Back to Instructor's Page

Past research has shown that when listeners are exposed to a repeated sequence of brief steady-state vowels (of the same duration and pitch) they experience "phonemic transformations" (Chalikia & Warren, 1991; Warren, Bashford, & Gardner, 1990), and report hearing words and phrases absent in the original stimulus.  Earlier studies found little agreement across listeners in the verbal forms reported, most likely because any individual listener's organization could begin at any given point in the sequence.  Subsequent work (Warren, Healy, & Chalikia, 1996) has shown that when a silent gap is inserted between repetitions of a sequence, listeners typically report more similar verbal forms.  This may happen because listeners start their perceptual organizations after the pause.  Warren et al. (1996) also found that the verbal forms are robust enough to be recognized at a later time (e.g., a week later).  When two listeners' responses differed they could identify the particular stimulus corresponding to each other's verbal forms.  These results were also found with whispered vowels (Chalikia, Warren, & Bashford, 1992), suggesting that the perceptual organization of vowel sequences is based upon objective acoustic characteristics resembling specific English syllables.  They also support the notion that the organization of vowel sequences into syllables and words is accomplished by matching the auditory input to linguistic templates employed for the identification of syllables and words (Chalikia & Warren, 1994).  These findings also suggest that the listeners use common principles for the comprehension of these speech stimuli, and that they can perceptually reorganize the sequences and recognize which of them corresponded to particular forms heard by others
              It has been suggested (Chalikia & Warren, 1994) that the organization of vowel sequences into syllables and words is accomplished by matching the auditory input to linguistic templates employed for the identification of syllables and words.  These templates have fuzzy boundaries and can be activated by voices having different accents and idiosyncratic characteristics.
              Interestingly, when listeners hear the sequences they are unaware of the order of the items, or the nature of the vowels in the sequence.  The individual vowels in each sequence form "temporal compounds" (Warren, 1974; 1982) that permit listeners to discriminate between different arrangements of the same sounds, without any need for identification of constituent elements or their order.
Other studies (e.g., Chalikia & Dresser, 1994; Chalikia & Parvey, 1995; Chalikia, Becker, & Laurila, 1996) have investigated the possible effects of vowel duration changes on phonemic transformations, using voiced or whispered vowels.  In these studies base-line stimuli typically contained vowels at the same duration (60 ms).  Additional blocks of stimuli were created with duration variations ranging randomly (Chalikia & Dresser, 1994) from 30 to 120 ms per vowel.  In other studies (Chalikia & Parvey, 1995; Chalikia et al., 1996) vowel durations were changed systematically in an attempt to create "speech compression" or "speech expansion" effects.  It was hypothesized that if these changes are perceived as prosodic cues (e.g., variations in syllabic stress and rhythm), the original organizations heard with the base-line stimuli may still be recognizable, and that listeners will be able to match their organizations with the appropriate sequences.  Results showed that listeners were able to perform above chance in all conditions by matching phonemic transformations heard with the base-line stimuli to those heard with all the additional stimuli.  However, matching performance decreased for the shorter vowel durations.
          The above results suggest that the perceptual mechanism involved treats these stimuli in a manner similar to the way one treats the same speech utterance produced by different speakers at different speech rates.  It seems that the verbal organizations heard remain robust and stable, suggesting that they are based upon objective acoustic characteristics.  The reduction in performance with stimuli where vowel durations were relatively short suggests that the duration manipulations employed probably approached the limits for template recognition and matching.  For a more extended discussion of the impact of vowel duration on the ability to form verbal organizations see Warren et al., 1990.
          The present study constitutes a replication of the original phonemic transformation effect, with six different arrangements of four vowels.
« Return to Top»

          The baseline stimuli are six different orders of a vowel sequence.  Each order is made up of the same four vowel sounds, followed by an approximately 300 ms silent gap.  Each vowel was originally produced by a male speaker at 100 Hz voicing frequency by matching the pitch of his production to a 100 Hz tone heard through headphones.  Individual 10 ms glottal pulses, each starting and ending at zero-cross to minimize acoustic transients, were excised from the extended statements of each vowel.  These single glottal pulses were iterated several times to yield each steady-state vowel burst used to create each sequence. The particular arrangement of vowels within each sequence was determined randomly. The set of six vowel sequences, shown in Table 1, is determined randomly for each participant.  The four vowels used are the ones corresponding to the vowels in the following words: had, hood, heed, and hid. 
« Return to Top»

         The experiment uses a one-way repeated measures design.  The manipulated variable is the number of vowel sequences (6 in this case).  The dependent variable is the proportion of correct matches per set.
« Return to Top»

Chalikia, M. H., & Dresser, T. (1994).  The effects of duration changes on the perception of vowel sequences.  Journal of the Acoustical Society of America, 96, 3825.

Chalikia, M. H., & Parvey, N. (1995).  The effects of systematic duration changes on phonemic transformations. Journal of the Acoustical Society of America, 98, 2967.

Chalikia, M. H., & Warren, R. M. (1991).  Phonemic transformations: Mapping the illusory organization of steady-state vowel sequences.  Language and Speech, 34 (2), 109-143. 

Chalikia, M. H., & Warren, R. M. (1994).  Spectral fissioning in phonemic transformations.  Perception & Psychophysics, 55, 218-226.

Chalikia, M. H., Warren, R. M., & Bashford, J. A. Jr. (1992).  The phonemic transformation effect: Intersubject agreement on verbal forms. Journal of the Acoustical Society of America, 91, 2422.

Warren, R. M. (1974).  Auditory temporal discrimination by trained listeners.  Cognitive Psychology, 6, 495-500.

Warren, R. M. (1982).  Auditory Perception: A New Synthesis. New York: Pergamon.

Warren, R. M., Bashford, J. A., Jr., & Gardner, D. A. (1990).  Tweaking the lexicon: Organization of vowel sequences into words.  Perception & Psychophysics, 47, 423-432.

       Warren, R. M., Healy, E. W., & Chalikia, M. H. (1996).  The vowel-sequence illusion: Intrasubject stability and intersubject agreement on syllabic forms. Journal of the Acoustical Society of America, 100, 2452-2461.

« Return to Top»

Last revised:January 22, 2010 09:04:06 AM
Copyright© 2010 The University of Mississippi. All rights reserved.
Questions about this page?PsychExps