Access to writing and the acquisition of obligatory liaison in second language French

Liaison is a phenomenon of external sandhi that involves the production of a latent coda consonant (liaisonconsonant, LC) in prevocalic contexts (e.g. [z] in des [de] + ânes [an] → des ânes [dezan], (some) donkeys,but not in des [de] + poules [pul] → des poules [depul], (some) hens. The segmental content of the LC isthus dependent on the first word but produced as the onset of the second word: [de.zan]. As a consequence,word and syllable boundaries do not align. It has been argued that this misalignment constitutes potentiallydifferent difficulties for L1 and L2 learners (Wauquier, 2009).In the early stages of L1 acquisition children typically misinterpret the LC as the lexical onset of word2,leading them to substitute the LC and to produce sequences such as [E˜zan] instead of un âne [E˜nan] (adonkey) (Wauquier & Shoemaker, 2013). These errors are unattested in adult L2 learners. Indeed althoughLC substitutions are present in adult L2 learners' productions, they can typically be attributed to the writtenform of the word ([gKAdaKbK ˜ ] for grand arbre [gKA˜taKbK]) (Thomas, 2004). At the same time L2 learnersproduce LCs without resyllabification (e.g. [dez.an] instead of [de.zan]), an error that has not been reportedin monolingual acquisition. These errors have also been interpreted as being influenced by the written form.However, to date, the vast majority of previous studies have focused on monolingual preliterate children orhighly literate adult L2 learners with primarily written exposure to French. To bridge the gap between thesetwo populations, we propose in this study a qualitative analysis of L2 learners without systematic writteninput.This presentation brings together data from two beginning groups of French: preliterate children (n = 3,L1: Swedish, age of onset of acquisition: 3;0-3;5) and adults who have had little to no formal instruction inFrench (n = 10, L1: Chinese and Bengali). All learners received predominantly oral input without systematicwritten support. Productions of un/deux+noun (cases of obligatory liaison) were elicited using a picturenamingtask.Results from both groups include productions previously reported for L2 learners (e.g. LC without resyllabification).However, both adults and children in this study also produce L1-like LC substitutions(e.g. [E˜zaKbK] or even [dølaKbK]). These and other results are discussed in the light of previous modelssuggested for L1 and L2 development of liaison as they bring into question previous assumptions aboutdiffering developmental paths for L1 and L2 learners.Thomas, A. (2004). Phonetic norm versus usage in advanced French as a second language. IRAL - InternationalReview of Applied Linguistics in Language Teaching, 42(4), 365–382.Wauquier, S. (2009). Acquisition de la liaison en L1 et L2 : stratégies phonologiques ou lexicales ? Acquisitionet interaction en langue étrangère. Aile... Lia, (2), 93–130.Wauquier, S., & Shoemaker, E. M. (2013). Convergence and divergence in the acquisition of French liaisonby native and non-native speakers: A review of existing data and avenues for future research. Language,Interaction and Acquisition/Langage, Interaction et Acquisition, 4(2), 161–189