Printer Friendly

Une theorie de l' organisation du lexique des langues semitiques.

Une theorie de l' organisation du lexique des langues semitiques: Matrices et etymons. By GEORGES BOHAS and MIHAI DAT. Collection Langages. Lyon: ENS EDITIONS, 2007. Pp. 235 (paper).

The book under review represents a synthesis of the research on word formation in Semitic and the organization of the Arabic lexicon undertaken by Georges Bohas in the last fifteen years. In addition to the Arabic material that formed the basis of Bohas's previous studies, it also includes Hebrew data provided by Mihai Dat. As such, the book, although it extends the analysis to another Semitic language, does not provide any new theories per se.

Bohas's work on word formation in Semitic is well known among Semitists. It has fueled the centuries-old debate of whether the Semitic lexicon is primarily built on biradical or triradical bases, that is, whether triradical roots are original or developed out of a more archaic biconsonantal base.

The concept of a triradical root was originally developed by Arab grammarians in the tenth century C.E. It commonly underlies the analysis of the Semitic lexicon, morphological descriptions, and the organization of the dictionaries of individual Semitic languages. In the last two decades, Bohas has vehemently criticized this concept of a triradical root, in particular its use for the organization of the lexicon. In the present volume, Bohas and Dat summarize arguments previously brought forth against an original triconsonantal root, of which I will mention the most significant in the following paragraphs.

According to Bohas and Dat, the assumption of a triradical root that underlies the organization of the lexicon fails to explain much phonetic and semantic regularity in Semitic, such as polysemy and homonymy (p. 10). The triconsonantal theory often associates words that share the same three basic consonants with the same root, even when its presumed derivatives express unrelated semantic notions, as in the case of sbr, whose meanings range from "to connect" to "pile of food" and "mountain" (p. 22). Furthermore, the assumption that the verb in Semitic in its unaugmented form consists of a triconsonantal root might be further challenged by evidence from weak verbs in which only two consonants are present, as in Arabic rama, da a (3ms perfect). Although the third consonant can be deduced by comparison with other forms in many instances, as in ramaytu and da'awtu (1cs perfect)--which show that the third root consonants are /y/ and /w/ respectively--this is supposedly not possible in all cases or languages (pp. 15-18). In addition, the triconsonantal root is not accessible to native speakers of Semitic languages such as Arabic in an immediate way, as has been shown by experiments conducted by Bohas and Razouk. In these experiments, native speakers were asked to extract the triconsonantal roots of certain words and to determine their patterns. In most cases, they failed to do so (pp. 12, 24f.). Based on this experiment, Bohas concludes that native speakers of Arabic make no conscious use of an underlying root (p. 26). Consequently, the triconsonantal root is no linguistic reality but merely a grammatical tool (p. 28).

Most importantly, the concept of a triconsonantal root cannot account for the phonetic and semantic relationship between words (pp. 22, 29). In order to prove this argument, Bohas and Dat list triconsonantal roots that have two consonants in common and a similar semantic range, such as Arabic bwh, bhh, and bhw, which all share the basic meaning "to calm down" (p. 29). Most of the roots listed either reflect geminates, roots containing a glide or guttural, and roots containing a sonorant (pp. 29-38). Although the examples provided are often similar in meaning, they are hardly ever real synonyms. Nevertheless, according to Bohas, the existence of such semantically and phonetically similar words can hardly be a coincidence. Consequently, he assumes that there exists a phonetic and semantic relationship between the individual members of each set and that it should be possible to propose a biconsonantal base that provides this phonetic and semantic link (p. 39). The main point of criticism Bohas brings forth against the common organization of the Semitic lexicon is that it fails to reflect this seemingly obvious relationship between the respective roots (p. 40). As an alternative to the traditional organization, Bohas proposes the following model, which is developed in detail in chapters 1-6 on the basis of Arabic and Hebrew. Since the description of Bohas's theory is spread throughout the present volume, I will not follow its outline directly but summarize the main points of his theory in the following paragraphs.

Bohas's main assumptions regarding word formation in Semitic is that the lexicon consists of three levels: the matrix / template ([mu]), the etymon ([member of]), and the radical (R).

The matrix is a combination of elements of phonetic character, such as [labial], [coronal], [+/-voice], [fricative], etc., that carry the basic semantic notion of a given stock of words (pp. 10, 46, 52). This means, unlike the common assumption that the phoneme is the most basic meaning differentiating unit, Bohas considers phonetic features the ultimate distinctive units of a language that cannot be divided further (p. 100). The matrix usually consists of two of these elements that can be characterized by one or more phonetic feature (pp. 44-46). The matrix itself is not linear, that is, the order of its elements is interchangeable. The assumption of an underlying matrix allows connecting surface forms that would not even be connected by the most adamant biradicalist. The matrix {[labial], [coronal, pharyngeal]}, for example, presumably accounts for the relationship between words such as sabara, dabba, 'ibadun, dafrun, taffa, tafana, zaffa, and rabata, which share the basic meaning "to bind." According to Bohas and Dat, all of these lexemes are surface representations of the same matrix (pp. 46f.).

