the perception of spanish lexical stress by french...

39
The perception of Spanish lexical stress by French speakers: stress identification and time cost Sandra Schwab & Joaquim Llisterri Universitat Autònoma de Barcelona Université de Genève May 3, 2010

Upload: nguyenhanh

Post on 16-Sep-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

The perception of Spanish lexical stress by French speakers:

stress identification and time cost

Sandra Schwab & Joaquim Llisterri

Universitat Autònoma de BarcelonaUniversité de Genève

May 3, 2010

Introduction

● Difficulties for French speakers to produce stress in L2

● Perceptual explanation● Phonologisches Sieb or Crible

phonologique (Polivanov, 1931; Troubetzkoy, 1939)

● Phonological system of L1 as a 'filter' through which all the sounds of L2 are perceived and classified.

● Stress ‘deafness’ (Dupoux et al., 1997, 2001, 2008)

Goals

● Examine the stress ‘deafness’ of the French speakers● Fixed stress (French) vs. free stress (Spanish)

● Assess the evolution of the ‘deafness’ along the L2 learning process

● Study the acoustic parameters which influence the perception of lexical stress, and their effect on the time needed to identify stress

Method - Participants

● 20 French speaking participants:● 10 "Advanced" learners of Spanish

● 6-11 years of study at university level● Raised in a monolingual French

environment

● 10 "With no knowledge" of Spanish● Raised in a monolingual French

environment● Without knowledge of other stress free

languages

Method - Material

● Llisterri et al. (2005)●Words (n=12) and pseudowords (n=12)

●Creation of test stimuli●Experimental procedure based on:

● Analysis● Manipulation● Resynthesis with the Praat PSOLA

algorithm

Method - Material

● Analysis of a corpus● 8 triplets (4 with words and 4 with pseudowords):

válido (adj.) - valido (V, pres.) - validó (V, past)

● Measurements for each vowel:●Duration● F0● Amplitude

� Base stimuli (with no manipulation)� Manipulated stimuli

Method - Material

● Creation of the manipulated stimuli● The physical parameter values of each vowel

of a proparoxytone word (e.g. válido) are replaced by the values of each vowel of a paroxytone word (e.g. valido).●PP>P

● The physical parameter values of each vowel of a paroxytone word (e.g. valido) are replaced by the values of each vowel of an oxytoneword (e.g. validó).●P>O

Method - Material

válido with the original F0 values válido with the F0 values from vali do

PP PP>P

Llisterri et al. (2005)

Method - Material

136

16Manipulation of F0, duration and amplitude

16Manipulation of F0 and amplitude

16Manipulation of duration and amplitude

16Manipulation of duration and F0

16Manipulation of amplitude

16Manipulation of F0

16Manipulation of durationStress shift

PP>P

P>O

24Base stimuliNo stress shift

N (words and pseudowords)

ManipulationCondition

Method – Procedure and Data analysis

● Procedure● Identification of the stressed syllable

● Data analysis● Percent correct identification●Reaction times for correct responses

E.g.: PP (medico) > P (medico)

� "correct" response = P (medico)

Results

● Reaction times in correct responses:

● For base stimuli

● For manipulated stimuli

Results - Reaction times in Base stimuli

● Mixed-effects model:● Random variables●Participants●Stimuli

● Dependent variable●Reaction time of correct responses

● Independent variables●Competence in L2●Pattern●Lexical status

Results - Reaction times in Base stimuli

800

850

900

950

1000

1050

1100

Competence in L2

Rea

ctio

n tim

e (m

s)

Advanced NoKnowledge

Figure 1. Reaction times (ms) in base stimuli as a function of Competence in L2(Advanced and No knowledge).

Effect of Competence in L2 (p<0.05)

Results - Reaction times in Base stimuli

800

850

900

950

1000

1050

1100

Competence in L2

Rea

ctio

n tim

e (m

s)

Advanced NoKnowledge

Figure 1. Reaction times (ms) in base stimuli as a function of Competence in L2(Advanced and No knowledge).

Faster detection forAdvanced

Effect of Competence in L2 (p<0.05)

Results - Reaction times in Base stimuli

400

500

600

700

800

Pattern

Rea

ctio

n tim

e (m

s)

PP P O

Figure 2. Reaction times (ms) in base stimuli as a function of Pattern (PP, P, O).

Effect of Pattern (p<0.05)

Results - Reaction times in Base stimuli

400

500

600

700

800

Pattern

Rea

ctio

n tim

e (m

s)

PP P O

Figure 2. Reaction times (ms) in base stimuli as a function of Pattern (PP, P, O).

Effect of Pattern (p<0.05)

Faster detection forPP > P and O

Results - Reaction times in Base stimuli

820

840

860

880

900

920

Lexical status

Rea

ctio

n tim

e (m

s)

Word PseudoWord

Figure 3. Reaction times (ms) in base stimuli as a function of Lexical status (Word, Pseudoword).

Effect of Lexical status (p<0.05)

Results - Reaction times in Base stimuli

820

840

860

880

900

920

Lexical status

Rea

ctio

n tim

e (m

s)

Word PseudoWord

Figure 3. Reaction times (ms) in base stimuli as a function of Lexical status (Word, Pseudoword).

