example, a verb might require a metonymical or metaphorical expression for one of its argument terms to comply with the selectional restrictions for the respective argument. Alternatively, a verb may be interpreted metaphorically in order to make a complement semantically compatible. These meaning shifts follow very general patterns and can analogously be determined for larger classes of lexical items – something not often observed with particular types of meaning variation for polysemous items. Furthermore, these general mechanisms of meaning accommodation can be observed crosslinguistically. Roughly, one might say that lexicalized polysemy induces idiosyncratic meaning variation, while systematic variation requires some knowledge of the general types of shifts. The types of meaning variation found in polysemous lexical entries seem to involve the same kinds of transitions between meaning variants: metaphor, metonymy and differentiation (along with others). How, then, can the working lexical semanticist distinguish between a case of, say, lexicalized metonymy and a case of a metonymical post-lexical shift due to context requirements? Note that meaning variants of polysemous items, too, are selected by the context, often by the very same mechanisms that coerce post-lexical meaning shifts. Evidence for the status of an alternation as lexicalized polysemy or as the result of post-lexical
. One talk immediately relevant here is [Abe] on the contextual re-interpretation (narrowing) of nouns in tautological or contradictory statements such as “a house is no house unless it has more than six rooms”. . See Löbner (2002: 46–52) for a general account. . Löbner (2002: 117f).
Sebastian Löbner
coercion can be found using statistical methods. Given that any use of a term undergoing coercion is loaded with a certain amount of cognitive effort, it can be assumed that coercion uses are less frequent. Conversely, one might expect that the language learners adjust their lexicons such that the most frequent types of uses do not require semantic repair by coercion. Thus, frequencies of use are indicative of regularities: more frequent readings will be regular, i.e., non-coerced; and regular meaning variation will be common across a broader class of lexical items. The problem of distinguishing between lexicalized readings and readings generated by post-lexical modifications is addressed in two papers, [Jezek] and [Nabeshima]. [Jezek] presents a study on the systematic alternations of Italian event nominals, which receive event, result or other readings. [Nabeshima] investigates the homonymy, polysemy and post-lexical variation of the English term(s) wave. He also presents evidence from language acquisition. This line of argument is important, because language learners, too, have to decide whether to store a certain ambiguity in their mental lexicon or to leave it to ad-hoc modifications by general mechanisms of meaning shift. 3.1.3 Global aspects of polysemy The question might be posed if polysemy, although certainly universal, is equally frequent and diverse in all languages. With homonymy, this is probably not the case. Languages having a comparatively small inventory of phonemes and a simple syllable structure evidence less phonological space for lexicalization, which may result in a higher quota of homonymy. Japanese appears to represent such a case. But the question is much less clear for polysemy. In a large number of cases polysemy is due to growth of the lexicon. Innovations in the world and/ or verbal communication about the world provide a constant demand for new vocabulary. In only a few cases will new expressions be invented (almost never); more frequently new expressions are imported or morphologically derived from others; but in the majority of cases, existent expressions with related meaning will be put to use for a newly lexicalized meaning variant. Being the major source of new lexicalizations, polysemy may accumulate over time in the lexicon of a particular language. The underlying processes that cause the growth of the lexicon are the result of historical cultural developments. It can therefore to be anticipated that some societies may have experienced a more rapid and extensive cultural development in their lexica than others, resulting in a higher rate of polysemy. [Kotorova] studies the question as to whether there is something like a polysemy index of individual languages. In a comprehensive comparison of dictionaries, the paper comes to the conclusion that Russian is more polysemous than German.
Lexical semantics
3.2 Meaning in context A number of studies are concerned with the role of lexical meaning in the contexts it enters. Within the lexicon itself, word meanings partake in word formation processes; beyond the lexicon, lexical items enter sentential context engaging combinatorial relations with other items; finally, in actual use lexical meaning is processed within the context of world knowledge relevant for this instance of use. In all of these levels of contexts, the lexical meaning of a given item interacts with the meanings of other items or with other information, and the way in which an item behaves in context provides evidence about its lexical meaning. 3.2.1 Meaning in the context of word formation A special subset of Japanese V-V-compounds is investigated from a general perspective in [Matsumoto], those having the verbs iku “go” and kuru “come” as the first verb of the compound. Although they should be expected to behave quite similarly, due to their similarity of meaning, “come”-compounds exhibit fewer readings than “go”-compounds. The asymmetry is shown to be caused by the difference in deictic orientation encoded in the meanings of the two verbs. [Hsieh & Kolodina] present a comparative account of Mandarin Chinese, Russian and English with respect to the ways in which expressions for eyes and hands are used for the formation of other concepts and lexicalized phrases. Hybrid English-Chinese words are investigated in [Riha]. [Nishina] discusses processes of lexicalization of grammatical phrases in Japanese. 3.2.2 Meaning in sentential context Corpus-linguistic techniques can be used for investigating more subtle differences in use, yielding evidence for the analysis of lexical meaning. [Yoo] presents a corpus-linguistic investigation into the uses English last and next in combination with a temporal noun such as year or week; these combinations sometimes occur with a definite article, sometimes without. The corpus-linguistic method enables the distinction of two different readings of the adjectives, largely coinciding with their two ways of being used. Similarly, the contribution by [Lu & Su] employs corpus-linguistic methods to clarify the difference in meaning between the nearsynonyms bother and worry. In recent years, the theory of construction grammar has attracted much attention. According to this approach the composition of sentence meaning may involve more than just lexical meanings and composition rules for a handful of very general syntactic rules of combination. According to constructionist theory, there is an intermediate level of so-called constructions – i.e., patterns of grammatical combinations much more specific than traditional phrase-structures. It has been argued that these constructions carry a meaning of their own, which is not derivable from lexical
Sebastian Löbner
meanings of the components of the construction by applying general composition rules. From this point of view it has to be investigated which constructions a lexical item, in a given meaning variant, can enter in, and what the contribution of the item is to the meaning of the construction. [Shen] has submitted a study of this nature. It investigates English and Mandarin Chinese double object constructions, the verbs that can enter these constructions and the meanings of the constructions, depending on the semantic sub-class the verb belongs to. 3.2.3 Meaning in the context of world knowledge It has often been asserted that a finer disambiguation or determination of word readings in sentential or even wider contexts requires taking world knowledge into account. For example, in the case of dimensional adjectives such as wide it may be necessary to relate to world knowledge about the argument in order to determine which dimension of the object the adjective refers to. [Chung] takes up the issue of the interpretation of spatial PPs, such as on the lake, which depending on the object to be located (e.g., a boat or a house) may be interpreted differently. The study presents an attempt to develop computational means of automatic disambiguation based on a system of world knowledge. [Chiang] reports on experiments with Taiwanese EFL learners to utilize bodily experience of emotions for teaching the meanings of English emotion words such as love and angry. 3.3 Models of lexical meaning: Decomposition Although the ultimate goal of lexical semantics should be the explicit description of lexical meanings, comparatively little work has been presented on lexical decomposition. This may be due to the fact that the community still lacks workable general models and methods of decomposition. Among the forty contributions to the session, only two are concerned with the decomposition of lexical meaning. These are [Lyutikova & Tatevosov] on the decomposition of verb meanings in the style of Ramchand (2008), and [Son & Roy] on denominal location verbs in Indonesian and French, applying the decompositional approach proposed by Jackendoff (1990). 3.4 Plowing ground A considerable number of contributions are engaged in plowing more ground in the field. Two studies are concerned with general aspects of noun meanings: [Job & Cubelli] on the semanticity of gender, and [Saltarelli] on the count/mass distinction. Most of the other studies that explore larger areas of the lexical landscape deal with classes of verbs or deverbal nouns. These include [Hsiao] on English mannerof-motion verbs; [Neðpore] on Latvian motion verbs; [Chuang] on control verbs in Mandarin. [Schwabe & Fittler] present an investigation into question-embedding
Lexical semantics
verbs in German; [Nazaire] explores the field of French deverbal nouns of what he terms adduction and abduction; [Blanco] contributes a general discussion of abstract nouns. Other studies are directed to more specific lexical fields: Korean emotion verbs [Song], expressions of intensification in French and English [Adler & Asnes], terms for sacred entities in Québécois [Tremblay], or causatives in polite phrases in Slovenian and Japanese [Shigemori Bucar]. 3.5 Working with semantic networks In a number of studies, semantic networks such as WordNet are used as secondary resources for semantic research by statistical methods. [Akama et al.] present a method for automatic evaluation of existing semantic networks using statistical instruments. [Bae] explores the possibilities and demands for a bilingual use of Korean and Chinese wordnets; similarly; [Prévot] presents a method for the structural comparison of French and Chinese verbs on the basis of semantic networks.
References Bierwisch, Manfred. 1982. Formal and lexical semantics. In Hattori, Shiro & Kazuki Inoue (eds), Proceedings of the XIIIth International Congress of Linguistics, August 29–September 4, 1982, Tokyo, 122–136. Tokyo. Croft, William & Alan Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press Dowty, David R. 1979. Word meaning and Montague grammar. Dordrecht: Reidel. Evans, Vyvyan. 2005. The meaning of time: Polysemy, the lexicon and conceptual structure. Journal of Linguistics 41. 33–75. Jackendoff, Ray. 1990. Semantic structures. Cambridge, MA: MIT Press. Lakoff, George. 1987. Women, fire, and dangerous things. What categories reveal about the mind. Chicago, IL: The University of Chicago Press. Löbner, Sebastian. 2002. Understanding Semantics. London: Hodder Arnold. Pustejovsky, James. 1995. The generative lexicon. Cambridge, Mass.: MIT Press. Pustejovsky, James. 2006. Lexical semantics: overview. In Keith Brown (ed.), Encyclopedia of language and linguistics, second edition, 98–106. Amsterdam: Elsevier. Ramchand, Gillian C. 2008. Verb Meaning and the Lexicon: A First Phase Syntax. Cambridge: Cambridge University Press.
Talks announced for the CIL 18 Parallel Session “Lexical Semantics” Abe, Hiroshi: La tautologie et la notion subjective de “desirabilite”. Adler, Silvia & Maria Asnes: Overtalented and Talented beyond measure: Bound and free morphemes as scalarity operators in English and French Akama, Hiroyuki, Maki Miyake & Jaeyoung Jung: New Method of Evaluation on the Graph Clustering of Semantic Networks Built on Lexical Co-occurrence Information. Bae, Sun-Mee: Representation of Korean-Chinese Conceptual Map Using Multilingual Lexical Semantic Network. Blanco, Xavier: Établissement et justification d’un inventaire de classes sémantiques de noms abstraits.
