Elision and liaison

[Jean Pierre Chevrot]

 >Dear Info CHILDES Members,
 >I am working on a project examining the first language acquisition of
 >French and I am trying to find out at what age French children begin to
 >show knowledge of phonological elision (e.g., je entends -> j'entends). Any
 >suggestions or references would be greatly appreciated.
 >Many thanks,
 >Cristina Dye


Dear Cristina

What do you mean when you say "knowledge of phonological elision". Which
developmental change do you think about?
- First, the schwa alternate or always appears in the final position of a
word and then it begins to disappear before a V (je entends, je prends -->
j'entends) ?
- First, the schwa never appears in the final position of a word and then it
begins to appear and alternate before C (j'prends, j'entends --> je prends,
j'prends) ?
- First, schwa and no-schwa alternate randomly in the final position of a
word and then, they begin to be context-dependent (schwa or no schwa before
a C, no schwa before a V)?
- etc.

The general problem is the acquisition of phonological alternations and
variations. Some questions about the acquisition of variable phonological
units ("free" variation or context dependant alternations) are :
- Are these units missing in the early lexical representations and then
added?
- Are these units memorized in the early lexical representations?
- Are these units variable as soon as they are memorized in lexical
representations?
- How do the factors that influence the selection of variants become
established? Linguistic factors (C or V in the initial position of the
right-hand word, etc.) and sociolinguistic factors (social status : what
the children hear in the family and the peer group, stylistic variation).

As Isabelle mentions, we have done some works about the acquisition of
liaison and acquisition of final post-consonantal /R/ ("quatre sous" with
[R] / "quat'sous" without [R] ). Our idea is that one way to answer these
questions is to consider the first stages of acquisition of variable
phonological units as an exemplar-based process. We have some data and we
propose a developmental scenario for liaison.
In this scenario, certains stages are more empirically documented
than others. Perhaps, we could apply some of these stages from liaison to
elision.

- Stage zero (not documented at this time) - the children store chunks of
frequently cooccuring pieces of verbal strings. The sequences clitic + verbe
and det+noun, stored as a whole, are such chunks of memorized verbal
material. No liaison consonant errors ("un zours", "des nours") should
appear at this stage.

- Stage 1 - The children make CV segmention in "determiner + liaison
consonant (LC)+ noun" sequences (un-n-ours, les-z-avions). The output of CV
segmentation is a CV intial lexical form : nours, zavion. But the problem is
that children hear the right-hand words in different liaison contexts. They
hear the word "ours" in "un-n-ours" with /n/ liaison, in "deux-z-ours" with
/z/ liaison, in "petit-t-ours" with /t/ liaison. So, for each word begining 
with
a vowel in a French dictionnary, the child memorizes two or three exemplars
: nours, zours, tours for the word "ours"
(see Chevrot & Fayol (1999) for
data, Dugua (2002), Chevrot & Fayol (2000) for data and exposition of this
part of the scenario). When the children use the word "un" before the
exemplar "zours", they produce a liaison error ("un zours").

- Stage 2 - In our develomental view, the second stage of acquisition
consists in the learning of association between certain left-hand words and
the exemplars of the right-hand word : the children learns to associate "un"
with the exemplar "nours", "deux" with the exemplar "zours", "petit" with the
exemplar "tours". The dynamics of this change is "statistical learning". We 
have some
data about this hypothese. We find early SES differences in the performance
for obligatory liaisons (from 28 to 45 months) and these differencies
disapear after 50 months. For optional liaisons, the SES differences appear
from 50 months (Nardy, 2002). In adults (and in child directed speech), the
frequency of optional liaison depends (among others) upon the Socioeconomic 
Status (SES)
of the speaker, but the obligatory liaison are realized 100 per cent by all 
speakers. A possible
explanation of the results for the obligatory liaisons should be that the 
amount of input is more
important for children from families with high SES. If so, the high SES
children should have more opportunities for the statistical learning of
associations.

Stage 3 - In a third stage (adult), this LC's (liaison consonant) evolve in
three directions, depending on (a) contraints concerning lexical unicity 
and lexical
invariance and (b) frequency effects (Côté, 2002; Chevrot, Aubergé, Côté,
Fayol, 2002) :
- the most important part of LC's detache form the right hand word and
become epenthetic (default case),
- others LC's remain at the begining of the right hand word, in case of
frequent combination of LC with a specific right-hand word.
- others LC's become fixed consonant encoded at the end of suppletive
variants of left hand words : adjectives with different vowel in their
liaison and non liaison forms ("bon", "ancien").

An other view challenging this epenthetic one : to consider liaison in
adults as parts of constructions or schemas (Bybee, 2001).


If we apply this scenario to elision, we have the following
stages :
0- Children memorize unsegmented heard clitic+verb or det+N sequences :
"j'arrive" ou "j'viens" (without schwa), "je viens" (with schwa), "l'ours", 
"le chien",
etc.
1 - After segmentation, children have two exemplars for the clitic "je"
(with schwa and without schwa). The more available is the more frequent (is 
"je"
more "frequent" than "j'' " in child directed speech?). We can observe
errors : "je entends".
2 - Children learn to associate the exemplar without schwa with the
vowel-initial words and the two exemplars with consonant-initial
words.
3 - Schwa becomes epenthetic

Refs
- Bybee, J. (2001). Frequency efects on French liaison, in Joan Bybee and
Paul Hopper (eds.), Frequency and the Emergence of Linguistic Structure,
John Benjamins, 337-359.
- Chevrot, J.-P. & Fayol, M. (2001). Acquisition of French liaison and
related child errors, in M. Almgren, A. Barreña, M.J. Ezeizabarrena, I. 
Idiazabal,
and B. MacWhinney (Ed.), Research on Child Language Acquisition :
Proceedings of the 8th Conference of the International Association for the
Study of Child Language, volume 2, Cascadilla Press, 760-774.
- Chevrot, J.-P. & Fayol, M. (2000). L'acquisition de la liaison : enjeux
théoriques, premiers résultats, perspectives, Lidil, 22, 11-30.
- Chevrot, J.-P., Beaud, L., Varga, R. (2000). Developmental data on a
French sociolinguistic variable : the word-final post-consonantal /R/, 
Language
Variation and Change 12(3), 295-319.
- Chevrot, Jean-Pierre, Aubergé, Véronique, Côté, Marie-Hélène & Fayol,
Michel (2002). La liaison : acquisition, théorie phonologique, traitement
automatique, Colloque Cognitique: bilan et perspectives, La Sorbonne, Paris,
6-7 décembre.
- Côté, Marie-Hélène (2002). Between Phonology and the Lexicon : French
liaison revisited, University of Toronto, 29 Novembre 2002.
- Dugua, Céline (2002). Liaison et segmentation du lexique en français :
vers un scénario développemental, mémoire de DEA, Grenoble 3.
- Nardy, Aurélie (2002). L'acquisition de la liaison : production, jugement
et environnement langagier, Maîtrise, Grenoble 3.

Jean-Pierre

***********************************************************************
Jean-Pierre Chevrot
UFR SCL, Université Stendhal, BP 25, 38040, Grenoble cedex, France
tel : (0)4 76 41 90 46
***********************************************************************