Institut de Lingüística Aplicada

Third Mediterranean Meeting on Morphology (MMM3)


J. D. Bobaljik (McGill University)
What's In A Paradigm?

The Question: Plank 1991 begins with the observation that "[t]he earliest extant grammatical texts are paradigms." (p.161) The long linguistic and philological tradition have established a wealth of knowledge about the properties of paradigms, notably regarding the issue of syncretism, but one fundamental question has not been definitively answered, namely (1):

(1) Does knowledge of language (grammar) include knowledge (memorization) of paradigms themselves or just of the pieces that constitute paradigms and rules for generating them?

The issue: That is, given that a set of features, and a set of "morpheme realization rules" or "vocabulary items" as in (2) is sufficient information to generate the nominative declensional paradigm for regular nouns (and pronouns, short adjectives and past participles) in Russian, is there any reason to posit knowledge of paradigmatic structure beyond (2)?

  3 genders   -/i/ <--> plural
  2 numbers   -/a/ <--> feminine
      -/o/ <--> neuter
      -/O/ <--> elsewhere ("O" = yer)
(Halle 1997:428)

A purported argument. Williams 1994:26, e.g., criticises theories such as (2) observing that "the pattern of syncretism is a quite abstract structure, standing above particular words, particular rules, particular suppletive relationships." The notation (2b) treats it as an accidental property of these particular morphemes that gender is not expressed in the plural, but this is clearly a quite general fact of Russian morphology, holding across declension class, case, part of speech (N, V, A), and regardless of the particular desinences expressing case, number and gender. The existence of such systematic syncretisms (i.e., as opposed to accidental homophony) is taken as an argument that (2) is insufficient and thus for the grammatical relevance of paradigms over and above the pieces that constitute them.

The incompleteness of this argument. The existence of such syncretisms establishes that (2) is insufficient, but it does not establish that paradigms are necessary. Such syncretisms are captured within frameworks such as Distributed Morphology via (e.g.,) impoverishment rules (see especially Bonet 1991); such rules delete features from the Morphosyntactic Representation prior to the application of Morphological Realization (or Vocabulary Insertion) Rules as in (2b). Impoverishment rules capture language-specific syncretisms and must be learned, but they are of no greater formal complexity than the devices employed within theories advocating paradigms. As the existence of syncretism follows directly from neither approach, and is statable in either, syncretism alone in principle does not distinguish between the theories.

Are the theories distinct? Williams 1994 offers an argument that can distinguish the classes of theories in principle. He argues that there are universal constraints on the patterns of syncretism that can exist in a given language. In particular, he proposes that every language must have an "instantiated basic (sub-)paradigm," such that all other (sub-)paradigms mark a subset of the distinctions of it. Thus, it follows from his theory that if two cells "A" and "B" are syncretic in some paradigm, and cells B and C are syncretic in a another, then there will be a third paradigm in which A?B?C (i.e., all three are distinguished). This requirement is not statable in frameworks such as Distributed Morphology; if substantiated as a universal, it would constitute a convincing argument for paradigmatic knowledge, i.e., a "yes" answer to (1).

Against basic paradigms. In the Russian nominal and adjectival declension, accusative (A) is systematically identical to either genitive (B) or nominative (C) (conditioned by animacy) in all plurals and masculine and neuter singulars. This syncretism is systematic and not merely a fact about individual morphemes. Williams's basic paradigm hypothesis predicts (correctly) the existence of another paradigm in Russian in which accusative, genitive and nominative are distinct; this happens to be the feminine singular. However, exactly in the feminine singular, the dative and prepositional cases (distinct elsewhere) are systematically syncretic. In essence, since all six cases are distinguished in Russian, but no single paradigm overtly distinguishes all six, the basic paradigm hypothesis cannot be maintained. Hence, this potential argument in favour of the existence of paradigms fails.

Conclusion. We do not claim to have answered (1) definitively in this paper. We hope merely to have shed further light on the issue by (a) clarifying formally what is at stake, i.e., what can be taken as an argument for paradigms as a theoretical primitive, and (b) showing that on one point where the paradigmatic theory can be made to be more restrictive, such restrictions are found to be counter-evidenced by even moderately inflected languages such as Russian.