-
Analysis of Texts and a Basic Lexicon of the Abkhaz Language by Tamio Yanagisawa, in cooperation with Ana Tsvinaria, and: Analytic Dictionary of Abkhaz by Tamio Yanagisawa, with the assistance of Anna Tsvinaria-Abramishvili
These two volumes represent the English debut of a Japanese linguist and student of literature, Tamio Yanagisawa, an expert on Russian and Abkhaz associated with Nagoya University, in the field of Caucasus studies. They represent an enormous amount of effort and scholarly commitment. Further, they are the culmination of a series of studies of Northwest Caucasian languages from 1997 on that are apparently in Japanese. Taken together, along with one of the available grammars (Hewitt 1979; Chirikba 2003), the investigator should be able to acquire reasonable insight into the structure of this remarkable language.
The books are the result of six years of intensive fieldwork, conducted in Georgia, with the late Ana Tsvinaria (Abramishvili), who was born in Ochamchira (Analysis of Texts, p. 532). She spoke the Abzhywa (Abzhwi) dialect upon which the literary language is based. The language in both books is essentially the literary one; neither other Abkhaz dialects nor the Abkhaz dialect situation in general are discussed. Yanagisawa uses the modified Cyrillic alphabet of the literary language, adding a transliteration into a Latin-based script in the analyzed texts and grammatical discussions in both volumes. His transcription follows Hewitt (1979), as does the overall conceptual structure of his grammatical account. This approach results in a thorough account of many verbal and syntactical details, but also couches many of these same details in a somewhat opaque form; I discuss some examples of this opacity below.
Overall, Abkhaz is remarkable at many levels. It has a large consonantal inventory (between fifty-seven and seventy-one consonants, depending upon the dialect) that contrasts with a minimal vertical vowel system that opposes an unspecified syllabic peak (∂) to an open one (a). The consonants can occur in seemingly unconstrained two-member clusters, with a few three-member clusters occurring; some clusters involve gemination. Nouns are largely built up by elaborate processes of compounding, while the polypersonal verbs show elaborate modification by affixation. Lacking case, Abkhaz is a morphologically ergative language, with the form and order of personal inflection on the verb (prefixal person markers, called “indices” by many scholars) manifesting a split ergativity that is dependent upon syntactic processes of transitivity reduction and enhancement. Transitive verbs are ergative unless the transitivity of the verb is reduced in some way, in which case they show antiergative (antipassive) inflection. The language has several copulas and, curiously, one of them requires its subject to appear in an oblique or ergative form (Hewitt 1979:105). In addition, reflexives are transitive (ergative), with the reflexive expression in the object slot, while reciprocals are detransitivized (antiergative), with the reciprocal expression in an indirect object slot, contrary to constraints on anaphors espoused in current linguistic theory. Abkhaz syntax follows subject-object-verb order and is verb-dominated, with subordination often marked by the inflection of [End Page 184] the embedded verb. Transitive verbs may be cliticized to direct objects, and intransitives with third person subjects can be cliticized to the subject as well (Chirikba 2003:42). Multiple embeddings may show verb chaining if the lowest verb has been cliticized to its object. Abkhaz is one of the few languages known with unambiguous rightward question movement. Clearly, for the theorist this language promises some valuable insights, and these volumes now make it accessible to the English speaker. By providing both texts and dictionaries, Yanagisawa has provided the linguistic community with a valuable resource.