Sebastian Löbner Chancharu, James Naruadol: Semantic extension of verbs of tactile sensation in Thai and Zhuang. Chen, Yi-Rung: Meaning Extension of the Basic Taste Words in Mandarin Chinese. Chiang, Yueh-Tzu: L2 Vocabulary Learning Through the Semantic Mapping of two Emotion Words, love and anger, in English. Chuang, Wei-Chi: When Two Kinds of Mandarin Collective Modifiers Meet Two Kinds of Control. Chung, Eugene & Richard Sproat: Use of World Knowledge for Interpreting Spatial Expressions. Hsiao, Hui-Chen Sabrina: On manner-of-motion verbs. Hsieh, Shelley Ching-yu & Elena Kolodina: Eyes and hands expressions: Embodiment of Lexical Meanings. Huang, Shuping: Applying Principled Polysemy to Adverbials: The case study of Taiwanese ma “also”. Jezek, Elisabetta: Polysemy of event nominals: the process-result ambiguity. Job, Remo & Roberto Cubelli: The place of gender in the lexicon: Somewhere between meaning and form. Kotorova, Elizaveta G.: Polysemy as lexico-semantic universal: quantitative aspect. Lin, You-Min: Conceptual network of Chinese xin. Lu, Louis Wei-lun & Lily I-wen Su: Synonym Choice as a Reflection of Subjectivity: A Quantitative Event Structure Analysis on Bother and Worry Lyutikova, Ekaterina & Sergei Tatevosov: Subevents and their relations: a rich predicate decomposition. Mari, Alda: Dealing with polysemy: the case of avec. Matsumoto, Noriko: V-V Compounds in Japanese and Deixis. Mueanjai, Suthatip: The Semantics of the Verbs of Giving in Thai and Vietnamese: A Contrastive Study. Nabeshima, Kojiro: The ocean can’t wave bye-bye: Polysemy of wave and methods of identifying and locating polysemy in a theory. Nazaire, Mbame: Adduction/abduction et catégorisation lexicale. Neðpore, Gunta: The Semantic Structure of Motion Verbs in Latvian: Paradigmatic and Syntagmatic Aspect. Nishina,Yoko: Lexikalisierung und Grammatikalisierung. Pavlov, Vladimir: The Role of Lexical Variation in Developing a Model of the Lexicon. Prévot, Laurent et al.: Building and aligning Chinese and French lexical graphs. Riha, Helena: The Morphology and Semantics of Hybrid Lettered Words in Chinese. Saltarelli, Mario: A Uniform Hypothesis of Count/Mass Expressions: New Syntactic Evidence. Schwabe, Kerstin & Robert Fittler: Semantic conditions for question-embedding predicates in German. Shen, Yuan: The role of verb meanings in constructions – a comparison between double object constructions in Chinese and English. Shigemori Bucar, Chikako: Causative and Politeness. Son, Minjeong & Isabelle Roy: Decomposition of Denominal Location Verbs. Song, Eui-Jeong: Les constructions des verbes d’émotion en coréen contemporain. Soraj Ruangmanee: The Verbs of Acquiring in Vietnamese: A Corpus-based Cognitive Semantic Study. Tremblay, Mireille: La lexicalisation des sacres en français québécois. Yoo, Isaiah WonHo: A Corpus Analysis of (the) last/next + Temporal Nouns. Zaefferer, Dietmar: Polysemy as monosemy with salient proxies.
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report Johan van der Auwera
University of Antwerp, Belgium
Hana Filip
University of Florida, USA
1. Introduction This chapter surveys some of the very recent work on tense, aspect, mood and modality. The recency is taken very strictly: we focus on the work published in and after 2000. The imperfection of the survey is such that it cannot possibly be exhaustive, largely because of the obligatorily modest size of the paper. For this reason, the paper restricts the survey to those areas that we think have attracted most intellectual energy. In functional theories, we believe that it is mood and modality, more so than tense and aspect, that have fascinated linguists. For formal theories, it is our impression that it is aspect, rather than tense and mood and modality, that has been enjoying pride of place. Of course, these impresssions are influenced by the authors’ own research interests (functional and modal for van der Auwera, formal and aspectological for Filip). 2. Aspect – formal approaches The interest generated by aspect, events and event structure is clearly evident in numerous publications devoted to it since 2000 (see also Binnick 2008), and to take just a few salient examples, several collections of papers can be mentioned: Tenny & Pustejovsky (eds. 2000), Higginbotham, Pianesi & Varzi (eds. 2000), Alexiadou, Anagnostopoulou & Everaert (eds. 2004), Verkuyl, de Swart & van Hout (eds. 2005), Kempchinsky & Slabakova (eds. 2005), Erteschik-Shir & Rapoport (eds. 2005), Dölling & Heyde-Zybatow (eds. 2007), Bowerman & Brown (eds. 2007), and Rothstein (ed. 2008). 2.1 Aristotelian categories, grammatical aspect and semantic universals “Aspect” is today established as an overt or a covert linguistic category, which applies to expressions at the lexical and phrasal level and, at its conceptual core, concerns
Johan van der Auwera & Hana Filip
their categorization with respect to the notion of some limit, boundary or end: namely, (i) its presence and contrastive absence, and (ii) just in case it is entailed or implicated, the question also arises whether it is also required that it be reached with respect to some reference/topic time or not, which brings them close to the study of tense and temporal logic. What is much debated is the question of how the distinctions in (i) and (ii) are encoded in the grammar of words and phrases in languages with very different grammatical structures, and whether drawing a line between (i) and (ii) is justified on empirical and theoretical grounds. There is an agreement that the distinctions under (i) at least concern “Aristotelian classes,” “aspectual classes,” or “Aktionsart(en)”, and also “lexical aspect”, just in case they are entailed by lexical predicates, and further subcategorized following Vendler’s (1957/67) four-way classification, or the tripartite state-process-event schema used in event semantics (cf. Bach 1981). Such Aristotelian categories are probably universal (Bach 2005), as is generally agreed. What is highly controversial is their relation to the formal categories of grammatical aspect – perfective, imperfective (and its subcategory progressive). The grammatical aspect is not a universal formal category. However, according to some at least, it is a universal semantic category, on a par with Aristotelian categories, and regarded as covert on the sentential (or propositional) level in languages that lack the grammatical category of aspect (see Kratzer 2004, among others). The main controversies revolve around the fundamental questions whether the semantic notions posited for the interpretation of the formal categories of the grammatical aspect in a particular language and cross-linguistically, are of the same ontological nature as the semantic notions of the Aristotelian type, and whether their contributions to the semantic interpretation of sentences can be captured by the same analytical tools. For some, drawing a line between (i) and (ii) makes no sense, and they use “aspect” for both, and possibly also interchangeably with “telicity.” On one implementation of this view, Romance and Slavic languages, for instance, are taken to manifest the following form-meaning correspondences: “perfectivity-telicity” and “imperfectivity-atelicity”. (For references see Sasse 2002, and also Section 2.2.2 below.) Others argue that the categories of the grammatical aspect concern the distinctions like those in (ii) and must be distinguished from the Aristotelian categories on formal and semantic grounds, and interpret the former as compositional operators taking scope over semantic structures characterized in terms of the latter (Filip 2000; Paslawska & von Stechow 2003). The field has reached a certain impasse with respect to these questions, which can be overcome by sharpening and extending the empirical basis against which the soundness of the currently advocated and deeply ingrained theoretical claims can be evaluated.
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
2.2 Theories of aspect The main distinguishing criterion runs along the semantics/syntax theoretical divide. Are the phenomena falling under aspect (in the broadest sense of (i) and (ii) above) ultimately grounded in lexical semantics of verbs or are they derived from a dedicated syntactic structure? 2.2.1 Semantics of aspect Semantic theories of aspect constitute a large and a fairly heterogeneous group that shares one key assumption: the telicity of sentences crucially depends on the meaning of their lexical head verbs. Much progress has been made in our understanding of the aspectually relevant meaning components lexicalized in verbs and how they influence the integration of verbs into sentences. To mention just a few examples: causation and agentivity (Davis & Demirdache 2000), measure function (Filip 2000, 2004), manner (Harley 2005; Levin & Rappaport Hovav 2008), path (Erteschik-Shir & Rapoport 2005), change of state (Levin & Rappaport Hovav 2005; Koontz-Garboden 2007). We also find new proposals for unaccusativity (Chierchia 2004), and event structure decompositional analysis (Rothstein 2004). At the same time, the focus has shifted to telicity data that go well beyond the phenonemon of aspectual composition, to telic sentences that are noncompositional and to the role played by pragmatic and cognitive factors in their derivation. Aspectual composition, as defined by Krifka (1992, 1998) and Dowty (1991) within a mereologically-based event semantic framework, is a very restricted phenomenon, if understood in its original sense as applying to sentences whose telicity interpretation is just a function of (strongly) compositional rules (i.e., rules that operate on independently motivated syntactic structures, and nothing else): cp. John ate an apple (telic) vs. John ate ice cream (atelic). Such sentences contain an argument (Strictly Incremental Theme) whose referent undergoes a gradual and permanent change of state in its extent/volume in the course of the event, and just in case it is also quantized, like the sortal term an apple, and just in case no other factors intervene, they are telic. This interaction is motivated via a homomorphism (a structure-preserving mapping) between the part structure of the denotation of the Strictly Incremental Theme argument and the part structure of the event, which is treated as an entailment of a verb. Aspectual composition in the above sense is a special, rather than a central, telicity case, because the vast majority of telic sentences are non-compositional. Nevertheless, some of the most promising new directions in the study of aspect emerged from the critical evaluation of Krifka-Dowty’s aspectual composition. It provides some of the basic analytical tools for modelling what is intuitively
Johan van der Auwera & Hana Filip
the main source of telicity: namely, a type of definite change of state evolving in separate stages, which can be distinguished and measured with respect to some “object” with a part structure: namely, the extent/volume of some participant (eat an apple), the extent of a path in the concrete physical space (to walk to the post office) and a property scale associated with some participant (to hammer the metal flat, to heat the water by 40 degrees). First, there is a substantial agreement that the telic interpretation of sentences involves an object-event mapping of some sort, but it must be freed from its exclusive tie to the Incremental Theme argument – and perhaps no verbs are to be lexically marked as incremental (cf. Levin & Rappaport Hovav 2005, and references therein) – allowing for its sources to be distributed over other parts of the grammar and also to be determined by external pragmatic and cognitive factors. Second, although Krifka’s notion of “quantization” is problematic (Filip 2000; Zucchi & White 2001), it capitalizes on the notion of an extensive measure function and other conceptual apparatus of the grammar of measurement, which help us understand what counts as an individuated discrete event in the denotation of telic predicates (Filip 2004, 2005b). These two ingredients come together, though they may not always be clearly separated or recognized as such, in proposals for a unified analysis of telic sentences. We find variants that are based either on a generalized topological path structure (Gawron 2005; Zwarts 2005; see already Krifka 1998) or on a scale (Winter 2006); and in some proposals, the homomorphism is explicitly taken to apply between the part structure of an event and the part structure of a scale (Beavers 2007; Wechsler 2005; Filip 2008 (based on Filip 1993/99), Filip & Rothstein 2005). Telicity, and closely related notions of “quantization”, “complete event”, “culmination”, “delimitation”, “boundedness” (and the like), is recast as an entailment or a conversational implicature that the maximal degree of the appropriate closed scale/path be reached. Such path/scale-based approaches to aspect are applied in the analysis of resultatives (Wechsler 2005; Beavers 2007), scalar verbs (cool, lengthen) (Kearns 2007), spatial predicates that have variable state/event interpretation (Gawron 2005), for example. They have at least three theoretical implications. First, the elements of meaning defining a scale/path can be lexicalized in verbs, in their roots or derivational morphology, or expressed by a variety of means externally to verbs. This seems to be an emerging parameter in the statement of semantically-based generalizations regarding the encoding of telicity in a particular language and across languages (for the Germanic/Slavic contrast see Filip & Rothstein 2005; Filip 2008). Germanic languages, for example, do not lexicalize them in their native root verbs, which may motivate the observation (see Kratzer 2004) that they have no native root lexical accomplishments, as Filip (2008) proposes. Second, patterns in which such aspectually relevant scalar meaning components and also arguably (but controversially)
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
