JOURNAL
OF COMMUNICATION
DISORDERS
8 (1975), 259-269
SOME PHONEMIC CHARACTERISTICS IN APRAXIA OF SPEECH LEONARD Audiology
and Speech Pathology,
L. LA POINTE
Veterans Administration
Hospital,
Gainesville,
Florida 32602
and DONNELL Southwestern
Medical
School,
F. JOHNS
lJniver&
of Texas, Dallas,
Texas 75235
The atticulatory performance of 13 left hemisphere-damaged adults who presented apraxia of speech was tabulated on confusion matrices and analyzed according to error pattern. Consonants were more susceptible to error than were vowels, as were consonant clusters when compared to single consonants. No significant differences existed among error percentages for the intial, medial, and final positions. When errors were analyzed according to manner of production, affricatives and fricatives were significantly more susceptible to error than all others. Analysis of errors according to place of production revealed lingua alveolar and bilabial phonemes to be significantly less impaired than all other categories. No differences were found in error percentages of voiced and unvoiced phonemes. The sequential nature of substitution errors was further analyzed by tallying and classifying errors as anticipatory (prepositioning), reiterative (postpositioning), or metathesis. Seven percent of the substitution errors in this study were sequential, with anticipatory errors outnumbering reiterative errors by a ratio of 6 to 1. Feature analysis of substitutions to determine distance from the target sound revealed that 38% of the substitutions were defective in two or more features. Some of these subjectively bore little resemblance to the target sound.
Introduction The subject of sensorimotor speech disorders associated with cortical lesions has proved to be a fertile area for controversy for more than a century. Theoretical and interpretive arguments have abounded, resulting in biased and inconsistent terminology which, for the most part, has reflected preconceived theoretical assumptions. The term ‘ ‘apraxia of speech” appears to be increasing in frequency of usage in recent literature. Some writers have objected to this term (Martin, 1974), but others have attempted to construct a rationale for its use. Johns ( 1968) related that: The term apraxia of speech, focuses on the underlying dynamics of the disorder. Apraxia of speech explicitly (1) directs one’s attention to the motor aspects of speech, (2) emphasizes the volitional execution of articulation, (3) excludes significant weakness, paralysis, and incoordination of the speech musculamre, and (4) indicates a discrepancy between execution of the speech act and relative linguistic intactness. D American
Elsevier Publishing
Company,
Inc.,
1975
259
260
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
Patients with apraxia of speech have been described as displaying a distinctive articulatory disorder, which is characterized by variable off-target approximations of phonemes despite absence of significant weakness, slowness, or incoordination of the speech musculature. As they speak, they struggle to position their articulators correctly; they grope to assume correct articulatory postures and to accomplish the sequence of postures in word production. They are frequently off-target, will often recognize that they are off-target, and will (with great effort) try again. Errors recur, but these errors are not always the same. Errors on a series of trials will be variable. In their struggle to avoid error, they will slow down, space their words and syllables evenly, and stress them equally, subsequently affecting the prosody of their speech (Johns and Darley, 1970). Renewed effort on the part of speech pathologists in the study of neurologically based communication disorders is now evident. Contemporary researchers are attempting (a) to outline the characteristics of distinctive disorders and classify them on some homogeneous basis, (b) to develop tests and instrumentation of differential diagnosis, and (c) to study these disorders in order to better understand language theories and neural processes involved in language function. Recent work at the Mayo Clinic and elsewhere contributed to our understanding of clusters of deviant speech dimensions in the dysarthrias and our understanding of some of the differentiating characteristics of apraxia of speech (Darley, Aronson, and Brown, 1969; Deal and Darley, 1972; Rosenbek, Wertz, and Darley, 1973). Before clear differentiation and subsequently more efficient remediation can be accomplished, however, characteristic patterns of behavior must be analyzed and described systematically. The purpose of this study was to investigate and describe specific phonemic characteristics in 13 subjects with apraxia of speech. Methods
and Procedures
Thirteen adults who sustained cortical damage and presented certain articulatory characteristics indicative of apraxia of speech were studied. These Ss were given a measure of speech and language deficit (Minnesota Test for Differential Diagnosis of Aphasia), a picture test of articulation (a 141-item modified version of the Templin-Darley Test), and two measures of non-speech oral movement. The Templin-Darley Test was modified by eliminating several plates related to phonemes previously elicited. No limitations were placed on sex, etiology, or hemispheric location of lesion, and allSs had to be at least 16 years of age. Only Ss who were at least one month postonset were selected, and individuals with severe auditory input deficit were excluded (55% or more errors on the auditory section of the MTDDA). The cause of brain injury, duration of involvement, and severity of language involvement as represented by total number of errors on the MTDDA are given in Table 1. Criteria used to establish the presence of apraxia of speech were the following (La Pointe, 1969):
SOME
PHONEMIC
Etiology,
CHARACTERISTICS
IN APRAXIA
TABLE 1 Months Postonset, and MTDDA
Subject
Etiology
1 2 3 4 5 6 7 8 9 10 11 12 13
CVA CVA CVA CVA CVA Trauma CVA Trauma CVA Trauma Trauma CVA Trauma
Mean Range ‘579 possible errors on modified
Months postonset 26 35 29 23 4 3 11 4 I1 6
1 3 2 12.15 l-35
OF SPEECH
261
Scores for Ss MTDDA No. of error@ 107 65 192 297 355 209 280 183 289 126 81 147 100 187 65-297
MTDDA.
