Speech sound errors in patients with conduction and Broca's aphasia

BRAIN

AND

LANGUAGE

20,

175-194 (1983)

Speech Sound Errors in Patients with Conduction Broca’s Aphasia

and

HISAKO MONOI AND YOKO FUKUSAKO Tokyo Metropolitan Geriatric Hospital MOTONOBU ITOH Tokyo Metropolitan Institute of Gerontology AND SUMIKO SASANUMA Yokohama National Universit)l Speech sound errors exhibited by three conduction and three Broca’s aphasic patients on naming and word-repetition tasks were subjected to phonemic and subphonemic analyses. In the conduction aphasic patients, errors occurred equally often on consonants and vowels in both the naming and word-repetition tasks, while in the Broca’s aphasic patients errors occurred selectively on consonants. Transposition errors occurred almost as often as substitution errors in the conduction aphasic patients, while substitution errors constituted the majority of errors in the Broca’s aphasic patients. The Broca’s aphasic patients, as compared to the conduction aphasic patients, exhibited a markedly higher number of substitution errors occurring between phonemes separated by a single subphonemic feature on the naming task. On the basis of these findings, it was hypothesized that the differences in the error patterns of the two types of aphasia reflected differences in the underlying mechanisms of the impairment in each type.

INTRODUCTION One of the salient features shared by both conduction aphasic patients and Broca’s aphasic patients is that they make various kinds of speech sound errors in their oral language. The patterns of these speech sound errors have drawn considerable interest from investigators because of This work was supported in part by a grant from the Adult Disease Clinic Memorial Foundation and a Grant-in-Aid for Scientific Research (No. 56570792),the Japanese Ministry of Education. Send requests for reprints to Sumiko Sasanuma. Present address: Tokyo Metropolitan Institute of Gerontology, 35-2, Sakaecho, Itabashiku, Tokyo-173, Japan. 17.5 0093-934X/83 $3.00 Copyright 0 1983 by Academic Press. Inc. All rights of reproduction in any form reserved.

176

MONO1 ET AL.

the possible information they could provide for hypothesizing the nature as well as the level of the underlying deficit in the speech production processes of these two types of patients. According to some of these investigators (Brown, 1975; Dubois, Hecaen, Angelergues, de Chatelier, & Marcie, 1973; Goodglass & Kaplan, 1972; Lecours & Rouillon, 1976), speech sound errors in conduction aphasia and other fluent types of aphasia are attributable to dysfunction at the level of phonological (and therefore linguistic) processing, where a set of phonological rules, such as the proper choice and sequencing of the sound units of the language, are applied to the signal. Figure 1 is a schematic representation of the various stages suggested by these and

LINGUISTIC ENCODING A, SEMANTIC 6. SYNTACTIC C. PHONOLOGICAL

SPEECH FIG. 1. A schematic representation of the various stages involved in speech production.

SPEECH SOUND ERRORS IN APHASIA

177

other authors as being involved in speech production. In the framework of this schema, the locus of impairment for the speech sound errors in conduction aphasia is at Level 2-C, which has been labeled variously by different authors-e.g., the level of “second articulation” (or phonemic level) by Lecours and Rouihon (1976), or the level of the “first articulation” by Dubois et al. (1973). As for speech sound production errors in Broca’s aphasia, at least three different levels of dysfunction have been proposed in the literature. According to some investigators (e.g., Darley, Aronson, & Brown, 1975; Johns & Darley, 1970; La Pointe & Johns, 1975), articulation errors in Broca’s aphasia are associated with impairment at the level of programming for articulatory movement or Level 3 in Fig. 1, which seems to correspond to what is called the level of “third articulation” (or phonetic level) by Lecours and Rouillon (1976) and the level of “second articulation” by Dubois et al. (1973). According to other investigators (e.g., Martin 1974; Martin & Rigrodsky, 1974a, 1974b; Dunlop & Marquardt, 1977) however, dysfunction at the level of phonological processing (Level 2-C in Fig. 1) is responsible for speech sound errors in Broca’s aphasia. According to still others (Sasanuma, 1971; Itoh, Sasanuma, & Ushijima, 1979), not only Level 3 but also Level 2-C is likely to be involved in some patients. These authors conducted an in-depth study of a patient with a left frontal lesion over a period of several years using direct observation of the articulatory movements of the patient, coupled with a phonemic analysis of his speech production errors. Their findings indicated that the speech sound errors of the patient relatively early in the postonset period (7 months after the onset of the illness) reflected an impairment which was not confined to the motor programming aspect of speech production (Level 3), but extended into the phonological sphere (Level 2-C) as well. However, at a later stage of recovery (6 years and 4 months post-onset) the number of errors reflecting an impairment in phonological processing had diminished to almost none. Most of the foregoing studies, however, have dealt with either nonfluent (or anterior) patients only, or fluent (or posterior) patients alone, precluding the possibility of a direct comparison of error patterns among different subtypes of aphasia. Only few studies have attempted a systematic analysis of speech production errors in both fluent and nonfluent aphasic patients. In one such study, Blumstein (1973) conducted a detailed phonological analysis of the speech output obtained from 17 aphasic patients belonging to three different diagnostic categories: Broca’s, conduction, and Wernicke’s aphasic patients. The data were collected while the patients were engaged in an open-ended conversation in an interview situation. The results, however, revealed no consistent differences among the three aphasic groups either in terms of the error types exhibited or the distinctive features and markedness characteristics of the substitution errors.

