BRAIN AND LANGUAFE 31, 215-244 (1987) On the Nature of the Foreign Accent Syndrome: A Case Study SHEILA E. BLUMSTEIN MICHAEL P. ALEXANDER JOHN H. RYALLS Centre de Ret-herche drr Centre Hospiiulier WILLIAM Core-des-Neigrs KATZ AND BARBARA DWORETZKY A detailed acoustic analysis was conducted of the speech production of a single patient presenting with the foreign accent syndrome subsequent to a lefthemisphere stroke in the subcortical white matter of the pre-rolandic and postrolandic gyri at the level of the body of the lateral ventricle. It was the object of this research to determine those changes which contribute to the perception of a “foreign accent.” A number of acoustic parameters were investigated. including features of consonant production relating to voice, place, and manner of articulation, vowel production relating to vowel quality and duration, and This research was supported in part by Grant NS22282 to Brown University and NS 06209 to the Boston University School of Medicine. Many thanks to W.Nelson Francis, Henry Kucera, Philip Lieberman, and Robert Meskill for many helpful discussions on the speech pattern of our patient, and to Margaret Naeser and Carol Palumbo for providing the CT scan and for their assistance in its interpretation. Present address of William Katz: University of California, San Diego. Present address of Barbara Dworetzky: Albert Einstein Medical School. Requests for reprints should be sent to Professor Sheila E. Blumstein. Brown University, Department of Cognitive and Linguistic Sciences, Providence. RI 02912. 215 0093-934X187$3.00 Copyright All rights G 1987 of reproductmn by Academic in any Press. form Inc. reserved. 216 BLUMSTEIN ET AL. speech melody relating to fundamental frequency. The results indicated that many attributes which might have contributed to the foreign quality of the patient’s speech were similar to those of normal English speakers. However, a number of critical elements involving consonant and vowel production and intonation were impaired. It was hypothesized that the acoustically anomalous features are linked to a common underlying deficit relating to speech prosody. It is suggested that the normal listener categorizes this speech pattern as a foreign accent because the anomalous speech characteristics, while not a part of the English phonetic inventory, reflect stereotypical features which are a part of the universal phonetic properties found in natural language. Q 1987 Academic Press. Inc. Speech output disorders are a common manifestation of unilateral leftbrain damage. The descriptive clinical characteristics of such disorders vary considerably depending upon the site of damage. One rare form of speech disorder is the “foreign accent syndrome.” The pathognomic characteristic of this syndrome is the emergence of a foreign accent in speech subsequent to a lateralized brain lesion. Little is known about the pathophysiological mechanisms which produce such a syndrome. We are aware of six previously reported cases of the foreign accent syndrome. The associated clinical findings are shown in Table 1. All were aphasic, but none severely. Each case had emerged from a nonfluent agrammatic aphasia, but at the time of the case reports all of the patients were relatively fluent. Articulation was impaired in four reported cases (Monrad-Krohn, 1947; Whitaker, 1982; Schiff, Alexander, Naeser, & Galaburda, 1983; Graff-Radford, Cooper, Colsher, & Damasio, 1986), but not markedly. Every patient had only a transient mild right face and/or arm paresis. There is also little known about the pathologic anatomy, as exact information about lesion site is available in only two previous cases (Schiff et al., 1983; Graff-Radford et al., 1986). In one case (Graff-Radford et al., 1986), the patient had an infarction demonstrated by CT scan in the deep frontal white matter anterolateral to the frontal horn and in the overlying posterior motor association cortex of the middle frontal gyrus. In the other (Schiff et al., 1983), there was an infarction demonstrated by CT scan in the inferior and middle motor strip. In other cases where localization data was available, patients were reported to have left frontal or frontoparietal lesions (Monrad-Krohn, 1947; Whitty, 1964). An autopsy was performed on Monrad-Krohn’s patient (Ryalls & Reinvang, 1985). Postmortem examination was complicated by the presence of metastatic bronchogenic carcinoma in the right hemisphere; but there was an old 9 x ,6-cm left cortical-subcortical infarction involving anterior inferior parietal lobule, inferior motor cortex, and pars opercularis. The defining characteristic of the “foreign accent” syndrome is that to the listener the patient’s speech seems to have the phonetic characteristics of a foreign language. Pick (1919) reports that his Czech-speaking patient sounded Polish. Monrad-Krohn (1947) reports that his Norwegian 60 62 Present case 56 30 27 Schiff et al.. 1983 et al.. 1986 1982 Graff-Radford Whitaker, 