an object-event homomorphism co-occur in particular verb classes cannot be easily aligned with Vendler’s classes or the tripartite event-process-state schema common in event semantics. Scalar verbs (e.g., cool, lengthen) and incremental verbs (e.g., eat, read) cannot be classified once and for all as either accomplishments/events or activities/processes. Third, what this approach also makes understandable is the well-known observation that virtually any atelic verb can be used as a basic building block of a telic sentence: namely, it is used in this way, provided some suitable scale/ path structure and its maximal endpoint can be identified based on the verb’s lexical meaning, linguistic context, possibly also in interaction with cognitive and pragmatic principles of interpretation. Related to the third point is the issue of non-compositionality of the vast majority of telic sentences, due to a variety of factors: for example, type-mismatches, variability verb meanings, a nontransparent meaning, among others. We have a number of competing proposals to address such non-compositional phenomena, including underspecification, ambiguity, which may be explained by general lexical rules or type-shifting operations (i.e., aspect shift and coercion), null morphology, and constructional approaches. Therefore, in order to arrive at a comprehensive account of telicity, scalar/path approaches must also address the following questions: What are the reliable descriptive generalizations about different types of compositional and non-compositional processes implicated in the derivation of the (a)telic interpretation of sentences in a particular language and cross-linguistically? What are the empirical predictions inherent in the proposals that capture them? What constitutes reliable empirical evidence that allows us to evaluate such proposals? 2.2.2 Syntax of aspect Syntactic theories of aspect are concerned with a narrower set of data than semantic (and pragmatic) ones, and mainly bearing on the nature of the categories that represent aspect, or “event structure”, within the VP. The result is a highly articulated VP structure, including VP shells and VP-internal subjects (Travis 2000) and lexical decompositions motivated by aspect (Folli & Ramchand 2005, among others), which can be viewed as representing “a return to Generative Semantics” (see Travis 2000); inner (VP) aspectuality structures, contrasted with outer (S) aspect(uality) (Verkuyl (2005, and many related proposals); a functional projection dedicated to aspect above the VP (Borer 2005), also implemented in terms of feature checking within Minimalism (Kratzer 2004; van Hout 2000). One of the key features they share is the prominence of the connection between telicity and direct object (DO), and possibly also unaccusativity (Schoorlemmer 2004). Related to the telicity-DO connection is the claim that languages parametrically differ in the syntactic (and morphological) strategies for the encoding of
Johan van der Auwera & Hana Filip
telicity, which fall into two main types: namely, (i) an object-marking strategy (e.g., Germanic languages, Finnish), with VP telicity taken to be predictably linked to morphology of the direct object (accusative case, presence of certain quantifiers, for example); (ii) a verb-marking strategy (e.g., Slavic languages), with VP telicity taken to be predictably linked to morphology of its perfective verb (prefixes). This general idea has been explored by a number of scholars (Borer 2005, and references therein), and support for it sought in first and second language acquisition (Kempchinsky & Slabakova 2005; Bowerman & Brown 2007). This hypothesis raises the following questions: Do languages only differ in the grammaticalization sources for the expression of telicity, without any substantial semantic and pragmatic differences? Is the parametric variation in the encoding of telicity limited to syntactic factors, or must semantic and pragmatic factors be taken into account, as well? Among the studies that bear on the plausibility of the syntactic telicity parameter, we may mention Borer (2005, and references therein) who argues in support of it, based on the English, Hebrew and Slavic (Russian and Czech) data, and Filip (2005a) against it, specifically based on the formal and semantic properties of Slavic verbal affixation. This discussion also generated much interest in the syntactic representation of Slavic affixation (cf. Svenonius 2004, and references therein). 2.3 Aspect and argument selection A lot of important debates continue to surround the role of aspect and the much narrower thematic property “Incremental Theme” (Dowty 1991) in argument selection (although the two may not always be clearly distinguished, causing considerable confusion) as well as the unergativity/unaccusativity (cf. Alexiadou et al. 2004; Erteschik-Shir & Rapoport 2005). Here, we find a difference in opinions that cuts across the semantic(-pragmatic) vs. syntactic theoretical divide in aspect theories, which is discussed in great detail in Levin & Rappaport Hovav (2005). (See also Hale & Keyser 2005.) 2.4 The notion of an “event” “Event” is a stock concept of aspect studies, but even the most basic ontological and referential questions related to its nature (e.g., individuation criteria, relation to objects and times) are still a source of much confusion. Syntacticians mostly avoid such questions, and explicitly define “event” and “event structure” in syntactic terms (Ritter & Rosen 2000, for example). Semanticists argue that there is no evidence supporting a syntactic category corresponding to aspect or event (Smith 2005), and instead try to get at their nature applying traditional philosophical
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
argumentation based on adverbial modification (Ernst 2000), causation (Eckardt 2000), coreferentiality (ter Meulen 2000), individuation of events with objects that do not persist through time (Pustejovsky 2000; Asher 2000), for example. Partee (2000) contributes to the clarification of the ways in which the notion of an “event” is used in linguistics, and how it differs from its understanding in philosophy, which presupposes a more conservative approach to its ontological commitments. 3. Mood and modality – functional approaches One feature of the functional scene is that there is a strong feeling that linguistics is getting more and better data. On the language-specific level we witness the exploitation of large electronic corpora, both diachronic and synchronic, allowing us to identify slow developments – a perfect example is Krug (2000), who documents the rise of a new class of English modals, including want to and have to. On the crosslinguistic level, corpora also exist, viz., parallel corpora, increasingly allowing finegrained frequency differences between languages, e.g., Nuyts (2001a) on English, Dutch and German, or Wärnsby (2006), who explores data mining techniques for a contrastive analysis of Swedish and English modals. Diepeveen et al. (2006) study modal expressions of variants of the same language (the Dutch of Belgium vs. the Dutch of the Netherlands). The last few years also saw the setting up larger crosslinguistic databases for typology, either for the world (WALS & Gusev 2005 for the imperative) or for a region (e.g., Enfield 2003 for mainland Southeast Asian modals) and just like typology saw an unsurge of the subfield of areal typology, some of the cross-linguistic work on modality takes this areal-typological perspective too (Hansen 2000; Enfield 2003; van der Auwera et al. 2005, In press). One could also see a growing amount of work on modality of languages other than the better described languages of Western Europe: e.g., Slavic (Hansen & Karlík eds. 2005), South-East Asia (Enfield 2003), Burmese (Vittrant 2004) or Japanese (Narrog 2007). The beginning of this century has seen an unsurge of interest in expression of modality other than modal verbs, esp. particles and adverbs (Nuyts 2001a; Aijmer & Simon-Vandenbergen. eds. 2006; Simon-Vandenbergen & Aijmer 2007) or adjectives (Nuyts 2001a; Vinzerich 2007), or in modals verbs other than the central ones (so not verbs like English may, must or can, but rather need (Loureiro Porto 2006; Taeymans 2006). Another focus of interest has been the more general concept of irrealis (Lander et al. (eds) 2004; Verstraete 2005); this further relates to the study of conditionality, which saw at least two studies of importance (Declerck & Reed 2001; Dancygier 2006). And it also connects to mood, with a focus of research on irrealis moods such as the French “conditionnel” (Dendale & Tasmowski eds. 2001b) or the German “Konjunktiv” (Mortelmans 2000).
Johan van der Auwera & Hana Filip
The other mood that came to the foreground is the imperative. 2001 saw the publication of the English translation of a 1992 collection on the typology of the imperative (Xrakovskij 2001). Later typological work includes van der Auwera et al. (2004), Gusev (2005), Dobrushina & Goussev (2005), and van der Auwera et al. (In press). A collection that focusses on Europe is Rothstein & Thieroff (2008). An in depth study language-specific study that makes maximal use of electronic corpora of English is De Clerck (2006), and another one comparing the use of the Dutch imperative with that of modal verbs is Nuyts et al. (2007). An important topic is evidentiality, mostly understood to be the expression of the source of the speaker’s knowledge. Two questions have received a lot of attention: (i) can evidentiality and modality form one system? (ii) should one restrict the notion of evidentiality to the grammaticalized expression of the source of knowledge. There is no consensus on this point. Thus both Aikhenvald (2004) and Diewald & Smirnova (In press) focus on what makes evidentiality different from modality, but whereas Aikenvald (2004) restricts the term “evidentiality” to grammatical markers, Diewald & Smirnova (In press) do not. In that respect, they are like Boye (2006), Cornillie (2007) and Guentchéva and Landaburu (eds). (2007), but these linguists stress what modality and evidentiality have in common (see also some contributions in Dendale & Tasmowski eds. 2001a). Just like for modality work started with verbs (e.g., Diewald 2001; Cornillie 2007) but has progressively dealt with other markers (Wiemer 2006; Squartini ed. 2007). A final focus of attention is the role of subjectivity and, from a diachronic perspective, subjectification. It relates to both modality (Verstraete 2001) and evidentiality (Nuyts 2001b). An overview volume is Athanasiadou et al. (eds. 2006). Though this survey does not deal with either tense and aspect, it may be mentioned that some of the recent work on these domains strongly stresses the relation to modality or the need for an integrated perspective (e.g., Brisard ed. 2002; Gosselin 2005; Langacker 2003; Pietandrea 2005; Trnavac 2006; Boogaart 2007). 4. Envoi The preceding overview has made a distinction between functional and formal approaches and this at least suggests that linguistics is a rather fragmented discipline. While it is correct that the functional – formal divide remains as important as it was in the last third of the preceding century, the dividing line is not always as clear as the above discussion suggests. That is rather good. From that point of view, the fact – assuming that we are roughly correct – that the first seven years of this century has seen formalist progress on aspect and functionalist progress on mood
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
and modality could be given a positive interpretation: to some extent the progress is that of linguistics as such, not only formalist or functionalist linguistics. It is also of interest to observe what publication channels have been used. The reference list counts 11 collections of articles (other than collections of which the articles get mentioned in their own right), 16 journal articles, 24 monographs and 49 papers in collections. This century increasingly uses bibliometrical evaluations, but the latter so far only count journal articles and this practice misrepresents at least part of the field.
References Aijmer, Karin & Anne-Marie Simon-Vandenbergen (eds). 2006. Pragmatic markers in contrast. Amsterdam: Elsevier. Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press. Alexiadou, Artemis, Elena Anagnostopoulou & Martin Everaert (eds). 2004. The unaccusativity puzzle: Explorations of the syntax-lexicon. Oxford: Oxford University Press. Asher, Nicholas. 2000. Events, facts, propositions, and evolutive anaphora. In James Higginbotham, Fabio Pianesi & Achille Varzi (eds), 123–150. Athanasiadou, Angeliki, Costas Canakis & Bert Cornillie (eds). 2006. Subjectification. Various paths to subjectificaton. Berlin: Mouton de Gruyter. Auwera, Johan van der, Nina Dobrushina & Valentin Goussev. 2004. A semantic map for Imperative-Hortatives. In Dominique Willlems, Bart Defrancq, Timothy Colleman & Dirk Noël (eds), Contrastive analysis in language. Identifying linguistic units of comparison, 44–66. Basingstoke: Palgrave. Auwera, Johan van der, Andreas Ammann with Saskia Kindt. 2005. Modal polyfunctionality and Standard Average European. In Alex Klinge & Henrik Høeg Müller (eds), Modality. Studies in form and function, 247–272. London: Equinox. Auwera, Johan van der, Ewa Schalley & Gunther de Vogelaer. In ������������������������������� press. Analogie und die Verbreitung der verbalen Kongruenz bei Imperativen, Konjunktionen und Antwortpartikeln. In Rüdiger Harnisch (ed.), Prozesse sprachlicher Verstärkung, Berlin: Mouton. Bach, Emmon. 1981. On time, tense and aspect: an essay in metaphysics. In Peter Cole (ed.). Radical pragmatics, 63–81. New York City: Academic Press. Bach, Emmon. 2005. Eventualities, grammar and language diversity. In Henk Verkuyl, Henriette de Swart, & Angeliek van Hout (eds), 167–180. Beavers, John. 2007. Scalar complexity and the structure of events. In Johannes Dölling & Tatjana Heyde-Zybatow (eds), 245–268. Binnick, Robert. 2008. Project on the bibliography of tense, verbal aspect, aktionsart, and related areas. http://www.utsc.utoronto.ca/~binnick/TENSE/ Boogaart, Ronny. 2007. The past and perfect of epistemic modals. In Louis de Saussure, Jacques Moeschler & Genoveva Puskás (eds), Recent advances in the syntax and semantics of tense, aspect and modality, 47–69. Berlin: Mouton de Gruyter. Borer, Hagit. 2005. Structuring sense. Oxford: Oxford University Press. Bowerman, Melissa & Penelope Brown (eds). 2007. Crosslinguistic perspectives on argument Structure. New York: Routledge.
