The liaison in written French Cecilia Gunnarsson Christiane

  • Slides: 27
Download presentation
The liaison in written French Cecilia Gunnarsson, Christiane Soum Favaro, Pierre Largy, Jean-François Camps

The liaison in written French Cecilia Gunnarsson, Christiane Soum Favaro, Pierre Largy, Jean-François Camps Université de Toulouse le Mirail

The peculiarities of French phonology Example of the consonant /t/: Fixed and stable word

The peculiarities of French phonology Example of the consonant /t/: Fixed and stable word initial consonant: /tisy/, always pronounced Fixed word finals: /mamut/, always pronounced In liaison context: /p titami/ vs. /p ti /

Where to attach the liaison consonant? To the word to the left (most frequent

Where to attach the liaison consonant? To the word to the left (most frequent point of view) To the word to the right (Gougenheim, 1938; Morin, 1981, 1986; Morin & Kaye, 1982) Epenthetic (Klausenberger, 1974; Tranel, 1981; Côté, 2005) Which are the implications in written French?

Experimental study of the liaison in writing Transversal study, 7 to 12 year-old children

Experimental study of the liaison in writing Transversal study, 7 to 12 year-old children Dictation of nominal syntagms with either liaison consonant (LC) or fixed word initial consonant (FIC) 3 most frequent liaison consonants, /z/, /n/ and /t/, 99, 7% of liaisons cases (Boë & Tubach, 1992)

Population 86 children q q 20 children in 2 nd year (7 -8 years)

Population 86 children q q 20 children in 2 nd year (7 -8 years) 24 children in 3 rd year (8 -9 years) 21 children in 4 th year (9 -10 years) 21 children in 5 th year (10 -11 years)

Materials 48 recorded nominal syntagms, French neutral accent + 2 to training For each

Materials 48 recorded nominal syntagms, French neutral accent + 2 to training For each consonant 8 with LC and 8 with FIC: un gros aigle – un gros zèbre / – /; un avion – un navire / – /; un petit éléphant – un petit téléphone / – / A note pad of 50 pages, one page/syntagm

Procedure Room apart Groupes of 6 -8 children In average 2 listenings per syntagm

Procedure Room apart Groupes of 6 -8 children In average 2 listenings per syntagm No going back to correct

General treatment of data 2 contexts: liaison and word initial 3 consonants: /z/, /n/

General treatment of data 2 contexts: liaison and word initial 3 consonants: /z/, /n/ and /t/ 3 error types: – regroupment: un grosélan – un grosouave – apostrophe: un gros s’élan– un gros s’ouave – insertion or elision: un gros zélan – un gros ouave

General phonogical hypothesis The liaison consonant is more difficult to detect than the initial

General phonogical hypothesis The liaison consonant is more difficult to detect than the initial fixed consonant (Wauquier-Gravelines, 1996, Nguyen et al. , to appear) Hypothesis 1: There will be more errors in the liaison context than in the initial fixed context

General phonological hypothesis According to acoustical-phonetic indicators LC ≠FIC for /t/; LC = FIC

General phonological hypothesis According to acoustical-phonetic indicators LC ≠FIC for /t/; LC = FIC for /n/ (W-G, 1996); LC = FIC for /z/ (Nguyen et al. , to appear) /t/ and /z/ LC are easier to detect than /n/ LC (W-G) (Nguyen et al. ) Hypothesis 2 There will be more errors for /n/ than for /z/ and /t/; and more errors for /z/ than for /t/

Phonological error hypothesis The liaison consonant is unstable/floating vs the initial fixed consonant (Encrevé,

Phonological error hypothesis The liaison consonant is unstable/floating vs the initial fixed consonant (Encrevé, 1988) Syllabification in French: CV Hypothesis 3: There will be more errors of insertion of a consonant to the right than elision of a fixed initial consonant

Results Hypothesis 1 F(1, 82)=5, 57 ; Cme=652, 8 ; p<0. 03 * More

Results Hypothesis 1 F(1, 82)=5, 57 ; Cme=652, 8 ; p<0. 03 * More errors in liaison context