1964 30 Monrad-Krohn, Whitty, 26 Pick. 1919 1947 Age ___- Etiology ___ ~ AVM with small hemorrhagic: L prerolandic gyrus Stroke: L hemisphere Infarct; L middle frontal gyrus and deep white matter Infarct in L prerolandic gyrus Infarct in L pre& postrolandic wus Stroke; L hemisphere Trauma: L frontal ~~ __ anomia: Mild agrammatism; mild semantic paraphasias Mild anomia Hesitant: rare paraphasia Mild agrammatism Agrammatism; paraphasia Agrammatism; paraphasia Normal Language -___~ __~ impaired Impaired Impaired Impaired Impaired Impaired Articulation ~~ __ Uncertain Accent -__ change -_ - Boston 4 ?French Portuguese-American + Chinese American midwest Spanish American midwest 4 Norwegian 4 German French English + German Czech -+ Polish __ Nordic E 2 3 G 218 BLUMSTEIN ET AL. patient sounded German, and the medical notes on this patient indicate that she also sounded to some as French or Finnish (Ryalls & Reinvang, 1985). Whitty (1964), Whitaker (1975), and Graff-Radford et al. (1986) report that their English-speaking patients sounded German, Spanish, and Nordic, respectively. And Schitf et al. (1983) report that their PortugeseAmerican patient sounded Chinese. Despite the variety of “foreign accents” reported, many of the descriptive characteristics of the speech of these patients are similar. Abnormalities in vowel production were described in nearly all cases, particularly involving changes in vowel length and tenseness (Whitaker, 1982; Graff-Radford, et al., 1986). Such features of consonant production as voicing, place of articulation, and manner of articulation also seemed affected (Pick, 1919; Whitaker, 1982). Failure to reduce unstressed syllables within words was common as were poor transitions across word boundaries (Monrad-Krohn, 1947; Whitaker, 1982; Graff-Radford, et al., 1986). Finally, both word level and sentence level intonation were impaired, with sentence level intonation contours often showing a rising pitch at the end of the sentence (Pick, 1919; Monrad-Krohn, 1947). It is not clear which parameters of the speech of these patients contribute to the perception of a “foreign accent,” and further, why these changes are perceived by the listener as a “foreign accent” and not simply as impaired speech output. A number of clinical syndromes are characterized by abnormal speech production, and yet these patients do not sound like foreign speakers. Patients with Broca’s aphasia commonly have articulatory disturbances which affect both the segmental and the prosodic aspects of speech (Blumstein, 1981). Nevertheless, the foreign accent syndrome has never been recognized in a patient with Broca’s aphasia. Patients with aphemia (articulatory disorder without aphasia after damage to the left hemisphere cortical-subcortical speech systems) also have articulatory impairments. Although some of the patients with the “foreign accent syndrome” have had aphemia (Whitaker, 1982; Whitty, 1964; Schiff et al., 1983), the speech output of most aphemic patients has not been reported to have a “foreign accent.” For instance, Lecours and Lhermitte (1976) in their detailed analysis of a patient with a lesion of the inferior three-fifths of the motor strip and an isolated speech disorder do not describe a foreign accent, despite many segmental and prosodic deficits. Speech was described as “slowed, syllabic, dysprosodic, and dysarthric” (p. 93); it was “laborious” with “anomalies in scansion” (p. 95). There were many phonetic distortions including increased duration, devoicing, denasalization, defricativization, and pseudodiphthongization. In short, this patient had the same lesion and many of the same deficits as “foreign accent” cases, but somehow the cluster of speech deficits did not combine to produce a “foreign accent.” One element of speech which contributes importantly to a speaker’s accent, foreign or other, is prosody. Prosody is the melody of speech FOREIGN ACCENT SYNDROME 219 including parameters of both stress and intonation. There are numerous reports of impairments in prosody after various brain lesions, and several distinct elements of prosody have been described (Ross, 1981), but in none of the common disorders of prosody has a “foreign accent” been described. In this study, we present a detailed acoustic analysis of the speech production of a single patient presenting with the foreign accent syndrome subsequent to a left hemisphere stroke. It is our objective to explore the changes in speech production of this patient which contributed to the perception of a “foreign accent.” We focused on a number of acoustic parameters corresponding to particular phonetic dimensions which seemed to be affected in this patient’s speech. These included features of consonant production relating to voice, place, and manner of articulation, vowel production relating to vowel quality and duration, and speech melody relating to fundamental frequency. CASE REPORT A 62-year-old right-handed woman noticed right-sided weakness