COMPARATIVE ANALYSIS OF NUMBER, DURATION AND LOCALIZATION OF PAUSES FOR STUTTERERS AND PEOPLE WITH NORMAL SPEECH
Federal Center of Speech Pathology and Neurorehabilitation, Moscow, Russia
For the first time, pauses appearing in reading of a given text (200 syllables) and in monologue speech (two-minute narration about one's job) were analyzed and compared for stutterers and people with normal speech. Speech was tape-recorded, transformed to clipped signals and worked on by computer, which selected and counted the pauses in 0.1-1 sec. range with 0.1 sec. intervals. The appearance of spasms was checked visually and aurally. The results were controlled using Student's criterion. In the experiment, 39 adult stutterers and 16 persons with normal speech took part. People with clonic and tonic spasms were studied separately. Our measurements disclose that in stutterers' speech, the following take place: 1) the diminishing of a number of syntaxic and hesitation pauses; 2) the breaking up of syntag-matic partitions of the text; 3) the existence of numerous specific pauses, not manifesting as an acoustic signal, but filled with anomalous motoric activity of muscles composing speech apparatus. The results are interpreted as disclosing: a) peculiarities in process of intra-speech programming and b) reduction of control upon the sense of speech for stutterers. Pauses are obligatory components of a speech. Their duration and characteristic distribution in speech current strongly determine the rhythmical and melodic side of intonation. The appearance of pauses is connected with several causes. The shortest pauses are simply the components of speech current and are due to specific pronunciation of occlusive consonants (up to 0.1 seconds) (Warshawski and Litvak, 1955). Pauses due to linguistic construction of text (linguistic pauses) appear in places of division of sense groups, i.e., between the syntagms and sentences (Kulakov, 1964); their duration is 0.5-1.5 seconds. (Goldman-Eisler, 1972). Here so called "speech" inhalation may be operated, located in accordance with structural divisions of text (Scheiken, 1966; Dalton and Hardcastle, 1977). Linguistic pauses appear in activities such as reading or during spontaneous speech.
Apart from linguistic pauses, hesitation pauses are also characteristic to spontaneous speech. They are placed not only between syntagms and sentences, but also inside them. They may have greater duration (Akhmanova, 1978) and may be accompanied by nonphonological vocal formations (Goldman-Eisler, 1966). Experimental investigation shows that hesitation pauses have very variable duration and reflect that aspect of speech which is automatic in the least degree. Strict regularity in the appearance of hesitation pauses before words, which are very informative and unpredictable in given context, was noticed. There was a connection between the duration of the pauses and the degree of speech problem of difficulty. This seems to mean that during the hesitation pauses, the planning of utterance is realized and the duration of pauses depends on the degree of difficulties and the choice of speech units (Goldman-Eisler, 1961; Nosenko, 1975).
It is well known from speech therapy practice that with stuttering and the breaking of speech intonation, speech rhythm also breaks. That is why working on pauses is an important part of any correction program. It is commonly believed that teaching stutterers to make pauses is useful from many points of view. For example, a pause is connected with the need to make an inhalation, to relax a little during the speech act, and possibly to draw the attention of an interlocutor by particular intonation. However, there have never been any experimental investigations permitting comparison between the pausing process of stutterers and that of persons with normal speech. This knowledge may disclose some new features regarding the mechanisms of stuttering; it will also help teach smooth speech to stutterers.
This paper discusses the comparative analysis of the number, duration and localization of pauses appearing in a text reading and a monologue speech. Our task was to investigate the peculiarities of pausing and make clear the connection between pauses and the appearance of spasms in speech apparatus muscles. To solve this problem we have measured the frequency of the appearance of pauses and their duration and place in speech current; simultaneously, we fixed the appearance of spasms during reading and monologue narration.
A tape-recorder fixed two specimens of speech: 1) a reading of a 200-syllable text, and 2) a two minute monologue about subjects' work. Preparation time was one minute. Speech was recorded openly; (tape-recorder was not concealed). Before tape-recording, about three minutes were dedicated to general talk for adaptation to the experiment. Tape-recorded specimens were passed through a specially programmed computer which transformed electric signals from the microphone to the sequence of rectangular pulses of the same height (clipp-signal). Pulses reflected the duration of sounding speech, and their absence reflected the duration of pauses (of 0.01 second or more long) with accuracy of +0.005 second As a result, we obtained the successive values of time intervals filled with sounding speech or silence (pauses).