Johan van der Auwera & Hana Filip Boye, Kasper. 2006. Epistemic meaning: a cross-linguistic study. Copenhagen: University of Copenhagen dissertation. Brisard, Frank (ed.). 2002. Grounding: the epistemic footing of deixis and reference. Berlin: De Gruyter. Chierchia, Gennaro. 2004. A semantics for unaccusatives and its syntactic consequences. In Artemis Alexiadou, Elena Anagnostopoulou & Martin Everaert (eds), 22–59. Cornillie, Bert. 2007. Epistemic modality and evidentiality in Spanish (semi-)auxiliaries. A cognitive-functional approach. Berlin: Mouton de Gruyter. Dancygier, Barbara. 2006. Conditionals and prediction: Time, knowledge and causation in conditional constructions. Cambridge: Cambridge University Press. Davis, Henry & Hamida Demirdache. 2000. On lexical verb meanings: Evidence from Salish. In Carol Tenny & James Pustejovsky (eds), 97–142. De Clerck, Bernard. 2006. The imperative in English: a corpus-based, pragmatic analysis. Gent: University of Gent dissertation. Declerck, Renaat & Susan Reed. 2001. On conditionals. A comprehensive empirical analysis. Berlin: Mouton de Gruyter. Dendale, Patrick & Liliane Tasmowski (eds). 2001a. Theme issue on evidentiality, Journal of Pragmatics 33: 339–464. Dendale, Patrick & Tasmowski, Liliane (eds). 2001b. Le conditionnel en français. Metz: Université de Metz. Diepeveen, Janneke, Ronny Boogaart, Jenneke Brantjes, Pieter Byloo, Theo Janssen & Jan Nuyts. 2006. Modale uitdrukkingen in Belgisch-Nederlands en Nederlands-Nederlands: Corpusonderzoek en enquête. Amsterdam: Stichting Neerlandistiek. Diewald, Gabriele. 2001. Scheinen-Probleme: Analogie, Konstruktionsmischung und die Sogwirkung aktiver Grammatikalisierungskanäle. In Marga Reis & Reimar Müller (eds), Synchronie und Diachronie der Modalverben, 87–110. (Linguistische Berichte, Sonderheft 9). Diewald, Gabriele & Elenan Smirnova. In press. The German evidential constructions and their origins. A corpus based analysis. In Frank Brisard, Tanja Mortelmans & Walter de Mulder (eds), Chronos 7. Amsterdam: Rodopi. Dobrushina, Nina & Valentin Goussev. 2005. Inclusive imperative. In Elena Filimonova (ed.), Clusivity. Typology and case studies of the inclusive-exclusive distinction, 179–211. Amsterdam: Benjamins. Dölling, Johannes & Tatjana Heyde-Zybatow (eds). 2007. Event structures in linguistic form and interpretation. Berlin: Mouton de Gruyter. Dowty, David R. 1991. Thematic proto-roles and argument selection. Language 67: 547–619. Eckardt, Regine. 2000. Causation, contexts, and event individuation. In James Higginbotham, Fabio Pianesi, & Achille Varzi (eds), 105–122. Enfield, Nick J. 2003. Linguistic epidemiology. Semantics and grammar of language contact in mainland Southeast Asia. London: Routledge Curzon. Ernst, Thomas. 2000. Manners and events. In Carol Tenny & James Pustejovsky, 335–358. Erteschik-Shir, Nomi & Tova Rapoport (eds). 2005. The syntax of aspect. Deriving thematic and aspectual interpretation. Oxford: Oxford University Press. Erteschik-Shir, Nomi & Tova Rapoport. 2005. Path predicates. In Nomi Erteschik-Shir & Tova Rapoport (eds), 65–86. Filip, Hana. 1993/1999. Aspect, Eventuality Types, and Noun Phrase Semantics. New York: Routledge. (A revised version of the Ph.D. Thesis University of California at Berkeley 1993.) Filip, Hana. 2000. The quantization puzzle. In Carol Tenny & James Pustejovsky (eds), 3–60.
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
Filip, Hana. 2004. Prefixes and the delimitation of events. Journal of Slavic Linguistics 11(1): 55–101. Filip, Hana. 2005a. The telicity parameter revisited. Semantics and Linguistic Theory 14: 92–109. Filip, Hana. 2005b. Measures and indefinites. In Gregory N. Carlson & Francis Jeffrey Pelletier (eds), Reference and quantification: The Partee effect, 229–288. Stanford, CA: CSLI Press. Filip, Hana. 2008. Events and maximalization: The case of telicity and perfectivity. In Susan Rothstein (ed.), Theoretical and crosslinguistic approaches to the semantics of aspect, 217–256. Amsterdam: Benjamins. Filip, Hana & Susan Rothstein. 2005. Telicity as a semantic parameter. In James Lavine, Steven Franks, Hana Filip & Mila Tasseva-Kurktchieva (eds), Formal approaches to Slavic linguistics XIV, 139–156. Ann Arbor, MI: University of Michigan Slavic Publications. Folli, Raffaella & Gillian Ramchand. 2005. Event structure composition: the case of goal of motion and resultative constructions in Italian and Scottish Gaelic. In Henk Verkuyl, Henriette de Swart & Angeliek van Hout (eds), 81–106. Gawron, Jean Mark. 2005. Generalized Paths. Semantics and Linguistic Theory 15: 76–94. Gosselin, Laurent. 2005. Temporality et modalité. Bruxelles: De boeck duculot. Guentchéva, Zlatka & Jon Landaburu (eds). 2007. L’énonciation médiatisée II. Le traitement épistémologique de l’information: illustrations amérindiennes et caucasiennes. Louvain: Peeters. Guéron, Jacqueline & Jacqueline Lecarme (eds). 2004. The syntax of time. Cambridge, MA: MIT Press. Gusev, Valentin Ju. 2005. Tipologija specializirovannyx glagol’nyx form imperativa (Typology of specialized verbal imperative forms). Moscow: Lomonosov University dissertation. Hale, Ken & Samuel Jay Keyser. 2005. Aspect and the syntax of argument structure. In Nomi Erteschik-Shir & Tova Rapoport (eds), 11–41. Hansen, Björn. 2000. The German modal verb “müssen’’ and the Slavonic languages: The reconstruction of a success story. Scando-Slavica 46: 77–93. Hansen, Björn & Petr Karlík (eds). 2005. Modality in Slavonic languages. New perspectives. München: Sagner. Harley, Heidi. 2005. How do verbs get their names? Denominal verbs, manner incorporation, and the ontology of verb roots in English. In Nomi Erteschik-Shir & Tova Rapoport (eds), 42–64. Higginbotham, James, Fabio Pianesi, & Achille Varzi (eds). 2000. Speaking of events. Oxford: Oxford University Press. Hout, Angeliek van. 2000. Projection based on event structure. In Martin Everaert, Peter Coopmans & Jane Grimshaw (eds). Lexical specification and insertion, 397–422. Amsterdam: Benjamins. Kearns, Kate. 2007. Telic senses of deadjectival verbs. Lingua 117: 26–66. Kempchinsky, Paula & Roumyana Slabakova (eds). 2005. Aspectual inquiries. Dordrecht: Kluwer. Koontz-Garboden, Andrew. 2007. States, changes of states, and the monotonicity hypothesis. Stanford CA: Stanford University dissertation. Kratzer, Angelika. 2000. Building statives. Proceedings of the 26th Annual Meeting of the Berkeley Linguistic Society. 385–399. Kratzer, Angelika. 2004. Telicity and the meaning of objective case. In Jaqueline Guéron & Jaqueline Lecarme (eds), The syntax of time, 389–424. Cambridge, MA: MIT Press. Krifka, Manfred. 1992. Thematic relations as links between nominal reference and temporal Constitution. In Ivan A.sag & Anna Szabolcsi (eds), Lexical Matters, 29–53. Stanford, CA: CSLI.
Johan van der Auwera & Hana Filip Krifka, Manfred. 1998. The origins of telicity. In Susan Rothstein (ed.), Events and grammar, 197–235. Dordrecht: Kluwer. Krug, Manfred G. 2000. Emerging English modals. A corpus-based Study of grammaticalization. Berlin: Mouton de Gruyter. Lander, Yuri Y., Vladimir Plungian & Anna Y. Urmanchieva (eds). 2004. Issledovanija po teorii grammatiki. 3: Irrealis i irreal’nost’. (Studies in the theory of grammar. 3. Irrealis and irreality) Moscow: Gnosis. Langacker, Ronald. 2003. Extreme subjectification: English tense and modals. In Hubert Cuyckens, Thomas Berg, René Dirven & Klaus-Uwe Panther (eds), Motivation in language: Studies in honor of Günter Radden, 3–26. Amsterdam: Benjamins. Levin, Beth & Malka Rappaport Hovav. 2005. Argument realization. Cambridge: Cambridge University Press. Levin, B. & M. Rappaport Hovav. 2008. Lexical Conceptual Structure. In Claudia Maienborn, Klaus von Heusinger & Paul Portner (eds), Semantics: An International Handbook of Natural Language Meaning, Mouton de Gruyter, Berlin. Loureiro Porto, Lucia. 2005. The semantic predecessors of need: From Old to Early Modern English. Santiago de Compostela: University of Santiago de Compostela dissertation. Meulen, Alice G.B. ter. 2000. How to tell events apart. In Carol Tenny & James Pustejovsky (eds), 377–392. Mortelmans, Tanja. 2000. Konjunktiv II and epistemic modals in German: a division of labour. In Ad Foolen & Frederike van der Leek (eds), Constructions in cognitive linguistics, 191–215. Amsterdam: Benjamins. Narrog, Heiko. 2007. Modality and grammaticalization in Japanese. Journal of HistoricalPragmatics 8: 269–294. Nuyts, Jan. 2001a. Epistemic modality, language, and conceptualization. A cognitive-pragmatic perspective. Amsterdam: Benjamins. Nuyts, Jan. 2001b. Subjectivity as an evidential dimension in epistemic modal expressions. Journal of Pragmatics 33: 383–400. Nuyts, Jan, Pieter Byloo & Janneke Diepeveen. 2007. Mogen en moeten en de relaties tussen deontische modaliteit en modus. Nederlandse Taalkunde 12. 153–174. Partee, Barbara. 2000. Some remarks on linguistic uses of the notion of “Event’’. In Carol Tenny & James Pustejovsky (eds), 483–495. Paslawska, Alla & Arnim von Stechow. 2003. Perfect readings in Russian. In Artemis Alexiadou, Monika Rathert & Arnim von Stechow (ed.), Perfect explorations, 307–362. Berlin: Mouton. Pietandrea, Paola. 2005. Epistemic modality. Functional properties and the Italian system. Amsterdam: Benjamins. Pustejovsky, James. 2000. Events and the semantics of opposition. In Carol Tenny & James Pustejovsky (eds), 445–482. Rappaport Hovav, Malka. 2008. Lexicalized meaning and the internal temporal structure of events. In Susan Rothstein (ed.). Theoretical and crosslinguistic approaches to the semantics of aspect, 61–82. Amsterdam: Benjamins. Rappaport Hovav, Malka & Beth Levin. 2001. An event structure account of English resultatives. Language 77: 766–97. Rappaport Hovav, Malka & Beth Levin. 2005. Change of state verbs: Implications for theories of argument projection. In Erteschik-Shir & Rapoport (eds), 274–286. Ritter, Elizabeth & Sara Rosen. 2000. Event structure and ergativity. In Carol Tenny & James Pustejovsky (eds), 187–238.