while riding a bicycle. She returned home and went to bed for the night. The next morning the right-sided weakness persisted, and she noticed that her speech was slurred. She was admitted to the Boston V.A. Medical Center. Past medical history was noteworthy. There was a several-year history of treated hypertension. A multisystem autoimmune disorder had been suspected for years because of migratory arthralgias and arthritis, recurrent erythematous macular desquamative rashes of the trunk and hands, and positive antinuclear antibody reactions. Myasthenia gravis had been diagnosed 15 years previously by a characteristic neurophysiological profile. Symptoms at the time of diagnosis and at the time of admission were restricted to fluctuating ptosis and minimal diplopia which were all controlled with pyridostigmine. The patient and her husband denied any speech disorder caused by the myasthenia. There had been no history of any disease of the central nervous system. Personal history revealed no “source” for her foreign accent. She was born in the Boston area of American-born parents of lrish extraction. She spoke only English, had never traveled overseas nor studied any foreign language in school. She studied 1 year of business subsequent to graduation from high school. She and her husband both stated that prior to her illness she had had a typical Boston accent. Neurological examination 9 days post-onset revealed a dense right central facial paresis, mild clumsiness of the right hand, increased rightsided reflexes, and a right Babinski sign. There was subjective decrease to touch and pinprick of the right lower face and hand. There was no akinesia, rigidity, tremor, left-sided motor deficit, or increased snout reflex or jaw jerk. 220 BLUMSTEIN ET AL. TABLE 2 BOSTON DIAGNOSTIC APHASIA EXAM SUMMARY Fluency Articulation rating (l-7) Phrase length (l-7) Melodic line (I-7) Grammatical form (l-7) Auditory comprehension Word discrimination (O-72) Commands (O-15) Complex material (O-l 2) Repetition High probability (O-8) Low probability (O-8) Writing Written naming (O-IO) Sentences to dictation (O-12) Naming Confrontation (O-l 14) 5.0 5.0 2.5 5.0 71.0 15.0 9.0 8.0 8.0 10.0 10.5 III Language assessment, performed 9 to 14 days after onset, revealed few abnormalities. Table 2 shows a summary of the Boston Diagnostic Aphasia Examination. Her speech output was slow with normal volume. Articulation seemed only mildly abnormal, whereas melodic line was very abnormal, characterized particularly by a sharply rising intonation at the ends of phrases and sentences. Language was generally fluent. Sentence length and sentence constructions were often lengthy, syntactically complex, and produced without error. Nevertheless, agrammatic structures occasionally emerged in speech output, particularly in the context of novel utterances. Modest word-finding difficulty was noted. However, paraphasias, particularly verbal paraphasias, were rarely heard. Repetition and recitation were normal. Writing was normal except for motor impairment. Reading aloud and comprehension were normal. The quality of speech output was not altered by repetition, recitation, oral reading, or singing. She was able to carry the tune of familiar songs. Rhythm tapping was normal with either hand. The patient’s ability to produce various affective intonations to command (e.g., happy, sad) seemed normal. Buccofacial praxis and limb praxis were normal. Neuropsychological assessment was performed 9 to 17 days postonset. Verbal IQ was 84 with low scores in arithmetic and comprehension. Performance IQ was 88 with greatest impairment in picture completion and block design. Drawings, right/left orientation, finger recognition, and memory were all normal. Calculations were adequate except for more complex multiplication and division. Affect was appropriately concerned. The patient was quite aware of her “foreign accent” and indicated that she was embarrassed by this new “accent.” FOREIGN ACCENT SYNDROME 221 F~ti. I. A nonenhanced CT scan done two years and three months post-onset demonstrates several lesions, The largest is immediately deep to the lower sensory-motor cortex area for face in the left hemisphere. Small subcortical white matter lesions are present deep to the left middle frontal gyrus and just lateral to the left putamen. There is a small, patchy lesion in the left middle temporal gyros. There is a very small lesion deep to the right middle frontal gyrus. A CT scan taken 2 months postonset demonstrated a small area of decreased density in the immediately subcortical white matter of the prerolandic and postrolandic gyri at the level of the body of the lateral ventricle, suggesting involvement of sensory and motor pathways for the face and hand. There was a possible slit of decreased density in the lateral left putamen. The left internal capsule was enlarged, suggesting clinically unsuspected prior left hemisphere disease. A second CT scan obtained 2$ years after onset