MONO1 ET AL.

In two separate studies, Trost and Canter (1974) and Bums and Canter (1977) investigated the speech production errors of 10 anterior (Broca’s) and 10 posterior (five conduction and five Wernicke’s) aphasic patients by means of a single-word articulation test made up of 130 monosyllabic words. The results indicated, among other things, that posterior patients exhibited more errors involving two or more subphonemic component alterations than errors involving a single component, while the anterior patients demonstrated the reverse tendency-i.e., more errors involving one subphonemic component than errors involving two or more components. These findings contradict, in part, the results of Blumstein (1973), who found more one-component errors than two- (or more) component errors in all of her three aphasic subgroups (i.e., Broca’s, conduction, and Wernicke’s aphasic patients). Thus, the results of research conducted so far are inconclusive as to the exact nature of the speech sound error patterns exhibited by different subgroups of aphasic patients. The major purpose of the present investigation, therefore, was to determine if there are any patterns of speech sound errors distinguishing two different types of aphasia: Broca’s and conduction aphasia. As a secondary purpose we wanted to know if the different types of tasks (i.e., repetition versus spontaneous speech production) would have different effects on the performance of Broca’s and conduction aphasic patients. METHODS Subjects The subjects were three conduction and three Broca’s aphasic patients with mild to moderate linguistic impairment. Table 1 gives their age, sex, type of aphasic syndrome, severity, and the period elapsed since onset. The etiology of the aphasia in every case was a single cerebrovascular accident in the left hemisphere. The locus and extent of the TABLE 1 SUBJECT DATA

Case

-he

Sex

Aphasic syndrome

Severity

POM

1

60

M

Conduction

27

2

45

M

Conduction

3 4

63 45

M M

Conduction Broca

5

44

F

Broca

6

49

F

Broca

mildmoderate mild. moderate mild mildmoderate mildmoderate mild

L?POM, the post-onset months.

3 5 7 4 5

SPEECH SOUND ERRORS IN APHASIA

179

lesions for all but one of the patients (Case 1, with conduction aphasia) were determined by CT scan. The findings were consistent with the aphasic syndromes. That is, the two patients with conduction aphasia showed an involvement of the supramarginal gyrus of the left hemisphere, while all the patients with Broca’s aphasia showed damage to the postero-inferior part of the left frontal lobe. Neurological examination disclosed that two of the Broca’s patients showed a right hemiparesis, while none of the patients showed any visual field defect, hearing loss, bilateral neuromotor involvement of the speech muscles, or impaired intellectual functioning. Classification of the patients into the two diagnostic categories of aphasia relevant to this study was based on the overall pattern of the linguistic impairment exhibited by each patient on our Diagnostic Aphasia Examination and, in particular, on those characteristics of speech production which are generally recognized as being diagnostically significant features in differentiating among different aphasic syndromes. The three subjects who were classified as having conduction aphasia had marked difficulty in repetition, with increasing difficulty for polysyllabic words and unfamiliar phrases. The longest sentences they could repeat were those with less than three words. On the other hand. they exhibited fluent and grammatical spontaneous speech with excellent articulation, as well as relatively wellpreserved naming, although with occasional sound errors. These sound errors tended to be most prominent in repetition. The subjects were well aware of their errors and tried to correct them with repeated approximations to the target, resulting in so-called “zeroingin” behavior. Their auditory comprehension was well preserved, attaining a score in the 90th percentile for auditory comprehension on our Diagnostic Aphasia Examination. The three subjects classified as having Broca’s aphasia, on the other hand, were characterized by nonfluent, effortful, and short-phrased speech with awkward articulation, although their spontaneous speech was functional with only a slight agrammatic tendency, and their naming well preserved. Their repetition was relatively good, too: they could repeat five- to seven-word sentences. Their auditory comprehension was almost intact, attaining a score in the 90th percentile for auditory comprehension on our Diagnostic Aphasia Examination.

Data Collection In order to obtain speech samples for analysis, two types of tasks-naming and repetition of three- to five-syllable words (nouns)-were given to each patient. There were three reasons for employing these two tasks with single-word stimuli: (1) we felt it necessary that the target words be known to us so as to be able to make an adequate analysis of the subject’s responses; (2) the preliminary observation of the speaking behavior of these patients had indicated that they tended to make speech sound errors most frequently on content words of three, four, and five moras (moras are roughly equivalent to CV syllables of equal temporal duration); and (3) we were interested in knowing whether the error pattern of these patients on the repetition task (i.e., when they were given a model to imitate) would be different from their performance on the naming task or the task of spontaneous production. The majority of these three- to five-mora test words were used in common for all of the patients and in both the naming and repetition tasks. However, since speech samples from each patient were gathered until what seemed to us a sufficient number of errors for analysis were obtained for each task (an arbitrary minimum of 70 errors) and since the rate of error occurrence was different from patient to patient, the number and the content of the test words used were not necessarily the same for all the patients. In both the naming and repetition tasks. the patients were seated in a sound-proof room with the examiner (H.M.). For the naming task, line drawings of the test words were used as stimuli. In the repetition task, the examiner read aloud each of the test words once and asked the subject to repeat them. The number of sessions required for each subject ranged

