A Language on the Border: Identifying Nipmuck HOLLY GUSTAFSON University of Manitoba Identifying an unknown and extinct language from an anonymous manuscript means facing unique challenges, including cross-border borrowing. Listed as number 32 in "Releve des ecrits indiens aux Archives de S. Sulpice" (Marinier 1973), an inventory of the native North American manuscripts in the Sulpician archives in Montreal, is one simply titled "Mots loups". This entry refers to an undated and anonymous notebook, comprised of 124 unnumbered pages containing no statement of either the place of its composition or the language it attempts to record (Day 1975:18). The title "Mots loups", literally "Wolf words", makes reference to the name applied to several Algonquian tribes by the French; for two of these tribes, no name besides "Loup" is known (Goddard 1978:71). Jean-Andre Cuoq attributed the manuscript of the "Mots loups" to the Sulpician missionary Jean-Claude Mathevet (1717-1781). In 1942, the manuscript was found at Oka by J. O. Lesieur; it had apparently survived a fire at the mission in 1877 that Cuoq believed had destroyed Mathevef s notebooks on the Loup language (Day 1975:21). Lesieur shipped this and two other manuscripts to the Sulpician archives in Montreal (Lesieur 1942), where it was microfilmed and identified as Algonquian by Victor Hanzeli (1961:237). In 1975, the manuscript was published as Hie Mots loups of Father Mathevet, edited by Gordon Day. Although the language of the "Mots loups" notebook has not been positively identified, it is generally accepted that it is the language of one of the central Massachusetts tribes, either Pocumtuck or Nipmuck (Goddard 1978:174). As Gordon Day (1975:44) explains, speakers of the "Loup" language appear to have originated in New England, but the dialect of the manuscript differs from all recorded New England languages. In order to determine the precise location of origin of Mathevef s "Loup", Day uses two basic sets of clues: references found in the manuscript, and morphological analysis of the Loup name for themselves. The most helpful clue in the manuscript to the location and identity of the Loup people, Day suggests, is found in the juxtaposition of two entries, the word (tak8aangan) 'the sawmill', and <8miskan8ag8iak) 'the Loup A LANGUAGE ON THE BORDER 123 nation'. Day goes on to propose that "the sawmill" is probably the French sawmill located on the Missisquoi River at Swanton, Vermont. The fact that this word is followed by (mak8sem), which is the word for 'wolf but was interpreted by Mathevet as the name of the Loup nation, suggests that "either he was writing at Missisquoi, or that his Loups had recently come from Missisquoi and he therefore identified them with the village" (Day 1975:51). From these references and from the word <8miskan8ag8iak), apparently the real name of the Loup nation, Day (1975:59) concludes that the Loup were indeed the Pocumtuck tribe. However, there are two main problems with this deduction. First, the word (8miskan8ag8iak) is morphologically analyzed by Day (1975:55) to mean "beaver-tail-hill people", and since the Pocumtuck range, according to legend, is said to be the petrified body of a huge beaver, it might be reasonable to make the tentative connection between the Loup and the Pocumtuck. However, Day's morphological analysis is quite tenuous, and he dismisses three major problems with this analysis as "mishearings" or "misspellings" by Mathevet. The second problem with considering the Loup to be the Pocumtuck is a phonological one. Using present-day place names and those recorded centuries ago, Goddard's review of James Hammond Trumbull's Indian names in Connecticut (1977) helps to "establish or confirm the locations of the phonological isoglosses that separated the rather poorly known languages of Connecticut". According to Goddard, Proto Eastern Algon­ quian */ became r in the area of the Housatonic Valley, became v in the Mohegan-Pequot region of southeastern Connecticut, became n to the east of this area (present-day Rhode Island and eastern Massachusetts), and remained / to the north, in central Massachusetts. He points out that no place names containing the / reflex occur west of the Connecticut River. This finding immediately rules out Pocumtuck as a possible candidate for the Loup language, since the Pocumtuck clearly occupied an area west of the Connecticut River, where an r-dialect was spoken. However, Mathe­ vef s manuscript contains no r's in Loup words, instead displaying a consistent / reflex. This eliminates all languages west of the Connecticut River, including Pocumtuck, as well as all languages south of the Massachusetts-Connecticut border. Thus there is only one possible candi­ date, the only language yet to be accounted for east of the Connecticut River and north of the Mohegan-Pequot region — Nipmuck. 