The Analysis of Texts volume, besides its front matter, contains four analyzed texts (pp. 9–144), a lexicon (pp. 145–524), a list of prefixal particles (pp. 526–33), and a reverse dictionary (pp. 534–64). The analyzed texts include a myth about the Atsan (ac’án ‘fairy-like beings’) and three folktales. Yanagisawa has published several additional analyzed texts as journal articles, one of which (evidently the last part of a folktale) is accessible online (http://www.lang.nagoya-u.ac.jp/proj/genbunronshu/30-1/yanagisawa.pdf). The lexical components of this volume provide a near-exhaustive collection of the data needed to work with these texts, though to a large extent they are superseded by the subsequent Analytic Dictionary. Texts are given both in the Abkhaz Cyrillic alphabet and in phonemic Latin-based transcription (the online text omits all Cyrillic); grammatical glossing is provided, and finally an English free translation. At the end of each text is a series of grammatical notes, which often give alternative examples of a root or a construction. The material is illuminating, but could be improved in a few respects. Occasional minor omissions or unclarities of glossing occur. For instance, line 1 of the first tale ends in an existential verb that has an anomalous personal inflection, a schwa. This verb is not cliticized to the preceding noun, so this inflection is unexplained; it is the sort of personal index seen in Ubykh forms. In lines 3 and 4, a personal prefix –a– is left unglossed on two weather verbs; this is simply the indirect object index (person marker) referring to ‘mountain’ in line 3, upon which the weather blows. Some of these minor problems are more pervasive. The grammatical process whereby verbs can be cliticized to the preceding noun is not noted anywhere in the texts, although examples of it occur—the two verbs in lines 3 and 4 of the first tale just mentioned have a zero initial personal index because they have been cliticized in this way. Another such problem is that, following Hewitt’s practice, geometrical preverbs are glossed simply as “PREV” (for “preverb”), whereas the different preverbs often supply important elements of the verb’s semantic content. So, for example, line 2 of the same text includes the verbs j–k°–la–n∂ (they-prev-climb-absolutive) ‘they climbed onto the top’, and j–k’á–r–č’a–wa–n (it-prev-they-cut.off.(a.branch.of)-dynamic-imperfective) ‘they would cut them down’; the preverb in the first verb means ‘top, on top’, while that in the second means ‘down’. (See Serdjučendo [1956] for a clear exposition of such preverbs in the closely related Abaza.) (The glossing of the final morpheme of the first verb as ‘absolutive (gerundive)’ is erroneous, too; rather, it is a “pro-tense” —that is, a morpheme occupying the position of tense. A verb ending with a pro-tense morpheme has its tense determined by that of the last verb in the sentence, in this case –wa–n, a dynamic imperfective or habitual.)
The glossing also provides no information on derivation. For example, in line 35 of text 4 (p. 135), the verb root –lac°q’°s– ‘wink at’, given as a unit, may be analyzed as –la–c°–q’°s– (eye-skin-flex.down-hit), where ‘hit’, –s–, lends the verb a sense of quickness and direction; without –s–, la–c°–q’° simply means ‘to blink’—that is, ‘to lower the skin of the eye (the eyelid)’. In line 5 of text 1 one finds járbanzaalák’–gy∂ ak’ (anything-also something). In fact, this expression shows the interesting way in which Abkhaz constructs indefinite pronouns. A derivational analysis would be já–rban=z–a–l–á–k’–gy∂a–k’– (what-pronominal.suffix =indefinite.index-hither-be.located-in-one-and the-one); já rban=z–a–l–á–k’–gy∂ by itself can be rendered as ‘someone somewhere’, and the [End Page 185] whole expression as ‘the one who is something/someone somewhere’. The root ja– [ye] is the nonhuman interrogative suffix for verbs, –ja ‘what’, and –rban makes it an independent pronoun, just as in the next sentence the human interrogative suffix for verbs, –da ‘who’, appears as a free pronoun with the same suffix, da–rban ‘whoever’. Yanagisawa has failed to note the parallel. This failure to supply information on derivational morphology extends to the lexicon as well. Thus, the causative forms of verbs, semantically transparent, are all listed under the the letter r simply because the causative prefix is –r–. Yanagisawa justifies this by noting that nothing can come between causative –r– and the root, but this assumes that a grammar is constructed purely on distributional criteria without any semantic competence.