(Tense,) Aspect, mood and modality – an imperfect 2008 state of the art report
Rothstein, Björn & Rolf Thieroff (eds). 2008. Mood in the languages of Europe. Amsterdam: Benjamins. Rothstein, Susan. 2004. Structuring events. Oxford: Blackwell. Rothstein, Susan (ed.). 2008. Theoretical and crosslinguistic approaches to the semantics of aspect. Amsterdam: Benjamins. Sasse, Hans-Jürgen. 2002. Recent activity in the theory of aspect: Accomplishments, achievements or just non-progressive state. Linguistic Typology 6. 199–271. Schoorlemmer, Maaike. 2004. Syntactic Unaccusativity in Russian. Artemis Alexiadou, Elena Anagnostopoulou & Martin Everaert (eds), 207–242. Simon-Vanderbergen, Annemarie & Karin Aijmer. 2007. The semantic field of modality certainty. A corpus-based study of English adverbs. Berlin: Mouton de Gruyter. Smith, Carlota. 2005. Navajo event structure and morphosyntax. In Nomi Erteschik-Shir & Tova Rapoport (eds), 190–211. Squartini, Mario (ed.). 2007. Theme issue on “Evidentiality between lexicon and grammar”, Rivista de Linguistica 19(1). Svenonius, Peter. 2004. Slavic prefixes inside and outside VP. Nordlyd 32: 205–253. Taeymans, Martine. 2006. An investigation into the emergence and development of the verb need from Old to Present-day English: a corpus-based approach. Antwerp: University of Antwerp dissertation. Tenny, Carol. 2000. Core events and adverbial modification. In Carol Tenny & James Pustejovsky (eds), 285–334. Tenny, Carol & James Pustejovsky (eds). 2000. Events as grammatical objects, from the combined perspectives of lexical semantics, logical semantics and syntax. Stanford, CA: CSLI Press. Travis, Lisa. 2000. Event structure in syntax. In Carol Tenny & James Pustejovsky (eds), 145–185. Trnavac, Radoslava. 2006. Aspect and subjectivity in modal constructions. Utrecht: Landelijke Onderzoeksschool Taalkunde. Vendler, Zeno. 1957. Verbs and times. Philosophical Review 56. 143–160. Reprinted in Linguistics in philosophy, 1967, 97–121. Ithaca, NY: Cornell University Press. Verkuyl, Henk. 2005. Aspectual composition: Surveying the ingredients. In Henk Verkuyl, Henriette de Swart & Angeliek van Hout (eds), 19–39. Verkuyl, Henk, Henriëtte de Swart & Angeliek van Hout (eds) 2005. Perspectives on aspect. Dordrecht: Springer. Verstraete, Jean-Christophe. 2001. Subjective and objective modality: Interpersonal and ideational functions in the English modal auxiliary system. Journal of Pragmatics 33: 1505–1528. Verstraete, Jean-Christophe. 2005. The semantics and pragmatics of compositive mood marking. The non-Pama-Nyungan languages of northern Australia. Linguistic Typology 9: 223–268. Vinzerich, Aude. 2007. La sémantique du possible: approche linguistique, logique et traitement informatique dans les textes. Paris: Univerisity of Paris dissertation. Vittrant, Alice. 2004. La modalité et ses corrélats en birman dans une perspective comparative. Paris: University of Paris dissertation. WALS = Haspelmath, Martin, Matthew S Dryer, David Gil & Bernard Comrie (eds). 2005. The world atlas of language structures. Oxford: Oxford University Press. Wärnsby, Anna. 2006. (De)coding modality. The case of must, may, måst, and kan. Lund: Lund University. Wechsler, Stephen. 2005. Resultatives under the event-argument homomorphism model of telicity. In Nomi Erteschik-Shir & Tova Rapoport (eds), 255–273. Wiemer, Björn. 2006. Particles, parentheticals, conjunctions and prepositions as evidentiality markers in contemporary Polish. Studies in Polish linguistics 3: 5–67.
Johan van der Auwera & Hana Filip Winter, Yoad. 2006. Closure and telicity across categories. Semantics and Linguistic Theory 16. Xrakovskij, Victor S. 2001. Typology of imperative constructions. München: Lincom Europa. Zucchi, Sandro & Michael White. 2001. Twigs, sequences and the temporal constitution of predicates. Linguistics and Philosophy 24: 223–270. Zwarts, Joost. 2005. Prepositional aspect and the algebra of paths. Linguistics and Philosophy 28(6): 739–779.
Syntax The state of the art Farrell Ackerman UCSD, USA
James P. Blevins
Cambridge University, UK The aim of science is not to open the door to infinite wisdom, but to set a limit to infinite error. (Bertolt Brecht, Galileo) … individuals are quite stupid compared to the complexity of the problems we aspire to solve … All anyone can hope to do is to make canny simplifications that do minimum damage to understanding. (Peter J. Richerson & Robert Boyd 2005: 248)
1. Introduction In the exploration of syntax, as in all scientific investigation, there are phenomena and there are theories that try to make sense of them, with the theories serving as filters on which phenomena seem central and instructive, and which phenomena seem peripheral or even epiphenomenal. Syntax, viewed from the perspective of basic phenomena, investigates the organization of “words” into larger units and constructions.1 Key properties of this organization include the arrangement of words into linear sequences and the hierarchical dependencies displayed by the units into which they are clustered. In order to generalize beyond individual instances, syntactic approaches abstract concrete phrasal combinations into exemplary types and hypothesize that these abstracted types serve as a model for new, specific combinations of words that define well-formed sentences. The . There is a long literature on the grammatically relevant notion of “word” (see Robins 1959 for an early discussion of the notion “grammatical word” and Dixon & Aikhenvald 2002 for one proposed typology). Syntacticians working within generative theory broadly construed, have tended to assume that lexical items can only be expressed by single wordforms. However, theorists working within other traditions recognize periphrastic expressions of lexical representations and distinguish these multiword expressions from ordinary syntactic phrases (see Curme 1935 and Ackerman & Stump 2004). To some degree the empirical domain of syntactic theory depends on how “words” are defined and, more generally, how the relation between syntax and morphology is construed.
Farrell Ackerman & James P. Blevins
primary evidence comes from the sequencing of elements, their aggregation into constituents, and the permutability of constituents. In the Samoyed language Forest Enets simple distributional evidence suggests the existence of Nouns and Verbs, Noun phrases, and a basic clausal structure of Subject Object Predicate: the Predicate can be a Noun as in (1) or Verb, as in (2),2 where both are inflected for person/number agreement with the subject, and for tense, yielding the morphotactic order V-agr-tns:3 (1) to d’od’igun kasa äči-d-uš that time.prol man youngster-2sg-pst In those days you were a young boy (2) uu ši’ toida-d-uš 2sg 1sg.acc ask-2sg-pst You asked me.
These are what Bloomfield referred to as ‘‘conventional grammatical arrangements’’ and, as Matthews (2007: 2) notes, such arrangements have been traditionally called constructions: The term ‘‘construction’’ was in origin the equivalent of the Ancient Greek ‘‘suntaxis’’ meaning literally ‘‘arrangement together’’… The sense is, in the words of Johnson’s dictionary twelve centuries later, ‘‘the putting of words, duly chosen, together in such a manner as is proper to convey a complete sense.’’ … The aim of syntax, as the study of words ‘‘chosen’’ and ‘‘put together’’, in a ‘‘proper manner’’, is to analyse the different constructions of a language and show how they work together to ‘‘convey the sense’’ of sentences as a whole.
Research on cross-species communication systems from a range of different theoretical perspectives (Anderson 2004, Deacon 1997) converges on the view that there is a fundamental split between humans and other, even cognitively impressive, creatures. What seems uniquely human is the capacity for combining meaningless sounds into words, i.e., meaningful combinations, as well as systematically combining these words into larger meaningful units. Humans don’t just do more, but do more quite differently. Many of the fundamental analytic assumptions of Post-Bloomfieldian structuralist analysis in conjunction with varying Chomskyan hypotheses and construals of syntax have been construed as conforming to Boyd and Richerson’s canny simplifications that do little damage to understanding. Moreover, there has been a tendency to interpret them as promising, in Brecht’s phrase, to “open the door to infinite wisdom” concerning the nature of the human
. See section 3 for discussion of tense on non-verbal elements as in (1). . Examples from Florian Siegl based on fieldwork with Leonid Bolin and Yelizaveta Bolina.
Syntax: The state of the art
language capacity, language learnability, and, more broadly, to offer insights into humans as distinctively linguistic creatures.4 The question for syntacticians, of course, is what particular ensembles of canny simplications achieve the correct balance between better insight and ineluctable error. Parallel to the Chomskyan tradition there have been other approaches to grammar with consequences for the interpretation of larger issues such as language acquisition, language change, connections with cognitive neuroscience: these have adopted quite different guiding assumptions. Given this, Newmeyer (2005: 197) assesses the state of syntax in the following stark terms: In syntax, the rival approaches are just that – for the most part at root incom patible. The move from Government and Binding to the Minimalist Program or from GPSG to HPSG entailed abandoning central hypotheses of the earlier models. Only an intellectual schizophrenic or anarchist write an LFG paper one year and a categorial grammar paper the next, without having severe qualms about the adequacy of the former and having been intrigued by the possibilities of the latter. And it goes without saying that there exists a plethora of ‘‘functionalist’’ and ‘‘cognitive’’ approaches to syntax that do not share even the questions that formal approaches ask, much less the answers. As a consequence, syntax is in a highly fragmented state.
We will argue that while the state of syntactic theory is “highly fragmented” with favored theoretical constructs, familiar methodologies, and consensus assumptions all in play, the prospects for syntactic research are arguably brighter than ever, providing data and problems for a new era of theory construction based on the detailed study of lesser studied languages and the utilization of quantitative, computational, and experimental techniques. These new empirical and methodological emphases may better facilitate, among other things, addressing such enigmatic problems as native language acquisition in accordance with earlier insights: In the natural acquisition of our native language, the rules as such are not given, only a number of examples. We hear little by little a number of sentences which are constructed in the same way and therefore constitute a group. The memory of the particular content of the individual sentences can become ever more faded, the common element is newly reinforced through repetition, and so it is that the rule is unconsciously abstracted from these models. (Paul 1970: 111) Language acquisition is based on the ability to abstract from a corpus of sentences sharing a certain structural pattern, and to construct, from the old materials, new sentences conforming to this pattern. (Chomsky 1975: 131)
. See P.H. Matthews (1993) and P. Seuren (1998) for extensive and critical discussion and Lasnik 2000 et al. for an alternative historical perspective that is more sympathetic to the development of Chomskyan theory.