(1) The presence of initiation difficulties (defined as repetitions of initial sound or syllable or retrials using different sound combinations on each attempt). (2) The presence of islands of error-free production (on serial speech tasks or spontaneous utterances). (3) More substitution errors than distortions and omissions combined. Articulation testing involved recording subject responses to all 141 stimulus cards on magnetic tape, then transcribing them phonetically. Phonetic transcription allowed errors to be tabulated on confusion matrices similar to those used by Shankweiler and Harris (1966) and Fry (1959).
Results and Discussion Analysis of the errors plotted on confusion matrices revealed several patterns of articulatory performance. Table 2 illustrates that significantly more errors were made on consonants than vowels (P < 0.001) and significantly more errors were made on consonant clusters than on single consonants (P =C0.00 1) as determined by a Wilcoxon Signed Rank Test. The percentages for each subject on Table 2 were determined by dividing the number of errors for each category (vowels, consonants, clusters, singles) by the number of occurrences of each of these within the articulatory test material. For example, subject 3 made errors on 15% of all vowels that occurred in the articulatory test material. Further, he made errors on 25% of all consonants tested,
262
Percentages Subject
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
TABLE 2 of Vowel vs. Consonant and Cluster vs. Single Errors for 13 Apraxia of Speech Subjects Vowels 0
1 2 3 4 5 6 1 8 9 10 11 12 13
15 12 3 26 25 2 71 5 7 4 17
Range Mean
o-71% 14%
0
Consonants 2 10 25 40 10 59 30 2 88 9 24 6 35
o-8896 26%
Clusters 6 27 35 70 20 87 34 7 96 18 48 11 55 7%-96% 40%
Singles
4 21 34 6 50 29 49 6 15 4 29 l%-50% 19%
Wilcoxon Signed Rank Test Vowels vs. consonants significant P = < 0.001 Clusters vs. singles significant P = < 0.001
35% of all clusters, and 21% of all single consonants. The data in Table 2 support the findings of Trost ( 1969) and Shankweiler and Harris ( 1966), and lend evidence to Darley’s (1969) contention that errors vary with the complexity of the articulatory adjustment. Analysis of the position of errors has resulted in different interpretations. Some authors (Trost, 1969; Shankweiler and Harris, 1966) have suggested that initial consonants are more susceptible to error than final consonants. As can be seen in Table 3, error percentages for initial, medial, and final positions in this sample were nearly equal, with no significant differences existing among them. This supports the finding of Johns and Darley ( 1970) that “No single position in the word emerged as characteristically more difficult . . .” Lack of agreement on this issue may be partly definitional, however, depending on whether retrials and repetitions (the difficulty in getting started) are counted as a separate type of error or are added to those errors made on the initial phoneme of words. In this study, retrials and repetitions were catalogued separately from errors made on the initial phoneme. Another method of viewing patterns of error is relative to phoneme type. Errors in manner of production and in place of production are presented in Tables 4 and 5. The data in both of these tables were analyzed by multivariate analysis of variance (Morrison, 1967). In Table 4 each subject has a vector of five error percentages, while in Table 5 each subject has a vector of seven error percentages. The overall
SOME
PHONEMIC
Percentage
CHARACTERISTICS
of Position
IN APRAXIA
TABLE 3 Errors in 13 Apraxia
263
OF SPEECH
of Speech Subjects
Subject
Position Initial
Medial
Final
1
1
2 3 4 5 6 7 8 9 10 11 12 13
1.5 21 57 16 54 29 2 82 9 25 6 37
3 4 34 31 9 56 18 3 87 10 21 9 40
2 7 26 35 2 67 39 2 94 8 24 4 31
Range Mean
I%-82% 27%
3%-87% 25%
2%-94% 26%
Friedman two-way analysis of variance Not significant P = > 0.05.
by ranks
Errors According
to Phoneme
Error percentage
Type
1. Affricatives 2. 3. 4. 5.
TABLE 4 Type in Apraxia
33 24 17 16 16
Fricatives Glides Nasals Plosives.
of Speech (Manner)
Paired comparisons l-2, l-3”, I-4a, l-5a 2-3O, 2-4”, 2-5a 3-4, 3-5 4-5
Overall F statistic (4 and 9 u” = 4.lOZn OSignificant
at
P < 0.05.
F test compared the equality of the population mean percentage error rates in the five and seven respective categories. Pairwise comparisons of means were also made. The five manner categories used in this study are shown in Table 4, along with the relative error percentage for each category, and the results of analysis of paired comparisons. JZrror percentages reflect the number of errors relative to the total occurrence of each phoneme type. For example, in Table 4, errors were made on 33% of the affricatives that occurred in the articulatory test material. The overall F statistic indicates that there were significant differences among manner categories.
264
LEONARD
Errors According
and DONNELL
F. JOHNS
TABLE 5 to Phoneme Type in Apraxia of Speech (Place) Error percentage
Type 1. 2. 3. 4. 5. 6. 7.
L. LAPOINTE
Linguapalatal Linguadental Linguavelar Labiodental Glottal Lingua alveolar Bilabial
31 27 23 20 17 16 13
Paired comparisons 1-2, l-3, 1-4, l-5, l-60, l-7” 2-3, 2-4, 2-5, 2-6, 2-7n 3-4, 3-5, 3-6, 3-7” 4-5, 4-6, 4-7n 5-6, 5-7 6-7
Overall F statistic (6 and 7 dJ) = 9.65 la ‘Significant
at P < 0.05.
