Iroquoian∗ Matthias Urban Max Planck Institute for Evolutionary Anthropology
1. Expert Classification
Table 1 is a fam...
162 downloads
458 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Iroquoian∗ Matthias Urban Max Planck Institute for Evolutionary Anthropology
1. Expert Classification
Table 1 is a family tree for Iroquoian, adapted from Mithun (1999:418), presenting a consensus expert classification. In this paper, similarities and differences between this tree and that produce by ASJP for the family (Figure 2) are discussed.
Table 1. Consensus genetic classification of Iroquoian languages. SOUTHERN IROQUOIAN: Cherokee NORTHERN IROQUOIAN: Tuscarora-Nottoway Tuscarora Nottoway Proto-Lake Iroquoian Huron-Wyandot Laurentian Iroquois Proper = Five Nations Iroquois Seneca-Cayuga Seneca Cayuga Onondaga Susquehannock Oneida-Mohawk Oneida Mohawk The grand scheme of the internal relations of the Iroquoian languages is straightforward and uncontroversial according to orthodox classification: The family consists of two ∗
I am indebted to Cecil H. Brown and Søren Wichmann for useful comments and suggestions on the paper.
primary branches, Northern Iroquoian (NI) and Southern Iroquoian (SI), mirroring linguistically the fact that “[p]erhaps four thousand years ago” (Mithun 1984: 263) an until then putatively homogenous group of people separated into two distinct subgroups, the latter of which migrated southward a considerable distance from the homeland to modern day Tennessee and North Carolina. Descendants of this migrating group are the present day speakers of Cherokee, which makes up the only member of the SI subfamily. The linguistic difference between Cherokee and the NI languages in general represents “[t]he widest cleavage in the Iroquoian family” (Lounsbury 1961: 11). Applying the method of glottochronology to a Swadesh-200 list of the relevant languages, Lounsbury (1961) arrived at an estimated time depth of this major split within the family of 3.5003.800 years. A map created using Bibiko (2005) showing Northern and Southern Iroquoian languages (concentrating on the languages for which comparison with AJSP classification is amenable) is shown in Fig. 1.
Figure 1. A map of the Iroquoian languages
NI itself is further divided into the Tuscarora-Nottoway subgroup and the so-called Lake Iroquoian. Nottoway is known only through early wordlists (analyzed in Rudes 1981).
2
Lake Iroquoian consists of the extinct language Laurentian1 (documented only in wordlists collected in the 16th century), the two extinct, but definitely closely related languages Huron and Wyandot, and a grouping variously called Iroquois proper or Five Nations Iroquoian, which contains five of the seven Iroquoian languages still spoken, namely Seneca, Cayuga, Onondaga, Oneida and Mohawk, as well as the extinct Susquehannock. This classification of NI languages is motivated by a number of sound changes from the phoneme system of the proto-language. While Tuscarora has undergone a number of rather drastic sound changes The Five Nation Languages are less changed phonologically. The reconstructed phoneme inventory for Proto Northern Iroquoian as well as a concise overview of changes leading to the individual languages is found in Mithun (1979: 161-180), discussion of developments in the vowel system as well as an extensive list of cognate sets (building on earlier work by Hewitt 1887) is in Rudes (1995). Intelligibility tests conducted by Hickerson et al. (1952) also point to a particular close affinity between the Five Nations languages, and among them in particular between Seneca and Cayuga (the westernmost languages) and Mohawk and Oneida (the easternmost languages). Together with independently noted grammatical and lexical affinities these lines of evidence have lead to the recognition of the Seneca-Cayuga and Oneida-Mohawk subbranches of Five Nations Iroquoian. On closer inspection, however, the precise internal relations between certain languages of NI can be shown to be of a much more complicated and intertwined nature than indicated by the standard tree classification. For instance, Cayuga and Tuscarora, languages not closely related otherwise within NI, can be shown to uniquely share innovations in the pronominal prefix system and the lexicon (Chafe & Foster 1981: 126, Chafe 1984: 303), which is explained by positing an (albeit short) stage of common development prior to the departure of the Proto-Tuscarora-Nottoway southward. That the ancestral Cayugas did not follow this migration, but instead remained in an area close to the mainstream Iroquois group is suggested by the presence of several phonological changes which must have originated in Cayuga and then spread to Seneca. In fact, the 1
Other extinct languages, for which little or no data are available, are Merrehin (supposedly close to Tuscarora-Nottoway), Petun, Neutral, Wenro, Erie and Susquehannock (see Mithun 1979: 138-48 for the information available).