The account of grammar in the Analysis of Texts volume (pp. 526—33) focuses on the prefixal particles of the verb, both personal particles (the indices) and geometrical particles (preverbs) with adverbial senses. The geometrical particles convey a sense of suddenness or satisfaction, reminiscent of the ostensibly geometric preverbs of Russian or Georgian, which have acquired an aspectual sense. Yanagisawa shows, however, that these particles, largely confined to oral narratives, act more like subordinators in that they occupy the second position from the front in the verbal complex and tend to convey that one action is a result of another (Analysis of Texts, p. 532; Chirikba 2003:43, 54).
A schematic phonological inventory is provided in the text volume (p. 6). The voiced pharyngeal is absent (see below) and the rounded voiced pharyngeal is represented by y°. No comment is made about the “rounded” alveolar stops t° d° t’°. This is unfortunate since it would be useful to inform the reader that these are actually realized with simultaneous labial and alveolar release, that is, that they are pronounced [], [], and []. Equally unusual is the realization of the rounded alveolar affricates, c° Ʒ° c’°. These have a form of labiodentalization in which the lower lip is brought up over the tips of the upper teeth. One should also know that unmarked č ǯ č’ are retroflexed or velarized, whereas those with diacritic yod (j) are laminal. In addition, no mention is made of how the vowels assimilate to adjacent consonants, one of the more interesting features of the vertical vowel system found in this language and its sisters.
The “User’s Guide” and “Grammatical Sketch” at the front of the Analytic Dictionary (pp. ix-xxxvii) begin by discussing stress and the issue of initial a–. This is all the phonology to be found in either volume. Stress is unpredictable in Abkhaz, and is typologically interesting because it is reminiscent of the stress patterns reconstructed for Proto-Indo-European (Fortson 2010:119-23). Yanagisawa determines stress (p. ix) by following the Abkhaz-Russian Dictionary (abbreviated ARD, p. xxxv) of V. A. Kaslanzdija, or by etymology, though how the latter is done is never explained. The phonological process of a-deletion is mentioned. Nouns, infinitives (“masdars”), adjectives, and adverbs all begin with an indefinite marker, a–, which deletes when the form begins with an underlying a–. The issue of initial a– is linked to an oversight in the phonological inventory (Analysis of Texts, p. 6). Yanagisawa notes that some preverbs with ḥ can yield [–aa–] as a variant by assimilation to a voiced consonant (pp. x, xxiv–xxv). Like the two other authors of Abkhaz grammars in English, he has overlooked the simple phonological process by which the voiceless pharyngeal ḥ can voice to —just as happens with other personal indices, e.g., s becoming z—to produce a surface [–aa–]. The output of this process is clearly a violation of the a-deletion rule-evidence that [aa] cannot represent a–a, and an alternative explanation is required. The result of this analytical oversight is that is omitted from the phonological inventory.
Yanagisawa sketches some aspects of the grammar of nouns, adjectives, adverbs, and verbs in the introductory matter. In dictionary entries, nouns are cited with their plural, indefinite, and possessed forms, and with some postpositions. Verbs are entered as infinitives, including any lexically bound prefixes and showing voicing variants of [End Page 186] personal indices where applicable. Many verbs are given not only in their present forms, but also in the present negative form, where negation is a suffix, or the aorist (past) negative form, where it is a prefix; emphatic forms are also often given. A columnar schema (pp. xiv–xv) is used to organize information about the complex verbal morphology. Column 1 (“C1”) contains absolutive personal indices (subjects of intransitives and objects of transitives). Column 2 (“C2”) contains the subjects of two-place transitive verbs, including the causer of a causative verb derived from an intransitive one. Column 3 (“C3”) contains the subject of a ditransitive verb or the causer of a causative verb derived from a transitive one; if either of these is present, column 2 is then used for indirect objects. (That is, column 2 contains oblique or ergative personal indices.) Preverbs occur only between columns 1 and 2. Some preverbs act like incorporated nouns and occupy column 2, forcing the subject into column 3. Abkhaz, like the related Ubykh, exhibits a special third person plural column 3 index, –na–.