Affricative and fricative phonemes were significantly more susceptible to error than all others. Results of paired comparisons reveal that when error percentages for affiicatives (1) or fricatives (2) were compared with any other manner category, statistically significant differences emerged. These findings support the trend suggested by Trost (1969) and by Shankweiler and Harris (1966). Similarly, analysis of errors according to place of production reveals significant differences among categories, as can be seen in Table 5. Again, error percentages reflect the number of errors relative to the total occurrence of each phoneme type. For example, errors were made on 13% of the bilabial phonemes that occurred in the articulatory test material. The overall F statistic indicates significant differences among place categories. Lingua alveolar and bilabial phonemes were the least impaired in this sample, and paired comparison analysis revealed significant differences in error percentages when these categories were compared with several of the categories more susceptible to error. These findings also offer support to previous suggestions in the literature (Trost, 1969). Analysis of errors in voicing revealed no significant differences between error percentages of voiced and unvoiced sounds. Sequential
Errors
Further analysis of the phonemic production of theseSs was concerned with the sequential nature of articulation errors. Lashley (1951), in his work on serial order, postulated that the realization of linguistic sequences requires the interplay of at least three relatively independent neurophysiological mechanisms: tendency” or “idea.” (1) The “determining or “arousal of expressive units.” (2) A mechanism of “priming” (3) A mechanism having to deal specifically with serial ordering of preactivated units.
SOME
PHONEMIC
CHARACTERISTICS
IN APRAXIA
OF SPEECH
265
Lecours and Lhermitte (1969) reviewed some of Lashley’s theories and suggested that many of the phonemic errors made by brain-damaged patients are sequential in nature. Some researchers (Darley, 1969; Johns and Darley, 1970) suggested that these sequential errors are “anticipatory” or ‘ ‘perseverative. ” In the present study, a tally was made of sequential errors, and examples of them are presented in Table 6. Sequential difficulty is only apparent by analyzing two categories of subject error, those of phoneme substitutions and those of initiation errors (difficulty in getting started). When initiation errors and substitution errors are analyzed relative to the target word, three types of sequential error become evident: “anticipatory,” “reiterative,” and “metathesis.” In substitution errors, if a phoneme were replaced by one that occurred later in the word, such as the example for “yellow,” it was tallied as an anticipatory or prepositioning error. Similarly, if in attempting to get started (initiation difficulty), the subject used a phoneme that occurred later in the stimulus word, the error was counted as anticipatory (for example, in the response to “bicycle,” It, = bai = bai = s = s = sai = s = a = bai = sikl]; here, the use of s = s = sai was tallied as an anticipatory error, even though the word was eventually produced correctly). If a phoneme were replaced by one that occurred earlier in the word, the error was called one of “postpositioning” or “reiterative,” as in the responses to “dress” and “December” in Table 6. If two phonemes switched places, as in “tefalon” for “telephone,” the error was also regarded as being sequential in nature and labeled a “metathesis.” The percentage of substitution and initiation errors of each subject, which were sequential in nature, can be seen in Table 7. The percentages
TABLE 6 Sequential Error Examples Stimulus
word
Subject response
Pre-positioning (anticipatory) Yellow Grasshopper Bicycle
“redul, ledul, (pause) 1190” “grap-popper, let’s see, - grass-hopper” “b-bai-bai-s-s-sai-s“uh - bai - ski”
Post-positioning (reiterative) Dress December
“dred” “ees, ees, esender”
Metathesis Telephone Sandwich
“teflone “wansin
- tefalone” - sanwich”
266
LEONARD
L. LAPOINTE
Percentage of Initiation and Substitution
Subject
2 3 4 5 6
8 9 10 11 12 13 Totals Range
and DONNELL
F. JOHNS
TABLE 7 Errors That Are Sequential in 13 Apraxia of Speech Subjects
Init. and sub. errors
Sequential 4
40 35 156 229 110 240 139 33 263 80 71 75 173
3 2 16 3 7 10
1644 33-263
95 1-16
12 13 12
errors
% Sequential errors 0.10
0.03 0.13 0.05 0.06 0.05 0.04 0.09 0.008 0.20 0.04 0.09 0.06
0.008%-0.20%
of sequential errors listed in Table 7 were determined by dividing the number of errors that were sequential (anticipatory, reiterative, or metathesis) by the total number of initiation and substitution errors that occurred for each subject. All subjects in this sample produced some sequential errors, but the percentage of those errors relative to total substitution and initiation errors is small. The group mean of sequential errors was 7% in this sample. In Table 8 it can be seen that metathesis (two’phonemes changing position within a word) was not prevalent. Prepositioning (or anticipatory) errors outnumber postpositioning errors by a ratio of 6 to 1, and this difference is significant. Sequential errors, as defined in this study, do exist but do not account for a significant proportion of the phonemic errors in our sample of subjects. Feature Analysis As indicated by Johns and Darley (1970), the substitution errors in apraxia of speech are of special interest and are many times unrelated to the attempted sound. Trost (1969) relates that “. . . additions (resulting in complex substitutions for singleton phonemes) and substitutions comprise roughly 78% of the total errors, clearly distinguishes apraxic speech from dysarthric speech phonetically.” She also states that “primary verbal apraxia is marked by salient phonetic features including substitutions, complex additions, and marked difficulty in forming articulatory transitions. ’ ’ In the present study, we determined if substitutions were errors of placement
SOME
PHONEMIC
CHARACTERISTICS
Sequential
IN APRAXIA
267
TABLE 8 Errors in 13 Apmxia of Speech Subjects
Subject
Total
Error Type Prepositioning (anticipatory)
Postpositioning (reiterative)
1 2 3 4 5 6 7 8 9 10 11 12 13
3 1 9 10 I 11 2 3 1 12 3 7 9
1 0 3 3 0 1 3 0 1
Totals Range
78 1-12
Wilcoxon
OF SPEECH
Metathesis
0 0 0
0 0 0 0 0 0 0 0 0 3 0 0 1
4 1 12 13 7 12 5 3 2 16 3 7 10
13 O-3
4 O-3
95 O-16
1
Signed Rank Test (pre- vs. post-). Significant
P = < 0.01
alone, of manner, voicing, or any combination of these. Figure 1 presents a summary of the percentages of these errors. When substitutions are defective in two or more features, as they were 38% of the time in this sample (total percentages of P-M, M-V, P-V, and PMV errors), many of them appear to bear little acoustic resemblance to the target sound. Conclusions regarding distances
50
40
Fig. 1.