180

MONO1 ET AL. TABLE 2

EXAMPLES

Case

OF SPEECH SOUND ERRORS MADE BY CASE 1 (WITH CONDUCTION CASE 4 (WITH BROCA’S APHASIA)

Target

APHASIA)

AND

Responses

ltebukuroi /kutsuSita/ /sakura/

“tebe, te, tebu, tebu, tebutsu, tebe, tebukuro.” “ku, kuji, kuji, kusu, kutsu, kutsujita” “haku, ha, a, sakura”

/jido:ja/ lsentakukil /toire/

“ito:ja” “tsentaki, tsentakuki” “toide”

from four to six, and the order of the two tasks was randomized across the sessions for each subject. All the speech error samples thus obtained and recorded on tape were transcribed using a broad phonemic transcription. There were a few instances of verbal paraphasias and jargon, but these were deleted from the sample. Table 2 shows some examples of the errors for two patients, one with conduction aphasia and the other with Broca’s aphasia. As can be seen from the table, there was a conspicuous tendency for the conduction aphasic patient (Case 1) to try to correct his errors with repeated approximations, while this was not marked in the Broca’s aphasic patient (Case 4). Since this “zeroing-in” behavior was a cardinal feature of every one of the three conduction aphasic patients, we decided to transcribe all the errors produced in the process of self-correction for every patient, including the Broca’s aphasic patients. All the errors thus transcribed were grouped into three classes according to the number of moras in the target words (three-, four-, and five-mora words). Lastly, the final selection of the errors to be analyzed was made by means of a random sampling, in such a way that the number of errors occurring for three-, four-, and five-mora words became roughly 3:2: 1, respectively, for each patient. The relative frequency of occurrence of three-, four-, and five-mora words decreases in this order in Japanese. A total of 959 errors appearing in 455 words on the naming task and a total of 931 errors appearing in 485 words on the repetition task were selected with these procedures and formed the data for analysis in this study. (See Appendix 1 for the breakdown of these errors.)

Analysis

of Data

It is crucial for a given method of error analysis to accurately capture the nature of the errors to be analyzed. An example will help clarify the point. As can be seen in Table 2, for the target word /kutsujita/ (“socks”), Case 1 with conduction aphasia responded with /ku, kuJi, kuj, kusu, kutsu, kutsu.fita/. The sequence /kuJi/, appearing as the second and the third trials in the series just quoted, can be interpreted in several ways. First, /kuji/ could be a verbal paraphasia, because /kuji/ means “comb” in Japanese, and Case 1 may have retrieved this word in place of /kutsujita/ (“socks”). But this is rather unlikely since Case 1 finally produced the target word correctly. Second, it is possible to interpret Case l’s /kuJi/ as a case of phoneme omission for the phonemes /ts, u, t, a/. This does not seem to be quite right either, because the patient was well aware of his errors and tried to correct them immediately. Third, it is also possible to regard the phonemes /.f, i/ of /kuJii/, which constitute the third mora of the target word /kutsujita/, as being transposed to the position of the second mora in his response-with the rest of the phonemes of the

181

SPEECH SOUND ERRORS IN APHASIA

target word “omitted.” The latter part of this third interpretation does not seem to be the case for the same reason mentioned for the second possible interpretation. The fourth possible interpretation is similar to the third in that the phonemes /J, i/ in /kuJi/ are interpreted as having been transposed, but it is different with respect to the interpretation of the missing phonemes. That is, according to the fourth possible interpretation, the patient’s speech production could have been “discontinued” right after /kuJi/ because he immediately realized his error and started to correct it. Since clinical observation of all the patients with conduction aphasia indicated to us that the last interpretation is closest to what was actually taking place, we adopted the fourth interpretation. Thus, each error produced by the patients was evaluated and analyzed with reference to a phoneme or phonemes in the target word. The analysis consisted of the following three steps: 1. Determination

of the relative ,frequency

of errors involving

consonants

and vowels.

The first step of the analysis was to determine whether the target phoneme on which the error occurred was a consonant, a vowel, or otherwise (e.g., syllabic nasals, contracted sounds, or long vowels).’ 2. CIassijication of error types. The second step of the analysis was to classify all the error sounds on consonants and vowels into different categories, i.e., substitution, transposition, addition, and omission. “Substitution” errors were those characterized by the production of a phoneme (consonant or vowel) which was not a constituent of the target word, in place of an actual constituent of the target word, e.g., /daikon/ “raddish” -+ /daiton/, /sampo/ “walk” --f /sampu/. “Transposition” errors had two subtypes. In the first subtype, errors were characterized by the production of a phoneme, in place of the target phoneme, that was a constituent elsewhere in the target word, e.g., /tamago/ “egg” --, /tamamo/, /tagamo/, and /Jimbun/ “newspaper” 4 /Jimbin/. The second subtype was characterized by transposition errors plus “discontinuation” as discussed above, e.g., /kutsujita/ “socks” -+ /kuJ/ and /asakusa/ “the name of a town” + lakaf. an extra consonant was inserted before the vowel in a mora In the case of “Addition,” consisting of only a vowel, e.g., /taiko/ “drum” + /tariko/. In “Omission,” a consonant in a mora was omitted, e.g., /hagaki/ “card” -+ /agaki/. There was no instance of the omission of only the vowel of a mora in the corpus. 3. Subphonemic feature analysis. The third step was to analyze the substitution errors on consonants in terms of a subphonemic feature scheme, so as to quantify the degree of the individual errors by means of the feature distance between the error sound and the target phoneme. We used a four-feature scheme of place, manner, voicing, and oral-nasal, similar to the one used by Trost and Canter (1974). There were six place categories: bilabial, dental, alveolar, palatal, velar, and glottal. Manner categories included stop, affricate, fricative, Rap, and semivowel. Voicing was either voiced or voiceless. Oral-nasal was either oral or nasal.

