Verbal suppletion: an analysis of Italian, French, and Spanish to go

Abstract

Analyses of suppletive verb conjugations have focused on aspect or tense divisions, type and token frequency, and lexicalization of individual forms at an earlier stage of the language. In this paper, I attempt to show that person-number suppletive replacements follow language-specific alternation patterns, or templates, which mark the positions of stem alternations often shared by several verbs. These paradigmatic frameworks are stored in the lexicon with their verb forms and guide verbal restructuring processes by serving as the base configuration for suppletive replacements. The relative strength of a pattern is directly related to the number of verbs associated with it such that the most productive template is represented by completely rule-driven patterns, the template of lowest productivity encompasses only one verb, and all intermediate stages of productivity are represented by templates shared by various numbers of verbs. As verbs acquire new alternations, the shift is consistently to more productive templates. Moreover, since Italian andare, French aller, and Spanish ir have patterns of paradigmatic alternations that are shared by other verbs, in this model they cannot be considered as isolated idiosyncrasies in their verbal systems.The extreme irregularity of suppletive verbs, or verbs whose paradigms include etymologically unrelated forms, poses a challenge for theories of morphology that offer neat explanations for systematic morphological alternations. Each model takes a different approach, ranging from categorizing suppletive elements as "marginal garbage" (Lass 1990: 82) or as completely unnatural (natural morphology), to envisioning the occurrence of suppletive forms as a natural result of phonological change (morphological economy). 1 However, it is generally agreed that elements

Allomorphy and the Autonomy of Morphology

“…13 More arguments for the paradigm are given in Börjars et al (1997). 14 The same observation is made in Aski (1995).…”

Two suppletive adjectives in Megleno-Romanian

“…Nor is there any sense in which the plurals of mik and ˈmari are likely to have been morphologically or phonologically 'deficient' , for example by being excessively reduced phonologically or somehow already irregular or archaic (cf. Rudes 1980, Aski 1995 and also the discussion in Börjars and Vincent 2011, p. 144). The plural of ˈmari, as we have seen, actually does exist and it is simply mar, while that of mik should be *mits (in fact attested by Weigand, as a noun).…”

“…), of uadere originally 'go' , make 'an impressive, terrifying, threatening, rapid, dangerous or showy advance (especially into dangerous situations)' , 2 in the singular and third person forms of the present masculine and feminine forms of mic 'small' . I reproduce the data in Table 1, transposed into IPA 5 (the place names are as given in the source): For comparison I give in Table 2 the -regular -paradigm of the word for 'dead' (see Atanasov 2009, maps 582 and 583): The example in Table 2 is fairly typical of Megleno-Romanian adjectives, most of which mark masculine singular by final -u (or zero, according to dialect and to phonological environment), feminine singular by -ǝ, and feminine plural by -i (< *-e); the original masculine plural ending, now generally deleted at the end of a phonological word, was -i, which originally produced various palatalizing and/or affricating effects on preceding root-final consonants (e.g., mworts < *ˈmortsi < *ˈmorti). There is a second, less numerous, inflexion class which only distinguishes number (and not gender) and is characterized by -i (< *-e) in the singular and usually zero in the plural (although originally the plural ending was *-i): e.g., sg ˈratsi 5.…”

“…351, 360, 384). The fact that Atanasov (2009) does not present maps for this word perhaps also implies that he considers it morphologically regular.…”

Prototype categorization and phonological split

“…Schemata are cognitive associations that represent the commonalities of the affected items, and their effects are similar to those of 'templates' , which are described by Bybee (1985) and Aski (1995). Schemata represent how language users may make sense of/categorize incoming data, and can be composed of any number of items, large or small.…”