Place-manner-voice
errors in 13 apraxia
of speech subjects
268
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
between error and target sound are dependent upon the number of features used in the analysis, however. For example, a researcher who uses a scale comprised of 13 feature categories might be tempted to infer that errors that differ in three features are “related” or “close” to the target phoneme. Another investigator, using a five feature analysis, might conclude that errors differing in three features are “unrelated” to the target. The relationship between number and type of replaced or omitted features and listener judgments of “nearness to target” must be clarified considerably before firm conclusions about error-target “relatedness” are warranted. Further distinctive feature analysis of phonemic errors in speech apraxic subjects appears to be a fruitful area for additional research.
Summary Speech pathologists have attempted to outline characteristics of neurogenic communication disorders and differentiate among them in order to facilitate study of various treatment strategies. Though several papers have described general characteristics of apraxia of speech, descriptive data of specific phonemic characteristics have been meager. The purpose of this study was to investigate and describe specific phonemic characteristics of apraxia of speech in 13 hemispheredamaged subjects. Patterns of error obtained from articulatory confusion matrices of these subjects revealed the following: (1) Consonants were more susceptible to error than were vowels, as were consonant clusters when compared to single consonants. (2) No significant differences were found among error percentages for initial, medial, and final positions. (3) Error analysis relative to manner and place of production revealed the following: (a) affricatives and fricatives were significantly more susceptible to error than other manner categories; (b) lingua alveolar and bilabial phonemes were significantly less susceptible to error than other place categories; (c) no differences existed between error percentages of voiced and unvoiced phonemes. (4) When substitution errors were tabulated as either anticipatory (prepositioning), reiterative (postpositioning), or metathesis, anticipatory errors outnumbered reiterative errors by a ratio of 6 to 1. (5) Feature analysis of substitution errors revealed that 38% were defective in two or more features.
References Darley, F. L. The classification of output disturbance in neurologic communication disorders. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969. Darley, F. L., Aronson, A.E., and Brown, J.E. Differential diagnostic patterns of the dysarthrias.J. Speech Hearing Res., 1%9, 12, 246-270. Deal, I. L., Darley, F. L. The influence of linguistic and situational variables on phonemic accuracy in apraxia of speech. J. Speech Hearing Res., 1972, 15, 639-653.
SOME
PHONEMIC
CHARACTERISTICS
IN APRAXIA
OF SPEECH
269
Fry, D. B. Phonemic substitutions in an aphasic patient. Lang. and Speech, 1959, 2, 52-60. Johns, D. F. Phonemic variability in apraxia of speech: A disorder distinct from dysarthria and aphasia. Paper presented at the American Speech and Hearing Association Convention, Denver, 1968. Johns, D. F., Darley, F. L. Phonemic variability in apraxia of speech./. Speech Hearing Res., 1970, 13, 556583. La Pointe, L. L. Non-speech oral movement and articulation abilities in brain-injured adults. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969. Lashley, K. S. The problem of serial order in behavior. In L. A. Jeffress (Ed.), Cerebral mechanisms in behavior. New York: John Wiley, 1951. Lecours, A. R. Lhermitte, F. Phonemic paraphasias: Linguistic structures and tentative hypotheses. Correx, 1969, (September), 193-228. Martin, A. D. Some objections to the term Apruxio of Speech. J. Speech Hearing Dis., 1974, 39, 53-64. Morrison, D. F. Multivariafe starisfical methods. New York: McGraw-Hill, 1967. Rosenbek, J. C., Wettz, R. T., Darley, F. L. Oral sensation and perception in apraxia of speech and aphasia. J. Speech Hearing Res., 1973, 16, 22-36. Shankweiler, D., Harris, K. S. An experimental approach to the problem of articulation in aphasia. Cortex, 1966, 2, 277-292. Trost, J. Patterns of articulatory deficits in patients with Broca’s aphasia. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969.
