Desktop version

Home arrow Language & Literature arrow Word and paradigm morphology

Morphomic stem syncretism

Part of the explanation for the contemporary neglect of analogy lies in the selective coverage of theoretical studies. Since Bloomfield’s time, the goal of comprehensive or even broad coverage has receded to the point that linguistic theories are justified in ignoring analogical patterns that challenge their analytical assumptions. This serves to reinforce the impression that such patterns are infrequent or marginal and somehow fall outside ‘core grammar’. For a more accurate assessment of the prevalence and status of these types of patterns one can, however, turn to linguistic traditions that retain a commitment to broad coverage. One pattern that has attracted philological, pedagogical and theoretical attention concerns the ‘supine’ stem in Latin. Gildersleeve and Lodge (1895:71) provides a standard description of the relation between the supine, the past passive participle and future active participle:

From the Supine stem as obtained by dropping final -m of the Supine, form

a. Perf. Part. Passive by adding -s.

b. Fut. Part. Active by adding -rus (preceding u being lengthened to u).

The regularity of this pattern is illustrated in Table 5.3, which isolates the supine stem in forms of the exemplary Latin verbs from Table 4.12.

As Matthews (1972) notes, this analysis is formulated, in almost identical terms, in the grammar of Priscian:

Priscian’s rule would actually derive the Nominative Singular Masculine ... by the addition of rus to the u: form of the ‘Supine’ and would moreover be limited to verbs in which the ‘Supine’ is not deficient... (Matthews 1972: 85)

Matthews’s own statements of this pattern differ only in treating r as a stem formative rather than as part of the future active participle ending:

Table 5.3 Analysis of Latin supine stems (Gildersleeve and Lodge 1895)

Verb

Supine

Past Pass Prtl

Fut Act Prtl

Gloss

AMO

amatu-m

amatu-s

amaatua-rus

‘to love’

MONEO

monitu-m

monitu-s

monitua-rus

‘to advise’

TEGO

tactu-m

tectu-s

teactua-rus

‘to cover’

CAPIO

captu-m

captu-s

captua-rus

‘to take’

AUDIO

auditu-m

auditu-s

auditu-rus

‘to hear’

ANALOGY

104

The stem of the Future Participle is derived from the stem of the Past Participle by the suffixation of u:r [orthographic ur JPB]. (Matthews 1972:86)

There are a few exceptions; but, in general, if the stem of the Past Participle is x, no matter how irregular it may be, that of the Future Participle is x with -Ur- added. (Matthews 1991:200)

These formulations determine the segmentations in Table 5.4, which depart from those in Table 5.3 in assigning uniform participial endings in -us.

The theoretical interest in this pattern resides in the fact that the common stem in the supine series does not make any consistent morphosyntactic contribution to the forms it underlies. The challenge that this poses for a model of morphemic analysis is set out clearly as early as Matthews (1965):

But in monitUrus, etc. the case is not so simple. Preceding the segment Ur (which can clearly be assigned to FU [the Future Active Participle JPB]) one finds precisely those segments which in monitus, etc. were assigned to PA [the Past Passive Participle JPB]: each form, apparently, includes the allomorph of a morpheme which is absent from its syntactic analysis! (Matthews 1965:143)

The supine stem represents a complex case of recurrent but non-redundant structure. Like the simpler Russian and Estonian patterns in Chapter 4.1, separating out the stem from the forms in which it occurs leads to ‘overextraction’, the isolation of a recurrent ‘unit of form’ that has no fixed meaning or function in isolation. The difficulties raised by overextraction cannot be overcome by further division. Analyses like those in Table 5.5 merely push the problem down a level. As Matthews (1972:83) notes, “ur could be handled as a quite unproblematic allomorph of Future Participle”. “But”, as he then asks, “what morpheme would the t or s [in lapsura JPB] then belong to?”.

Table 5.4 Forms based on the supine stem (Matthews 1991:200)

Verb

Supine

Past Pass Prtl

Fut Act Prtl

Gloss

AMO

amat-um

amat-us

amaatuar-us

‘to love’

MONEO

monit-um

monit-us

monituar-us

‘to advise’

TEGO

tect-um

tect-us

taectuar-us

‘to cover’

CAPIO

capt-um

capt-us

captuar-us

‘to take’

AUDIO

audit-um

audait-us

audaituar-us

‘to hear’

Table 5.5 Morphomic structure of Latin monitus and moniturus

Root

Theme Vowel

?

Fut

Masc Nom Sg

Past Pass Prtl

mon

i

t

_

us

Fut Act Prtl

Ur

In short, the supine stem cannot be associated with properties that either make a consistent contribution to the meaning of the forms in which it occurs or determine its distribution in those forms—rather than in other forms—of an item. Recurrent ‘units of form’ that cannot be associated with consistent ‘units of meaning’ have come to be known as morphomes. The term originates with Aronoff (1994), who initially applies it to ‘purely morphological’ functions that mediate between morphosyntactic properties and phonological form:

Let us call the level of such purely morphological functions morphomic and the functions themselves morphomes. What is novel about this level, and what warrants giving it a special name, is that it embodies an empirical claim: the mapping from morphosyntax to phonological realization is not direct but rather passes through an intermediate level. (Aronoff 1994:25)

In the continuation of this passage, Aronoff acknowledges that an intermediary level is only motivated in cases where the mapping between morphosyntax and phonology is indirect. Hence, although “all... mappings technically involve morphomes”, whether direct or indirect, “it is morphomes like Fen [which maps Perfect and Passive onto the same participial form in English JPB] that truly earn their name” (25). In his subsequent description of Latin conjugational stems, Aronoff (1994) construes ‘morphomic’ as ‘non-morphemic’, and even as “abstract and unmotivated morphological machinery”:

On the other hand, unlike morphosyntactic entities, they do not directly reflect the semantic or syntactic system. Instead they are morphomes, part of the abstract and unmotivated morphological machinery of the language. In this respect, stems are much closer to theme vowels than to more orthodox affixal markers of morphosyntactic properties. (Aronoff 1994: 58)

From a Post-Bloomfieldian standpoint, a negative characterization of ‘morphomes’ is entirely apt. If morphemic relations can be taken to be normative, morphomic patterns will seem like noise. But from a classical WP point of view, classifying morphomes as “part of the abstract and unmotivated morphological machinery of the language” concedes too much. Categorizing patterns as ‘nonmorphemic’ is informative only if morphemic patterns do, in fact, enjoy some privileged status. Not only is this far from established, but, as discussed in Chapter 2.4.1, it is unclear that contemporary notions of ‘morpheme’ are of any classificatory value. Even in their heyday, morphemic models never represented more than an extreme and programmatic approach to morphological analysis. Hence the justification for singling out morphomic patterns must rest on a positive characterization of their function or distribution.

An obvious candidate is their predictive value. The systematic character of mor- phomic patterns makes one member in a morphomic relation a reliable predictor of another. This is precisely the role that morphomic patterns play in classical and pedagogical grammars. Moreover, this practical conception can be extended naturally to a more general theoretical position. If one takes predictability to be the fundamental relation that binds together the elements of a morphological system, then morphomic patterns can be interpreted in terms of predictive relations and morphomes defined as ‘recurrent units of predictive value’. Morphemic relations, where they exist, can be construed as limiting cases in which the predictive value of a pattern derives from or is enhanced by a stable feature-form correspondence. Morphomic patterns are then of interest not because they fail to exhibit this simple feature-form correspondence but because they provide a pure expression of predictive relations.

 
Source
< Prev   CONTENTS   Source   Next >

Related topics