Neuroscience Letters 293 (2000) 115±118 www.elsevier.com/locate/neulet Stutter-free and stutter-®lled speech signals and their role in stuttering amelioration for English speaking adults Joseph Kalinowski a,*, Vikram N. Dayalu a, Andrew Stuart a, Michael P. Rastatter a, Manish K. Rami b a Department of Communication Sciences and Disorders, School of Allied Health Sciences, East Carolina University, Greenville, NC 27858-4353, USA b Department of Communication Sciences and Disorders, University of North Dakota, Grand Forks, ND-58202-8040, USA Received 1 June 2000; received in revised form 30 August 2000; accepted 31 August 2000 Abstract This study examined the power of an exogenously generated stuttered speech signal on stuttering frequency when compared to an exogenously generated normal speech signal. In addition, we examined the speci®c components of the second speech signal, which might be responsible for the inducement of ¯uency in people who stutter. Eight males and two females who stuttered participated in this study. Experiment I involved meaningful speech: normal continuous speech, normal interrupted speech, stuttered continuous speech, and stuttered interrupted speech, whereas Experiment II involved vowels and consonants: /a/, /a-i-u/, /s/, /s-sh-f/. The results indicated that stuttered and normal speech signals were equally effective in reducing stuttering frequency. Further, the vowels were more powerful than consonants in inducing ¯uency for people who stutter. It is suggested that acoustic manifestations of stuttering, rather than a problem, may be a natural compensatory mechanism to bypass or inhibit the `involuntary block' at the neural level. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Stuttering; Auditory signals; Auditory feedback; Speech gesture; Choral speech A second speech signal comes from another speaker in unison, choral, semi-choral, or shadow forms, where the speech signal from the other speaker is temporally either ahead, behind, or in perfect synchronization with the person who stutters. In terms of probability, perfect synchronization of two speech signals is relatively rare. In its electronic manifestations, a second speaker is `derived' via electronic means to produce a signal, which is temporally out of phase and oftentimes shifted in frequency. A substantial number of replicated empirical reports coupled with the near total absence of opposing reports indicate that endogenous alterations of the speech signal output (e.g. prolonged or slowed speech, rhythmic speech, singing, and lipped speech) or exogenous alterations of speech signal input induces relatively ¯uent speech in people who stutter [6]. Exogenous auditory speech signals in the form of a `second speech signal' (e.g. chorus reading, shadow speech, delayed auditory feedback, frequency altered feedback) [2±6,8,14,15], or * Corresponding author. Tel.: 11-252-328-1986; fax: 11-252328-4469. E-mail address: kalinowskij@mail.ecu.edu (J. Kalinowski). speech signals via the visual modality (e.g. visual choral speech) [16], produce more powerful and natural-sounding reductions in stuttering than incongruous non-speech auditory (e.g. masking noise, clicks) [1,8,10,19] or visual (e.g. ¯ashing lights) inputs [17,18]. All exogenous speech signals, either produced by a second speaker or produced via electronic means, should be considered as `second speech signals'. In other words, when persons who stutter are speaking in the presence of an exogenous second speech signal, they always receive two speech signals due to the impossibility of removing the bone conducted signal generated by the endogenous speech signal (i.e. the person's own speech), and the incomplete masking of the air-borne speech signal. Generally, the reduction in stuttering frequency under alterations of the second speech signal has been attributed to such factors as entrained rhythm [14], distraction [5,6], modi®ed vocalization, and rate reduction [20]. These explanations are based on the assumption that the peripheral manifestations of stuttering (e.g. repetitions, prolongations, and audible struggle behaviors) are the problem. There has been, however, no examination as to the possibility that 0304-3940/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S03 04 - 394 0( 0 0) 01 50 9- 3 116 J. Kalinowski et al. / Neuroscience Letters 293 (2000) 115±118 stuttering behaviors may be a natural compensatory mechanism for an `involuntary block' at the central level, rather than a manifestation of a mechanistic problem at the peripheral level. If so, then the immediacy and naturalness of the ¯uency enhancement via the second speech signal may be explained in a dramatically different manner. In order to investigate these notions, we examined the power of an exogenous stuttered speech signal on stuttering frequency when compared to an exogenous normal speech signal. We used incongruent second speech signals in order to compare the inherently incongruent nature of exogenous stuttered speech to that