Atelier de Phonologie - Neige Rochant

The discussion about the definition of clitics has centred on whether it should incorporate prosodic criteria. The conventional definition characterizes clitics as syntactically free but prosodically bound (e.g., Halpern 1998: 101; Dixon 2007: 574; Booij 2012: 290; Bonet 2019; Ionova 2019: 22; Pescarini 2021: §1.1, cited by Haspelmath 2023: 30). Haspelmath (2023: 2) challenges this by defining the clitic as “a bound morph that is neither an affix nor a root,” explicitly excluding any prosodic characteristics. Influenced by Selkirk’s (1996) research on the prosodization of function words, which offers a suitable framework for describing various prosodic phenomena in Pukur (Atlantic < Niger-Congo), I contend that Haspelmath’s (2023) grammatical definition of clitics is aptly mirrored by a strictly prosodic definition. Hence, in the debate on whether clitics should be defined using both morphosyntactic and prosodic criteria, I propose a definition that incorporates both but separately, distinguishing grammatical from prosodic clitics. The prosodic clitic, which I define as a bound phonological sequence that is neither internal to a single prosodic word (i.e., a prosodic affix) nor a free prosodic word itself, corresponds to either of the following configurations (proposed by Selkirk’s (1996) as two of the three prosodizations of function words):

a. (clitic (word)ω )φ                

[The prosodic clitic belongs to the same phonological phrase as a prosodic word, while being external to any prosodic word]

b. ((clitic (word)ω )ω )φ       

[The prosodic clitic is located inside a higher-rank, yet outside a lower-rank prosodic word]

From a 6,5-hour transcribed and translated corpus of naturalistic Pukur field data, I demonstrate the utility of this split approach, which accommodates elements that are grammatical clitics without being prosodic clitics, and vice versa. In Pukur, numerous elements exclusively qualify as clitics at one level, either prosodic or grammatical, while being overlooked by a standard definition as they are not free grammatical words prosodized as affixes. For instance, I argue that the phonological phenomena affecting verbal inflectional morphemes, such as lenition, nasal sandhi and vowel assimilations, are aptly explained by a configuration where the stem, which cannot be a grammatical clitic by definition, functions as a prosodic clitic. This talk will provide several examples of prosodic clitics in Pukur with a focus on phonological phenomena allowing to discern a prosodic clitic structure.