OF COMMUNICATION
DISORDERS
8 (1975), 259-269
SOME PHONEMIC CHARACTERISTICS IN APRAXIA OF SPEECH LEONARD Audiology
and Speech Pathology,
L. LA POINTE
Veterans Administration
Hospital,
Gainesville,
Florida 32602
and DONNELL Southwestern
Medical
School,
F. JOHNS
lJniver&
of Texas, Dallas,
Texas 75235
The atticulatory performance of 13 left hemisphere-damaged adults who presented apraxia of speech was tabulated on confusion matrices and analyzed according to error pattern. Consonants were more susceptible to error than were vowels, as were consonant clusters when compared to single consonants. No significant differences existed among error percentages for the intial, medial, and final positions. When errors were analyzed according to manner of production, affricatives and fricatives were significantly more susceptible to error than all others. Analysis of errors according to place of production revealed lingua alveolar and bilabial phonemes to be significantly less impaired than all other categories. No differences were found in error percentages of voiced and unvoiced phonemes. The sequential nature of substitution errors was further analyzed by tallying and classifying errors as anticipatory (prepositioning), reiterative (postpositioning), or metathesis. Seven percent of the substitution errors in this study were sequential, with anticipatory errors outnumbering reiterative errors by a ratio of 6 to 1. Feature analysis of substitutions to determine distance from the target sound revealed that 38% of the substitutions were defective in two or more features. Some of these subjectively bore little resemblance to the target sound.
Introduction The subject of sensorimotor speech disorders associated with cortical lesions has proved to be a fertile area for controversy for more than a century. Theoretical and interpretive arguments have abounded, resulting in biased and inconsistent terminology which, for the most part, has reflected preconceived theoretical assumptions. The term ‘ ‘apraxia of speech” appears to be increasing in frequency of usage in recent literature. Some writers have objected to this term (Martin, 1974), but others have attempted to construct a rationale for its use. Johns ( 1968) related that: The term apraxia of speech, focuses on the underlying dynamics of the disorder. Apraxia of speech explicitly (1) directs one’s attention to the motor aspects of speech, (2) emphasizes the volitional execution of articulation, (3) excludes significant weakness, paralysis, and incoordination of the speech musculamre, and (4) indicates a discrepancy between execution of the speech act and relative linguistic intactness. D American
Elsevier Publishing
Company,
Inc.,
1975
259
260
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
Patients with apraxia of speech have been described as displaying a distinctive articulatory disorder, which is characterized by variable off-target approximations of phonemes despite absence of significant weakness, slowness, or incoordination of the speech musculature. As they speak, they struggle to position their articulators correctly; they grope to assume correct articulatory postures and to accomplish the sequence of postures in word production. They are frequently off-target, will often recognize that they are off-target, and will (with great effort) try again. Errors recur, but these errors are not always the same. Errors on a series of trials will be variable. In their struggle to avoid error, they will slow down, space their words and syllables evenly, and stress them equally, subsequently affecting the prosody of their speech (Johns and Darley, 1970). Renewed effort on the part of speech pathologists in the study of neurologically based communication disorders is now evident. Contemporary researchers are attempting (a) to outline the characteristics of distinctive disorders and classify them on some homogeneous basis, (b) to develop tests and instrumentation of differential diagnosis, and (c) to study these disorders in order to better understand language theories and neural processes involved in language function. Recent work at the Mayo Clinic and elsewhere contributed to our understanding of clusters of deviant speech dimensions in the dysarthrias and our understanding of some of the differentiating characteristics of apraxia of speech (Darley, Aronson, and Brown, 1969; Deal and Darley, 1972; Rosenbek, Wertz, and Darley, 1973). Before clear differentiation and subsequently more efficient remediation can be accomplished, however, characteristic patterns of behavior must be analyzed and described systematically. The purpose of this study was to investigate and describe specific phonemic characteristics in 13 subjects with apraxia of speech. Methods
and Procedures
Thirteen adults who sustained cortical damage and presented certain articulatory characteristics indicative of apraxia of speech were studied. These Ss were given a measure of speech and language deficit (Minnesota Test for Differential Diagnosis of Aphasia), a picture test of articulation (a 141-item modified version of the Templin-Darley Test), and two measures of non-speech oral movement. The Templin-Darley Test was modified by eliminating several plates related to phonemes previously elicited. No limitations were placed on sex, etiology, or hemispheric location of lesion, and allSs had to be at least 16 years of age. Only Ss who were at least one month postonset were selected, and individuals with severe auditory input deficit were excluded (55% or more errors on the auditory section of the MTDDA). The cause of brain injury, duration of involvement, and severity of language involvement as represented by total number of errors on the MTDDA are given in Table 1. Criteria used to establish the presence of apraxia of speech were the following (La Pointe, 1969):
SOME
PHONEMIC
Etiology,
CHARACTERISTICS
IN APRAXIA
TABLE 1 Months Postonset, and MTDDA
Subject
Etiology
1 2 3 4 5 6 7 8 9 10 11 12 13
CVA CVA CVA CVA CVA Trauma CVA Trauma CVA Trauma Trauma CVA Trauma
Mean Range ‘579 possible errors on modified
Months postonset 26 35 29 23 4 3 11 4 I1 6
1 3 2 12.15 l-35
OF SPEECH
261
Scores for Ss MTDDA No. of error@ 107 65 192 297 355 209 280 183 289 126 81 147 100 187 65-297
MTDDA.