Farrell Ackerman & James P. Blevins
Section 2 identifies some essential assumptions and constructs in the approaches to syntactic analysis. We delineate a landscape of competing views that are likely to be transformed by the types of directions evident in empirical research discussed in section 3 and the conceptual and methodological issues discussed in section 4. Section 3 presents two phenomena and uses them to typify the sorts of challenging issues that arise from detailed cross-linguistic investigation. In the face of such data many conflicts among contemporary syntactic theories begin to look somewhat parochial, suggesting that a radical reconceptualization of assumptions, methods, and explanatory constructs may be required. Thus, section 4 focuses on some new directions that appear to be transforming theoretical syntax, reconnecting it with directions that anteceded American structuralist tradition. These earlier traditions were informed by insights which were difficult to develop without modern quantitative and experimental tools, but were also foreclosed from consideration by powerful analytic prejudices against the need to appeal to such methods. In some ways, the present vigor in the field of syntax can be seen as revisiting and building upon Kaplan & Bresnan’s (1982) desideratum for a “strong competence hypothesis” which was central to the early breach between generative lexical versus non-lexicalist approaches. On their interpretation linguistic theories should be evaluated not only in terms of their ability for parsimonious and comprehensive analysis of language data (however this is finally calculated), but with respect to how the ingredients of such analyses relate to research results in domains independent of language analysis, such as acquisition and sentence processing. This, they suggested, would encourage the development of a syntactic theory responsive to relevant results in other disciplines and less inclined toward increasingly abstract constructs based solely on linguistic considerations and, therefore, removed from plausibility evaluations from other disciplines. Throughout this short overview we focus on shifting ways of seeing syntax and suggest that they are more important than any technical assumptions, since, in fact, they ultimately motivate such assumptions. 2. S yntactic ways of seeing and consequences for interdisciplinary research A predominant and influential perspective in linguistic theory construction, particularly as it relates to syntax, has been the notion that human language is utterly special, the phylogenetic product of some specific biological endowment that also guides ontogenetic development in every child. This research program is associated with the transformationalist work of Noam Chomsky and its most recent hypothesis is formulated as the Minimalist Program. We will refer to this tradition as constituting the central dogma of language analysis: it represents what
Syntax: The state of the art
Culicover and Jackendoff (2005), hence C & J, refer to as mainstream generative grammar (MGG). They characterize it as consisting of: Structural Uniformity: An apparently defective or misordered structure is actually a distorted regular form. Interface Uniformity: The syntax-semantics interface is maximally simple, in that meaning maps transparently onto syntactic structure; and it is maximally uniform, so that the same meaning always maps onto the same syntactic structure. Derivational Uniformity: Where possible, the derivations of sentences are maximally uniform. (2005: 46 & 47)
Table 1 One of the abiding goals of MGG has been to identify classes of (abstract) s tructures and principles which provide the deductive basis for the generation of language specific surface patterns. A central role in this approach is played by what C & J term ‘‘hidden’’ levels of analysis. C & J’s argument against ‘‘hidden’’ levels echoes Hockett’s (1987) discussion of ‘‘the great agglutinative fraud’’ with respect to morphological analysis. Hockett came to regard ‘‘item and arrangement’’ analysis5 (which he had been instrumental in developing) as an essentially formal exercise in which an idealized agglutinative structure was foisted onto a language by ‘devising an “agglutinative analog” of the language and formulating rules that would convert expressions in that language into the shapes in which they are actually uttered’ (Hockett 1987: 83). He observes that the establishment of the agglutinative ideal within the generative paradigm encouraged a more abstract and theory-internal style of analysis. Theoretical analyses ceased to address the facts of a given language directly, but instead constructed idealized analogues, which could then be brought into correspondence with the actual patterns in the language in ways presently construed as in Table 1. Language-particular patterns, especially those not denominated as “core” and often found in lesser-studied languages, exerted a less formative influence on the primitives and constructs of linguistic theories. Hence, the focus of linguistic analysis shifted from the study of patterns of cross-linguistic variation to the study of the properties of abstract analogues and the nature of the devices needed to mediate between these idealized forms and attested surface patterns. . A morpheme-based perspective, though somewhat more abstract, is still assumed in much of contemporary morphological work within the Chomskyan research paradigm. Departures from structuralist models of morphemic analysis tend to concern matters of implementation, as in the case of the adoption of a realizational view of morphemic exponence within Distributed Morphology (Halle & Marantz 1993). For a general taxonomy of morphological approaches, see chapter 1 of Stump 2001.
Farrell Ackerman & James P. Blevins
The effect of this was to adapt Post-Bloomfieldian structuralist assumptions and contemporize them into what Elman et al. (1996) refer to as representational nativism. Syntactic structures and principles for relating them to one another are posited as providing core properties of grammar, accessible without learning, but triggered by experience.6 The singularity of human languages as it is filtered through the central dogma has led to a host of folkloric constructs within cognitive neuroscience. One prominent example concerns the arguments from the Poverty of the Stimulus that are proposed in order to explain how humans can develop their subtle knowledge of complex language phenomena despite the presumed absence of cues and clues from their experience. More general still is the postulation of Universal Grammar, which is taken to supply the grammatical distinctions in the child’s startstate that experience allegedly cannot provide.7 A perennial source of explanation in this tradition has been hierarchical, configurational representations, which are extended from classic distributional analyses of words into phrases and phrases into clauses, into other domains of linguistic analysis such as morphology, semantics and discourse. This representational reductionism, when augmented by increasingly abstract contraints on the licensing and permutation of structures defines what Jackendoff (2002: 107) has referred to as the syntactocentric gambit. Theories designed to address questions derived from this perspective have sought a range of answers. Recently, proponents have begun to speculate that their results, putatively facilitated by the repudiation of language specific and construction specific assumptions, are concordant with research in developmental and evolutionary biology:8
. See Stiles 2008 for a parallel contrast between native versus nurture assumptions in psychology and arguments for how concerns for biological plausability can clarify similar stalemates in that discipline. . It is important to distinguish between claims about a distinctively human language capability or faculty versus claims about the content attributed to this faculty in the form of universal grammar. Where views differ considerably is what specifically can be attributed to this faculty (Scholz & Pullum 2006; Clark & Lappin; To appear; Goldberg 2006 & Snyder 2007.) . While we are not competent to engage in compelling speculation, it does seem that biolinguistics offers only one perspective among many concerning the relevance of biology to language analysis. The stance of the central dogma of linguistics appears to parallel what has been called the central dogma of molecular biology (Gottlieb 2000a: 93). On Gottlieb’s account the central dogma is associated with two assumptions that theoretical linguists will find familiar and appears to be the perspective adopted by Chomskyan linguists when articulating their program of biolinguistics (Chomsky 2005): there is a deterministic relation between the
Syntax: The state of the art
Abandonment of the construction-based view of earlier (and traditional) work had further consequences for the biolinguistic program. In the earlier framework, not only rules but also UG principles were expressed in terms of grammatical constructions (islands, specified-subject and other constraints on operations, Emonds’s structure-preserving hypothesis, filters, etc.), all inherently specific to language, without even remote counterparts in other biological systems. Within the P&P framework, the basic computational ingredients are considerably more abstract (locality, minimal search, basic recursion, etc.), and it becomes quite reasonable to seek principled explanation in terms that may apply well beyond language, as well as related properties in other systems. (Chomsky 2005: 7)
An unquestionable virtue of this paradigm has been the identification of particular objects of inquiry, i.e., specific phenomena such as scopal effects, binding, filler gap dependencies, headedness, etc., and the articulation of research questions associated with these constructs. As must happen, this has encouraged certain ways of seeing and discouraged, even foreclosed, others. The adaptability of this theoretical perspective in the face of apparently recalcitrant facts and critical evaluation is impressive, paralleling the adaptability of the Modern Synthesis in biology, exemplary of the central dogma of molecular genetics concerning the relation of genes to phenotypes, as described by Robert G.B. Reid. He offers the following genome and phenotypic or behavioral expression and the variety of types is associated by what is referred to as its reaction-range: The reaction-range concept … presumes that the genotype imposes a priori limits (a range) on the expression of a phenotype’ (Platt & Sanislow 1988), such that the phenotype has upper and lower bounds that cannot be transcended. (Gottlieb 2000b: 7) Gottlieb contrasts this traditional perspective with the current views that are yielding a modern reconceptualization of the developmental and evolutionary sciences (see Stiles 2008; Reid 2007 & Oyama et al. 2001 for detailed discussion): the new synthesis assumes a notion of probabilistic epigenesis, i.e., the multidirectional interaction between genetic, behavioral, and environmental factors on ontogenetic and phylogenetic phenotypic outcomes associated with a norm of reaction (NOR): The NOR holds that, if we know the phenotypic outcome of two genotypes under one rearing (environmental) condition, we cannot predict their relative standing when these genotypes (actually, organisms) are reared in a different environment. (Gottlieb 2000b: 7) This perspective seems synergistic with linguistic approaches that are construction-theoretic. Thus, recent appeals to biology in the linguistic literature refer to competing views of this field, often without acknowledging its heterogeneity: one group tends to reflect the conventional genocentric reductive perspective where small bits aggregate to create larger units without relevant remainder, while the other reflects a more holistic perspective where different dimensions interact to produce probabilistic outcomes (see Lewontin 2000; Weber & Depew 2003; Jablonka & Lamb 2005, and Oyama, Susan et al., 2001).
Farrell Ackerman & James P. Blevins
observations about responses to substantive challenges to the idea that natural selection is the ultimate cause, rather than the possible effect of evolution and, by hypothesis, mediated by proximate non-genetic causes: In the mess of the Modern Synthesis, the evolutionary elite at the captain’s table keep beating the gong of selection. The dictum that ideas grow old and die with their originators does not apply: new evolutionary authorities with more extreme versions of the same old ideas have grabbed the vacant privileged seats. At the captain’s table, accomodation of any previously antithetical idea, referred to as a ‘‘change of emphasis,’’ is accompanied by rounds of self-congratulation for the vitality of the evolutionary debate. (Reid 2007: 41)
Early and prescient worries of similar elasticity employed to “save the appearances” within the Chomskyan paradigm were expressed by Uriel Weinreich: By this mysterious power to change his theory without changing it, Katz seeks to guarantee the perennial correctness of his approach, abstracted from any particular formulation of it. (Weinreich 1967: 286)
As in both psychology and biology where new empirical findings and new methods of analysis have led to reconceptualizations of theory and reappraisals of old data to better conform to new ideas, syntactic theory is undergoing a reshaping, while keeping true to certain perennial concerns. In this regard, the basic goals of modern theoretical grammarians were well articulated by W.D. Whitney: That science [linguistic science] strives to comprehend language, both in its unity, as a means of human expression and as distinguished from brute communication, and in its internal variety, of material and structure. It seeks to discover the cause of the resemblances and differences between languages, and to effect a classification of them, by tracing out the lines of resemblance, and drawing the limits of difference. It seeks to determine what language is in relation to thought, and how it came to sustain this relation; what keeps up its life and what has kept it in existence in past time, and even, if possible, how it came into existence at all. (W.D. Whitney 1875: 4)
Likewise, the basic theoretical tasks engaged in by MGG are, naturally, shared by all theoretical approaches. In particular, all theoretical syntacticians are passionate about identifying the constructs that can account for similarities and differences in language particular and cross-linguistic expression. They want to develop the toolkits that facilitate the identification of instructive data sets and to know how these toolkits can be used to explain the observed grammatical phenomena. Accordingly, there have always been alternatives concurrent with MGG. Though they may collectively constitute a majority of language researchers, each individual community has been relatively separate, with some exerting an influence on research outside of linguistics parallel to that exterted by MGG. On the other hand, narrow preoccupations with theory-specific assumptions and beliefs
Syntax: The state of the art
about the neccesity of responding to the predominant central dogma combined to dilute and fracture their effectiveness in providing a cohesive alternative. Some of these approaches have been lexicalist (Head-Driven Phrase Structure Grammar, Lexical Functional Grammar, Categorial Grammar) and some have been construction-theoretic (Construction Grammars of either a formal or more cognitive/ functional type). Many alternative theories of grammar have come, gone, are going, and, for our purposes, are now developing. Each represents its own gambits about the nature of human language. Recently, many of the doubts and criticisms raised separately about the central dogma by these alternative approaches have begun to cohere across frameworks and to resonate across disciplinary boundaries. This emerging perspective was reflected in the Linguistic Society of America’s 2007 Summer Institute at Stanford which relegated syntactic theory to a minor role, devoting two-day courses to them, and reserved main session four week courses for detailed inquiry into lesser-studied languages as well as new methods and perspectives in linguistic analysis. Morphosyntax is in a promising state of flux, both concerning the toolkits for language analysis as well as, inevitably, the nature of the objects that need to be investigated and the sorts of explanations required to account for them. In the concluding two sections we will try to identify some trends. 3. The influence of lesser studied languages on syntactic research A central issue that distinguishes MGG from numerous other approaches concerns the privileged status assigned to structural hierarchical representations over any alternative and the attribution of a primary explanatory role to these structures. This type of representation is then used as a template for explanation in other domains of language analysis. Given the usual adoption in scientific inquiry of Ockham’s razor Pluralitas non est ponenda sine necessitate – Plurality is not to be posited without necessity, there are, at least, two important issues to keep in mind concerning these assumptions, as well as those associated with all theories. First, have the representative and relevant empirical phenomena been identified so that theoretical constructs are firmly grounded and well motivated? Second, how does one calculate parsimony so that one can be reasonably sure that the representational reductionism exemplified by syntactocentrism is advantageous? With respect to the first issue, much of MGG can be regarded as finding surface forms largely interesting or useful inasmuch as they help theoreticians to “discover” the nature of abstract underlying representations and the operations required to (re)achieve surface expression. Neither the surface forms themselves nor their organization into networks of related forms are taken to be interesting or
Farrell Ackerman & James P. Blevins
instructive about the nature of grammar. Hence, the phenomena themselves have little effect on challenging fundamental theoretical assumptions, such as whether there are hierachically arranged binary branching functional categories or whether all languages have a verb phrase. We will briefly consider two phenomena in this connection. Both are by any definition “core” grammatical phenomena and both have been examined in MGG and elsewhere. The first concerns the distribution of tense and agreement marking, while the second concerns the typology of relative clauses. A recent set of papers (Lecarme (2004); Nordlinger & Sadler (2004); Tonhauser (2007), among others) has examined constructions in which morphosyntactic marking such a tense, aspect, agreement, ordinarily associated with verbs, appears with nouns. This is typified by Paraguayan Guarani as discussed in (Tonhauser 2007), where the markers kue and rã convey a sense of temporality:9 (3) a.