3
Cayuga and the Seneca must have gone through a number of different periods of relative separation (attested by independent developments in the languages) followed by relative proximity (evidenced by common changes or diffusion, as in the case of the changes in the system of stress assignment and syllable structure and the loss of resonants at a later period of contact). However, contact influence is not restricted to Seneca and Cayuga. The Onondaga language can be shown to have relatively recently undergone considerable phonological changes involving resonants due to contact with Seneca (see Chafe & Foster 1981 for details of those processes). NI Languages also differ considerably as to which grammatical categories are distinguished in their pronominal systems, due to languagespecific semantic narrowing and widening of the meaning of proto-Iroquoian prefixes (see Chafe 1984 for full discussion). An alternative classification of the Iroquoian languages, which overlies and supplements the tree-based classification, is that of peripheral vs. inner languages proposed in Lounsbury (1961). This classification is supported both by grammatical evidence, specifically, the distribution of certain pronominal prefixes, as well as some lexical evidence. Lounsbury noted that it is sometimes necessary to posit two proto-forms for the mother language to account for dissimilarities among languages. For instance, for ‘lake’, a root *-ōtar- is reflected in Cherokee, Wyandot, Huron, and Laurentian, whereas the terms in the Five Nations Languages (Seneca, Cayuga, Onondaga, Oneida, Mohawk) go back to *-nyatar. This plus a similar distribution of differing roots for the numerals ‘four’, ‘six’ and ‘seven’ as well as ‘paternal aunt’ yields a grouping of outer languages, consisting of Cherokee, Laurentian, Huron-Wyandot and Tuscarora, which surround the inner languages of the Five Nations, with a center of cognacy in the relevant forms for the latter in the eastern Five Nations Languages. A map based on this classification, adding Nottoway to the peripheral languages due to its close affinity with Tuscarora, would look as in Fig. 2.
4
Figure 2. Inner vs. Peripheral Iroquoian languages
These additional complicating factors do not render the traditional tree representation wrong. They simply point to the fact that something more can be said about the internal relations of the Iroquoian languages, thus enriching the classic tree structure with additional information. A deep genetic unity of the Iroquoian languages with the Siouan and Caddoan families, forming a unit often called ‘Macro-Siouan’, has been suggested at various points, but remains unproven by the comparative method (see Chafe 1973 for discussion).
2. Comparison with ASJP tree
Sufficient data for a comparison using the ASJP method has been obtained for the seven extant Iroquoian languages, Seneca, Cayuga, Onondaga, Oneida, Mohawk, Tuscarora, and Cherokee. Additional data has been acquired for Wyandot and Nottoway, although both language datasets fail to pass the threshold of having words for at least 70% of 40 referents necessitated by ASJP. All of the Iroquoian languages of the ASJP sample are not branched directly together on the ASJP tree of Müller et al. (2009) produced for all of the world’s
5
languages (at least those world languages currently in the ASJP database, i.e., approximately 3,500). Instead, Iroquoian languages are found in two different segments of the tree. In one segment the Northern Iroquoian languages are gathered, and in another Cherokee appears as a sister of Cuitlatec (an isolate of Mesoamerica), with the next closest relatives being Atakapa (a North American isolate), and Pyu (a Papuan language of Papua New Guinea). The failure of Cherokee to directly branch with the other Iroquoian languages on the ASJP tree reflects the fact that, although there are a fair number of clearly cognate lexemes in Cherokee and the other Iroquoian languages (see Rudes 1995, appendix 2), these often involve multiple changes in both vowels and consonants within the same form. The failure of ASJP to get Cherokee to group with Iroquoian is therefore most likely a conspiracy of the tendency for both vowels and consonants to change and the fact that ASJP, unlike traditional historical linguistics, works exclusively with lexical data and does not consider grammatical evidence. When only languages which are extant, known to be Iroquoian by orthodox methods of historical linguistics, and pass the 70%-treshold imposed by ASJP, are used in the generation of an ASJP tree, the result is as in Fig 3.