of the disconsonant ¯uent speech (in incongruent or disconsonant speech, the second speech signal contains different phonemic material than that read aloud by the participants). If ¯uency reduction was observed as would be expected in the aforementioned compensatory mechanism model, it would be essential to determine the speci®c components of the incongruent second speech signal, which might be responsible for the inducement of ¯uency. Speci®cally, the natural classi®cation scheme of vowels and consonants were examined in both dynamic and relatively static vocal tract positions as this provides an appropriate continuum for examination. Ten normal-hearing adults who stutter (eight males, two females, mean 27.9 years, SD 9.4) participated in both experiments. Participants did not present with any other speech and language disorders. All participants had a history of therapy, but were currently not receiving any formal therapeutic intervention. Participants read different junior-high level passages of 300 syllables with similar theme and syntactic complexity in both experiments. Based on informal clinical measures, severity of stuttering for the participants in this study ranged from moderate to severe. The two experiments were counterbalanced while the experimental conditions and the passages were randomized. They were instructed throughout the experiment to read at a normal rate and not to use any controls to reduce or inhibit stuttering. In both experiments participants listened to the exogenous auditory speech signals via supra-aural earphones (headphones that completely cover the external ear) at a most comfortable listening level. The participants listened to the exogenous speech signals during the various reading tasks in both experiments. It should be reiterated that the auditory signals used in these experiments did not fully mask the auditory feedback generated by the participants' own speech. Participants would perceive their endogenously produced speech via bone conduction as well as portion of the airborne signal due to the incomplete masking capabilities of the exogenous speech signal. The ®rst experiment required participants to listen to incongruous ¯uent or stuttered speech samples presented continuously or intermittently (50% duty cycle). Both speech samples were incongruent recorded text. The stuttered speech sample contained discrete stuttering acts on all words. Thus in total, there were four listening conditions. In the second experiment, participants listened to four contin- uous speech signals: a steady state neutral vowel /a/, a three vowel train representing the three corners of the vowel triangle /a-i-u/, a steady state consonant /s/, and a three consonant train /s-sh-f/. The consonants were selected as these could be presented in the absence of a vowel. Steady vowels and consonants and trains of each were used to represent different levels of proximity with the speech act. Participants also read a control passages for each experiment. This was the non-altered auditory feedback (NAF) condition, wherein the participants read a passage in the absence of any external auditory stimuli other than the auditory feedback normally received while speaking (i.e. no exogenous speech signals). The stimuli for these samples were recorded in a soundtreated room with a digital tape-recorder (Sony model 8819). A normal ¯uent American-English speaking adult male produced the vowel, consonant, and ¯uent speech samples for both experiments. An American-English adult male who stutters produced the stuttered speech sample for the ®rst experiment. Both speakers produced speech samples at a normal vocal effort. The text for the ®rst experiment was junior-high level passages with similar theme and syntactic complexity as that read by the participants of the experiments. The recorded signals were then fed into a personal computer (Apple Power Macintosh 9600/300) via an apple sound input port. Sampling was at 44 kHz. Sound analysis software (Sound Edit version 2) was used to introduce the silence, select the various stuttering moments, and loop the signals. Silent intervals randomly varied from 2±5 s. These were then recorded onto a compact disk that would be used to deliver the signal via a compact disk player (Sony model CFD-S28). The signals were delivered binaurally via headphones (Optimus model PRO.50MX) at the participants' most comfortable level. All participants spoke into a lapel microphone (Radio Shack model 33-3003) af®xed no more than 15 cm from their mouths with an approximate orientation of 08 azimuth and 21208 altitude. The microphone output was fed into a video camera (Sony model CCDTVR 75). Stuttering episodes were calculated from the participants' videotape recorded passages. Stuttering was de®ned as partword repetitions, part-word prolongations, and/or inaudible postural ®xations. Interjudge syllable-by-syllable agreement, as indexed by Cohen's kappa [9] was 0.79. Intrajudge Cohen's kappa syllable-by-syllable agreement was 