Kova ha’e peteĩ óga-kue. This cop one house-kue “This is a former house.”
b. Kova ha’e peteĩ óga-rã. This cop one house-ra “This is a future house.”
Theories have generally been designed to restrict the distribution of certain morphosyntactic categories to particular lexical categories and this, naturally, has encouraged certain representational assumptions and explanations. Expanding empirical inquiry to lesser studied languages reveals that distributions based on better studied languages needs to be revised, since the expectations derived from the latter are falsified by the former. The challenge that arises for all theories is how organically they can accommodate these new results and how much such a broader distribution of markers affects a theoretician’s beliefs about grammar design. A similar issue for grammar design is exemplified by externally headed prenominal relative clauses in Western Osytak (Uralic), since relative clauses are generally considered core grammatical phenomena.10
(4)
NP MODIFYING CLAUSE xans-әm write-
‘the book I wrote’
NP ne:pәk-e:m book-1
. See also the Forest Enets example (1). . We have utilized a simple phrase structure expression for expository purposes.
Syntax: The state of the art
In Western Ostyak, and numerous other languages of Eurasia, a person/number marker within the domain of the relativized nominal, the suffixed 1st singular marker in (4), expresses the subject relation associated with the verbal mixed category that heads the modifying clause. From a descriptive perspective the person/ number marker exhibits an anomalous distribution: it does not appear locally within the domain of the verb with which it bears a relation, but appears on the relativized nominal outside of that domain. To our knowledge no syntactic theory predicts the existence of this distribution of pronominal markers within relative clauses, though, there is no absence of toolkits and constructs in syntactic theory that, if deployed imaginatively, are equal to the task of redescribing it with favored assumptions.11 The question is whether some assumptions are more explanatory than others and how phenomena such as these might bear on determining what counts as explanatory assumptions. The second issue involving Ockham’s Razor concerns whether parsimony should be calculated only with respect to internal linguistic considerations and ensuing abstraction of constructs, or whether the parsimony of linguistic constructs should be calculated in the context of how it relates to other aligned disciplines. In particular, how do syntactic constructs and methods of analysis relate to independently motivated constructs and proven methods of analysis in other cognitive (neuro)sciences? Here too the Western Ostyak relative constructions are instructive. One can argue, following contruction-theoretical proposals (Kay & Fillmore 1999; Sag 1997, among others) that relative clauses are not the predictable and transparent result of interacting principles operating over tree-theoretic universal structures, despite assumptions to the contrary. Constructions, on this analysis, are not epiphenomenal nor are they peripheral to grammar or to the theoretical grammarian’s task of analyzing language. Instead, constructions may represent a crucial level of analysis in their own right, a level in which objects are not merely reducible to their constitutive component parts. However, there is nothing more mysterious in this irreducibility than what is accepted as common operating assumptions for the analysis of complex systems within the biological sciences, where phenomena can be seen as participating in a dynamic, that is, changeable system in which possible endstates are best understood by a systemic analysis identifying the probabilistic outcomes of multiple interacting factors. Applied to grammar, the relevant relative clauses can be interpreted as the product of language change within systems where the interactions between clausal constructions and their properties define pathways with probabilistic outcomes (see Dahl 2004). This,
. See Ackerman, Nikolaeva & Malouf (To appear) for a construction-theoretic proposal of this relative clause type and a review of alternative analyses such Kornfilt 2005.
Farrell Ackerman & James P. Blevins
f ollowing Bybee (to appear), is compatible with the view “that the true universals of language are not synchronic patterns at all, but the mechanisms of change that create these patterns.” Thus, the parsimony concerns for a theory that permits an exploration of this type may best be considered against the backdrop of tools and methods from other established sciences studying complex dynamic systems. 4. New methods All theoretical syntacticians seek empirically responsible and explicitly formulated theories. One way of achieving this, as suggested in the previous section, is by permitting “exotic” constructions to impact on the ways we ask questions. A second is in developing toolkits that permit the identification of empirical phenomena that need to be modeled as well as permit syntacticians to formulate new questions, or old ones in new ways. In this latter respect, syntax from all theoretical perspectives is undergoing a radical reconceptualization which finds early antecedents, among other places, in Weinreich, Labov & Herzog (1968). It has committed itself to the employment of quantitative (statistical) measures, psycholinguistic and neurolinguistic experiment based inquiry, and computational simulation that will have consequences for theory construction that are only dimly visible at the moment. Mentioning a few specific areas will give the general flavour of these changes. There has been increasing recognition concerning the explanatory role of analogy or patterned knowledge in grammar (see the articles in Blevins and Blevins to appear), on gradience considerations with respect to grammatical phenomena (Keller 2000; Sorace & Keller 2005; Hay & Baayen 2005; Aarts 2007; Bresnan & Nikitina to appear, among others) and on statistical methods of analysis (Manning 2003; Baayen 2003; Levy 2008 among others). The perspectives developed in this research replace the simplifying categorical distinctions frequently assumed in syntactic proposals of the past. The emphasis on human cognitive pattern-identifying capacities, in both language and other domains, also brings the phylogenetic research on human development and evolutionary biology into direct contact with ontogenetic research in the emergence of language and cognition over real time in the transition from infancy to adulthood. Indeed, in arguing for research strategies informed by tighter relations between language and other domains, Tomasello (2003: 4) observes that skills in human pattern-finding derive, among other things, from: the ability to form perceptual and conceptual categories of “similar” objects and events; the ability for perform statistically based distributional analyses on various kinds of perceptual and behavioral sequences: the ability to create analogies (structure mappings) across two or more complex wholes, based on the similar functional roles of some elements in these different wholes.
Syntax: The state of the art
Finally, research has been notably active and productive in the areas of sentence processing and discourse factors concerning prosody and context for the analysis of syntactic structure. This is evident from the increasing selectivity, size and importance of the annual CUNY sentence processing conference. It is striking that theoreticians building upon very different and often incommensurable foundational assumptions have begun to share many of these methodological and conceptual inclinations. Despite the vitality of this research, the results must still be regarded as indeterminate with respect to the evaluation of current theories. 5. Conclusion All of these new empirical and methodological emphases and consequent results testify to the richness of the landscape in contemporary syntax. Some of these results will seem to support certain foundational assumptions of MGG, while some will seem to support alternative theories. On the other hand, it strikes us as most likely that the directions of these changes will demand a re-evaluation of the biases that inform much of modern theory, permitting a productive and transformative reappraisal of ideas about grammar that antedate Post-Bloomfieldian structuralism. This will require that the latter be replaced or complemented by a more subtle understanding of (morpho)syntax as a complex system, analyzed as other complex systems in nature, but still unique in its specificity to humans.
References Aarts, Bas. 2007. Syntactic gradience. Oxford: Oxford University Press. Ackerman, Farrell & Gregory T. Stump. 2004. Paradigms and periphrastic expression. In Louisa Sadler & Andrew Spencer (eds), Projecting Morphology, 111–158. Stanford: CSLI Publications. Ackerman, Farrell, Irina Nikolaeva & Rob Malouf. To appear. Descriptive typology and grammatical theory: A construction-theoretic study of morphosyntax. Stanford: CSLI Publications. Anderson, Stephan R. 2004. Doctor Doolittle’s delusion: Animals and the uniqueness of human language. New Haven: Yale University Press. Baayen, Harald. 2003. Probabilistic approaches to morphology. In Rens Bod, Jennifer Hay & Stefanie Jannedy (eds), Probabilistic linguistics. Cambridge: Cambridge University Press. Blevins, James P. & Juliette Blevins (eds). To appear. Analogy in grammar: Form and acquisition. Oxford: Oxford University Press. Brecht, Bertolt. 1994. Galileo. New York: Grove Press.
