Desktop version

Home arrow Language & Literature

Referral and directionality

The rules in Figures 6.10 and 6.14 follow Matthews (1991) in treating Priscianic dependencies as directional. This reflects a general view of the difference between symmetrical neutralization and asymmetrical referral. In the case of neutralization, features that are distinctive elsewhere in a system are collapsed, so that there is just one cell realizing the neutralized features.[1] In cases where form identity cannot be treated as neutralization, the single word or stem form is typically identified as the realization of one of the cells in the referral relation. The realization of the other cell or cells is then taken to ‘refer to’ or be ‘parasitic on’ this cell. The directional character of parasitic derivations and morphological transformations is shared by the referral rules of Zwicky (1985) and Stump (1993b, 2001), and the ‘takeovers’ of Carstairs (1987).

A challenge that confronts directional accounts is the need to identify the primary and dependent cells in a referral relation. In the case of the present infinitive- imperfect subjunctive pattern in (6.2a), it might seem obvious that the subjunctive is dependent on the infinitive, since the infinitive form defines the base for the subjunctive paradigm. However, given that a realizational account cannot refer directly to the form that realizes a different paradigm cell, the infinitive form must be assembled independently in the spell-out of infinitive and imperfect subjunctive forms. These realizations invoke the same rules, i.e., those in Figure 6.11, which, one can argue, primarily realize the features of the present infinitive, and derivatively realize the imperfect subjunctive.

Yet an argument of this kind is difficult to make for the past passive-future active participial syncretism in (6.2b). As Aronoff (1994:38) argues, “there is no way to choose either one of the supine or the perfect participle as underlying the other or to choose either one as underlying the future participle”. Aronoff’s point is illustrated by the arbitrary character of the rules in Figures 6.14 and 6.15. These rules stipulate that the future active participle is dependent on the past passive, and that -at realizes past passive rather than future active features. However, the directionality of this dependency is not only unmotivated, but also has no independent effects. A realizational account that reversed the dependency would have exactly the same empirical coverage.

Cases in which a syncretic form has a shape that would make it appropriate to fill one of the cells in a referral relation are sometimes taken to provide prima facie

Table 6.2 Dual/plural syncretisms in Slovene (Herrity 2000:49)

Sg

Dual

Plu

Sg

Dual

Plu

Nom

potnik

potnika

potniki

clovek

cloveka

ljudje

Acc

potnika

potnika

potnike

cloveka

cloveka

ljudi

Gen

potnika

potnikov

potnikov

cloveka

ljudi

ljudi

Dat

potniku

potnikoma

potnikom

cloveku

clovekoma

ljudem

Inst

potnikom

potnikoma

potniki

clovekom

clovekoma

ljudmi

Loc

potniku

potnikih

potnikih

cloveku

ljudeh

ljudeh

First declension genitive and locative plural rules in Slovene

Figure 6.19 First declension genitive and locative plural rules in Slovene

evidence for a directional referral relation. This intuition underlies the notion of directional ‘takeovers’ proposed by Carstairs (1987):

the realisation of two or more morphosyntactic properties (A and B) in some context by an inflexion which elsewhere realises only one of these properties. In such circumstances we can say that B takes over A (Carstairs 1987:117)

The Slovene paradigms in Table 6.2, which are discussed by Baerman et al. (2005:1756.), among others, illustrate some takeover-type patterns. The paradigm of the regular first declension masculine potnik ‘traveller’ exhibits the formal identity of the genitive forms potnikov and the locative forms potnikih. In the suppletive paradigm of clovek ‘person, the syncretic forms ljudi and ljudeh also pattern with other plural forms, rather than with dual forms.

As with the infinitive-subjunctive syncretism in Latin, one should be alert to the possibility that the directionality in Table 6.2 records the historical origin of these patterns. From a purely synchronic perspective, these paradigms exhibit two patterns. The first is that nouns have a complete set of distinct forms in the plural. In a realizational model, this pattern implies a full set of plural rules, including the genitive and locative rules in Figure 6.19.

The second pattern, marked by the bold forms in Table 6.2 involves the plural and dual syncretism in the genitive and locative. Given that the plural does appear to ‘take over’ the dual in these paradigms, directional rules that define the duals with reference to the plurals might appear motivated. Yet within a realization-based approach, the directionality in these rules would be completely redundant. The fact that genitive and locative duals are dependent on plurals, rather than vice versa, does not need to be stipulated in referral rules, since it is already expressed by an independent asymmetry in the rule inventory of Slovene. There are rules realizing the genitive and locative plural, but none realizing the duals. Hence referrals need only identify the genitive and locative duals with the corresponding plurals, as in Figure 6.20. In the course of realizing dative and locative duals, the referral

Symmetrical referral rules expressing dual/plural syncretism

Figure 6.20 Symmetrical referral rules expressing dual/plural syncretism

rules in Figure 6.20 will introduce plural features, which can be interpreted by the exponence rules in Figure 6.19. The realization of genitive and locative plurals will be unaffected, because the dual features trigger no new rules.

All of the usual realizational strategems for controlling rule application remain available for regulating rule interactions involving symmetrical referrals. In particular, a fully developed realizational analysis would associate the class-specific rules described in this chapter with ‘class features’ that control their interaction. Thus the first declension exponence rules in Figure 6.19 and the referral in Figure 6.20 would be assigned a specification such as [dc 1], which would make the rules more ‘specific’ than any class-neutral rules and allow them to override general or default patterns of exponence in Slovene.

More broadly, the general (or class-specific) asymmetries in rule inventories assumed by realizational accounts eliminate the need for directional referrals. The Latin referrals in Figures 6.10 and 6.14 can be reformulated as symmetrical rules, given the independent assumption that stem formation rules exist for the ‘primary’ cell in each of these referrals but not for the ‘dependent’ cell. Directionality seems equally redundant in other types of referral rules, such as the ‘bidirectional’ referrals described initially by Stump (1993b):

These facts suggest the accusative singular/dative singular referral in Old Icelandic is bidirectional: for nouns in Class A, the dat sg takes on the form of the acc sg, while for nouns in Class C, the Acc sg takes on the form of the dat sg. (Stump 1993b: 469)

The logic of this reduction is the same as in previous cases. A referral analysis assumes that in class A there is a rule realizing the dative singular but no rule realizing the accusative singular, and that in class C there is a rule realizing the accusative singular but no rule realizing the dative singular. Given this difference in the rule inventory of Icelandic, a symmetrical referral that applies in classes A and C will again determine an asymmetrical dependency.

In sum, a rule that identifies the realization of distinct paradigm cells will determine an asymmetrical dependency if, as in the cases considered in this section, only one of those cells is associated with an exponence rule (or with a rule that is more specific than the referral rule). Hence directionality is a redundant and excisable property of referral rules in these types of cases.

  • [1] The neutralization of gender contrasts in the plural of German or Russian adjective paradigmsprovides familiar examples of symmetrical neutralization.
 
Source
< Prev   CONTENTS   Source   Next >

Related topics