Furthermore, Bohas attributes semantic notions to specific feature combinations in the attempt to establish a close relationship between sound and meaning. For example, the matrix {[labial, -sonorant], [pharyngeal]} presumably has a basic notion "to constrict." A pharyngeal constricts the pharynx when it is pronounced, and thus there exists a connection between the invariant semantic notion of the matrix and the manner of articulation of its elements (p. 118). This connection of meaning and articulation is not limited to sounds. Certain matrixes presumably are connected to forms, such as [mu] {[labial, -sonorant], [dorsal]} with the presumed notion of converse [intersection]. concave [union], and round O forms based on the manner of articulation of the contained dorsal (p. 124). Examples for this specific matrix include rukba 'knee', 'aqibun 'heel', butnun 'belly' for 'convex', jubbun 'pit, cistern' for 'concave', and ja-bdjibun "drum" and falakun 'globe' for 'round' (pp. 125-30). Other examples of matrixes that reflect notional or semantic fields include {[labial], [coronall]} 'to strike a blow' and {[labial, -sonorant], [-voice, + fricative]) 'air movement, wind, blowing'(p. 197).

The concept of a matrix, whose meaning is derived from the manner of articulation of its elements, supposedly shows the mimophonic origin of the lexical forms (p. 11), that is, the most basic meaning of a lexeme is derived by the phonetic imitation of the real world object. According to Bohas, it is the relationship of different surface forms and the relationship of phonetic features and meaning that cannot be detected with the usual concept of a triconsonantal root. In his work, Bohas tries to create a system that assigns a conceptual value to combinations of phonetic features and thus to establish the relationship between sound and meaning.

The second level, the etymon ([member of]), is a combination of usually two consonantal phonemes that are surface representations of the underlying matrix. These two phonemes supposedly do not operate on the same level as the previously assumed biradical roots (pp. 10, 52). The matrix {[coronal], |dorsal]}. for example, yields etymons such as {q, $}, {k, i}, {g, d}, and {k, $}, all of which share the conceptual field 'cut, decide, tear, destroy, death' derived from the underlying matrix (p. 147). The etymon not only shares the basic semantic notion with the matrix, but also its non-linearity (p. 48). The non-linearity of the etymon, in turn, explains many instances of metathesis that occur in the Arabic lexicon. In fact, according to Bohas, metathesis in a root does not exist since different phoneme orders in semantically related words simply reflect the possible variants of the underlying etymon (p. 67). The words butta, 'inbata'a, balaka, batata, batata, and sabala, which all express the basic notion 'to cut, separate, to be isolated', can therefore be derived from a single etymon {b. t}, meaning 'to cut' (pp. 41f.).

The third level, the radical (R), is the etymon extended by a third consonantal phoneme. It is the radical that constitutes an autonomous lexeme. The extension of the biconsonantal etymon can be achieved in various ways: by the reduplication of the second etymon phoneme, as in bu from {b, t}; by the addition of a sonorant /r. 1, m, n/ or guttural /', h, h, '/ in any position, as in 'inbata'a and batata; by addition of a glide /y, w/; by prefixation, as in sabata; by adding a final consonant, mostly /f/ or /b/; or by the merger of two independent etymons, as in bataka, which presumably derives from the etymons {b, t} and {t,k}, both with the underlying notion 'to cut' (pp. 56f.). These augmentations have no significant effect on the semantic range of a given etymon (pp. 64, 106). Instead, the addition of a third phoneme is supposedly caused by the fact that the morphology of Semitic languages is organized around triconsonantal patterns, to which the biconsonantal etymon has to adapt (p. 53). Semantic changes develop through the extension of the original notional field, metaphoric uses, derivation of abstract concepts, and associative extensions. Bohas provides a very detailed description of how different meanings develop out of an underlying basic notion. However, he never explains why a specific element is added to a given etymon, that is, why in a given case we find the addition of /y/ vs. /'/ etc. (p. 107).

Semantic changes can also be the result of the merger of etymons. In this case, the resulting etymon acquires a meaning that is derived from the notions of both original etymons. The radical nataka 'to pull something violently to oneself, to be at the breaking point', for example, supposedly reflects the merger of {n. t} 'to pull' and {t. k} 'to cut, break' (p. 59). When a radical has two contrasting meanings, this presumably likewise goes back to a merger of etymons. For example, sa'aba 'to collect, disperse' presumably goes back to the etymons {s, '} 'to disperse' and {'. b} 'to gather' (p. 60). The concept of a radical as being an extended etymon also explains homonyms in Semitic. Homonyms originate in different matrixes that acquired the same consonantal phonemes by the addition of a third consonant (p. 199).

