Desktop version

Home arrow Language & Literature arrow Word and paradigm morphology

Priscianic deduction

Indeed, ifwereturntoclassical andphilological treatmentsofmorphomic patterns, we find that they are not characterized as ‘units of pure form’, but as something more akin to ‘units of predictive value’. This perspective is fully explicit in Matthews (1972), in which the reference to “knowledge” of form variation anticipates the information-theoretic proposals in Chapter 7:

What one is saying... is that the Future Participle formation is ‘parasitic’, in a sense on the Past Participle formation: if one ‘knows’ the latter then one can use this ‘knowledge’ to derive the former (Matthews 1972: 86)

Strikingly, when Maiden (2005) summarizes Aronoff’s analysis, he formulates it in implicational terms that are largely absent from Aronoff (1994):[1]

The third [‘supine’] stem constitutes an allegedly inviolable distributional regularity—what Aronoff terms a morphome—in that its presence in any one member of the specified, idiosyncratic, set of cells, always implies its presence in all of the other members of the set. (Maiden 2005:137)

The other cases of ‘Priscianic’ or ‘parasitic’ syncretism cited by Matthews (1972, 1991) are likewise described in terms of predictive or implicational patterns. The correspondence between present active infinitives and imperfect subjunctives is again expressed as a predictive relation between surface forms:

For any Verb, however irregular it may be in other respects, the Present Infinitive always predicts the Imperfect Subjunctive. For the Verb ‘to flower’, florere ^ florerem; for the irregular Verb ‘to be,, esse ^ essem, and so forth without exception. (Matthews 1991:195)

Although Latin (and Romance) stems have attracted particular theoretical attention, similar patterns are found in other languages with rich stem inventories. Both Whitney (1889) and Macdonnell (1927) describe the formation of the perfect active participle in Sanskrit in morphomic terms, as involving the addition of the possessive suffix to the past passive participle:

Table 5.6 Morphomic structure of Sanskrit krta and krtavat (Whitney 1885:21)


Past Pass


Past Pass Prtl



Perf Act Prtl


From the past passive participle, of whatever formation, is made, by adding the possessive suffix vant, a secondary derivative having the meaning and construction of a perfect active participle: for example tat krtavan ‘having done that’; tam nigirnavan ‘having swallowed him down’. Its inflection is like that of other derivatives made with this suffix...; its feminine ends in vatl; its accent remains on the participle. (Whitney 1889:344).

By adding the possessive suffix vat to the past pass. part., a new form of very common occurrence is made, which has the value of a perfect active participle;—e.g. kr-ta, ‘done’: krta-vat, ‘having done’. It is generally used as a finite verb, the copula being omitted ... (Macdonnell 1927:136).

The participles in Table 5.6 present the same problem as those in Table 5.5. The past passive participle krta can be divided into a root and passive exponent ta. Neither the participle nor the exponent preserve past or passive features when they occur within the perfect active participle krtavat. However, they do reliably predict the form of the perfect active participle.[2]

This predictive relation is naturally expressed by proportional analogies. The proportion in (5.4a) licences the deduction of the Latin imperfect subjunctive essem from the present infinitive. Proportion (5.4b) sanctions the deduction of the past passive moniturus ‘advise’ from the future active participle.

(5.4) Priscianic analogy in Latin conjugations

a. florere : florerem = esse : essem

b. amatus : amaturus = monitus : moniturus

c. amatum : amatus : amaturus = monitum : monitus : moniturus

The proportion in (5.4c) likewise expresses the full deductive import of the original generalization in Gildersleeve and Lodge (1895:71). Given an exemplary item that exhibits the three-way dependency, the supine of a new verb sanctions the deduction of both the past passive and future active participle.

Significantly, these analogies do not require that a common meaning be assigned to shared ‘units of form’ such as amat/monit or florere/esse. The value of recurrent elements resides in their predictive force, and it is only the full word forms in

(5.4) that must be assigned grammatical features. From a classical WP perspective, segmentations are also principally motivated by their predictive value. The proportion in (5.4a) predicts the future active participle form monitoorus, irrespective of whether -r- is grouped with the stem or ending.

Bases such as florere, amat and krta are all clearly ‘morphomic’ in the sense of Aronoff (1994). But in this respect they are not fundamentally different from the stems in any other proportional analogy. The exemplary forms and principal parts in a proportional analogy are associated with the features of a paradigm cell (or ‘morphosyntactic representation’). However, there is no kind of ‘derivational’ relation between these features. Instead, they serve solely to locate co-varying forms within exemplary paradigms and to match principal parts against exemplary forms. Because the proportional format imposes no analysis below the word level, it can exploit the predictive value of the stems in (5.4) without creating the problem of assigning grammatical features to them. In other cases it may be possible to associate properties with stems, just as single morphs may sometimes realize a simple property. However, within a classical WP model, morphemic stems are no more normative morphosyntactically than stems that can stand alone as words are normative morphotactically. Both just represent limiting cases whose isolability is of significance only insofar as it enhances predictions about other forms.

  • [1] The main implications that Aronoff considers relate syntactic categories and morphological units,as in, for example, the suggestion that “there is indeed a mutual implication between the complexsyntactic category Perfect Active and the perfect stem type” (Aronoff 1994: 55f.).
  • [2] Another morphomic correspondence in Sanskrit holds between indicative verbs and active participles. As Whitney (1889:246) remarks, “the active participle-stem may be made mechanically fromthe 3rd pl. indic. by dropping i”.
< Prev   CONTENTS   Source   Next >

Related topics