RESULTS 1. Relative

Frequency

of Errors on Consonants

and Vowels

Figure 2 is a graphical representation of the percentage of errors for each patient in terms of whether they involved consonants, vowels or other sounds. (See Appendix 1 for the relative number of errors on each category .) It can be seen that the conduction aphasic patients (i.e., Cases 1, 2, and 3) made roughly the same number of errors on both consonants and ’ Errors on other than consonants or vowels were omitted from further analysis since the number of these errors was negligibly small.

182

MONO1 ET AL.

NMNG 1 REPETI

NAMING

2 REPETI

NANING

3 REPETI

NAMING

4 REPETI

NAMING

5 REPETI

6

NAMING REPETI

CONSONANTS VOWELS

OTHERS

FIG. 2. Relative percentage of errors on consonants, vowels, and other segments in the naming and repetition tasks for each of the three conduction (Cases 1, 2, and 3) and three Broca’s (Cases 4, 5, and 6) aphasic patients.

vowels on the naming task as well as on the repetition task. The mean percentages of errors involving consonants and vowels on the naming task were 53.6 and 42.6%, respectively, and those on the repetition task were 56.7 and 42.3%, respectively. This indicates that consonants and vowels were almost equally susceptible to error for our patients with conduction aphasia. Only a negligible number of errors involved other than consonants or vowels. The Broca’s patients (i.e., Cases 4, 5, and 6), on the other hand, made most of their errors on consonants. That is, they produced a mean of 92.1% consonant error on the naming task and 97.3% on the repetition task, with only a small percentage of errors on vowels and other sounds, indicating that for this group of patients consonants were selectively susceptible to error.

183

SPEECH SOUND ERRORS IN APHASIA 50

II I

LASE Lxx

I

100 I

NAMING 1 REPETI

NAMING

2 REPETI

NAMING

3 REPETI

NAMING

4 REPETI

NAMING

5 REPETI

I@

NAMING

w

REPETI

I@

6

q SIJEST.

TRANS,

ADD.

OMI.

FIG. 3.

Distribution of the consonant error types (substitution, transposition, addition, and omission) in the naming and repetition tasks for each of the three conduction (Cases 1, 2, and 3) and three Broca’s (Cases 4, 5, and 6) aphasic patients.

2. Distribution of Error Types Figure 3 shows the distribution of the errors in percentageson consonants among the four error types: substitution, transposition, addition, and omission. For the conduction aphasic patients, substitution and transposition errors constituted the two major error types on both the naming and repetition tasks-accounting for 42.9 and 46.6% of the errors on the naming task, and 46.1 and 41.1% on the repetition task. (See Appendix 2 for the relative number of errors on each type.) For the patients with Broca’s aphasia, on the other hand, the most frequently occurring errors were those of substitution on both the naming and repetition tasks-accounting for 65.8 and 74.8% of all the errors, respectively. The rest of the errors were transposition and other types of errors, roughly in equal proportions.

184

MONO1 ET AL.

NAAING 1 REPETI

NANING

2 REPETI

NANING

3 REPETI

SUBST.

TRANS,

ADD,

OMI,

FIG. 4. Distribution of the vowel error types (substitution, transposition, addition, and omission) in the naming and repetition tasks for each of the three conduction aphasic patients.

Figure 4 shows the distribution of the vowel error types in percentage for the patients with conduction aphasia, indicating that the incidence of transposition errors was somewhat higher than that of substitution errors. (See Appendix 3 for the relative number of errors on each type.) No analysis of the vowel error types was made for the patients with Broca’s aphasia because of their low incidence of errors involving vowels. 3. Subphonemic Feature Analysis Figure 5 illustrates the results of the subphonemic feature analysis, plotted separately for the naming and repetition tasks for the conduction and Broca’s aphasic patients. The abscissa represents the number of subphonemic features separating the error sound from the target phonemes. The ordinate represents the percentage of these error responses. The dotted line in each figure indicates an approximation of the natural chance distribution.’ It can be seen that there was a difference in the distribution patterns between the conduction and the Broca’s aphasic patients. The distributions obtained from the Broca’s aphasic patients, both on the naming and repetition tasks, markedly deviated from the natural chance ’ This is an approximation of the natural chance distribution in the sense that it represents the frequency of occurrence of one-, two-, three-, and four-feature substitutions between all the possible pairs of Japanese consonants when the relative frequency of occurrence of these consonants in daily speech is assumed to be equal.