The main conclusions of Bohas and Dat's study are that the lexicon of Arabic and Hebrew can be derived from a finite number of morpbosemantic structures, namely feature templates or matrixes. This type of organization presumably not only leads to a more profound organization of the lexicon, but also reveals the mimophonic character of the linguistic sign. The assumption that the linguistic sign is not arbitrary but phonetically related to the object it depicts is a decisive step in lexical theory according to Bohas and Dat since the arbitrariness of the linguistic sign has been one of the basic tenets of linguistics since Saussure (p. 135). Bohas and Dat reject this principle by claiming that mimophony, not arbitrariness, is the basic characteristic of the linguistic sign (p. 216). Furthermore, since the matrix and etymon are not linear, they argue against the commonly assumed linearity of the linguistic sign (p. 93). Especially in Semitic languages, the meaning presumably is not dependent on the spatial order of auditory or graphic signs (p. 94). Lastly, since the etymon is biconsonantal and the radical triconso-nantal, the old dispute about the basic form of the root has lost its basis (p. 132).

It hardly needs to be mentioned that Bohas's approach has been widely criticized. Many of the "arguments" brought forth against the triradical origin of Semitic words have long been known and have been convincingly refuted by scholars such as R. M. Voigt in his book Die infirmen Verbaltypen des Arabischen und das Biradikalismusproblem (Stuttgart: Franz Steiner. 1988). The commonly cited examples of roots that differ in one consonant and have similar meanings can more convincingly be explained by "Reimwortbildung," a concept already developed at the end of the nineteenth century. "Reimwortbildung" means that words that have similar meanings tend to assimilate on a phonetic level. Similarly, words that are similar in meaning and sound also tend to become closer in meaning (Voigt, 81). Other cases that involve geminates and roots containing a glide can often be explained as a mixing of root types. For example, in certain paradigms geminates and hollow roots can have the same pattern, as in Hebrew yasob (3ms imperfect of sbb) and yaqom (3ms jussive of qwm). These overlapping paradigms can lead to a reanalysis of the underlying root from an original geminate to a hollow root and the other way round, a process that is particularly observable in Hebrew (see also Voigt, 85f.).

The most important argument against the assumed development of root augmentation as proposed by Bohas and Dat is that the addition of the third consonant is completely random. Despite the fact that root augmentations can be traced in other language families such as Indo-European, for which the concept was developed, it has not been possible so far to prove that such a process ever operated in Semitic. This is caused by two main factors: first, in IE the augmented element can only occur in final position, and, more importantly, the elements originally had a specific semantic function that was later lost (Voigt, 50). No such original meaning of any root augment has ever been convincingly suggested in the case of Semitic. Bohas and Dat do not suggest any functional explanation either; on the contrary: according to Bohas and Dat, there is no functional difference between adding a guttural, glide, or other consonant. That is, the augments themselves have no semantic or grammatical purpose that could explain why each is added in a specific instance. As mentioned above, Bohas and Dat assume that a third phoneme was added to the biconsonantal etymon in order to assimilate the etymon to the triconsonantal structure of Semitic morphology. Even if we were to accept that the Semitic lexicon went through a biconsonantal stage, there exists no reason why the morphology of Semitic should not have developed on the basis of biconsonantal bases, as happened in other languages such as Sumerian. The lack of any convincing explanatory model that can account for root augmentations in Semitic makes the development suggested by Bohas and Dat highly unlikely.

Furthermore, there are some methodological problems in the analysis of certain weak roots in Semitic as described by Bohas and Dat. For example, Bohas claims that there is no regular rule governing how diphthongs behave in Syriac and thus it is not possible to assume an underlying triradical root. He quotes the example of gba (3ms perfect G)--which has a comparable form containing lyl as third root letter in the 2ms perfect gbayt--in which the supposed original diphthong /ay/ of the 3ms * gabay contracted to /a/. In bayta, however, the diphthong is preserved and in 'elgabbi (3ms perfect Dt), it presumably contracted to /i/ (p. 19). Synchronically, Bohas has a valid point in doubting an underlying triconsonantal root in this case. Gba has a synchronic parallel in the 3ms perfect of strong roots gbar that suggests an underlying form ** gbay, while the Dt of strong roots has the form 'etgabbar that equally suggests an underlying form with final *ay, **'etgabbay. Based on these synchronic parallels, it is indeed difficult to suggest regular phonological processes that would produce the attested surface forms from an assumed underlying triconsonantal root containing /y/.