(1) The presence of initiation difficulties (defined as repetitions of initial sound or syllable or retrials using different sound combinations on each attempt). (2) The presence of islands of error-free production (on serial speech tasks or spontaneous utterances). (3) More substitution errors than distortions and omissions combined. Articulation testing involved recording subject responses to all 141 stimulus cards on magnetic tape, then transcribing them phonetically. Phonetic transcription allowed errors to be tabulated on confusion matrices similar to those used by Shankweiler and Harris (1966) and Fry (1959).
Results and Discussion Analysis of the errors plotted on confusion matrices revealed several patterns of articulatory performance. Table 2 illustrates that significantly more errors were made on consonants than vowels (P < 0.001) and significantly more errors were made on consonant clusters than on single consonants (P =C0.00 1) as determined by a Wilcoxon Signed Rank Test. The percentages for each subject on Table 2 were determined by dividing the number of errors for each category (vowels, consonants, clusters, singles) by the number of occurrences of each of these within the articulatory test material. For example, subject 3 made errors on 15% of all vowels that occurred in the articulatory test material. Further, he made errors on 25% of all consonants tested,
262
Percentages Subject
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
TABLE 2 of Vowel vs. Consonant and Cluster vs. Single Errors for 13 Apraxia of Speech Subjects Vowels 0
1 2 3 4 5 6 1 8 9 10 11 12 13
15 12 3 26 25 2 71 5 7 4 17
Range Mean
o-71% 14%
0
Consonants 2 10 25 40 10 59 30 2 88 9 24 6 35
o-8896 26%
Clusters 6 27 35 70 20 87 34 7 96 18 48 11 55 7%-96% 40%
Singles
4 21 34 6 50 29 49 6 15 4 29 l%-50% 19%
Wilcoxon Signed Rank Test Vowels vs. consonants significant P = < 0.001 Clusters vs. singles significant P = < 0.001
35% of all clusters, and 21% of all single consonants. The data in Table 2 support the findings of Trost ( 1969) and Shankweiler and Harris ( 1966), and lend evidence to Darley’s (1969) contention that errors vary with the complexity of the articulatory adjustment. Analysis of the position of errors has resulted in different interpretations. Some authors (Trost, 1969; Shankweiler and Harris, 1966) have suggested that initial consonants are more susceptible to error than final consonants. As can be seen in Table 3, error percentages for initial, medial, and final positions in this sample were nearly equal, with no significant differences existing among them. This supports the finding of Johns and Darley ( 1970) that “No single position in the word emerged as characteristically more difficult . . .” Lack of agreement on this issue may be partly definitional, however, depending on whether retrials and repetitions (the difficulty in getting started) are counted as a separate type of error or are added to those errors made on the initial phoneme of words. In this study, retrials and repetitions were catalogued separately from errors made on the initial phoneme. Another method of viewing patterns of error is relative to phoneme type. Errors in manner of production and in place of production are presented in Tables 4 and 5. The data in both of these tables were analyzed by multivariate analysis of variance (Morrison, 1967). In Table 4 each subject has a vector of five error percentages, while in Table 5 each subject has a vector of seven error percentages. The overall
SOME
PHONEMIC
Percentage
CHARACTERISTICS
of Position
IN APRAXIA
TABLE 3 Errors in 13 Apraxia
263
OF SPEECH
of Speech Subjects
Subject
Position Initial
Medial
Final
1
1
2 3 4 5 6 7 8 9 10 11 12 13
1.5 21 57 16 54 29 2 82 9 25 6 37
3 4 34 31 9 56 18 3 87 10 21 9 40
2 7 26 35 2 67 39 2 94 8 24 4 31
Range Mean
I%-82% 27%
3%-87% 25%
2%-94% 26%
Friedman two-way analysis of variance Not significant P = > 0.05.
by ranks
Errors According
to Phoneme
Error percentage
Type
1. Affricatives 2. 3. 4. 5.
TABLE 4 Type in Apraxia
33 24 17 16 16
Fricatives Glides Nasals Plosives.
of Speech (Manner)
Paired comparisons l-2, l-3”, I-4a, l-5a 2-3O, 2-4”, 2-5a 3-4, 3-5 4-5
Overall F statistic (4 and 9 u” = 4.lOZn OSignificant
at
P < 0.05.
F test compared the equality of the population mean percentage error rates in the five and seven respective categories. Pairwise comparisons of means were also made. The five manner categories used in this study are shown in Table 4, along with the relative error percentage for each category, and the results of analysis of paired comparisons. JZrror percentages reflect the number of errors relative to the total occurrence of each phoneme type. For example, in Table 4, errors were made on 33% of the affricatives that occurred in the articulatory test material. The overall F statistic indicates that there were significant differences among manner categories.
264
LEONARD
Errors According
and DONNELL
F. JOHNS
TABLE 5 to Phoneme Type in Apraxia of Speech (Place) Error percentage
Type 1. 2. 3. 4. 5. 6. 7.
L. LAPOINTE
Linguapalatal Linguadental Linguavelar Labiodental Glottal Lingua alveolar Bilabial
31 27 23 20 17 16 13
Paired comparisons 1-2, l-3, 1-4, l-5, l-60, l-7” 2-3, 2-4, 2-5, 2-6, 2-7n 3-4, 3-5, 3-6, 3-7” 4-5, 4-6, 4-7n 5-6, 5-7 6-7
Overall F statistic (6 and 7 dJ) = 9.65 la ‘Significant
at P < 0.05.