We analyzed only the pauses which had a duration of 0.1 second or more, because it is known that shorter interruptions in speech signals are the result of the pronunciation of occlusive consonants (Warshawski and Litvak, 1955). Comparison of speech specimens with clipp-signals gives the exact picture of the localization of the pauses in speech. The total number of pauses of the following intervals of duration was calculated from each speech specimen (in seconds): 0.1-0.2; 0.2-0.3; 0.3-0.4; 0.4-0.5; 0.5-0.6; 0.6-0.7; 0.7-0.8; 0.8-0.9; 0.9-1.0 and 1 second and more. The certainty of estimates was controlled by Student's criterion. The manifestations of spasm activity of the speech apparatus muscles during the reading of the text and during the monologue was determined visually and audibly. Thirty-nine adult stutterers (24 men, 15 women) and 16 persons with normal speech (seven men, nine women) took part in the experiments .
Results and Discussion
Pauses made by persons with normal speech. Distribution of the number of pauses with different duration is shown on the Fig. 1. We see that the shortest pauses of 0.1-0.2 second are the most frequent ones. Their relative number was practically the same in reading and monologue specimens (20.8% and 21.5% accordingly; difference is uncertain p > 0.1). These pauses are mostly the result of pronouncing occlusive consonants; they take place inside the words. They appeared more rarely between the words, but never between the sentences. In the process of the text reading, all pauses with 0.1 -1.0 second duration appeared generally on the border between two syntagms and rarely between two sentences; i.e., they corresponded to the syntagmic partition of the text. Pauses within the 0.2-0.6 second range were more characteristic to the process of reading (54.9%). Longer pauses of 1 second and more (the longest didn't surpass 1.3 seconds) appeared rarely (5.3%). They were recorded only between the sentences and corresponded to dots. The observed regularity in localization and duration of pauses for persons without speech pathology were syntactically explicable: it witnessed about significative text's organization and about adequate understanding of the text.
In monologue speech 0.2 to 0.6 second pauses were less common (39.1 %), than in reading; (difference is statistically certain for 0.4 to 0.5 second pauses with uncertainty p < 0.25 and for 0.5 to 0.6 second pauses with p < 0.005). In contrary, long pauses (from 1 second and up to 3 seconds) appeared 3.2 times more frequently in monologue than in reading (with p < 0.01). The long pauses were the most characteristic for monologue speech (17%). Sometimes these pauses appeared on the borders between sentences and syntagms. Generally, the long pauses took place within the syntactically whole divisions of the speech. In these cases, pauses were preceded by specific intonations of uncertainty by prolongation of the last consonant in the word before a pause, or a pause was interrupted by unphonic vocal formations like "E-E-E" or "M-M-M;" i. е., the phenomena appeared that testified about difficulties, connected with choice of linguistic units in the process of speech realization. Such pauses were absent within the cliche habitual speech turns and between words having a high degree of association. These data show, that long pauses exceeding one second were not only the hesitation pauses, but also reflected difficulties in the process of speech programming at semantic level.
Pauses made by stutterers. Distribution of pauses is shown in Fig. 2. The shortest pauses (0.1-0.2 second) were on an average as frequent as in norm (difference is uncertain with p > 0.1), but they took place generally inside the words. At text reading by stutterers the duration of pauses was more differentiated than in norm. The longest pause reached 15.58 seconds! The most characteristic were either very short pauses (0.1-0.3 second, 37.2 %) or rather long ones (one second and more, 19.7 %). It should be noted that long pauses appeared 3.7 times more frequently than in norm. Other pauses, in particular of 0.3-0.6 second in text reading were considerably more rare than in norm (difference is certain with p < 0.1; p < 0.002 and p < 0.001). Such a significant reduction of pauses number, whose duration in norm is influenced by sy ntagmatic structure of the text, and, in contrary, increasing of the number of long pauses signaled about the deformations in the sense division of text by stutterers. Comparison of tape-records with etalon text offered for reading showed that stutterers made lots of mistakes in the pronunciation of words. The following mistakes were most frequent: (a) changes in grammatical forms of the word; (b) inadequate substitution of words on the basis of phonetic proximity, (c) omission of words, (d) insertion of extra words, and (e) incorrect stresses.
The mistakes made by stutterers were not corrected, and this lead to significant distortions of the text's sense. This suggests that there is no full unity of external and internal sides of the process of reading and that perception of the text is incomplete or deformed. One may think that the principal worry for stutterers reading aloud is securing the pronunciation side of the process. All the conscious attention is directed on the articulation of the text, but not on the perception and transmission of its sense. This leads to the reduction of control over the sense of text. In monologue speech the duration of pauses of the same stutterer has even higher dispersion than in reading. The longest pause reached even 111,85 second In the distribution of pauses we noted the same rule as in reading: the most frequent were either very short pauses (0.1-0.3 second, 35%) or long ones (one second or more, 24.5%).