185

SPEECH SOUND ERRORS IN APHASIA

12 S~BPHONEHIC

12

3

4

FEATURE DISTANCE

3

4

SUBPHONEMIC FEATURE DISTANCE

12

3

4

SUBPHONEMIC FEATURE DISTANCE

12

3

4

SUBPHONEMIC FEATURE DISTANCE

FIG. 5. Relative percentage of one-, two-, three-, and four-feature errors in the naming and repetition tasks for each of the three conduction (Cases 1, 2, and 3) and three Broca’s (Cases 4, 5, and 6) aphasic patients.

distribution-i.e., substitution errors occurring between phonemes separated by a single feature were by far the greatest in number on both the naming repetition tasks. In contrast, the distributions obtained from the patients with conduction aphasia on the naming task deviated less from the chance distribution.

186

MONO1

ET AL.

On the repetition task, the frequency of one-feature errors for these subjects clearly deviated from chance distribution. 4. Naming vs. Repetition Tasks As has been reported above, the two types of tasks, i.e., the naming and repetition tasks, had no differential effect on the performance of our aphasic subjects, except that the patients with conduction aphasia showed slightly more one-feature errors on the repetition task than on the naming task. One possible interpretation of this latter finding is that the presentation of auditory stimuli in the repetition task might have aided the patients’ selection and/or sequencing of appropriate phonemes to a certain extent, resulting in an increased approximation to the target phonemes. The exact nature of this process, however, remains to be explored in further research. DISCUSSION There were three major findings in this study. First, the conduction aphasic patients exhibited almost as many errors on vowels as on consonants, whereas the majority of errors exhibited by the Broca’s aphasic patients were on consonants. Second, substitution errors and transposition errors constituted the two major error types for the conduction aphasic patients, while for the Broca’s aphasic patients substitution errors accounted for the majority of errors. Third, an analysis of the substitution errors on consonants in terms of a subphonemic feature scheme disclosed that in the Broca’s aphasic patients a high proportion of single-feature errors was exhibited on both the naming and repetition tasks. For the conduction aphasic patients, the single-feature errors on the naming task were somewhat less, although in the repetition task the proportion of single-feature errors increased considerably. Taken together, these findings indicate some differences rather than similarities in speech sound production performance between conduction and Broca’s aphasic patients, which is in conflict with previous findings of similarities rather than differences in performance among various aphasic syndromes (e.g., Blumstein, 1973). There may be some reasons to account for the differences. The first reason which immediately suggest itself is the nature of the speech samples used. In Blumstein’s study, for instance, the speech material used for analysis was spontaneous speech obtained by having the patients engage in an open-ended conversation, whereas in the present study the speech material for analysis consisted of isolated words of three to five moras obtained by having the patients engage in the tasks of naming and repeating these words. According to our clinical impression, the degree to which target-oriented behavior is required of patients with conduction aphasia

SPEECH SOUND ERRORS IN APHASIA

187

may significantly affect their performance. For instance, if a conduction aphasic patient makes a sound-production error on a given word in his conversational speech, it is very likely that he will substitute a less difficult word for the target, and just go on talking, rather than trying to correct the error with repeated approximations-as our conduction aphasic patients did on the naming and repetition tasks. Moreover, since the examiner may not always know the exact target that the patient is trying to get at in conversational speech, error words that contain multiple sound errors are likely to be taken as jargon (or something else) and thus be deleted from the speech sample. The second, and perhaps more crucial, reason may have to do with the classification scheme for the error types. There are some differences between Blumstein’s classification scheme and ours. Table 3 shows two analysis schemes, one used by Blumstein (1973) and the other used in the present study, but rearranged in such a way that a comparison can be made between the two schemes. It can be seen that a relatively good correspondence exists between Blumstein’s scheme and ours for the error types A, B-l, C, and D; However, for the error type B-2 there are some important differences. The first two lines in Table 4 are examples of “syllable simplification” quoted from Blumstein. The third line shows what we think is an equivalent type of error exhibited by one of our conduction patients. That is, the underlined responses /kusa/ and /sakusa/ for /asakusa/ seem to correspond to /pEnslv/ for /EkspEnsIv/, and /ploiya/ for /Emploiya/. If we follow Blumstein’s system, /kusa/ and /sakusa/ should be classified as instances of simplification because the moras /asa/ and /a/ are omitted from the word /asakusa/. According to our classification scheme, however, each of these represents an error of sequencing, or of an anticipatory transposition of phonemes, in which a later appearing phoneme is transposed to the position of a phoneme appearing earlier in the target word. In our conduction aphasic patients, errors of this type were as prevalent as substitution errors (comprising almost half of the total errors), while

TABLE 3 THE ERROR CLASSIFICATION SCHEMES USED IN BLUMSTEIN’S (1973) STUDY AND IN THE PRESENT STUDY

A B-l B-2 C D

Blumstein’s study

Present study

Phoneme substitution Phoneme simplification Syllable simplification Addition Environment

Substitution Omission Transposition Addition Transposition

188

MONO1 ET AL. TABLE 4

EXAMPLES

1 2 3

OF “SYLLABLE

/EkspEnsIv/ /Empbiya/ IAsakusal

SIMPLIFICATION” (BLUMSTEIN, (PRESENT STUDY)

1973) AND

+ (expensive) (employer) (The name of a well-zown +

OF “TRANSPOSITION”