ONONDAGA CAYUGA SENECA ONEIDA MOHAWK TUSCARORA CHEROKEE
Iro.NORTHERN IROQUOIAN
Iro.SOUTHERN IROQUOIAN
10
Figure 3. An ASJP tree limited to just the Iroquoian languages for which 70% or more on the 40-item list are attested.
The ASJP classification in fig. 3 is very similar to the classification advocated by experts (Table 1). The procedure has adequately detected the particularly close relationship between Oneida and Mohawk. It has also correctly positioned Tuscarora within NI, but crucially outside Proto-Lake-Iroquoian. This nice result can be seen as evidence that the ASJP routine, if fed with qualitatively appropriate lexical data, indeed produces results
6
which come close to expert classifications, despite the fact it does not involve other discriminating grammatical data. Within NI, the ASJP tree differs from expert classification with respect to the positioning of Onondaga since Onondaga is grouped as a direct sister of Cayuga, with Seneca branching off from a higher node of the tree. This is unexpected, but two explanations suggest themselves: First, as shown by Chafe & Foster (1981), the relationship between the three languages in question is characterized by mutual influence due to various periods of language contact. This scenario pertains primarily to Seneca and Cayuga, but to some extent also to Onondaga, as discussed above. A closer look at the actual datasets used for the ASJP classification reveals the more direct reasons for the program’s choices: There are a fair number of cognates in which the segmental similarity in fact is (for whatever reasons) superficially closer between Cayuga and Onondaga than between Cayuga and Seneca. Striking examples are shown in Table 2 below.2
‘fish’ ‘blood’
Seneca
Cayuga
Onondaga
kɛjɔh,
o·jǫ7da7, ,
ojyų́ʔdaʔ,
-ij-
-(i)j ǫ7d(a)-
-ijyųʔd-
ʔakɛ́tkwɛhsaʔ3,
-tgwęhs(a)-
otgwę́hsaʔ,
-tkwɛ́hs(a)‘horn’ ‘liver’ ‘see’
-tgwęhs-
ʔonɔ́kæ:ʔ,
o·ná7ga:7,
onáʔga·ʔ,
-nɔ:ʔka(æ)-
-na7g(a:)-
-naʔgaR-
ʔakéthwɛhsaʔ 4,
o·twę́h·sa 7,
othwę́hsaʔ,
-thwɛ́hs(a)-
-twęhs(a)-
-thwęhs-
-ká:ne-/-ká:n(æ)-,
-gę-
-gę-
-kɛ-, -tokTable 2: Some cognates for Seneca, Cayuga and Onondaga.
2
The data are from Chafe (1967), Froman et al. (2002), and Woodbury (2003). Nouns are represented first as inflected forms, with prefixes and suffixes, as they appear as minimally freestanding forms in discourse and by the underlying root. Original orthography is employed rather than the ASJP code transcriptions used in making automated similarity judgements (Brown et al. 2008, Appendix C). 3 ‘my blood’ 4 ‘my liver’
7
In light of the fact that the ASJP wordlist consists of no more than 40 items, and does not include grammatical factors in determining relations between languages, the program’s choice is not so much a surprise as one might initially think, given examples such as the five above, where, at least on a superficial level, there are close affinities between Onondaga and Cayuga. We now turn to see what happens if the insufficient data for Wyandot and Nottoway are added. The datasets for both languages fall below the 70%-criterion, but are not too far away from it. Nottoway is represented by 23 out of 40 items (57,5%) and Wyandot even by 25 out of 40 items (62,5 %), so the exercise could be worthwhile and interesting. The extended family tree is displayed in Fig. 4.