Phenomena like these can be worth investigating when dealing with the language from a purely synchronic perspective. The problem with Bohas's analysis is that he claims that it is valid on both a synchronic and a diachronic level (pp. 42, 52). The forms of Syriac just quoted can, of course, easily be explained from a diachronic perspective. The final /a/ of the 3ms perfect does not derive from a diphthong but from a triphthong *gabaya, in which *aya >as elsewhere in Aramaic. Original diphthongs as in bayta only contract in a few dialects, and the final /i/ in the Dt is derived from a form containing the original active theme vowel of the t-stem /i/, * 'itgabbiy, which was subsequently leveled to the original passive theme vowel /a/ when the t-stems in Aramaic acquired a primarily passive function. Historically, these forms thus behave perfectly regularly with regard to the phonological rules they underwent and Bohas's criticism of the triradical root on the basis of certain weak roots has no basis from a diachronic perspective.

The case of III-y roots in Syriac shows an important problem with Bohas's approach in general, namely the distinction between a synchronic and diachronic analysis. It is on the synchronic level that parts of Bohas's theory might actually have some interesting implications, as in the case of Syriac mentioned above, or in the case of lexemes and roots in Arabic and Hebrew that share certain sounds and have similar meanings. In these cases, it would be interesting to know whether native speakers connect roots such as bwh, bhh, and bhw to an underlying biconsonantal base. This question could easily be answered by the same experiment with which Bohas and Razouk tested the ability of native speakers to extract triconsonantal roots and patterns. In other words, Bohas's etymon should be experimentally verifiable. In the hypothetical case that such an investigation would yield positive results, one might reconsider the analysis of the lexicon--although it needs to be stressed that this would be from a purely synchronic perspective.

Another major problem of Bohas's theory is the assumed mimophonic origin of the matrix and thus of the linguistic sign. Bohas even suggests that, on a purely diachronic level, the combinations of phonetic features and notions seem to be tightly linked to the capacities of the first speakers to imitate real-world objects phonetically (p. 193). As mentioned above, this assumption violates one of the most basic and accepted principles in the study of language, the arbitrariness of the linguistic sign. The present reviewer, like the majority of linguists, is therefore not inclined to follow this essential part of Bohas's analysis.

But even if we were to follow Bohas's theory and admit the existence of matrixes that are derived by imitating objects phonetically, the attribution of a semantic field to a given matrix still remains problematic. For example, the aforementioned matrix {[labial, -sonorant], [pharyngeal)} acquires its basic notion of 'constriction' through the constriction of the pharynx when the pharyngeal is pronounced. The matrix, however, also includes a non-sonorant labial /b, f/, which, as far as the present author understands, does not contribute to the semantic field of the matrix. In the matrix ([labial, -sonorant], [dorsal]}, which expresses concave, convex, and round shapes, the labial presumably causes external roundness of the muscles around the mouth, while the dorsal causes internal roundness of the tongue when it touches the palate (p. 123). Despite the fact that the pronunciation of /f/ does not cause any rounding, it is not clear why the same presumed rounding is not part of the semantics of the previous matrix which contains the same [labial, -sonorant] element. The attribution of meaning to certain elements of a given matrix and their effect on the semantic field of the matrix as a whole is therefore not regular and thus fails to be convincing.

Furthermore, assuming notional fields that can be expanded through various processes allows us to connect almost every term that is only remotely associated with a given notion. It would be no problem to connect English verbs like "to note, write, inscribe, incise, scratch," all of which contain a coronal /n/ or /t/, to the notional field 'to make a sign/ put into writing' that was extended by semantic association. It hardly needs mentioning that such an approach would not be very efficient for the ordering of the English lexicon and the same is true for Semitic.

Lastly, the discussion of previous literature here is rather selective and primarily includes the work of proponents of the biradical theory. In particular, studies in favor of a triradical origin of Semitic lexemes, such as Voigt's important book (1988) and proponents of a milder biradicalism such as Za-borski and others, are missing from the bibliography (e.g., A. Zaborski, "Biconsonantal Roots and Triconsonantal Root Variation in Semitic: Solutions and Prospects," in Semitic Studies in Honor of Wolf Leslau on the Occasion of his Eighty-fifth Birthday, ed. Alan S. Kaye [Wiesbaden: Harrassowitz, 1991], vol. 2, 1675-1703).

In short, the book under review is a nice summary of the theories developed by Bohas in the last two decades. The theories themselves, however, should be treated with caution.

REBECCA HASSELBACH

UNIVERSITY OF CHICAGO
COPYRIGHT 2008 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:The Journal of the American Oriental Society
Article Type:Book review
Date:Apr 1, 2008
Words:3532
Previous Article:The Modal System of Earlier Egyptian Complement Clauses.
Next Article:A Ninth Century Bookman in Baghdad.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters