Categorical Speech Perception in Cerebellar Disorders

60, 323–331 (1997) BL971826

BRAIN AND LANGUAGE ARTICLE NO.

Categorical Speech Perception in Cerebellar Disorders Hermann Ackermann,* Susanne Gra¨ber,*,† Ingo Hertrich,* and Irene Daum† *Department of Neurology and †Institute for Medical Psychology and Behavioral Neurobiology, University of Tu¨bingen, Tu¨bingen, Germany Keele and Ivry (1991) considered the cerebellum an ‘‘internal clock’’ responsible for temporal computations both in the motor and in the perceptual domain. These authors, therefore, expected that the processing of durational parameters of the perceived acoustic speech signal such as voice onset time (VOT) depends upon the cerebellum as well. However, a preliminary investigation of Ivry and Gopal (1992) revealed unimpaired phoneme-boundary effects in cerebellar patients along a continuum of monosyllabic stimuli with systematically varied VOT (/ba/–/pa/). Since the energy of the aspiration noise provides additional cues for the discrimination of voiced and voiceless stops, the present study used a series of disyllabic stimuli differing in a purely durational parameter. Both controls and patients with unilateral cerebellar lesion identified the endpoints of this continuum in nearly all instances as the counterparts of a minimal pair (Boten, /bo:tn/, ‘‘messengers’’ versus Boden, /bo:dn/, ‘‘floor’’). Subjects with bilateral pathology of the cerebellum, in contrast, did not show a comparable phoneme-boundary effect. Our data corroborate the hypothesis of the cerebellum as an internal clock and implicate a role of this structure in speech perception.  1997 Academic Press

INTRODUCTION

A widely recognized view in clinical neurology implies that cerebellar lesions and diseases compromise motor functions such as the regulation of muscle tone, the coordination of voluntary movements, as well as the control of posture and gait, while sensory and cognitive functions are spared (Gilman, Bloedel, & Lechtenberg, 1981; Dichgans, 1984). This assertion has, however, been challenged by recent studies (see Daum & Ackermann, 1995, for a review). As concerns perceptual processes, Ivry and Keele (1989) documented for the first time that cerebellar disorders may interrupt the ability This study was supported by grants from the DFG (SFB 307) and the BMFT (01 KL9001 O). The authors thank Prof. Dr. Johannes Dichgans and Dr. M. M. Schugens for critical discussion and helpful comments. Address correspondence and reprint requests to Hermann Ackermann, Department of Neurology, University of Tu¨bingen, Hoppe-Seyler-Strasse 3, D-72076 Tu¨bingen, Germany. 323 0093-934X/97 $25.00 Copyright  1997 by Academic Press All rights of reproduction in any form reserved.

324

ACKERMANN ET AL.

to estimate durations demarcated by pairs of auditory clicks: Patients with cerebellar damage performed significantly worse than controls when asked to compare two successive time intervals, the second stimulus being either shorter or longer than the first one which had a duration of 400 ms. A later experiment revealed impaired judgments of the velocity of moving stimuli— a task which also requires precise timing—in cerebellar subjects (Ivry & Diener, 1991). Since at least some cerebellar motor deficits can be considered to reflect disordered central timing mechanisms, Keele and Ivry (1991) conceptualized the cerebellum as an ‘‘internal clock’’ required for temporal computations both in the motor and the perceptual domain. Perceived phonetic features depend, among others, upon durational parameters of the acoustic signal such as the voice onset time (VOT; Lisker & Abramson, 1964, 1967). It is well established that continua of speech utterances varying in VOT of the initial stop consonant are perceived in a categorical manner giving rise to an abrupt transition between two response categories (e.g., Liberman, Harris, Kinney, & Lane, 1961). When, e.g., Englishspeaking adults are asked to label monosyllabic stimuli ranging in VOT from 2150 to 1150 ms as /da/ or /ta/, perception abruptly shifts from a voiced to a voiceless stop at a VOT of approximately 135 ms (Lisker & Abramson, 1970). On the basis of their timing model, Keele and Ivry (1991) expected distorted phoneme-boundary effects with respect to the voicing contrast in patients with cerebellar disorders. Using a series of monosyllabic stimuli with systematically varied VOT (endpoints: 210 and 170 ms), Ivry and Gopal (1992) found, however, a quite similar crossover from one response category, i.e., /ba/, to the other, i.e., /pa/, in cerebellar subjects and their controls. Thus, categorical perception of the voiced–voiceless distinction of initial stops seemed to be preserved.1 These authors, therefore, concluded that the identification of speech sounds by means of durational parameters does not depend upon the assumed cerebellar clock. It should be noted, however, that listeners integrate a variety of acoustic cues in order to extract phonetic features from the speech signal (Repp, Liberman, Eccardt, & Pesetsky, 1978). For example, the energy of the release burst (Moslin, 1979) as well as the amplitude of the aspiration phase (Repp, 1979) seem to contribute to the perceived voicing contrast of stop consonants. Using synthetic stimuli, Miller, Wier, Pastore, Kelly, and Dooling (1976) demonstrated, moreover, that the categorical voicing distinction might reflect a threshold effect of backward masking with respect to the detection of aspiration noise. Thus, the identification of voiced and voiceless stops, respectively, does not neces1 Categorical perception typically gives rise to identification functions characterized by an abrupt transition from one response category to the other. In the absence of discrimination data, however, this pattern does not prove categorical perception according to the criteria established by Studdert-Kennedy, Liberman, Harris, and Cooper (1970). For convenience, the notion ‘‘categorical perception,’’ nevertheless, will be used in the present context.

SPEECH PERCEPTION AND CEREBELLUM

325

FIG. 1. Acoustic speech signal of the utterance Boten (/bo:tn/, ‘‘messengers’’; A) and Tick (/tik/, ‘‘tic’’; B). The row of arrows at each oscillogram indicate the nine time intervals which were—from right to left—successively removed in order to generate 10 stimuli with systematically varied occlusion time (A) and voice-onset-time (VOT; B), respectively. The VOT values of the 10 stimuli extend from 56 to 20 ms, the occlusion times from 110 to 20 ms.

sarily require the computation of durational parameters but might depend on the detection of unharmonic acoustic energy preceding vowel onset. Cerebellar patients may, thus, be able to perceive voicing contrasts despite of impaired internal clock mechanisms. In order to test whether the cerebellum contributes to temporal computations during speech perception, stimuli varying in a purely durational parameter are required. Liberman, Harris, Eimas, Lisker, and Bastian (1961) found a phoneme-boundary effect on a series of synthesized words with intersyl-

326

ACKERMANN ET AL.

SPEECH PERCEPTION AND CEREBELLUM

327

labic silence extending from 20 to 130 ms. These stimuli were perceived as rabid at one end and as rapid at the other. Comparable results were obtained for German disyllabic words in that the duration ratio of vowel/(vowel 1 closure) determined, among others, the perceived voicing category of a stop consonant preceding nasal plosion (Kohler, 1979, 1984). For example, a ratio above 0.7 indicates a voiced stop such as in leiden (/laedn/, ‘‘to suffer’’) whereas a ratio below 0.6 seems to be related to its voiceless minimal pair cognate leiten (/laetn/, ‘‘to lead’’). The present study systematically manipulated the occlusion time of the utterance Boten (/bo:tn/, ‘‘messengers’’) in order to obtain purely durational speech contrasts. It was expected that a pattern of categorical perception would emerge in healthy control subjects along a continuum of stimuli with Boten and its minimal pair counterpart Boden (/bo:dn/, ‘‘floor’’) as perceived endpoints. Assuming that the cerebellum represents an universally employed clock, disorders of this structure should yield distorted identification functions. A second set of utterances differing in VOT of the initial alveolar stop was used as a control condition. PATIENTS AND METHODS Ten patients with cerebellar pathology participated in the present study. One subject (male, age 27 years) had undergone extirpation of nearly the entire left cerebellar hemisphere for tumor removal about 10 years ago. Cerebellar ischemia had occurred in another patient (male, age 69 years) 5 years prior to the present study. Magnetic resonance imaging performed shortly after that event revealed a hypodense area within the region of blood supply of the right superior cerebellar artery (Fig. 3 in Ackermann, Vogel, Petersen, & Poremba, 1992). Both patients had an unilateral ataxic syndrome in the acute stage after surgery and infarction, respectively. At the time of examination of speech perception these individuals, however, did not show ataxic signs any more. The remaining eight individuals (five males, three females; age: 28 to 72 years, mean: 58.4 years) had a diagnosis of diffuse atrophy of the cerebellum. All of them presented—more or less pronounced—with ataxia of gait, dyscoordination of voluntary arm movements, oculomotor disorders, and ataxic dysarthria; none showed impaired cognitive abilities or emotional disturbances at clinical examination. All patients with cerebellar atrophy had undergone extensive clinical, electrophysiological, and neuroradiological investigations (magnetic resonance imaging) in order to ensure that the degenerative process did not extend to extracerebellar structures. The control group comprised 10 age- and sex-matched subjects. The utterance Boten produced by a certified speech pathologist was recorded on a DAT recorder and displayed on the screen of a PC. Using commercially available software (CSL

FIG. 2. (A) Identification functions on the series of stimuli varying in VOT averaged across groups. The ordinate shows the number (in percent) of perceived Tick utterances for each of the 10 stimulus categories. VOT of stimulus 1 5 56 ms, of stimulus 10 5 20 ms, steps by 4 ms (NC 5 controls, CERE 5 cerebellar patients). (B) Identification functions for the series of stimuli varied in occlusion time (number of perceived Boten utterances in percent). Cerebellar subjects with unilateral lesion (CERE-1, n 5 2) and with bilateral pathology (CERE-2, n 5 8) are plotted separately (NC 5 controls). Occlusion length of stimulus 1 5 110 ms, of stimulus 10 5 20 ms, steps by 10 ms.

328

ACKERMANN ET AL.

4300; Kay Elemetrics) 10 stimuli with an occlusion time of the medial stop consonant extending from 110 to 20 ms in steps of 10 ms were edited as exemplified in Fig. 1A: stimulus No. 1 corresponds to the original utterance (occlusion time 5 110 ms), stimulus No. 2 has an occlusion time of 100 ms, stimulus No. 3 of 90 ms, etc. As a control, the minimal pair Tick/dick (/tik/, ‘‘tic’’; /dik/, ‘‘thick’’) was considered. These two utterances differ in VOT of the initial alveolar stop. An earlier study had revealed a VOT above 47 ms in /t/ productions of German control speakers, the values of the respective voiced cognate were below 21 ms (Ackermann & Hertrich, 1997). The perceptual phonemeboundary effect can be expected, thus, to occur between these two endpoints. As in the Boten/ Boden series, 10 equally spaced stimuli covering the relevant range of VOT were generated. To obtain these items, the VOT of the acoustic signal of the word Tick (VOT 5 56 ms) was reduced in steps of 4 ms (Fig. 1B). The lowest VOT value tested, thus, amounted to 20 ms. A master tape—comprising 10 tokens each from the 10 stimuli of the respective series in randomized order—was played to the patients and their controls using loudspeakers adjusted to a comfortable loudness. Subjects were asked to indicate the perceived item on a sheet. A training interval preceded formal testing. The four endpoint items (VOT 5 56 ms, VOT 5 20 ms; occlusion duration 5 110 ms, occlusion duration 5 20 ms) were presented five times each in randomized order, but blocked for minimal pair. Following each single stimulus, the experimenter provided an orthographic representation of the appropriate category. Apart from feedback, training and test phase did not differ from each other.

RESULTS AND CONCLUSIONS

Figure 2 displays the obtained results in terms of identification functions (Studdert-Kennedy et al., 1970). As concerns the continuum of stimuli differing in VOT of the initial alveolar stop, both the cerebellar patients and their controls produced—in accordance with the findings of Ivry and Gopal (1992)—a rather abrupt transition from one response type to the other, i.e., a pattern indicating categorical perception (Fig. 2A). Disorders of the cerebellum, thus, do not seem to compromise categorical perception on a continuum of VOT values. With respect to the identification of stimuli varying in occlusion time (Boten/Boden series), the subjects with cerebellar atrophy, i.e., bilateral pathology, did not show a discernible phoneme-boundary effect (Fig. 2B). Accordingly, statistical analysis revealed neither a significant linear (F(1, 7) 5 2.69, p 5 .145) nor a quadratic trend (F(1, 7) 5 .036, p 5 .855). However, the respective effects achieved significance in the control group (quadratic trend: F(1, 9) 5 14.94, p 5 .004; linear trend: F(1, 9) 5 59.29, p , .001). As can be seen in Fig. 2B, the two subjects with unilateral cerebellar lesion had similar identification curves as the controls. Perception of linguistically relevant intervals, thus, seems to be impaired in bilateral cerebellar disorders if the stimuli do not provide further task-relevant cues. These findings are compatible with the assumption that the cerebellum represents an internal clock prerequisite for temporal computations also in the perceptual domain (Keele & Ivry, 1991). In addition to perceptual functions, cognitive abilities such as problem solving and set shifting also have been related to the cerebellum (Schmah-

SPEECH PERCEPTION AND CEREBELLUM

329

mann, 1991; Leiner, Leiner, & Dow, 1993). The observed discrepancies between the Tick/dick and the Boten/Boden series, thus, might be explained by the assumption that identification of the latter stimuli depends upon cognitive skills. Indeed, Mattingly and Liberman (1988) have argued that categorical perception reflects the functions of a phonetic module impenetrable by other cognitive activities but these suggestions do not necessarily hold true for all paradigms of phoneme-boundary effects. However, an earlier neuropsychological study had provided evidence that lesions and diseases restricted to the cerebellum do not compromise cognitive abilities (Daum, Ackermann, Schugens, Reimold, Dichgans, & Birbaumer, 1993). Deficits in this domain rather indicate the extension of the disease process to extracerebellar structures. Moreover, five subjects of the present study had participated in that previous investigation. At formal testing they did not show any significant impairment with respect to intelligence, verbal and visuospatial memory, socalled frontal lobe functions, and acquisition of cognitive skills. It seems unlikely, thus, that the observed impairment of speech perception in cerebellar subjects with bilateral pathology reflects a general decline of cognitive abilities. The two subjects with unilateral lesion showed normal performance with respect to the identification of stimuli varying in occlusion time. Conceivably, perceptual deficits in this task, thus, depend upon bilateral pathology of the cerebellum. In accordance with this assumption, animal studies indicate an at least partially bilateral projection of acoustic input to the cerebellum (Snider & Stowell, 1944). Behavioral studies in infants (Eimas, Siqueland, Jusczyk, & Vigorito, 1971) and animals (Kuhl & Miller, 1975) have provided evidence that categorical perception with respect to VOT reflects innate constraints of the auditory system. As indicated by the present study, processing of occlusion times might require a more complex mechanism comprising, among others, a cerebellar clock. A further observation corroborates this assumption. As compared to the stimuli varied in VOT, both the control group as well as the two subjects with unilateral cerebellar lesion of the present study showed a less steep identification function regarding the Boten/Boden series. Using the minimal pair leiten/leiden, Kohler (1979) found quite similar shapes as the ones obtained with Boten/Boden. Conceivably, the discrepancies in steepness of the identification functions on VOT and occlusion length, thus, are due to different underlying decoding mechanisms. REFERENCES Ackermann, H., & Hertrich, I. 1997. Voice onset time in ataxic dysarthria. Brain and Language, 56, 321–333. Ackermann, H., Vogel, M., Petersen, D., & Poremba, M. 1992. Speech deficits in ischaemic cerebellar lesions. Journal of Neurology, 239, 223–227.

330

ACKERMANN ET AL.

