Desktop version

Home arrow Language & Literature

Principal parts

The predictive structure of inflectional systems is also what allows classical WP accounts to factor them into exemplary ‘patterns’ and lexical ‘identifiers’. Given the interdependence of inflectional ‘choices’, a system can be described by exemplary paradigms that exhibit the morphological patterns in the system, together with principal parts that identify the patterns followed by non-exemplary items. As Matthews notes, this organization lends cohesion to individual paradigms and underlies the efficiency of inflectional systems:

The most general insight is that one inflection tends to predict another... This insight can be incorporated into any model. Traditionally, it is the basis for the method of exemplary paradigms. If the alternations were independent, these would have to be numerous... But since they are interdependent, their number can be very small... It is more attractive to learn paradigms as wholes than each alternation separately. (Matthews 1991:197b)

The leading idea developed in the “method of exemplary paradigms” is that form variation in a grammatical system can be exhibited by exemplary patterns that serve a dual role. On the one hand, exemplary paradigms specify the forms of particular lexical items. On the other hand, they provide a model for the inflection of non-exemplary items. Hence one and the same paradigm exhibits the forms of an actual lexical item while representing morphological patterns characteristic of the class to which that item belongs. In the same way that stems are not basic units in a classical WP model, but are instead abstracted from a set of forms, classes are not ‘properties’ of items but are abstractions over sets of paradigms that exhibit congruent patterns of form variation. The class of an item is exhibited via characteristic patterns of form alternation. Unlike Bloomfieldian approaches, classical WP models do not dissociate lexis or ‘form’ from grammar or ‘combinatorics’. Also unlike realizational models, they do not radically separate properties from forms. The notion of a set of abstract paradigm cells (or morphosyntactic representata- tions) that are ‘filled in’ by spell-out rules arises later, in the models developed by Matthews (1972), Anderson (1992), Aronoff (1994) and Stump (2001). Instead, classical WP models embody a distinctively ‘integrative’ exemplar-based conception.

The patterns exhibited by exemplary paradigms are extended by matching ‘principal parts’ against cells in exemplary paradigms, and deducing additional forms by analogy to the forms that realize other exemplary cells. The diagnostic value of principal parts is not due to the features that they realize or to any aspect of their own form, such as the presence or absence of a particular exponent, or whether they serve as a ‘base’ for other forms, etc. Rather, a form is of diagnostic value to the extent that it has a distinctive (morphologically conditioned) shape corresponding to each class. In Russian, as in many Slavic languages, the nominative singular can be regarded as “the basic case [form]” (Corbett 1991:35) from which other forms can be deduced. As noted in Section 4.2, the form of the nominative singulars in Table 4.1 is informative about the declension class of the corresponding nouns. A nominative singular in -a identifies a second declension noun. A nominative singular that ends in a soft consonant identifies a first declension noun if the noun is masculine and a third declension noun if the noun is feminine. Alternatively, the class of a noun can be identified from a nominative singular that ends in a soft consonant, together with any other singular case form.

It is conventional to assign items to classes by holding a paradigm cell constant and comparing the forms that realize that cell. However, the opposite deduction, which classifies items based on the properties they associate with a given form, is equally effective and, in fact, more economical in Russian. As again remarked in Section 4.1.1, the forms in -u are fully diagnostic. A form in -u that realizes dative singular identifies a first declension noun, one that realizes accusative singular identifies a second declension noun, and one that realizes instrumental singular identifies a third declension noun.

Russian declensions thus approach a kind of pedagogical ideal, in which one principal part can identify the class of a noun and sanction reliable inferences about other forms of the noun. The paradigms in Table 4.1 also illustrate the pedagogical insight that not all forms are equally informative in the same way. In particular, the shape of the dative, instrumental and locative plural is so uniform across the paradigms in Table 4.1 that these forms merely serve to identify a soft stem noun (a hard stem noun would end in unpalatalized -am, -ami and -ax). The diagnostic (and morphosyntactic) value of other case forms in Table 4.1 lies between these extremes. The lack of any general metric for measuring diagnostic values between ‘fully informative’ and ‘fully uninformative’ is an acknowledged limitation of classical WP approaches, one which is again addressed by information-theoretic formalizations of these models.

Variation in the diagnostic value of cells and forms may reflect a range of factors, including how recently the forms have been morphologized, whether they have been subject to levelling, and other aspects of their origins and history. In Russian, as in many flectional systems, this variation also exhibits a fundamental tradeoff between different types of information. For example, the diagnostic value of a form such as nedelja is offset by the fact that it is a highly ambiguous marker of grammatical features. The shape of nedelja does not immediately identify it as a nominative singular, but once it has been classified as a nominative singular form, it unambiguously identifies class. In contrast, a form such as nedeljami unambiguously realizes the instrumental plural but is of no direct value in identifying class. Contemporary morphological descriptions often recognize only one dimension of this information matrix, representing grammatical information, such as case, number and gender, while neglecting class-identifying aspects of form or leaving them implicit in the structure of exemplary paradigms and the choice of principal parts. Part of the problem lies in the fact that predictive value is not a property of a form in the same way that case is and hence cannot readily be expressed as a ‘feature’, even if one accepts the use of diacritic features for expressing class affiliation.

< Prev   CONTENTS   Source   Next >

Related topics