/pEnsIv/ /plDiya/ area in Tokyo) lkusa, sakusa, asakusal

in our Broca’s aphasic patients, transposition errors occupied only a minor portion of the total errors. As for substitution errors, our definition seems to be roughly equivalent to Blumstein’s “phoneme substitutions.” Indeed the results of our subphonemic feature analysis of the substitution errors for the Broca’s aphasic patients did agree with her results, in that the frequency of occurrence for one-feature errors was highest, although a direct comparison may not be feasible here because of the differences in the feature analysis systems used in Blumstein’s study and ours. The results obtained for our conduction aphasic patients, on the other hand, were again somewhat different from the results obtained for our Broca’s aphasic patients, in that the proportion of single-feature errors exhibited by our conduction aphasic patients was moderately depressed as compared to our Broca’s aphasic patients (Fig. 5). This slight depression of single-feature errors, although not very marked (for Cases 1 and 3), seems to be a rather consistent phenomenon in that we have observed it repeatedly in our previous cases of fluent aphasia whose speech sound errors have been studied using a similar feature analysis system (six patients thus far). The finding of this difference in performance between the conduction and the Broca’s aphasic patients is again in conflict with the findings of Blumstein (1973), but it is in agreement with the results of Burns and Canter (1977). These authors found that for their posterior aphasic patients (both conduction and Wernicke’s aphasic patients) approximately 60% of the subphonemic errors involved two or more components, while only 40% of the errors involved one component alone. As Fig. 5 indicates, the proportion of the multiple-feature to single-feature errors for our conduction aphasic patients was approximately 60 to 40% on the naming task and 50% each on the repetition task. How do we account for this difference in feature distribution between conduction and Broca’s aphasic patients? Our belief is that there are differences in the nature of the underlying deficits in the two types of aphasia. The following is an attempt at interpreting our present findings in terms of the functional loci of the impairments of these types of aphasic patients in a speech production model (Fig. 1).

SPEECH

SOUND

ERRORS

IN APHASIA

189

Implications of the Findings for a Speech Production Model A. Errors on consonants vs. vowels. The findings that our patients with conduction aphasia exhibited almost as many errors on vowels as on consonants is in sharp contrast to the findings reported for fluent aphasic speakers of Indo-European languages. Blumstein (1973) and Burns and Canter (1977), for example, reported that their fluent aphasic patients showed only a small proportion of errors on vowels as compared to single consonants and consonant clusters. Since Japanese phonological structure is quite different from Indo-European languages in that the majority of moras are of the CV type, it might be more reasonable to expect that the probability of errors occurring for vowels would be equal that for consonants at the level of phonological processing. The present findings, therefore, can be interpreted as being compatible with our contention that the disturbance of speech sound production exhibited by our conduction aphasic patients is located at the level of phonological processing. In our three patients with Broca’s aphasia, on the other hand, a predominantly greater number of errors did occur for consonants rather than for vowels as has been almost unanimously reported for other Broca’s patients in the literature. This indicates that in Broca’s aphasia the impairment at the level of articulatory programming tends to affect the more complicated articulatory gestures of consonants to a greater extent than the less complicated articulatory gestures of vowels. Thus, the locus responsible for the impairment of speech sound production exhibited by these patients is probably not limited to the level of phonological processing but rather extends to the level of speech production where programming for articulatory movement is carried out. B. Substitution vs. transposition errors. We have seen that a greater proportion of transposition errors were exhibited by our patients with conduction aphasia than by our patients with Broca’s aphasia. Since transposition errors reflect a deficit in the proper sequencing of phonemes, this finding seems to be another support for the contention that the major impairment of patients with conduction aphasia is at the level of phonological processing. Substitution errors, on the other hand, were exhibited by both conduction and Broca’s patients, and the functional loci of the occurrence of these errors can be traced to at least two different levels of phoneme production: Level 2-C in Fig. 1 for phoneme selection and sequencing, and Level 3 for the motor programming of articulatory gestures. We feel that the substitution errors exhibited by our conduction aphasic patients reflect an impairment which is more or less confined to Level 2-C, whereas the substitution errors exhibited by our Broca’s aphasic patients reflect an impairment which is not confined to Level 2-C but extends to Level 3. The basis for this inference will be discussed below.

190

MONO1

ET AL.

