A number of early grammatical descriptions of Tlingit, a non-Athapaskan Na-Dene language spoken in southeast Alaska and in adjacent areas of Canada, describe the language as being OSV. Evidence is presented in this paper that Tlingit is not OSV.

Until recent years, it was thought that of the six logically possible word orders of subject, verb, and object, only SOV, SVO, VSO, and VOS existed. (See for example Pullum 1977). Since then, however, evidence for the existence of object-initial languages has been presented by Derbyshire (1977) and Derbyshire and Pullum (1981). Almost all of the object-initial languages discussed by Derbyshire and Pullum are spoken in or near the Amazon basin in South America. There is to date no clear case of an object-initial language spoken outside of South America.

One of the principles of grammar introduced in Chomsky (1981) is the Empty Category Principle, which can be stated as follows:

• (i) [α e] must be properly governed.

Proper government, as defined by Chomsky, is of two types (see footnote 2). First, there is proper government by a lexical category of the [±N±V] type. In (2a), for example, the empty category (ei) is properly governed by the verb see, a lexical element of the [-N+V] type. Second, there is proper government by coindexation. In (2b), for example, the empty category in subject position (ei) is properly governed by the coindexed trace in COMP (ti).

Paradis (1980:501) observes that

The prosody of Acadian French has received relatively little attention in formal analyses of the dialect, though few who are acquainted with this variety of French, whether linguist or not, would hesitate to give an impressionistic judgement when asked to describe it. The delivery (débit) is characteristically slow, the melody “chantante” (‘singsong’). V. Lucci (1972:121) attributes this impression to the large number of accents, represented by repeated short, rapid rises in pitch. At the same time, he notes a feature that has been referred to in other descriptions of Canadian varieties of French (cf.J.-D. Gendron 1966:142–146; M. Boudreault 1968:87–99) that results in rhythmic patterns not encountered, in theory, in standard French. This dialectal feature is pretonic syllable lengthening, which, according to Lucci, although it appears to be an accent, is in reality a preaccent, the articulatory strengthening of the syllable preceding the accented syllable, which, of course, is the final one. This feature, whether in Acadian or in other varieties of French (cf. F. Carton 1980:85), is in fact generally ascribed to the durational characteristics of the vowel; whether or not the resulting syllable prominence should be interpreted as accent (displacement of the tonic accent, pretonic stress) has been a matter of dispute. Both Boudreault and Gendron, it is true, in describing the phenomenon, refrain for the most part from referring either to syllable prominence (perceptual judgement or physical measurement) or to accent. The former, however, does attribute one group of examples to the presence of an accent d’insistance; these examples are, in fact, stressable monosyllables, pretonic in the word group. The remainder, according to him, are a residue of intrinsic, etymologically motivated vowel length, the explanation preferred by Gendron. According to Gendron, the tonic accent still falls on the final syllable, although it is weak (in relation to a stronger pretonic syllable), and he criticizes J.-P. Vinay (1955:75), who speaks of displacement of the tonic accent.