Reflexives are listed as variants under verb entries in the dictionary portion (but not reciprocals, which are antiergatives; see Chirikba 2003:154). The picture here is obscured to some extent by Yanagisawa’s use of the concept of version (“voice”), which was developed to deal with the details of the Georgian verb and is best suited to that language. Subjective version is simply a reflexive in which the anaphor (‘self’) is an indirect object (column 2). What Yanagisawa categorizes as two types of object version (p. xii) are merely an adjunct benefactive, e.g., j–l–z –q’a–s–c’a–jt’ (them-her-for-on.top-I-set-past) ‘I set them on top for her (benefit)’, and an adversative, e.g., j–l–c°–q’a–s–c’a–jt’ (them-her-despite-on.top-I-set-past) ‘I set them on top despite her’, respectively. The last two verb forms appear to have “extra” preverbs between C-1 and C-2, –l–z∂– ‘-her-for-’ and –l–c°∂ ‘-who-despite-’. In fact, only arguments can occupy columnar slots, and these “extra” forms are not arguments, but adjuncts, a benefactive and detrimentive, respectively. They have been incorporated into the verbal inflection. Such adjunct inflections always follow the initial C-1 slot. The columnar system of Yanasigawa does not account for these adjuncts.
The grammatical sketch in the Analytic Dictionary (pp. xiii–xxix) deals entirely with the verb and its syntactical variations. Here, forms are cited both in Cyrillic and Latin-based transcriptions (with j used for yod), whereas Cyrillic alone is used throughout the rest of the volume. Yanagisawa presents an array of twelve morphological slots for the verb (Chirikba gives seventeen [2003:37–39]), and a display of the personal affixes (often called “indices” by other scholars). Then follow examples of “non-derivational” verb formation, that is, inflected verbs that are not causatives and lack other modifications of their arguments. The primary division is into statives and dynamic forms. Statives can be either one-place (e.g., ‘be sitting’) or two-place (e.g., possessive verbs such as ‘have’, with possessor coded as dative). Dynamic verbs show more possibilities—one-place intransitives (e.g., ‘jump’), two-place intransitives (e.g., antiergative forms related to transitive verbs), two-place transitives (e.g., ‘kill’), and three-place transitives (e.g., ‘give’). Also discussed here are forms with preverbs, such as those in bold type in the following examples: d–l–x°a–pš–wa–jt’ ((s)he-her-at-look-dynamic-finite) ‘(s)he is looking at her’, d–na–la–šyt–wa–jt’ (him/her-thither-she-send- dynamic-finite) ‘she is sending him/her thither (away)’, j–[a–]ta–s–zy–wa–jt’ (it/them-[it-]into-I-throw-dynamic-finite) ‘I am throwing it/them into it’. (I have simplified the glosses.)
The section on derived verbs (pp. xvii–xxii) deals with causative, potential, and nonvolitional forms. The causatives simply adds an extra argument to the verbal complex, as in d–l–r–pa–wa–jt’ (him/her-she-causative-jump-dynamic-finite) ‘she made him/her jump’. The potential suffix –z– follows the agent index. If the verb is transitive, the complex of agent plus potential affix is also moved into an adjunct position (Yanagisawa’s column 2). This is most clearly seen in three-place transitives; thus, note [End Page 187] the different positions of the agent index in d–b–s–ta–wa–m (him/her-you.feminine -I-give-dynamic-negative) ‘I do not give him/her to you (feminine)’ and the potential form d–s∂–z–ba–ta–wa–m (him/her-I-potential-you.feminine-give-dynamic-negative) ‘I cannot give him/her to you’. Similarly, the marker of involuntary agent, –ámxa– (which attracts stress), is placed after the agent index, with the two morphemes fronted into the adjunct (column 2) position when the verb is transitive, as can be seen by comparing the ordinary transitive j–p–s–q’a–Ø–jt’ (it/them-sever-I-cut-aorist-finite) ‘I cut [through] it/them’ (Yanagisawa’s gloss is simply ‘I cut it/them’), where the agent s– ‘I’ follows the preverb p–, with the involuntary transitive j–s–ámxa–p–q’a–Ø–jt’ (it/them-I-involuntary -sever-cut-aorist-finite) ‘I accidentally cut [through] it/them’, where s– and –amxa– precede the preverb. Reflexives are shown to be simple transitives. Reflexives of two-place intransitives can be rendered periphrastically as ‘reflexive possessorhead, e.g., l–x d–á–s–Ø–jt’ (her-head (s)he-it-hit- aorist-finite) ‘she hit herself, instead of d–l–č–s–Ø–jt’ (she-her-self-hit-aorist-finite).