C. Subphonemic feature distance. Another difference found between our conduction and Broca’s aphasic patients had to do with the results of the subphonemic feature analysis of their substitution errors. Compared to the conduction aphasic patients, the Broca’s aphasic patients exhibited a markedly higher number of “substitution” errors occurring between phonemes separated by a single feature on the naming task. Referring to the speech production scheme again, this very high proportion of onefeature errors in our Broca’s aphasic patients may be accounted for by postulating that they retained a substantial ability to select and sequence target phonemes properly, but that (due to an impairment at the level of articulatory programming) they tended to produce deviation (or phonetic) errors. Nevertheless, these deviation errors were often perceived as and classified as single-feature “substitution” errors, because the errors were not allophones of the target phonemes in Japanese but were allophones of other neighboring phonemes. The so-called “substitution” errors exhibited by Broca’s patients, therefore, can in fact, be of two (or even three) types with different origins: (I) those taking place at the level of phoneme selection, or “true” substitution errors; (2) those taking place at the level of articulatory programming, and thus, in fact, phonetic distortions which tend to be perceptually classified as one-feature substitutions; and perhaps (3) those taking place at both of these levels which tend to be classified as multiple-feature errors. The second of errors-those which are, in fact, these three types of “substitution” “phonetic” in nature-appears to constitute an important factor contributing to the markedly high number of single-feature errors in the Broca’s patients as shown in Fig. 5. The phonetic nature of the speech sound errors produced by patients with frontal lesions has been verified by a series of experimental studies in recent years investigating such parameters as voice onset time in the speech production of fluent and nonfluent aphasic patients (Blumstein, Cooper, Zurif, & Caramazza, 1977; Itoh, Sasanuma, Tatsumi, Murakami, Fukusako, & Suzuki, 1982), velar movements by means of fiberscopic observation, and the temporal organization of different articulators by X-ray microbeam observation in a patient with “apraxia of speech” (Itoh et al., 1979; Itoh, Sasanuma, Hirose, Yoshioka, & Ushijima, 1980). The error pattern exhibited by conduction aphasic patients, in contrast, may indicate an impairment which is relatively confined to the level of phonological processing with few, if any, errors of a phonetic nature, such as those exhibited by Broca’s aphasic patients. In other words, it is quite likely that the major defect in conduction aphasic patients has to do with “a very early stage” of speech production processing (Dubois et al., 1973; Yamadori & Ikemura, 1975). In conclusion, on the basis of the findings in the present study, it is hypothesized that the differences in error patterns between conduction

191

SPEECH SOUND ERRORS IN APHASIA

and Broca’s aphasia reflect real differences in the underlying mechanisms of the impairment in each type. The error pattern in the conduction aphasic patients reflects an impairment at the level of phonological processing (i.e., phoneme selection and sequencing); while the error pattern in Broca’s aphasic patients reflects an impairment at the phonetic level of the speech production process (i.e., programming of articulatory movement) in addition to an impairment at the phonological level. APPENDIX RELATIVE

Case

NUMBER

1

OF ERRORS ON CONSONANTS, VOWELS, AND OTHER SEGMENTS IN THE NAMING AND REPETITION TASKS FOR EACH SUBJECT

Naming -. Consonants Vowels

Conduction I 2 3 Total

140(50.4)” 127(45.7) 98(51.3) 84(44.0) 140(59.3) 89(37.7) 378(53.6) 300(42.6)

Broca’s I 2 3 Total

96(96.0) 73(93.6) 65(85.5) 234(92.I)

Repetition Others Total Consonants Vowels I l(4.0) 9(4.7) 7(3.0) 27(3.8)

278 I91 236 705

4(4.0) O(O.0) 100 5(6.4) O(O.0) 78 9(l I .8) 2(2.6) 76 18(7.1) 2(0.8) 254

fl Percentage of errors in parentheses

Others Total

160(57.3) I lS(42.3) l(O.4) 137(53.1) 117(45.3) 4(1.6) 85(62.0) 50(36.5) 2(1.5) 382(56.7) 285(42.3) 7( I .O) l14( 100.0) 70(98.6) 66(91.7) 250(97.3)

O(O.0) l(l.4) 6(8.3) 7(2.7)

279 258 137 674

O(O.0) 114 O(O.0) 71 O(O.0) 72 O(O.0) 257

O(O.0) l(l.5) l(O.4)

16(21.9) ll(16.9) 44(18.8)

40(54.8) 48(73.8) 154(65.8)

2 3 Total

49(38.6) 29(34.5) 36VO.4) 114(38.0)

78(61.4) 55(65.5) 52(58.4) 185(61.7)

” Percentage of errors in parentheses.

Case 1 2 3 Total

Transposition

Substitution

O(O.0) O(O.0) o(O.0) O(O.0)

Addition O(O.0) O(O.0) l(l.1) l(O.3)

Omission

ERROR TYPES IN THE NAMING

Naming

OF THE VOWEL

DISTRIBUTION

2

3

127 84 89 300

Total

41(34.7) 51(43.6) 18(36.0) l10(38.6)

Substitution

21(18.4) 6(8.6) 1l(16.7) 38(15.2)

52(32.5) 70(51.1) 35(41.2) 157(41.1)

Transposition

APHASIC

W7.0) l(l.4) 2(3.0) 1l(4.4)

14(8.8) 4(2.9) 7(8.2) 25(6.5)

Addition

l(O.8) O(O.0) O(O.0) l(O.4)

Addition

Repetition Transposition 74(62.7) 66(56.4) 32(64.0) 172(60.4)

SUBJECT

Repetition

TASKS FOR EACH

TASKS FOR EACH CONDUCTION

79(69.3) 55(78.6) 53(80.3) 187(74.8)

84(52.5) 56(40.9) 36(42.4) 176(46.1)

Substitution

AND REPETITION

AND REPETITION

APPENDIX

73 65 234

17(23.3) 5(7.7) 35(15.0)

” Percentage of errors in parentheses.