Affricative and fricative phonemes were significantly more susceptible to error than all others. Results of paired comparisons reveal that when error percentages for affiicatives (1) or fricatives (2) were compared with any other manner category, statistically significant differences emerged. These findings support the trend suggested by Trost (1969) and by Shankweiler and Harris (1966). Similarly, analysis of errors according to place of production reveals significant differences among categories, as can be seen in Table 5. Again, error percentages reflect the number of errors relative to the total occurrence of each phoneme type. For example, errors were made on 13% of the bilabial phonemes that occurred in the articulatory test material. The overall F statistic indicates significant differences among place categories. Lingua alveolar and bilabial phonemes were the least impaired in this sample, and paired comparison analysis revealed significant differences in error percentages when these categories were compared with several of the categories more susceptible to error. These findings also offer support to previous suggestions in the literature (Trost, 1969). Analysis of errors in voicing revealed no significant differences between error percentages of voiced and unvoiced sounds. Sequential
Errors
Further analysis of the phonemic production of theseSs was concerned with the sequential nature of articulation errors. Lashley (1951), in his work on serial order, postulated that the realization of linguistic sequences requires the interplay of at least three relatively independent neurophysiological mechanisms: tendency” or “idea.” (1) The “determining or “arousal of expressive units.” (2) A mechanism of “priming” (3) A mechanism having to deal specifically with serial ordering of preactivated units.
SOME
PHONEMIC
CHARACTERISTICS
IN APRAXIA
OF SPEECH
265
Lecours and Lhermitte (1969) reviewed some of Lashley’s theories and suggested that many of the phonemic errors made by brain-damaged patients are sequential in nature. Some researchers (Darley, 1969; Johns and Darley, 1970) suggested that these sequential errors are “anticipatory” or ‘ ‘perseverative. ” In the present study, a tally was made of sequential errors, and examples of them are presented in Table 6. Sequential difficulty is only apparent by analyzing two categories of subject error, those of phoneme substitutions and those of initiation errors (difficulty in getting started). When initiation errors and substitution errors are analyzed relative to the target word, three types of sequential error become evident: “anticipatory,” “reiterative,” and “metathesis.” In substitution errors, if a phoneme were replaced by one that occurred later in the word, such as the example for “yellow,” it was tallied as an anticipatory or prepositioning error. Similarly, if in attempting to get started (initiation difficulty), the subject used a phoneme that occurred later in the stimulus word, the error was counted as anticipatory (for example, in the response to “bicycle,” It, = bai = bai = s = s = sai = s = a = bai = sikl]; here, the use of s = s = sai was tallied as an anticipatory error, even though the word was eventually produced correctly). If a phoneme were replaced by one that occurred earlier in the word, the error was called one of “postpositioning” or “reiterative,” as in the responses to “dress” and “December” in Table 6. If two phonemes switched places, as in “tefalon” for “telephone,” the error was also regarded as being sequential in nature and labeled a “metathesis.” The percentage of substitution and initiation errors of each subject, which were sequential in nature, can be seen in Table 7. The percentages
TABLE 6 Sequential Error Examples Stimulus
word
Subject response
Pre-positioning (anticipatory) Yellow Grasshopper Bicycle
“redul, ledul, (pause) 1190” “grap-popper, let’s see, - grass-hopper” “b-bai-bai-s-s-sai-s“uh - bai - ski”
Post-positioning (reiterative) Dress December
“dred” “ees, ees, esender”
Metathesis Telephone Sandwich
“teflone “wansin
- tefalone” - sanwich”
266
LEONARD
L. LAPOINTE
Percentage of Initiation and Substitution
Subject
2 3 4 5 6
8 9 10 11 12 13 Totals Range
and DONNELL
F. JOHNS
TABLE 7 Errors That Are Sequential in 13 Apraxia of Speech Subjects
Init. and sub. errors
Sequential 4
40 35 156 229 110 240 139 33 263 80 71 75 173
3 2 16 3 7 10
1644 33-263
95 1-16
12 13 12
errors
% Sequential errors 0.10
0.03 0.13 0.05 0.06 0.05 0.04 0.09 0.008 0.20 0.04 0.09 0.06
0.008%-0.20%
of sequential errors listed in Table 7 were determined by dividing the number of errors that were sequential (anticipatory, reiterative, or metathesis) by the total number of initiation and substitution errors that occurred for each subject. All subjects in this sample produced some sequential errors, but the percentage of those errors relative to total substitution and initiation errors is small. The group mean of sequential errors was 7% in this sample. In Table 8 it can be seen that metathesis (two’phonemes changing position within a word) was not prevalent. Prepositioning (or anticipatory) errors outnumber postpositioning errors by a ratio of 6 to 1, and this difference is significant. Sequential errors, as defined in this study, do exist but do not account for a significant proportion of the phonemic errors in our sample of subjects. Feature Analysis As indicated by Johns and Darley (1970), the substitution errors in apraxia of speech are of special interest and are many times unrelated to the attempted sound. Trost (1969) relates that “. . . additions (resulting in complex substitutions for singleton phonemes) and substitutions comprise roughly 78% of the total errors, clearly distinguishes apraxic speech from dysarthric speech phonetically.” She also states that “primary verbal apraxia is marked by salient phonetic features including substitutions, complex additions, and marked difficulty in forming articulatory transitions. ’ ’ In the present study, we determined if substitutions were errors of placement
SOME
PHONEMIC
CHARACTERISTICS
Sequential
IN APRAXIA
267
TABLE 8 Errors in 13 Apmxia of Speech Subjects
Subject
Total
Error Type Prepositioning (anticipatory)
Postpositioning (reiterative)
1 2 3 4 5 6 7 8 9 10 11 12 13
3 1 9 10 I 11 2 3 1 12 3 7 9
1 0 3 3 0 1 3 0 1
Totals Range
78 1-12
Wilcoxon
OF SPEECH
Metathesis
0 0 0
0 0 0 0 0 0 0 0 0 3 0 0 1
4 1 12 13 7 12 5 3 2 16 3 7 10
13 O-3
4 O-3
95 O-16
1
Signed Rank Test (pre- vs. post-). Significant
P = < 0.01
alone, of manner, voicing, or any combination of these. Figure 1 presents a summary of the percentages of these errors. When substitutions are defective in two or more features, as they were 38% of the time in this sample (total percentages of P-M, M-V, P-V, and PMV errors), many of them appear to bear little acoustic resemblance to the target sound. Conclusions regarding distances
50
40
Fig. 1.