Effect of Lexical status (p<0.05)

Faster detection forWords

Discussion - Base stimuli

● Compared to participants with no knowledge of Spanish, advanced French speakers:

● identify stress more accurately. (Schwab & Llisterri, to appear)

●are faster in correctly identifying stress. ● Stress in proparoxytones is perceived:

●more accurately than in paroxytones and oxytones. (Schwab & Llisterri, to appear)

● faster than in paroxytones and oxytones.● Stress is perceived:

●as accurately in words as in pseudowords. (Schwab &

Llisterri, to appear)

● faster in words than in pseudowords.

Results – Reaction times in Manipulatedstimuli

● Mixed-effect model:● Random variables

● Participants● Stimuli

● Dependent variable●Reaction time of correct responses

● Independent variables●Competence in L2●Manipulation● Pattern● Lexical status

● Effect of manipulation only and no interaction

Results – Reaction times in Manipulatedstimuli

● Examination of subsets of manipulations in comparison with base stimuli

● Mixed-effects model●Random variables●Participants●Stimuli

● Dependent variable●Reaction time of correct responses

● Independent variable●Manipulation

Results – Reaction times in Manipulatedstimuli

me d i co me di co

F0 + Duration + Amplitude

No manipulation F0 + Duration + Amplitude Manipulation

Original stress

P base stimulus PP>P manipulated stimulus

Stress shift

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base F0_Dur_Ampl

Manipulation

Figure 4. Reaction times (ms) in base stimuli and in F0+Dur+Ampl manipulatedstimuli .

No effect of Manipulation

Combined manipulation ofF0, duration and amplitude

No time cost

Results – Reaction times in Manipulatedstimuli

F0 versus Duration and Amplitude

me di co

F0

Duration + Amplitude

me di co

Duration + Amplitude

F0

F0 Manipulation Duration + Amplitude Manipulation

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base F0 Dur_Ampl

Manipulation

Figure 5. Reaction times (ms) in base stimuli , in F0 manipulated stimuli and in Dur+Ampl manipulated stimuli .

Effect of Manipulation (p<0.05)

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base F0 Dur_Ampl

Manipulation

Figure 5. Reaction times (ms) in base stimuli , in F0 manipulated stimuli and in Dur+Ampl manipulated stimuli .

F0 manipulationTime cost

Effect of Manipulation (p<0.05)

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base F0 Dur_Ampl

Manipulation

Figure 5. Reaction times (ms) in base stimuli , in F0 manipulated stimuli and in Dur+Ampl manipulated stimuli .

Combined manipulation ofduration and amplitude

No time cost

Effect of Manipulation (p<0.05)

Results – Reaction times in Manipulatedstimuli

me di co

F0

Duration + Amplitude

���� Time cost

me di co

Duration + Amplitude

F0

���� No time cost

Results – Reaction times in Manipulatedstimuli

Amplitude versus F0 and Duration

me di co

Amplitude

F0 + Duration

me di co

F0 + Duration

Amplitude

Amplitude Manipulation

F0 + DurationManipulation

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base Ampl F0_Dur

Manipulation

Figure 6. Reaction times (ms) in base stimuli , in Ampl manipulated stimuli and in F0+Dur manipulated stimuli .

Effect of Manipulation (p<0.05)

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base Ampl F0_Dur

Manipulation

Figure 6. Reaction times (ms) in base stimuli , in Ampl manipulated stimuli and in F0+Dur manipulated stimuli .

Effect of Manipulation (p<0.05)

Amplitude manipulationNo time cost

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base Ampl F0_Dur

Manipulation

Figure 6. Reaction times (ms) in base stimuli , in Ampl manipulated stimuli and in F0+Dur manipulated stimuli .

Effect of Manipulation (p<0.05)

Combined manipulation ofF0 and duration

Time cost

Results – Reaction times in Manipulatedstimuli

me di co

Amplitude

F0 + Duration

���� No time cost

me di co

F0 + Duration

Amplitude

���� Time cost

Results – Reaction times in Manipulatedstimuli

Duration versus F0 and Amplitude

me di co

Duration

F0 + Amplitude

me di co

F0 + Amplitude

Duration

DurationManipulation

F0 + Amplitude Manipulation

Results – Reaction times in Manipulatedstimuli

950

1000

1050

Rea

ctio

n tim

e (m

s)

Base Dur F0_Ampl

Manipulation

Figure 7. Reaction times (ms) in base stimuli , in Dur manipulated stimuli and in F0+Ampl manipulated stimuli .

No effect of Manipulation

No effect of Manipulation

Results - Identification rate in Manipulatedstimuli

me di co

Duration

F0 + Amplitude

���� No time cost

me di co

F0 + Amplitude

Duration

���� No time cost

Discussion – Manipulated stimuli

● Different role of the acoustic parameters:● in the identification of stress position (Schwab &

Llisterri, to appear)

● in the time needed to perceive stress

● More specifically:● Stress position is more accurately identified when

F0 is involved (Schwab & Llisterri, to appear).● When the accentual change is perceived, the

amplitude appears to condition the amount of time needed for stress to be detected.

Conclusion

● French speakers' stress ‘deafness’:● not really deaf to stress (Muñoz et al., 2009; Schwab & Llisterri, to

appear)

● Are they slower than native Spanish speakers?● Exposure to L2: more able and faster in identifying

stress● Acoustic parameters affecting the perception of

stress by French speakers:● F0 and amplitude

● Combination of identification rate and time cost to get a more complete picture of the processes involved in the perception of stress in an L2.

Thank you for your attention!

http://liceu.uab.cat/~joaquim/applied_linguistics/NS_10/NS_10.html