In connection with these peculiarities, we tried to trace the influence of spasm's type on pause distribution. It is known that clonic and tonic spasms differ first of all by duration. We examined whether duration of speech spasms influenced the duration of pauses appearing in speech production. I or this purpose those persons from our group of stutterers were selected who had mainly clonic or tonic spasms (six persons in each group). All of them were heavy stutterers. Comparing the results, we can see that for persons with clonic spasms, more than half of all spasms were of relatively short duration: from 0.1 second to 0.3 second (57.8% for reading, 54.4% for monologues) (Fig 3).
The analysis showed that these pauses were the result of speech spasms which led to multifold alternations of pauses and parts of syllables or words with nondifferentiated sounds. The longest pauses of one second or more duration appeared very rarely (4.5% for reading, 6.4% for monologues). On the contrary in cases of persons with mainly tonic spasms, (Fig. 4) most characteristic were long pauses of one second or more duration (43% for reading, 46% for monologues). The shorter pauses appeared rarely (22.2% in reading, 19% in monologues). These distinctions between clonic and tonic cases are statistically certain (p < 0.01). Analysis revealed that long pauses in the speech of stutterers with tonic spasms were accompanied by the processes opposed to the processes that are characteristic to hesitation pauses. For example, for spasms located in voice section of speech apparatus, the stops of speech were sudden, the words preceding a pause were uttered in accelerated manner and stopped abruptly. The appearance of nondifferentiated sounds was connected with the effort to begin the speech interrupted by spasms and expressed in specially strained or stressed phonation of the voice. Spasms of 1 second or more duration localized in articulatory apparatus frequently appeared at the attempt to pronounce the occlusive consonant sound; they tallied with the duration of occlusion of the articulative organ before "explosion." An absence of hesitation pauses and pauses connected with syntactic construction of the text and appearance of pauses due to spasmatic speech activity explains peculiarities of intonational side of stutterers speech: monotony, absence of emotional expressiveness and intonational completeness in reading as much as in monologue speech. Besides that, the peculiarities of speech pausing of stutterers may be considered as the singularity in realization of interspeech programming and its essential deviation from the norm as well as the reduced control over the sense of speech. In such a way, for stutterers it was typical to have a good deal of pauses, filled with spasmatic activity of speech apparatus muscles not manifesting itself as a speech signal.
On the contrary, considerably less than in norm was a number of syntagmatic and hesitation pauses needed for normal speech process. If in normal speech pause is the evidence of finishing one part of motoric realization of expression and preparation and planning of the subsequent piece of speech, at stuttering in time of pause, as our experiments show, the motoric activity is going on. This shows that in contrast to normal speakers, for stutterers speech programming develops simultaneously with high motoric activity of speech muscles.
Akhmanova O. S. (1978). About psycholinguistics. P. 55. Moscow, (in Russ.)
Dalton P., Hardcastle W. J. (1977). Disorders of fluency and their effects on communication . London.
Goldman-Eisler F. (1961) The distribution of pause duration in speech. Language and Speech, v. 4, 3-4, pp. 232-237.
Goldman-Eisler F. (1966) A comparative study of two hesitation phenomena. Language and Speech, v. 4, 1-2, pp. 18-26.
Goldman-Eisler F. (1972). Pauses, clauses, sentences. Language and Speech, v. 15, N 3-4, pp. 103-113.
Kulakov K. N. (1964). Some functions of speech pause. 17 conference of Minsk pedagogical institute of foreign languages. Part II. Minsk, (in Russ.)
Nosenko E. L. (1975). The peculiarities of speech in state of emotional stress. Dnepropetrovsk, (in Russ.)
Scheikin RS. (1966). Analysis of mechanisms of pauses in speech. Mechanisms of speech formation and perception of complex sounds. Moscow, Leningrad, pp. 31-44. (in Russ.)
Warshawski L. I., Litvak I. M. (1955). Investigation of formants and some other characteristics of sounds in Russian speech. Problems of physiological acoustics, v. 3, Leningrad (in Russ.)
© Дьякова Е.А. Текст и фотоматериалы, 2006-2014.
© Блосфильд Е.О. Дизайн и разработка, 2006-2014.