Acquisition of French Liaison and Related Child Errors

Jean-Pierre Chevrot (1) , Michel Fayol (2)

1 LIDILEM - LInguistique et DIdactique des Langues Étrangères et Maternelles
2 LAPSCO - Laboratoire de psychologie sociale et de psychologie cognitive

Although French liaison is a major topic in the fields of adult French phonology and sociolinguistics, its acquisition remains a terra incognita. Moreover, very frequent liaison consonant substitutions ("les- n-ours" instead of the adult form "les ours" with a /z/ liaison) or insertions ("papa-n-ours" instead of "papa ours") are well-known stereotypes of the French "babytalk", but their relationship with the acquisition of liaison has never been considered. Three kinds of data addressing these issues are presented. First, 300 such errors in the speech of a girl (from 2;0 to 3;6) during daily interactions will be analyzed. They suggest that /n/ is the most intrusive liaison consonant (compared with /z/ and /t/). Second, an experimentation has been carried out with 4-year old children (N=24) with the purpose of eliciting errors in words with obligatory liaisons. The results provide confirmation that /n/ is the most intrusive liaison consonant and suggest that some children are prone to use the liaison consonant they have just heard : they say more ofen "un-z-ours" (with a /z/ liaison instead of the adult /n/) after they have heard "des ours" with an adult-like /z/ liaison. Third, an experiment with three age groups [age means: 3;5 (N=15), 4;6 (N=24), 5;8 (N=15)] investigates whether implicit knowledge about probabilistic phonotactic constraints is used in segmenting ambiguous "determiner-noun" sequences. For example, subjects are hearing "determiner + non-word" sequences including a /n/ or a /z/ consonant, whose lexical and syllabic status is ambiguous : in such a sequence as "un (n)apil", one cannot hear whether /n/ is the coda liaison consonant of "un" (un apil), or whether it is the consonantal onset of the non word (un napil). Subjects are then asked to replace the singular determiner "un" with the plural determiner "des" (and vice versa when "des" + non-word sequences are presented). Responses like [dezapil] suggest that they consider /n/ to be a liaison consonant. Responses like [denapil] suggest that they view /n/ as the initial onset of the non-word "napil". Results show that responses of the youngest group do not fit phonotactic constraints (in French, /n/ is more frequent than /z/ as a word initial consonant and /z/ more frequent than /n/ as a liaison consonant) . The discussion addresses two issues : (1) What is the lexical status of liaison in the phonological representation of young children ? (2) If knowlegdge about phonotactic constraints is not available before the age of 4, how could we explain that /n/ is the most intrusive liaison consonant as early as the age of 2 ?

Mots clés: psycholinguistics; laison; language acquisition; phonology; phonological alternation

Domaines: Linguistique; Psychologie

Jean-Pierre Chevrot, Michel Fayol. Acquisition of French Liaison and Related Child Errors. M. Almgren, A. Barreña, M.J. Ezeizabarrena, I. Idiazabal, and B. MacWhinney. Research on Child Language Acquisition, vol. 2, Cascadilla Press, pp.761-775, 2001. ⟨hal-00706711v2⟩