The next section (pp. xxii–xxiv) deals with verb classes. There is a class of verbs whose agent is an experiencer of some sort. Typically, the syntactic and morphological patterns diverge here, with the syntax maintaining the usual subject-object-verb order, while in the verb the syntactic object is represented by a subject index while the agent or syntactic subject is represented by an indirect object index (which fails to show any voicing) followed by a morpheme indicating the experiencer relation (which is invariably –g°a–), the indirect object index following the subject index as usual; this can be seen in amra murát d∂–l–g°a–pxa–wá–jt’ (Amra Murat he-her-heart-please-dynamic-finite) ‘Amra likes Murat’. Other verbs, called “labile” (from a term used by Russian linguists, labilnyj) form intransitive-transitive pairs with constant agent, with normal syntactic and morphological accord; compare the intransitive d∂–Ʒ°Ʒ° a–Ø–jt’ ((s)he-do.washing-aorist-finite) ‘she/he did washing’ with the transitive y–l–Ʒ°Ʒ°á–Ø–jt’ (it/them-she-wash-aorist-finite) ‘she washed it/them’. Another set of verbs also falls into intransitive-transitive pairs, but in these the patient of the action is constant; for example, the intransitive a–mč’ bzja–n∂ j–b∂l–wá–jt’ (the-firewood good-adverbial it-burn-dynamic-finite) ‘the firewood burns well’ is opposed to the transitive á–mra s–bγaØ–á–b∂l–wa–jt’ (the-sun my-back (it-)it-burn-dynamic-finite) ‘the sun burns my back’ (the initial index because the verb is cliticized).
The final section of the grammatical sketch (pp. xxiv–xxvii) deals with criteria for distinguishing transitive and intransitive verbs. Some of these are phonological: for example, the agent index before a root with an initial voiced consonant voices in a transitive verb, but fails to voice in an intransitive one. Others are morphological: preverbs that govern a personal index never occur immediately before a root in transitives, but can do so in intransitives. A similar distinction holds with regard to potential and nonvolitional suffixes. The second person singular is overt in imperatives of intransitives, as in b∂–s–s (you.feminine-me-hit) ‘hit me!’ (an intransitive verb with an indirect object), but not in imperatives of transitives, as in d–Ø–bá (him/her-(you-)see) ‘see him/her!’ A similar distinction is found in gerundive (“absolutive”) forms, e.g., transitive d–ba–n (him/her-see-absolutive.past) ‘having seen him/her’, as opposed to s–b–s–n∂ (I-you. feminine-hit-absolutive.past) ‘I having hit you (female)’. The final criterion, and the one probably most significant for speakers, is the relation between the order of syntactic constituents, namely, subject–(indirect object)–object–verb, and the order of inflectional elements. For intransitive verbs, the order of inflectional indices parallels the order of syntactic arguments: the subject index precedes the object index. (Object indices can appear as “demoted” indirect objects in antiergatives and certain “indirect verbs,” such as ‘to hit’.) For transitive verbs, the order of indices is the mirror image of the syntax-the object index, the indirect object index, and the subject index occur in [End Page 188] that order. I might add that efforts to explain Abkhaz verbal inflection in terms of simple cliticization processes would find such variation to be a significant difficulty.
