Allomorphy: when, how, why

In recent year, much effort in theoretical linguistics has devoted to defining the grammatical restrictions governing morphological alternations, in which the exponence of an element x depends on the context in which x occurs. While the context may be of different nature – i.e., morpho-syntactic, phonological, and/or morpho-phonological – my research focuses primarily on identifying the grammatical conditions triggering morpho-syntactically conditioned and morpho-phonologically conditions alternations.

In morpho-syntactically conditioned alternations, the context is purely morpho-syntactic. For example, the English alternation go/went is usually posited as morpho-syntactically-conditioned, as consisting of a suppletive alternation exclusively triggered by a specific morpho-syntactic trigger (in our example, past tense). One question that researchers have been interested in is whether there are locality constraints acting upon allomorphy and what the relevant constraints are (Embick 2010; Bobaljik 2012; Merchant 2015; Moskal 2015; Moskal & Smith 2016). A strong hypothesis defended in Embick (2010) claims that, the trigger and the target of allomorphy must be linearly adjacent to each other (Linear Adjacency Hypothesis; LAH). Merchant (2015) has however pointed out that Greek root-allomorphy presents a counterexample to LAH, and proposes that vocabulary items are rather sensitive to spans (i.e., larger portions of morpho-syntactic structures; Span Adjacency Hypothesis, SAH). Christopoulos & Petrosino (2017) take issue with Merchant’s claim and propose a revisited analysis of Greek verbal morphology that seems to further support the LAH.

Morpho-phonologically conditioned alternations involve a phonological change triggered by morpho-syntactic and/or morpho-phonological trigger. Traditionally, these alternations are accounted for via application of morpho-phonological rules (i.e., the so-called readjustment rules). By the way of mixing phonological and morphological alphabets, these operations though make unfalsifiable predictions; as a consequence, a recent trend in the field has been to get rid of them as a whole (Bermùdez-Otero 2012, Haugen & Siddiqi 2013). In Petrosino (in prep.; available upon request), I discuss the morpho-phonologically conditioned allomorphy of the definite determiner in Neapolitan (southern Italian dialect) and argue for an account which is effortlessly able to explain the patterns as framed into language-wide generalizations governed by morpho-phonologically sensitive operations.