Place-manner-voice
errors in 13 apraxia
of speech subjects
268
LEONARD
L. LAPOINTE
and DONNELL
F. JOHNS
between error and target sound are dependent upon the number of features used in the analysis, however. For example, a researcher who uses a scale comprised of 13 feature categories might be tempted to infer that errors that differ in three features are “related” or “close” to the target phoneme. Another investigator, using a five feature analysis, might conclude that errors differing in three features are “unrelated” to the target. The relationship between number and type of replaced or omitted features and listener judgments of “nearness to target” must be clarified considerably before firm conclusions about error-target “relatedness” are warranted. Further distinctive feature analysis of phonemic errors in speech apraxic subjects appears to be a fruitful area for additional research.
Summary Speech pathologists have attempted to outline characteristics of neurogenic communication disorders and differentiate among them in order to facilitate study of various treatment strategies. Though several papers have described general characteristics of apraxia of speech, descriptive data of specific phonemic characteristics have been meager. The purpose of this study was to investigate and describe specific phonemic characteristics of apraxia of speech in 13 hemispheredamaged subjects. Patterns of error obtained from articulatory confusion matrices of these subjects revealed the following: (1) Consonants were more susceptible to error than were vowels, as were consonant clusters when compared to single consonants. (2) No significant differences were found among error percentages for initial, medial, and final positions. (3) Error analysis relative to manner and place of production revealed the following: (a) affricatives and fricatives were significantly more susceptible to error than other manner categories; (b) lingua alveolar and bilabial phonemes were significantly less susceptible to error than other place categories; (c) no differences existed between error percentages of voiced and unvoiced phonemes. (4) When substitution errors were tabulated as either anticipatory (prepositioning), reiterative (postpositioning), or metathesis, anticipatory errors outnumbered reiterative errors by a ratio of 6 to 1. (5) Feature analysis of substitution errors revealed that 38% were defective in two or more features.
References Darley, F. L. The classification of output disturbance in neurologic communication disorders. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969. Darley, F. L., Aronson, A.E., and Brown, J.E. Differential diagnostic patterns of the dysarthrias.J. Speech Hearing Res., 1%9, 12, 246-270. Deal, I. L., Darley, F. L. The influence of linguistic and situational variables on phonemic accuracy in apraxia of speech. J. Speech Hearing Res., 1972, 15, 639-653.
SOME
PHONEMIC
CHARACTERISTICS
IN APRAXIA
OF SPEECH
269
Fry, D. B. Phonemic substitutions in an aphasic patient. Lang. and Speech, 1959, 2, 52-60. Johns, D. F. Phonemic variability in apraxia of speech: A disorder distinct from dysarthria and aphasia. Paper presented at the American Speech and Hearing Association Convention, Denver, 1968. Johns, D. F., Darley, F. L. Phonemic variability in apraxia of speech./. Speech Hearing Res., 1970, 13, 556583. La Pointe, L. L. Non-speech oral movement and articulation abilities in brain-injured adults. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969. Lashley, K. S. The problem of serial order in behavior. In L. A. Jeffress (Ed.), Cerebral mechanisms in behavior. New York: John Wiley, 1951. Lecours, A. R. Lhermitte, F. Phonemic paraphasias: Linguistic structures and tentative hypotheses. Correx, 1969, (September), 193-228. Martin, A. D. Some objections to the term Apruxio of Speech. J. Speech Hearing Dis., 1974, 39, 53-64. Morrison, D. F. Multivariafe starisfical methods. New York: McGraw-Hill, 1967. Rosenbek, J. C., Wettz, R. T., Darley, F. L. Oral sensation and perception in apraxia of speech and aphasia. J. Speech Hearing Res., 1973, 16, 22-36. Shankweiler, D., Harris, K. S. An experimental approach to the problem of articulation in aphasia. Cortex, 1966, 2, 277-292. Trost, J. Patterns of articulatory deficits in patients with Broca’s aphasia. Paper presented at the American Speech and Hearing Association Convention, Chicago, 1969.
