Context Effects in Speech Perception
The perception of speech is incredibly context-sensitive. Neighboring stimuli have a great impact upon how speech sounds are perceived. These effects occur at a number of levels including phonetic, lexical, and semantic.
Phonetic Context Effects
Lindblom (1967) first reported an effect of adjacent speech sound upon vowel perception. Since then, there have been many demonstrations of the way in which speech sounds influence perception of their neighbors.
reference: Repp, B. H. (1982). Phonetic trading relations and context effects: New experimental evidence for a speech mode of perception. Psychological Bulletin, 92, 81-110.
This perceptual effect may counteract the constraints of coarticulation, in that, a syllable is acoustically more /da/-like when produced following /al/ and more /ga/-like when produced following /ar/. Thus, this context effect is likely important for veridical speech communication.
Whereas, these effects have traditionally been offered as evidence for the existence of specialized modules for speech perception, recent data suggest that the effects are the result of more general auditory processes.
In one classic example, Mann (1980) reported an influence of a preceding vowel-consonant (/al/ or /ar/) on how a following consonant-vowel syllable (/ga/ or /da/) is perceived.
The task of listeners was something like this. Listen to the sounds below and identify the second consonant as /da/ or /ga/.
Identify the second syllable:
 /al/ + ?
/ar/ + ?
It is likely that you heard the first as /ga/ and the second as /da/.
However, they were identical sounds!
The /al/ and /ar/ stimuli preceding this sound changes how it is heard.
This phenomenon has also been found for very young infants:
reference: Fowler, C. A., Best, C. T., & McRoberts, G. W. (1990). Young infants’ perception of liquid coarticulatory influences on following stop consonants. Perception & Psychophysics, 48, 559-570.
Moreover, a preceding /al/ or /ar/ influences /ga/ to /da/ perception in Japanese listeners who (owing to the characteristics of their native Japanese language) cannot discriminate /al/ from /ar/.
The chart to the right demonstrates this phenomenon.
In general, consonants are more often heard as /ga/ when preceded by /al/. Physically identical speech sounds are heard more often as /da/ when preceded by /ar/.
Generality of Phonetic Context Effects
Are these effects specific to speech sounds?
Lotto and Kluender (1998) investigated whether a preceding nonspeech tone that models some of the acoustical characteristics of /al/ or /ar/ produces a similar effect on /ga/ and /da/ identification.
tone + ?
tone + ?
Just like above, it is likely that you heard the first as /ga/ and the second as /da/.
However, they were identical sounds!
The graph at the right reproduces the findings of Lotto and Kluender (1998).
reference: Lotto, A. J., & Kluender, K. R. (1998). General contrast effects in speech perception: Effect of preceding liquid on stop consonant identification. Perception & Psychophysics, 60, 602-619.
Birds Show Context Effects Too
To test whether the mechanisms behind these effects are specific to humans, Lotto, Kluender, and Holt (1997) investigated this context effect using Japanese quail as subjects.
Quail trained to peck to /da/ or /ga/ syllables showed a shift in peck rates
dependent on the preceding syllable. More "/ga/ responses" were obtained
when CVs were preceded by /al/ and more "/da/ responses" were obtained when
CVs were preceded by /ar/. This shift in response is similar to what is
witnessed in humans responses to these syllables. A summary of those findings is shown below.
reference: Lotto, A. J., Kluender, K. R., & Holt, L. L. (1997). Perceptual compensation
for coarticulation by Japanese quail (Coturnix coturnix japonica). Journal
of the Acoustical Society of America, 102, 1134-1140.
What is the Mechanism?
Holt and Lotto (2002) investigated whether this effect occurs at a peripheral or central level of auditory processing by splitting sound presentation between the ears. The context sound was presented to one ear, the target sound to the opposite ear. The context effect remains.
|
In order to experience this element of the phonetic context effect, you must wear headphones. When you are ready, click the buttons below for the sounds. Like before, try to identify the second syllable.
|
/al/ + ?
|
/al/ + ?
|
/ar/ + ?
|
/ar/ + ?
|
It is likely that you identified the second syllables in the first row of sounds as /ga/ and the second syllables in the second row of sounds as /da/.
Notice that these are the same two-syllable pairings that were presented earlier in the page. The only change made to these sound files was that the syllables were presented to different ears. Regardless of this change, the effect still occurs.
These context effects do not appear to originate solely from the peripheral auditory system (e.g., due to cochlear masking). In further support of this conclusion, the effect appears to be long-lasting. When a silent interval of varying duration is inserted between the context and target stimuli, the effect of context remains for durations on the order of hundreds of milliseconds. This time course is inconsistent with purely peripheral processing.
| Try these sounds, and see if the second syllable changes |
(1)/al/ + ?
|
(2) /al/ + ?
|
(3) /al/ + ?
|
(4) /ar/ + ?
|
(5) /ar/ + ?
|
(6) /ar/ + ?
|
Hopefully, the second syllable in 1, 2, and 3 all were identified as /ga/, whereas the second syllable in 4, 5, and 6 were identified at /da/. Notice that between each set (1,2,3 or 4,5,6) the only difference between the sound clips was the length of time between syllables. However, the extra time did not retract from the auditory phenomenon. Another interesting note on the sounds - the second syllable was the same for all 6 clips!
|
reference: Holt, L. L. & Lotto, A. J. (2002). Behavioral examinations of the neural mechanisms of speech context effects. Hearing Research, 167, 156-169.
Does the auditory system treat nonspeech precursors in a similar manner?
When a single-formant nonspeech stimulus modeling the acoustic
characteristics of /al/ or /ar/ F3 preceded /ga/ to /da/ series stimuli,
similar effects were observed.
reference: Lotto, A. J., Sullivan, S. C., & Holt, L. L. (2003). Central locus for non-speech context effects on phonetic identification. Journal of the Acoustical Society of America, 113, 53-56.
|
These context effects do not appear to originate solely from the peripheral auditory system (e.g., due to cochlear masking). In further support of this conclusion, the effect appears to be long-lasting. When a silent interval of varying duration is inserted between the context and target stimuli, the effect of context remains for durations on the order of hundreds of milliseconds. This time course is inconsistent with purely peripheral processing.
Does the auditory system treat nonspeech precursors in a similar manner?
When a single-formant nonspeech stimulus modeling the acoustic
characteristics of /al/ or /ar/ F3 preceded /ga/ to /da/ series stimuli,
similar effects were observed.