Results Hypothesis 2 F(2, 81)=56, 26 ; Cme=185, 05 ; p<0. 001 * More

Results Hypothesis 2 F(2, 81)=56, 26 ; Cme=185, 05 ; p<0. 001 * More errors for /z/ and /n/ than for /t/

Results Hypothesis 3 * More Insert/Elide errors in Liaison conexte

Results Hypothesis 3 * More Insert/Elide errors in Liaison conexte

Results Hypothesis 3 bis F(2, 81)=7, 74 ; Cme=282, 44 ; p<0. 01 *

Results Hypothesis 3 bis F(2, 81)=7, 74 ; Cme=282, 44 ; p<0. 01 * More difference between liaison and fixed initial context for /t/ than /n/, no difference for /z/

Frequency error hypothesis The more obligatory the liaison context, the more the syntagm with

Frequency error hypothesis The more obligatory the liaison context, the more the syntagm with the LC will be treated as ONE unit (W-G, 1996) Det +noun more obligatory than adj + noun Hypothesis 4 More errors of regroupment (unacha for un achat) for /n/ than for /t/ and /z/

Frequency error hypothesis A frequent context type will cause errors in it’s direction. (Bybee,

Frequency error hypothesis A frequent context type will cause errors in it’s direction. (Bybee, 1995, 2001; Ellis, 2002) /z/ the most frequent LC, followed by /n/, followed by /t/ (Durand & Lyche, 2008) When non-words, almost all FIC /z/ and /n/ treated as LC (Stridfeldt, 2005) Hypothesis 5 /z/ FIC > LC more than /t/ and /n/ FIC > LC more than /t/

Result errors – Hypothesis 4 -5 * Not more regroupement for /n/ * More

Result errors – Hypothesis 4 -5 * Not more regroupement for /n/ * More FIC > LC for /n/ than /z/ and /t/ * Not more FIC > LC for /z/ than /n/ * LC > FIC > for /t/

To summarize The floating nature of the LC… Liaison context more difficult to treat

To summarize The floating nature of the LC… Liaison context more difficult to treat than word initial context Other errors than Insert/Elide are marginal. More Insert than Elide. And… /n/ most difficult to detect then /z/ then /t/. Acoustical-phonetic factors? Frequency? Or other factors?

Developmental treatment of data 4 school levels: 2 nd, 3 rd, 4 th and

Developmental treatment of data 4 school levels: 2 nd, 3 rd, 4 th and 5 th year 2 contexts: liaison and word initial 3 consonants: /z/, /n/ and /t/ No error types, regroupment and apostroph being marginal as errortype

Developmental Hypotheses The frequence of errors will decrease more in FIC (less problematic context)

Developmental Hypotheses The frequence of errors will decrease more in FIC (less problematic context) than in LC The frequence of errors will decrease more for the C: s /t/ and /z/ (easier to detect) , than for /n/; and more for /t/ (FIC≠LC) than for /z/ and /n/ (FIC = LC) The frequence of error will decrease in the liaison context where it is more frequent : in /z/ and /n/ but less in /t/.

Error frequence and age * Decrease first 3 years than slight increase, U-shaped curve

Error frequence and age * Decrease first 3 years than slight increase, U-shaped curve

Error frequence and context * Not more decrease en FIC than in LC

Error frequence and context * Not more decrease en FIC than in LC

Error frequence and consonant * More decrease for /t/ * Similar decrease for /z/

Error frequence and consonant * More decrease for /t/ * Similar decrease for /z/ and /n/

Error frequence age, consonant and contexte /z/ /n/ * /z/ and /n/ more decrease

Error frequence age, consonant and contexte /z/ /n/ * /z/ and /n/ more decrease in CL than FIC * /t/ not more decrease in FIC than LC /t/

To summarize The frequence of errors decreases form 7 to 12 years, but u-shaped

To summarize The frequence of errors decreases form 7 to 12 years, but u-shaped curve More decrease in the complicated liaison context for all 3 consonants, floor effect for FIC? More decrease for the /t/ where FIC ≠ LC and which is the easiest to detect.

Perspectives Analyze the temporal data collected on a digitizing tablet The impact of the

Perspectives Analyze the temporal data collected on a digitizing tablet The impact of the liaison on morphology, singular vs. plural. Widen the acquisitional perspective by populations of learners of L 2 French