was of much higher quality, and demonstrated several hypodense lesions in addition to the lesion immediately deep to left sensory-motor cortex (see Figure I). There were no events in the interim clinical history nor signs on examination to pinpoint the time when these lesions, presumably small infarcts, might have occurred. SPEECH ANALYSIS Drscriptivr Characteristics Speech output, while generally fluent and grammatically and lexically full, sounded to listeners like a foreigner speaking English. There was, however, little agreement among listeners about the type of accent the patient had. Suggestions ranged from Eastern European, perhaps Slavic, 222 BLUMSTEIN ET AL. to French, Dutch, or Scandinavian. Nevertheless, there was general agreement about the characteristics of her speech output which gave rise to the perception of a foreign accent.’ First, speech melody was changed. Productions sounded more syllable-timed than stress-timed. Thus, syllables seemed to receive relatively equal stress, resulting in changes not only in the rhythm of speech, but also in vowel quality. With more equal stress on each syllable, vowels seemed to be more equal in duration and less reduced than is usual in English. Vowels also seemed to be more tense. Intonation, another important component of speech melody, was marked by a rising inflection at the end of phrases and sentences. Second, anomalous consonant production also seemed to be a contributing factor to the “foreign accent.” There was a great deal of voicing assimilation, e.g., Chicago + [iIkagow], “yes, I know” + [yEzainow]. A further abnormality was that consonants which in normal American English undergo changes in particular phonetic environments were not changed in this patient’s speech. For example, although in English the phonemes /t/ and /d/ become flapped [r] in medial position following a stressed vowel, e.g. pretty + [priri], butter * [bArrI, the patient produced them as full stops, e.g., pretty + [prIthi], butter --f [bAthr]. While these descriptive characteristics provide some indication about the quality of the patient’s speech output, they do not demonstrate whether the perceptual impressions of those aspects of speech which have changed are those which actually did change, nor does it show how or why these constellations of changes are perceived by the listener as a foreign accent. In the following sections, we present a detailed acoustic analysis of a number of attributes of speech to determine whether in fact systematic changes have occurred in speech production, and to ascertain what the nature of these changes might be. ACOUSTIC ANALYSIS OF STOP CONSONANTS There are a number of differences in the production of stop consonants between speakers of English and speakers of Romance, Germanic, or Slavic languages. These differences involve the production of both voicing and place of articulation. In particular, in English, syllable-initial voiced stops are produced with prevoicing or a short-lag voice-onset time, and syllable-initial voiceless stops are produced with aspiration and a consequent long-lag voice-onset time. In contrast, in a number of languages such as the Romance languages, e.g., French or Spanish, or Slavic languages, e.g., Russian or Polish, voiced stops are produced with prevoicing, and I The general descriptive characteristics were based upon observations by a number of individuals including the authors, other members of the Aphasia Research Center including Norman Geschwind and Harold Goodglass, and a number of the staff of the Brown University Linguistics Department including W.N. Francis, Philip Lieberman, and Robert Meskill. FOREIGN ACCENT 223 SYNDROME TABLE 3 LIST OF STIMULI USED FOR ANALYSIS OF VOICING AND PLACE OF ARTICULATION pond Pot palm POP part tart top tot tar Tom cop car cod con calm barn bog bar bomb bob doll dark dot dock don god got gone guard gosh voiceless stops are produced with a short-lag voice-onset time. Moreover, in English, the place of articulation for the production of [t] and [d] is alveolar, whereas it is dental in many of the languages spoken in Europe, e.g., Romance and Slavic languages. Thus, acoustic analysis of voiced and voiceless stops produced by the patient should reveal whether there are either voicing or place changes which contribute to the impression that she is speaking with a foreign accent, and in particular an accent of a speaker from Eastern or Western Europe. A second purpose of the acoustic analysis is to investigate the production of medial stops. Rather than being produced as flaps, they seemed to be produced as full stops. Acoustic analysis of a number of utterances should confirm whether this is in fact the case. Stimuli and Procedures Stimuli used for the analysis of voicing and place of articulation in stop consonants consisted of 30 monosyllabic words containing the initial stops [p t k b d g] followed by the vowel [a] and either one or two consonants. Table 3 shows a list of these