0.81. Kappa values above 0.75 represent excellent agreement beyond chance [12]. The means and standard deviations for stuttering frequency as a function of the exogenous auditory speech signals delivered in Experiment I are as follows: (a) NAF (mean 35.4, SD 25.11), (b) ¯uent interrupted (mean 15.4, SD 17.11), (c) stuttered interrupted (mean 14, SD 15.59), (d) stuttered continuous (mean 10.4, SD 10.09), and (e) ¯uent continuous (mean 8.6, SD 13.2). A one-factor-repeated-measure-analysis-of- J. Kalinowski et al. / Neuroscience Letters 293 (2000) 115±118 variance revealed a signi®cant main effect of exogenous auditory speech signal on stuttering frequency (F(4,36)14.35, Greenhouse-Geiser P 0.0004, h 2 0.62). A post hoc single-d.f. comparison revealed there was a signi®cant reduction in stuttering frequency for all forms of exogenous auditory speech signals relative to NAF (P , 0:0001). No statistically signi®cant differences were observed between ¯uent and stuttered speech signals (P 0:76), or continuous and interrupted speech signals (P 0:10). The means and standard deviations for stuttering frequency as a function of exogenous auditory speech signals delivered in Experiment II ± are as follows: (a) NAF (mean 37.5, SD 27.23), (b) /a/ (mean 10.3, SD 15.98), (c) /a-i-u/ (mean 8.2, SD 11.92), (d) /s/ (mean 18, SD 20.49), and (e) /s-sh-f/ (mean 24.9, SD 25.53). A one factor repeated measure analysis of variance reveled a signi®cant main effect of the exogenous auditory speech signal on stutter-ing frequency Greenhouse±Geiser P 0:0001, (F 4;36 17:77, h2 0:66). A post hoc single-d.f. comparison revealed there was a signi®cant reduction in stuttering frequency for all forms of exogenous auditory speech signals relative to NAF (P , 0:0001). There were also statistically signi®cant fewer stuttering episodes when the exogenous auditory signals were vowel(s) vs. consonant(s) (P , 0:0001). Nonsigni®cant differences in stuttering frequencies were found between single versus trains of speech signals (P 0:40). Before this set of experiments, to the best of our knowledge, there has been no empirical documentation that both exogenously generated stutter-®lled and stutter-free incongruous speech signals could induce same levels of ¯uency enhancement in people who stutter. Our results indicated that stuttering frequency was substantially reduced, irrespective of whether the exogenous auditory speech signal were stuttered or stutter free. What are the possible mechanisms that could be involved in the generation of true ¯uency via the incongruous auditory speech signal presented in the form of a second speech signal? The ¯uency enhancement produced by a congruent but temporally shifted speech signal might suggest a timing or entraining support mechanism (i.e. choral speech, DAF, FAF). However, asynchronous speech signals (i.e. signals used in this study) produce the same level of ¯uency enhancement as the signals mentioned above, suggesting some form or type of `inhibitory response'. That is, the second speech signal provides the required compensation to aid the system in effectively executing the motor plan and generates ¯uent speech in persons who stutter. Our data suggests that a sonorant, voiced speech signal, stuttered or ¯uent, intermittent or continuous, is an effective acoustic mechanism to release the system from the involuntary block at the neural level. However, these second speech signals do not totally eliminate the involuntary block as they only occur in compensation to the formation of a `neural block' that has already occurred. For a second signal to be a prophylaxis to 117 the overt manifestations of stuttering, it must impact the neural system before the block occurs- as is evidenced in identical choral speech. We also suggest that this data might illuminate the purpose of producing oscillations and sustained vowel gestures by a person who stutters (i.e. repetitions and prolongations). It is hypothesized that the person produces the overt manifestations of stuttering (i.e. repetitions and prolongations), in an attempt to generate an auditory release mechanism from an `involuntary block' in speech execution at the neural level. Simply put the overt manifestations of stuttering are an attempt to compensate at the peripheral level for a loss of control at the central level-albeit a conspicuous compensation. Thus the overt stuttering manifestations are hypothesized to be a form of compensation rather than a problem in itself. The overt manifestations of stuttering appear to be analogous to the role of fever in an infectious disease state. Oftentimes, the fever has been confused as the problem, when in fact it is an inherent compensatory mechanism of the body to a disease state. Thus, a lack of an appropriate ¯uency enhancing gesture is hypothesized to be the predominate etiological factor that is exhibited or manifested due to a lack of inhibition on the part of the auditory cortex in assimilating the appropriate plan required for the smooth execution