96

13(13.5)

O(O.0)

140 98 140 378

Total

17(17.7)

3(2.1) g(8.2) 3(2.1) 14(3.7)

Omission

66(68.8)

8(5.7) 8(8.2) lO(7.1) 26(6.9)

Addition

Broca’s 1

76(54.3) 47(48.0) 53(37.9) 176(46.6)

APPENDIX ERROR TYPES IN THE NAMING

53(37.9) 35(35.7) 74(52.9) 162(42.9)

Transposition

Naming

OF THE CONSONANT

Conduction 1 2 3 Total

Substitution

DISTRIBUTION

2(1.7) O(O.0) O(O.0) 2(0.7)

Omission

SUBJECT

6(5.3) 8(11.4) O(O.0) 14(5.6)

10(6.3) 7(5.1) 7(8.2) 24(6.3)

Omission

118 117 50 285

Total

114 70 66 250

160 137 85 382

Total

3

5

i-2

SPEECH SOUND ERRORS IN APHASIA

193

REFERENCES Blumstein, S. E. 1973.A phonological investigation of aphasic speech. The Hague: Mouton. Blumstein, S. E., Cooper, W. E., Zurif, E. B., & Caramazza, A. 1977. The perception and production of voice-onset time in aphasia. Neuropsychologia, 15, 371-383. Brown, J. W. 1975. The problem of repetition: A study of “conduction” aphasia and the “isolation” syndrome. Correx, 11, 37-52. Burns, M. S., & Canter, G. J. 1977. Phonemic behavior of aphasic patients with posterior cerebral lesions. Bruin and Language, 4, 492-507. Darley, F. L., Aronson, A. E., & Brown, J. R. 1975.Motor Speech Disorders. Philadelphia: Saunders. Dubois, J., Hecaen, H., Angelergues, R., de Chatelier, A. M., & Marcie, P. 1973. Neurolinguistic study of conduction aphasia. In H. Goodglass, & S. Blumstein (Eds.), Psycholinguistics and uphasia. Baltimore: Johns Hopkins Press. Dunlop, J. M., & Marquardt, T. P. 1977. Linguistic and articulatory aspects of single word production in apraxia of speech. Cortex, 13, 17-29. Goodglass, H., & Kaplan, E. 1972. The assessment of aphasia and related disorders. Philadelphia: Lea & Febiger. Green, E. 1969. Phonological and grammatical aspects of jargon in an aphasic patient: A case study. Language and Speech, 12, 103-118. Guyard, H., Sabouraud, O., & Gagnepain, J. 1981.A procedure to differentiate phonological disturbances in Broca’s aphasia and Wernicke’s aphasia. Bruin and Language, 13, 19-30. Itoh, M., Sasanuma, S., & Ushijima, T. 1979. Velar movements during speech in a patient with apraxia of speech. Brain and Language, 7, 227-239. Itoh, M., Sasanuma, S., Hirose, H., Yoshioka, H., & Ushijima, T. 1980.Abnormal articulatory dynamics in a patient with apraxia of speech: X-ray microbeam observation. Bruin and Language,

11, 66-75.

Itoh, M., Sasanuma, S., Tatsumi, I. F., Murakami, S., Fukusako. Y., & Suzuki, T. 1982. Voice onset time characteristics in apraxia of speech. Bruin and Language, 17, 193210. Johns, D. F., & Dailey, F. L. 1970. Phonemic variability in apraxia of speech. Journal of Speech and Hearing

Research,

13, 556-583.

Keller, E. 1978. Parameters for vowel substitutions in Broca’s aphasia. Bruin and Languuge, 5, 265-285. La Pointe, L. L., &Johns, D. F. 1975. Some phonemic characteristics in apraxia of speech. Journal

of Communication

Disorders,

8, 259-269.

Lecours, A. E., & Lhermitte, F. 1969. Phonemic praraphasias: Linguistic structures and tentative hypotheses. Cortex, 5, 193-228. Lecours, A. R., & Rouillon, F. 1976. Neurolinguistic analysis of jargonaphasia and jargonagraphia. In H. Whitaker, & H. A. Whitaker (Eds.), Studies in neurolinguistics. New York: Academic Press. Vol. 2, pp. 95-144. Martin, A. D. 1974. Some objections to the term Apruxia of Speech. Journul of Speech and Hearing

Disorders,

39, 53-64.

Martin, A. D., & Rigrodsky, S. 1974. An investigation of phonological impairment in aphasia, Part 1. Cortex, 10, 317-328. (a) Martin, A. D., & Rigrodsky, S. 1974. An investigation of phonological impairment in aphasia, Part 2: Distinctive feature analysis of phonemic commutation errors in aphasia. Cortex, 10, 329-346. (b) Sands, E. S., Freeman, F. J., & Harris, K. S. 1978. Progressive changes in articulatory patterns in verbal apraxia: A longitudinal case study. Brain and Language, 6, 97-105. Sasanuma, S. 1971. Speech characteristics of patient with apraxia of speech: A preliminary case report. Annual Bulletin. Research Institute of Logopedics and Phoniatrics. 5, 85-89.

194

MONO1 ET AL.

Shankweirler, D., & Harris, K. S. 1966. An experimental approach to the problem of articulation in aphasia. Cortex, 2, 277-292. Trost, J. E., & Canter, G. J. 1974. Apraxia of speech in patients with Broca’s aphasia: A study of phoneme production accuracy and error patterns. Bruin und Language, 1, 63-79.

Yamadori, A., & Ikemura, G. 1975. Central (or conduction) aphasia in a Japanese patient. Cortex,

11, 73-82.