124 HOLLY GUSTAFSON Little is known about the Nipmuck tribe, apart from what can be gained from Mathevef s manuscript. With an original population of about 500, by the epidemic of 1634 rendered the Nipmuck virtually extinct (Shepherd 1988:185). Thus it might be said that Mathevef s manuscript is the only information still available about the Nipmuck language and culture. One of the major problems in identifying an unknown language like "Loup" is cross-border borrowing. Borrowing between languages within the same subgroup is not uncommon in Algonquian; for example, two languages within the Southern New England subgroup, such as Massachusett and Narragansett, are bound to borrow phonological, morphological, lexical, and even syntactic features from one another. Less common, and consequently less recognized, is borrowing that occurs between languages which are in close geographic proximity but are separated by a major isogloss. This is the case with Nipmuck and Western Abenaki. Not surprisingly, Nipmuck shares many characteristics with its fellow Southern New England languages. A list of selected diagnostic phonologi­ cal innovations in some Eastern Algonquian languages, adapted from Goddard (1978:75), is given in 1. The first four languages — Micmac, Maliseet-Passamaquoddy, Eastern Abenaki, and Western Abenaki — are the Northern New England languages, while the remaining five — Loup/Nipmuck, Massachusett, Narragansett, Mohegan-Pequot-Montauk, and Quiripi-Unquachog — are Southern New England languages. The fact that "Loup" fits much better with the Southern New England languages, at least according to these diagnostics, is further evidence that it is indeed Nipmuck as opposed to Pocumtuck, which is located further north and should resemble Northern New England languages, at least in part. (1) Selected diagnostic phonological innovations. Mc MaPs EAb WAb Loup Ms Nr MoPqMn QpUq Abenaki syncope / / / / / *s > s / / / / / / / / / *nOC / / / / / / / y / *a- > a / / / / / / y *e- > a- / / / / / / ^-palatalization / / / / s / The five Southern New England languages share five features as shown in 1. In these five languages, Proto-Algonquian *s and *s merged A LANGUAGE ON THE BORDER 125 to become s; in a consonant cluster in which the first consonant is a nasal, that nasal will delete; Proto-Algonquian long *a- becomes nasalized a, and long *e- becomes a-; and finally, Proto-Algonquian *k palatalizes before long *e- and unstressed short *i. However, cross-border borrowing by Nipmuck from Western Abenaki, its northern neighbour, makes identification using diagnostics such as these somewhat tenuous. According to the table in 1, "Loup" (Nipmuck) has all the features found in the four other Southern New England languages listed, thus showing it to be a clear example of a Southern New England language. However, I have called it a border language, since detailed study of the manuscript shows that "Loup" also possesses features that only the Northern New England languages should exhibit. One of these is Abenaki syncope. Abenaki syncope is a pattern of vowel loss in which original short vowels (vowels which were short in Proto-Algonquian) are lost before a consonant cluster. This excludes consonant clusters of which the first member is a nasal, since in all these languages the nasal deletes, as shown in the table. Nipmuck seems to possess a similar trait of dropping unstressed vowels, particularly in the first syllable. Since Mathevet did not mark stress, this provides internal evidence to the language's stress pattern, which seems to match the Proto-Algonquian alternating stress rule. In Proto-Algonquian, every long vowel, and every other short vowel (starting at the left of the word, and then restarting after a long vowel) receives stress (Pentland 1992). Syncope of an initial short vowel (which can never receive stress according to the alternating stress rule) before a consonant cluster is particularly prevalent throughout the manuscript. Examples of this are given in 2. The first word of each line is Proto-Algonquian, followed by a phonemic form of the Nipmuck word, followed by the word as it appears in Mathevef s orthography in the manuscript, in angled brackets. The symbol (8) that appears in Mathevet's orthography represents the missionaries' orthographic device of an omicron with an upsilon above it, for what the French would write <ou>. In Nipmuck, this generally represents o- or we between consonants, w before vowels, and a short vowel + w at the end of a word. The page on which the word may be found in the manuscript is in parentheses: 126 HOLLY GUSTAFSON (2) a. *espemenki > spemek (spemik) 'up above, up high' (90) b. *a6koka > skok <sc8g) 'snake' (53) c. *ahtemani > hteman (temanes) (pi.) 'snowshoe harness, lace' (78) Similarly, Proto-Algonquian short *e deletes between a consonant and a consonant cluster at the beginning of a word: (3) *e > 0/#C CC This is recognizable in the manuscript by the lack of both the *e and the preceding nasal consonant, which Mathevet apparently did not hear, and therefore did not record: (4) a. *mexkwamya > mhkwamiy (k8ami) 'ice, piece of ice' (107) b.