Daum, I., & Ackermann, H. 1995. Cerebellar contributions to cognition. Behavioural Brain Research, 67, 201–210. Daum, I., Ackermann, H., Schugens, M. M., Reimold, C., Dichgans, J., & Birbaumer, N. 1993. The cerebellum and cognitive functions in humans. Behavioral Neuroscience, 107, 411–419. Dichgans, J. 1984. Clinical symptoms of cerebellar dysfunction and their topodiagnostical significance. Human Neurobiology, 2, 269–279. Eimas, P. D., Siqueland, E. R., Jusczyk, P., & Vigorito, J. 1971. Speech perception in infants. Science, 171, 303–306. Gilman, S., Bloedel, J. R., & Lechtenberg, R. 1981. Disorders of the cerebellum. Philadelphia, PA: Davis. Ivry, R. B., & Diener, H. C. 1991. Impaired velocity perception in patients with lesions of the cerebellum. Journal of Cognitive Neuroscience, 3, 355–366. Ivry, R. B., & Gopal, H. S. 1992. Speech production and perception in patients with cerebellar lesions. In D. E. Meyer & S. Kornblum (Eds.), Attention and performance. Vol. XIV. Hillsdale, NJ: Erlbaum. Pp. 771–802. Ivry, R. B., & Keele, S. W. 1989. Timing functions of the cerebellum. Journal of Cognitive Neuroscience, 1, 136–152. Keele, S. W., & Ivry, R. 1991. Does the cerebellum provide a common computation for diverse tasks? A timing hypothesis. Annals of the New York Academy of Sciences, 608, 179– 211. Kohler, K. J. 1979. Dimensions in the perception of fortis and lenis plosives. Phonetica, 36, 332–343. Kohler, K. J. 1984. Phonetic explanation in phonology: The feature fortis/lenis. Phonetica, 41, 150–174. Kuhl, P. K., & Miller, J. D. 1975. Speech perception by the chinchilla: Voiced-voiceless distinction in alveolar plosive consonants. Science, 190, 69–72. Leiner, H. C., Leiner, A. L., & Dow, R. S. 1993. Cognitive and language functions of the human cerebellum. Trends in Neurosciences, 16, 444–447. Liberman, A., Harris, K. S., Eimas, P., Lisker, L., & Bastian, J. 1961. An effect of learning on speech perception: The discrimination of durations of silence with and without phonemic significance. Language and Speech, 4, 175–195. Liberman, A. M., Harris, K. S., Kinney, J. A., & Lane, H. 1961. The discrimination of relative onset-time of the components of certain speech and nonspeech patterns. Journal of Experimental Psychology, 61, 379–388. Lieberman, P. 1984. The biology and evolution of language. Cambridge, MA: Harvard Univ. Press. Lisker, L., & Abramson, A. S. 1964. A cross-language study of voicing in initial stops: Acoustical measurements. Word, 20, 384–422. Lisker, L., & Abramson, A. S. 1967. Some effects of context on voice onset time in English stops. Language and Speech, 10, 1–28. Lisker, L., & Abramson, A. S. 1970. The voicing dimension: Some experiments in comparative phonetics. In B. Ha´la, M. Romportl, & P. Janota (Eds.), Proceedings of the Sixth International Congress of Phonetic Sciences, Prague 1967. Prague: Academia. Pp. 563–567. Mattingly, I. G., & Liberman, A. M. 1988. Specialized perceiving systems for speech and other biologically significant sounds. In G. M. Edelman, W. E. Gall & W. M. Cowan (Eds.), Auditory function: The neurobiological bases of hearing. New York: Wiley. Pp. 775–792. Miller, J. D., Wier, C. C., Pastore, R. E., Kelly, W. J., & Dooling, R. J. 1976. Discrimination and labeling of noise-buzz sequences with varying noise-lead times: An example of categorical perception. Journal of the Acoustical Society of America, 60, 410–417. Moslin, B. 1979. The role of phonetic input in the child’s acquisition of the voiced-voiceless contrast in English stops: A voice-onset time analysis. Ph.D. Dissertation. Brown University. [Cited in Lieberman, 1984]

SPEECH PERCEPTION AND CEREBELLUM

331

Repp, B. H. 1979. Relative amplitude of aspiration noise as a voicing cue for syllable-initial stop consonants. Language and Speech, 22, 173–189. Repp, B. H., Liberman, A. M., Eccardt, T., & Pesetsky, D. 1978. Perceptual integration of acoustic cues for stop, fricative, and affricate manner. Journal of Experimental Psychology: Human Perception and Performance, 4, 621–637. Schmahmann, J. D. 1991. An emerging concept: The cerebellar contribution to higher function. Archives of Neurology, 48, 1178–1187. Snider, R. S., & Stowell, A. 1944. Receiving areas of the tactile, auditory and visual systems in the cerebellum. Journal of Neurophysiology, 7, 331–358. Studdert-Kennedy, M., Liberman, A. M., Harris, K. S., & Cooper, F. S. 1970. Motor theory of speech perception: A reply to Lane’s critical review. Psychological Review, 77, 234– 249.