ONONDAGA CAYUGA SENECA ONEIDA Iro.NORTHERN IROQUOIAN MOHAWK NOTTOWAY TUSCARORA CHEROKEE Iro.SOUTHERN IROQUOIAN WYANDOT Iro.NORTHERN IROQUOIAN 10
Figure 4. An ASJP tree of Iroquoian, including Nottoway and Wyandot
Again, there is both agreement and divergence between the standard expert classification and the tree computed by ASJP: According to the latter, but not the first, Wyandot is grouped with Cherokee in the SI subfamily. Note, however, that Wyandot is positioned differently in the tree in fig. 3, when other languages, including unrelated ones, are taken into account: In this tree, Wyandot is correctly placed within NI. The alternative grouping in fig. 4 might reflect the grouping into peripheral vs. inner languages discussed above, where both Cherokee and Wyandot belong to the peripheral languages. On the other hand, the classification of Nottoway as a direct sister of Tuscarora is in perfect agreement with orthodox classification.
8
References
Bibiko, Hans-Jörg (2005): “The Interactive Reference Tool.” CD-ROM accompanying Haspelmath, Martin, Matthew S. Dryer, David Gil & Bernard Comrie (eds.): The World Atlas of Language Structures. Oxford: Oxford University Press. Brown, Cecil H., Eric W. Holman, Søren Wichmann & Viveka Velupillai (2008): “Automated Classification of the World’s Languages: A Description of the Method and Preliminary Results.” Language Typology and Universals (STUF) 61, 285-308. Chafe, Wallace L. (1967): Seneca Morphology and Dictionary. Washington: Smithsonian Press. Chafe, Wallace L. (1973): “Siouan, Iroquoian, and Caddoan.” In Sebeok, Thomas (Ed.): Current trends in linguistics Vol. 10, 1164-1209. The Hague: Mouton de Gruyter. Chafe, Wallace L. (1984): “How to say ‘They Drank’ in Iroquois.” In: Foster et al. (eds), 301-311. Chafe, Wallace L. & Michael K. Foster (1981): “Prehistoric Divergences and Recontacts between Cayuga, Seneca, and the Other Northern Iroquoian Languages.” International Journal of American Linguistics 47, 121-142. Foster, Michael K, Jack Campisi & Marianne Mithun (eds.) (1984): Extending the rafters: Interdisciplinary approaches to Iroquoian studies. Albany: SUNY Press. Froman, Frances, Alfred Keye, Lottie Keye & Carrie Dyck (2002): English-Cayuga / Cayuga-English Dictionary. Toronto: University of Toronto Press. Hewitt, John Napoleon Brinton (1887): The Cheroki an Iroquoian Language: A Critical Study & Comparison of Etymologies, Words, Sentence-words, Phrase-forms, & Conjugations Common to the Cherokian and Iroquoian Tongues. To Establish Their Common Origin. Made in July & August, 1887. Manuscript 447, National Anthropological Services, Smithsonian Institution, Washington. Hickerson, Harold, Glen D. Turner & Nancy P. Hickerson (1952): “Testing Procedures for Estimating Transfer of Information Among Iroquois Dialects and Languages.” International Journal of American Linguistics 18, 1-8.
9
Lounsbury, Floyd (1961): Iroquois-Cherokee Linguistic Relations. Bureau of American Ethnology Bulletin 180. Mithun, Marianne (1979): “Iroquoian.” In: Campbell, Lyle and Marianne Mithun (eds.): The Languages of Native America, 133-212. Austin: University of Texas Press. Mithun, Marianne (1984): “The Proto-Iroquoians: Cultural Reconstruction from Lexical Materials.” In: Foster et al. (eds.), 259-281. Mithun, Marianne (1999): The Languages of Native North America. Cambridge: Cambridge University Press. Müller, André, Viveka Velupillai, Søren Wichmann, Cecil H. Brown, Pamela Brown, Eric W. Holman, Dik Bakker, Oleg Belyaev, Dmitri Egorov, Robert Mailhammer, Anthony Grant & Kofi Yakpo (2009): ASJP World Language Tree of Lexical Similarity: Version 2 (April 2009). Rudes, Blair A. (1981): “A Sketch of the Nottoway Language from a HistoricalComparative Perspective.” International Journal of American Linguistics, 47, 2749. Rudes, Blair A. (1995): “Iroquoian Vowels.” Anthropological Linguistics 37, 16-69. Woodbury, Hanni (2003): Onondaga-English / English-Onondaga Dictionary. Toronto: University of Toronto Press.
10