The entries themselves are a gold mine of forms. Finite, nonfinite, and absolutive (gerundive) forms are given for all verbs; potential, nonvolitional, and negative forms are also provided if they appear in the texts. This exhaustive listing makes the dictionary enormously useful. Each entry specifies the complete potential morphological structure of a verb by means of the column-based schemata discussed above. Thus, for example, the potential form of an antiergative verb is given as “C1-Pot-C2-R” (where “C” stands for “column”, “Pot” for “potential marker”, and “R” for the root). Every inflected form is thus implied, as the column labels “C1,” “C2,” etc., stand for the paradigms of personal indices that represent the verb’s arguments. Anyone working with the texts in the 2006 volume will have little trouble finding verb forms in the dictionary portion of that text or the full dictionary of 2010. Together these two volumes constitute the most extensive grammatical attestation yet provided for this language.
Having praised Yanagisawa’s extraordinary thoroughness, I must point out, however, that in one important respect his dictionary is opaque. Related verbs with different preverbs are entered separately under their preverbs, as though their senses were unrelated. Moreover, these preverbs are sometimes unglossed. In fact, preverbs typically have clear senses that are added to the basic sense of the root in a transparent manner. For example, both the verbs a–n–la–ra ‘to penetrate the boundaries/limits of; to be reflected in; to be printed’ and a–tá–la–ra ‘enter/go into and enclosed/fenced-in area; to bathe in the river/sea; to enter (university, etc.)’ are based on a root –la– ‘be in’ (which in turn is built from the preverb –l– ‘in’ plus an inessive or illative derivational suffix –a–), but with different preverbs preceding it; –n–l+a– (-in/on-in+inessive -) is ‘to be on’ (less often, ‘to be in’), and –t–a–l+a– (-inside-illative-be.in-illative-) is ‘to go into’. (For evidence of the internal structure of the preverb –t–a– ‘into’, compare –t– ‘out of, through’.) But the two verbs appear at widely separated points in the Analytic Dictionary (p. 290 and p. 369, respectively), without the connection between them being noted. Yanagisawa’s policy obscures an interesting and systematic aspect of Abkhaz verb formation.
A practice followed in all dictionaries of Abkhaz, including Yanagisawa’s, is that the noun marker a– is included in the headwords of all noun entries (including infinitives, which, of course, are nouns), while entries are alphabetized by the consonant that follows a–. (The motivation for this practice is that it is ungrammatical for a noun to appear without a– in Abkhaz.) This has, however, led to some errors in the listings because some words begin with a nonmorphological a, such as áž°a ‘word’, which is listed under ž°[Analytic Dictionary, p. 87], not under a. In fact, this word is always áž°a, whatever its referential value and should be entered after p. 16 in that dictionary. Unfortunately, Yanagisawa does not explain this convention of pervasive initial a–. Readers unfamiliar with the convention might be excused for finding it odd, if not downright misleading.
Overall, despite minor shortcomings like those discussed above, these two volumes are a valuable and lasting contribution to the documentation of this endangered language, as well as a valuable resource for the typologist and theoretical linguist.
References
2003 Abkhaz. Munich: LINCOM-Europa.
2010 Indo-European Language and Culture: An Introduction. 2d ed. Malden, Mass.: Wiley-Blackwell. [End Page 189]
1979 Abkhaz. Lingua Descriptive Studies 2. Amsterdam: North-Holland.
1956 Kratkij grammatičeskij očerk abazinskogo jazyka. In Russko-abazinskij slovar’, edited by X. D. Žirov and N. B. Èkba, 593–646. Moscow: Gosudarstvennoe Izdatel’stvo Inostrannyx i Nacional’nyx Slovarej.