1

HOW HIGH PITCH AFFECTS LEXICAL PROCESSING IN GERMAN ADULTS –

EVIDENCE FROM EYE-TRACKING Katharina Zahner, Bettina Braun and Muna Pohl (all University of Konstanz)

To resolve lexical competition, listeners use both segmental (e.g., Allopenna et al., 1998) and

suprasegmental information (e.g., Reinisch et al., 2010; Sulpizio & McQueen, 2012) as soon

as it becomes available. Even though in intonation languages pitch is not lexically contrastive,

listeners rely on the pitch contour to decide between lexical candidates (D'Imperio et al.,

2007; Ladd & Schepman, 2003) and do not readily accept words with mismatching f0-

contours (Friedrich et al., 2001). More importantly, recent data suggest that German infants'

speech segmentation is guided more by the position of the pitch peak than by metrical

prominence (Braun et al., 2014). Using the visual world eye-tracking paradigm (Tanenhaus et

al., 1995), we tested whether this association between high pitch and stress is still present in

adulthood, such that high-pitched but unstressed syllables lead to temporary lexical activation

of cohort competitors with initial stress.

We selected 48 frequency-matched trisyllabic cohort pairs (24 for experimental and 24 for

filler trials) that were segmentally identical up to at least the first consonant of the second

syllable, but differed in stress position (e.g., KaVERne-KAviar, capitals indicate stress). The

cohort pairs and two distractors were presented on screen. In experimental trials, auditory

stimuli referred to the cohort member with penultimate stress (e.g., "Bitte klicke KaVERne

an"). The target word was produced with a medial-peak (L+H*) or an early-peak pitch accent

(H+L*, see Figure 1). For each cohort pair, each participant heard the target in only one of the

two intonation conditions. 60 participants (44 female, 25.5 years on average) were tested.

Fixations were extracted in 4ms steps and coded as pertaining to the auditory target (e.g.,

KaVERne) or the stress competitor (e.g., KAviar). Empirical logits were analyzed using

linear-mixed effects regression models with intonation condition as fixed factor and

participants and items as crossed random factors (Baayen et al., 2008; Barr et al., 2013). As

expected, participants fixated the initial-stress competitor more in the early-peak condition

(average elogs -0.66) than in the medial-peak condition (average elogs -1.20, ß = -0.54

[0.02;1.08]1, SE = 0.27, t = 2.04, p < 0.05) (see Figure 2 for fixation proportions to the

competitor in both intonation conditions), prior to segmental disambiguation. This suggests

that initial high-pitched but unstressed syllables activate words with initial stress. Hence, high

pitch appears to be an important stress cue for German adults.

                                                                                                                         1 Square brackets denote the 95% confidence interval for the estimate ß.

 

  2

Figure 1: Intonational realization of auditory targets (e.g., KaVERne)

Medial-peak condition

Early-peak condition

Figure 2: Fixation proportions to competitor with initial stress (e.g., KAviar)

Bitte klicke Kaverne an

Ka VER ne

H+L* L-%

Time (s)0 1.361

Bitte klicke Kaverne an

Ka VER ne

H+L* L-%

Time (s)0 1.361

Bitte klicke Kaverne an

Ka VER ne

H+L* L-%

Time (s)0 1.361

Bitte klicke Kaverne an

Ka VER ne

H+L* L-%

Time (s)0 1.361

Bitte klicke Kaverne an

Ka VER ne

L+H* L-%

Time (s)0 1.313

0  

0.1  

0.2  

0.3  

0.4  

0.5  

0.6  

4   84  

164  

244  

324  

404  

484  

564  

644  

724  

804  

884  

964  

1044  

1124  

Fixa%o

n  prop

or%o

n  

Time  from  target  onset,  e.g.,  KaVERne  (ms)  

early-­‐peak    (H+L*)  

medial-­‐peak    (L+H*)  

 

  3

References

Allopenna, P. D., Magnuson, J. S., & Tanenhaus, M. K. (1998). Tracking the time course of spoken word recognition using eye movements: evidence for continuous mapping models. Journal of Memory and Language, 38, 419-439.

Baayen, H. R., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59(4), 390-412.

Barr, D. J., Levy, R., Scheepers, C., & Tily, H. (2013). Random-effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 36, 255-278.

Braun, B., Pohl, M., & Zahner, K. (2014). Speech segmentation is modulated by peak alignment: Evidence from German 10-month-olds. Proc. Speech Prosody (SP7), Dublin.

D'Imperio, M., Petrone, C., & Nguyen, N. (2007). Effects of tonal alignment on lexical identification in Italian. In C. Gussenhoven & T. Riad (Eds.), Tones and Tunes (pp. 79-106). Berlin: Mouton de Gruyter.

Friedrich, C. K., Alter, K., & Kotz, S. A. (2001). An electrophysiological response to different pitch contours in words. Cognitive Neuroscience and Neurophysiology, 12(15), 3189-3191.

Ladd, D. R., & Schepman, A. (2003). "Sagging transitions" between high pitch accents in English: experimental evidence. Journal of Phonetics 31, 81-112.

Reinisch, E., Jesse, A., & McQueen, J. M. (2010). Early use of phonetic information in spoken word recognition: Lexical stress drives eye movements immediately. The Quarterly Journal of Experimental Psychology, 63(4), 772-783.

Sulpizio, S., & McQueen, J. M. (2012). Italians use abstract knowledge about lexical stress during spoken-word recognition. Journal of Memory and Language, 66, 177–193.

Tanenhaus, M. K., Spivey-Knowlton, M. J., Eberhard, K., & Sedivy, J. (1995). Integration of visual and linguistic information in spoken language comprehension. Science, 268, 1632-1634.