Farrell Ackerman & James P. Blevins Bresnan, Joan & Tatiana Nikitina. To appear. The gradience of the dative alternation. In Linda Uyechi & Lian Hee Wee (eds), Reality Exploration and Discovery: Pattern Interaction in Language and Life. Stanford: CSLI Publications. Bresnan, Joan & Ron Kaplan. 1982. Lexical Functional Grammar: A formal system for grammatical representation. Joan Bresnan (ed.), The mental representation of grammatical relations. Cambridge Mass.: MIT Press. Bybee, Joan. To appear. Mechanisms of change as universals of language. In Ricardo Mairal and Juana Gil (eds), En torno a los universales lingüísticos. Cambridge: Cambridge University Press. Chomsky, Noam. 1975. The logical structure of linguistic theory. New York: Plenum Press. Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry 36(1): 1–22. Clark, Andy & Shalom Lappin. To appear. Linguistic Nativism and the Poverty of the Stimulus. Wiley-Blackwell. Culicover, Peter & Ray Jackendoff. 2005. Simpler syntax. Oxford: Oxford University Press. Curme, George. O. 1935. A grammar of the English language. Vol. 1: Parts of speech. Boston: Heath. Dahl, Östen. 2004. The growth and maintenance of linguistic complexity. Amsterdam; Philadelphia: John Benjamins. Deacon, Terence W. 1997. The symbolic species. New York: W. W. Norton. Dixon, Robert M.W. & Alexandra Y. Aikhenvald. 2002. Word: A cross-linguistic typology. Cambridge: Cambridge University Press. Elman, Jeff (et al.). 1996. Rethinking innateness: A connectionist perspective on development. Cambridge, Mass.: MIT Press. Gottlieb, Gilbert. 2000a. Environmental and behavioral influences on gene activity. Current Directions in Psychological Science 9(3): 93–97. Gottlieb, Gilbert. 2000b. Probabilistic epigensis. Developmental Science 10(1): 1–11. Goldberg, Adele. 2006. Constructions at work: The nature of generalization in language. Oxford: Oxford University Press. Halle, Morris & Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In Kenneth Hale and Jay Keyser (eds), The view from building 20. Cambridge: MIT Press. Hay, Jennifer & Harald Baayen. 2005. Shifting paradigms: Gradient structure in morphology. Trends in Cognitive Science. 9(7): 342–348. Hockett, Charles, F. 1987. Refurbishing the foundations. Philadelphia: John Benjamins. Jablonka, Eva & Maron J. Lamb. 2005. Evolution in four dimensions. Cambridge, Mass.: MIT Press. Jackendoff, Ray. 2002. Foundations of language. Oxford: Oxford University Press. Kay, Paul & Charles J. Fillmore. 1999. Grammatical constructions and linguistic generalizations: The what’s doing Y? construction. Language 75(1): 1–33. Keller, Frank. 2000. Gradience in grammar: experimental and computational aspects of degrees of grammaticality. Ph.D. Thesis. University of Edinburgh. Kornfilt, Jaklin. 2005. Agreement and its placement in Turkic non-subject relative clauses. In Guglielmo Cinque & Richard S. Kayne (eds), Handbook of Comparative Syntax, 513–541. Oxford: Oxford University Press. Lasnik, Howard with Marcela Depiante & Arthur Stepanov. 2000. Syntactic Structures revisited: Contemporary lectures on classic Transformational Theory. C ambridge, Mass.: MIT Press. Lecarme, Jacqueline. 2004. Tense in Nominals. In Jacqueline Gúeron & Jacqueline Lecarme (eds), The syntax of time, 440–475. Cambridge, Mass: MIT Press. Levy, Roger. 2008. Expectation-based syntactic comprehension. Cognition 106: 1126–1177.
Syntax: The state of the art
Lewontin, Richard. 2000. The triple helix. Cambridge, Mass: Harvard University Press. Manning, Chris. 2003. Probabilistic syntax. In Rens Bod, Jennifer Hay & Stefanie Jannedy (eds), Probabilistic linguistics. Cambridge: Cambridge University Press. Matthews, Peter H. 2007. Syntactic relations: A critical survey. New York: Cambridge University Press. Matthews, Peter H. 1993. Grammatical theory in the United States from Bloomfield to Chomsky. Cambridge: University of Cambridge Press. Nordlinger, Rachel & Louisa. Sadler. 2004. Nominal Tense in Cross-Linguistic Perspective. Language 80: 776–806. Newmeyer, Frederick, J. 2005. Possible and probable language: A generative perspective on lingistic typology. Cambridge: Cambridge University Press. Oyama, Susan, Paul E. Griffiths & Russell D. Gray (eds). 2001. Cycles of contingency: Developmental systems and evolution. Cambridge, Mass.: MIT Press. Paul, Hermann. 1970. Principles of the history of the language. College Park, Md., McGrath Publishing Co. Platt, S.A., & Sanislow, C.A. 1988. Norm-of-reaction: Definition and misinterpretation of animal research. Journal of Comparative Psychology 102: 254–261. Reid, Robert, G.B. 2007. Biological emergences: Evolution by natural experiment. Cambridge, Mass.: MIT Press. Richerson, Peter J. & Robert Boyd. 2005. Not by genes alone: How Culture transformed human evolution. Chicago & London: University of Chicago Press. Robins, Robert. H. 1959. In defence of WP. Philological Society of London. 116–144. Sag, Ivan. 1997. English relative clause constructions. Journal of Linguistics 33(2): 431–484. Scholz, Barbara & Geoffrey K. Pullum. 2006. Irrational nativist exhuberance. In Rob Stainton (ed.), Contemporary debates in cognitive science, 59–80. Oxford: Basil Blackwwell. Seuren, Pieter, A.M. 1998. Western linguistics: An historical introduction. Malden, Mass: Blackwell. Snyder, William. 2007. Child language: The parametric approach. Oxford: Oxford University Press. Sorace, Antonella & Frank Keller. 2005. Gradience in linguistic data. Lingua 115(11): 1497–1524. Stiles, Joan. 2008. The fundamentals of brain development: Integrating nature and nurture. Cambridge, Mass.: Harvard University Press. Stump. Gregory T. 2001. Inflectional morphology. Cambridge: Cambridge University Press. Tomasello, Michael. 2003. Constructing a language. Cambridge, Mass.: Harvard University Press. Tonhauser, Judith. 2007. Nominal tense? The meaning of Guaraní nominal temporal markers. Language 83(4): 831–869. Weber, David & Chauncy M. Depew (eds). 2003. Evolution and learning: The Baldwin effect re-considered. Cambridge, Mass.: MIT Press. Weinreich, Uriel. 1967. On arguing with Mr. Katz: A brief rejoinder. Foundations of Language 3: 284–287. Weinreich, Uriel, William Labov & Marvin I. Herzog. 1968. Empirical Foundations for a Theory of Language Change. In Winfred P. Lehmann (ed.), Directions for Historical Linguistics: A Symposium. 95–195. Austin: University of Texas Press. Whitney, William D. 1875. The life and growth of language: An outline of linguistic science. New York: D. Appleton and Company.
Index of terms
A alignment constraint, 44, 46 Aristotelian, 29, 39, 201, 202 Aristotle, 31, 38 aspect, 12, 98–106, 185, 200–206, 208–14, 224 B Bloomfield, 216, 229 brain, 17, 19–21, 25–27, 59, 60, 62–64, 66–72, 83, 111, 112, 114–21, 229 C Chomsky, 7, 8, 17, 18, 25, 26, 44, 56, 176, 217, 218, 220, 221, 228, 229 Chomskyan, 216, 217, 219, 220, 222 clone, 35–39 coercion, 73–76, 79, 80–84, 196, 205 Common Ground (CG), 30, 125, 126, 133 complementizer, 43, 47–56 compositionality, 73–76, 83, 192, 205 constraint, 35, 43–50, 52–56, 83, 134, 162, 179–81, 194, 221 alignment~, 44, 46 economy~, 45 faithfulness~, 48–50, 52–54 markedness~, 46, 47, 49, 50, 54 universal~, 43, 44 conversational implicature, 29, 40, 204 D delimitation, 123, 128, 204, 211 discourse, 10, 26, 35, 47, 81, 93, 94, 118, 125, 127, 128, 155–63, 168–70, 172, 220, 227
discursive construction, 153, 163 E economy constraint, 45 economy of structure, 43, 46, 47, 55, 56 endangered languages, 96, 140, 146–48, 183 evidentiality, 166, 168, 208–10, 213 F faithfulness constraint, 48–50, 52–54 fMRI (functional magnetic resonance imaging), 59, 63–65, 67–71, 117 focus, 36, 40, 55–57, 63–65, 70, 123–28, 130–35 formal theory, 201 frame setting, 123, 127, 128 functional theory, 40, 201 G Generative Lexicon (GL), 73, 76, 77, 83, 199 givenness, 123, 125–28, 131 globalization, 86, 111, 139, 146, 156, 158, 171, 172 globalization of discourse, 156, 158 grammaticalization, 18, 22–24, 26, 158, 177, 184–86, 206, 212 Greenberg, 8 Grice, 29, 30, 31, 39, 40 H harmonic bounding, 43, 44, 46, 49, 51–53, 56, 57 historical linguistics, 175–78, 183–86, 187 historical pragmatics (HP), 155, 161–63, 168, 170–72
I information structure (IS), 123–26, 128, 129, 131–35 inherent polysemy, 79 interculturality, 155, 164, 165, 167, 169, 171, 173 international workplace communication, 159 L language change, 23, 175–87, 217, 225, 229 language contact, 169, 172, 175, 177, 182, 184, 187, 210 language management, 139, 143, 145, 146, 151–53 language origin, 17, 18, 20, 22, 25, 27 language planning, 10, 137–42, 146, 149–53 language policy, 96, 137–45, 147–53, 160 language rights, 85–87, 95, 143–45, 151 language shift, 85, 87, 139, 143, 147, 150 lesser studied languages, 217, 219, 223, 224 lexical frequency, 114, 115 lexical meaning, 189–93, 195, 197, 198, 200, 205 lexical pragmatics, 29, 35, 38–41 lexical semantics, 29, 38, 189–93, 198, 199, 203, 213 linguistic human rights (LHR), 86–88, 92, 93, 144 lingua franca, 140, 145, 155–60, 168–71 linguistic diversity, 85, 86, 88, 89, 92, 94–96, 144, 152 localization, 59, 60, 64, 65, 67, 70, 72
Index of terms M Manichaean, 29, 34, 39, 40 markedness constraint, 46, 47, 49, 50, 54 MEG (magnetoencephalography), 59, 60, 67–71, 117 Mainstream Generative Grammar (MGG), 219, 222–24, 227 mind, 13, 20, 25, 26, 30, 111–16, 120, 172, 173, 184, 194, 199 modality, 15, 108, 201, 207–13 mood, 201, 207, 208, 213 morphological family size, 114, 115, 121 morphosyntax, 213, 223, 227 multimorphemic words, 111–16, 119 N neighborhood, 97, 98, 100, 105–07 neural, 59, 60–62, 68–70, 117 neuroimaging, 62, 69, 70 neurolinguistics, 59, 69, 114 O Optimality Theory (OT), 43, 46, 56, 179, 181 OWL-Time, 97, 99–101, 104–07 P Paul, 29, 30, 31, 39, 40, 217, 229 polysemy, 29, 73, 74, 76, 79, 83, 190, 193–96, 199, 200
inherent~, 79 selectional~, 73, 74, 76, 195 Post-Bloomfieldian, 216, 220, 227 pragmatics, 27, 29, 34, 35, 38–41, 108, 134, 155, 161–63, 167, 168, 170–72, 209, 213 historical~ (HP), 155, 161–63, 168, 170–72 lexical~, 29, 35, 38–41 principle of compositionality, 74, 192 prosody, 9, 57, 61–66, 68–70, 128–30, 132, 134, 227 psycholinguistics, 25, 59, 68–70, 114 Q Q, 29, 31, 33, 34, 36, 39, 40 qualia, 73, 77, 78, 80, 81, 189, 190 R R, 29, 31, 33–36, 39, 40 ranking, 44–47, 49, 50–55, 179 Roschian, 29, 37, 39 S selectional polysemy, 73, 74, 76, 195 semantics, 9, 29, 38, 56, 74–76, 80, 81, 83, 97, 100, 101, 103, 106–08, 131, 133, 134, 172, 179, 189–93, 198, 199, 202, 203, 205, 209–13, 219, 220 sign language, 3–15, 18
speech, 7, 14, 17–19, 23–26, 30, 33, 60, 61, 64–71, 80, 82, 112, 120, 128, 145, 146, 162, 163, 180, 183, 185, 191, 228 spoken language, 3, 4, 6–10, 13, 59–61, 68, 69 Stokoe, 9, 15 syntax, 9, 15, 21, 24, 35, 56, 57, 70, 73, 108, 135, 177, 186, 190, 203, 209–11, 213, 215–19, 226–29 T telicity, 202–06, 211–14 tense, 50, 56, 98, 99, 101–08, 185, 201, 202, 208, 209, 212, 216, 224, 228, 229 TimeML, 97–99, 101, 102, 106–08 topic, 12, 55, 123, 125–34, 157, 166, 202 U universal conceptual metaphor, 18 universal constraint, 43, 44 un-noun, 38, 39 un-word, 35, 40 V violation, 44–47, 49, 50–53, 55 W Whitney, 17, 27, 222, 229 written language, 12, 13, 157 Z Zipf, 30, 31, 33, 39, 41