of the speech act. Recent brain imaging procedures have employed choral speech condition to induce ¯uent speech in adults who stutter and have compared the brain images obtained, to those attained during the stuttering behavior [13,21]. A lack of activation in the auditory areas during the motor planning of stuttered speech was observed, but an essential normalization under the choral speech condition was noted, indicating its inherent ¯uency enhancing capabilities. Simply put, the second speech signals provide the required inhibition that is essential for the system of a person who stutters, to smoothly execute the speech act. The resultant inhibition via the second speech signal eliminates the system from requiring the production of oscillations or stuttering manifestations (i.e. repetitions and prolongations) that was previously required to produce speech. Thus the resultant speech is almost free of the overt manifestations of stuttering. It should be noted that a parallel between the manifestations of stuttering and that of expressive aphasia has been identi®ed previously and appears to be support, at least in part in the data revealed in these experiments [7,11]. That is, the stuttering and expressive aphasia may exist on a functional-morphological continuum and have similar central explanations. The power of auditory cueing or priming in both disorders provides an impetus for further investigation of the possible parallels of these disorders. [1] Adams, M.R. and Moore Jr., W.H., The effects of auditory masking on the anxiety level, frequency of dys¯uency, and selected vocal characteristics of stutterers, J. Speech Hear. Res., 15 (1972) 572±578. 118 J. Kalinowski et al. / Neuroscience Letters 293 (2000) 115±118 [2] Andrews, G., Craig, A., Feyer, A., Hoddinott, S., Howie, P. and Neilson, M., Stuttering: a review of research ®ndings and theories circa 1982, J. Speech Hear. Disord., 45 (1983) 226±246. [3] Andrews, G., Howie, P.M., Dozsa, M. and Guitar, B.E., Stuttering: speech pattern characteristics under ¯uency-inducing conditions, J. Speech Hear. Res., 25 (1982) 208±215. [4] Armson, J., Kalinowski, J. and Stuart, A., A model of stuttering remediation: multiple factors underlying ¯uency enhancement, In C.W. Starkweather and H.F.M. Peters (Eds.), Stuttering: Proceedings from The First World Congress on Fluency Disorders, University Press, Nijmegen, 1995, pp. 296±300. [5] Barber, V., Studies in the psychology of stuttering: XV, Chorus reading as a distraction in stuttering, J. Speech Disord., 4 (1939) 371±383. [6] Bloodstein, O., A Handbook on Stuttering, 5th edn., Singular, San Diego, CA, 1995. [7] Boller, F., Vrtunski, P.B., Kim, Y. and Mack, J.L., Delayed auditory feedback and aphasia, Cortex, 14 (1978) 212±226. [8] Cherry, E. and Sayers, B., Experiments upon total inhibition of stammering by external control and some clinical results, J. Psychosom. Res., 1 (1956) 233±246. [9] Cohen, J., A coef®cient of agreement for nominal scales, Edu. Psychol. Meas., 20 (1960) 37±46. [10] Conture, E.G., Some effects of noise on the speaking behavior of stutterers, J. Speech Hear. Res., 17 (1974) 714±723. [11] Eisenson, J., Stuttering as preservative behavior, In J. Eisenson (Ed.), Stuttering: A Symposium, Harper & Row, New York, 1975, pp. 403±452. [12] Fleiss, J.L., Statistical Methods for Rates and Proportions, 2nd edn., Wiley, New York, 1981. [13] Fox, P.T., Ingham, R.J., Ingham, J.C., Hirsch, T.B., Downs, J.H., Martin, C., Jerabek, P., Glass, T. and Lancaster, J.L., A PET study of the neural systems of stuttering, Nature, 382 (1996) 158±161. [14] Johnson, W. and Rosen, L., Studies in psychology of stuttering: VII. Effect of certain changes in speech pattern upon frequency of stuttering, J. Speech Disord., 2 (1937) 105±109. [15] Kalinowski, J., Armson, J., Roland-Mieszkowski, M., Stuart, A. and Gracco, V.L., Effects of alterations in auditory feedback and speech rate on stuttering frequency, Lang. Speech., 36 (1993) 1±16. [16] Kalinowski, J., Stuart, A., Rastatter, M.P., Snyder, G. and Dayalu, V., Inducement of ¯uent speech in persons who stutter via visual choral speech, Neurosci. Lett., 281 (2000) 198±200. [17] Kuniszyk-Jozkowiak, W., Smolka, E. and Adamczyk, B., Effect of acoustical, visual, and tactile reverberation on speech ¯uency of stutterers, Folia Phoniatr. Logop., 48 (1996) 193±200. [18] Kuniszyk-Jozkowiak, W., Smolka, E. and Adamczyk, B., Effect of acoustical, visual, and tactile echo on speech ¯uency of stutterers, Folia Phoniatr. Logop, 49 (1997) 26±34. [19] May, A.E. and Hackwood, A., Some effects of masking and eliminating low frequency feedback on the speech of stammerers, Behav. Res. Ther., 6 (1968) 219±223. [20] Wingate, M.E., Stuttering: Theory and Treatment, Irvington, New York, 1976. [21] Wu, J.C., Maguire, G., Riley, G., Fallon, J., LaCasse, L., Chin, S., Klein, E., Tang, C., Cadwell, S. and Lottenberg, S., A positron emission tomography [18 F] deoxyglucose study of developmental stuttering, NeuroReport, 6 (1995) 501± 505.