Why are there more pataquès in French folk songs? The example of postverbal pataquès
Résumé
In an ongoing project we focus on French liaison, a “sandhi phenomenon which involves the presence or absence of a consonant at the boundary between two words […]: more specifically, the consonant may appear when Word-2 is vowel initial, but is absent when Word-1 or Word-2 is uttered in isolation” (Durand & Lyche, 2016: 363-364) and pataquès, i.e. the realization of liaison in unexpected contexts such as: Je reviendrai [z] au pays ‘I will return to my home country’, where no consonant is expected, or Le plus jeune des trente chantait [z] une chanson ‘The youngest of the thirty was singing a song’, where another consonant, [t], could have been expected. Our work is based on a corpus of traditional folk songs, the Jean Dumas collection, recorded in the 1960s in Central France, which come from an ethnomusicologist database. The songs are progressively being transcribed on CLAN (Computerized Language ANalysis, MacWhinney, 2014), then pre-annotated for liaison based on orthographic clues (later manually checked) thanks to a tool developed for the purpose of our research (Badin, 2019). This time-saving tool also provides lemmatization, PoS-tagging and metadata containing singers’ personal information for each token.
Liaison has been extensively studied in various theoretical frameworks that can be briefly summarized in two different approaches: a formal phonological one (Encrevé, 1988; Wauquier-Gravelines, 2005; Scheer, 2006) and a constructionist one (Bybee, 2001; Côté, 2005; Eychenne & Laks, 2017). The latter approach that is adopted here. Our aim is to understand why pataquès are more common in folk songs (3.55%; 105/2959) compared to corpora previously investigated (French speakers worldwide: MoDyCo & RUG, 2017 (0.06% (25/41946) in conversations and 0.24% (29/12296) in a reading task); child productions in parent-child interactions: LRL, 2017 (2.61%; 60/2296); etc.). In a usage-based perspective, we see pataquès as a peripheral phenomenon that is part of the same system as liaison, the core phenomenon. In this framework, grammar is not seen as an abstract structure underlying language use but as “the cognitive organization of one's experience with language”, strongly tied to the person's previous linguistic experience, in which every speaker takes part in organizing the grammar of the whole speech community (Bybee, 2006). Our singers’ experience (as singers/hearers) can thus influence their productions, or more specifically, their realization of liaison and pataquès.
Observing the organization of the liaison system, we decided to first focus on a specific context: the construction Verb+X, in which not only the liaison consonants [z] and [t] can be produced but also pataquès occurrences. Liaison realization is higher in postverbal contexts in folk songs (57%; 312/547) than in contemporary conversations (17%; 2055/12051) as well as pataquès realization (10%; 57/547 vs 0.25%; 30/12051). This appears to be a common phenomenon in French traditional folk songs, which were orally transmitted without the normative pressure that contemporary singers may face. Outside of any sociolinguistic considerations, it thus seems that because the singers hear more liaisons than in other corpora, they also produce these liaisons in unexpected postverbal contexts.