stimuli. The words were written in large capital letters on 3 x 5 cards for presentation to the patient. She was asked to read each word twice, preceded by the word this, e.g., “this cop, this cop.” The entire list of words was presented twice giving a total of four productions for each test stimulus. The carrier phrase “this” had been used in an earlier investigation of VOT in aphasia (Blumstein, Cooper, Goodglass, Statlender, & Gottlieb, 1980) to provide subjects who had difficulty initiating speech, in particular, anterior aphasics, with a word other than the test word to begin the utterance. In this way any difficulty with producing initial voiced or voiceless stop consonants could not be attributed to a deficit in initiating speech. Contrary to what might be expected, this carrier phrase had no shortening effect on the VOTs of voiceless stops produced by normal 224 BLUMSTEIN ET AL. controls (Blumstein et al., 1980), presumably because this effect does not emerge (or is reduced) across morpheme boundaries, i.e. {this} {pot}, although it does emerge when [s] and [p] occur within the same morpheme, e.g. {spot}. The stimuli used for the analysis of flaps vs. full stops consisted of six words with a [t] or [d] occurring in medial position after a stressed vowel. One of the words, “Dotty,” occurred during free conversation with the patient. The remaining five words, “relative,” “butter,” “pretty,” “writer, ” “rider,” were read by her. As with the previous set of words, she read each of these words a minimum of two times. ANALYSIS OF RESULTS AND DISCUSSION Voice-Onset Time Stimuli were transferred from tape to a PDPl l-34 computer at the Brown University Phonetics Laboratory for acoustic analysis. Voiceonset time (VOT) was determined from the waveform display, measuring from the burst release of the stop closure to the onset of voicing and, in the case of prevoiced stops, from the onset of voicing until the onset of the burst release. The frequency distribution of the VOT responses was then plotted separately for the labial, alveolar, and velar stops. These results are presented in Fig. 2. The results indicate that the distribution of VOT productions is similar to that of an English speaker, especially for the voiceless utterances. In particular, she produces voiceless stop consonants with a long-lag VOT ranging from 60 to 100 msec for labials, 50 to 120 msec for alveolars, and 80 to 120 msec for velars. There is a tendency, as is found in English, for VOT values to increase slightly as the stop consonant is produced further back in the mouth, i.e., labial, alveolar, velar. Although most of her voiced productions are prevoiced, as is the case for many speakers of English, they are atypical in that the duration of prevoicing for many of the tokens is unusually long. While it is common to find prevoiced utterances ranging from - 100 to -5 msec VOT, it is uncommon to find prevoicing as long as 100 to 200 msec in duration. As Fig. 2 indicates, the patient often had such productions, particularly for labial and alveolar stops. To the listener, these utterances sounded as though a short syllable consisting of a neutral vowel was produced preceding the stop release. In order to determine whether these utterances were produced with long prevoicing or with a neutral vowel, additional acoustic analyses were conducted. The acoustic characteristics of prevoicing are quite different from those of the production of a neutral vowel. In prevoicing, there is a low first formant with a narrow bandwidth, and there is virtually no formant structure above 700 Hz. In contrast, a neutral vowel has defined formant peaks throughout the spectrum. Linear predicting coding (LPC) analyses 225 FOREIGN ACCENT SYNDROME 7’ 65432- 5- t g 4- f 3- L 2- ‘0 . z 1 7654321- 210 180 150 120 90 60 VOT 30 0 30 60 90 120 (ms) FIG. 2. Frequency distribution of voice-onset time (VOT) responses for labial, alveolar, and velar stop consonants. The dark lines represent voiced targets, the striped lines voiceless targets. The top panel shows the distribution of labial responses, the middle panel alveolar responses, and the bottom panel velar responses. of 30 utterances were conducted using a 25.6-msec full-Hamming window placed in the period of the waveform preceding the stop release. Fig. 3 shows the LPC spectra for two of these utterances. As the spectra show, the patient produces a vowel-like syllable prior to the release of these voiced stop consonants. Such changes affect the syllable structure of her words and hence the perceived rhythm of her speech output. Place of Articulation Recent research with normal speakers has shown that the spectral properties measured at the moment of consonant release can uniquely distinguish place of articulation in stop consonants (Stevens & Blumstein, 1978; Blumstein & Stevens, 1979). Alveolar consonants show a diffuse spread of energy with greater energy in the higher frequencies compared 226 BLUMSTEIN 90 - 80 - ET AL. ‘DOT’ 70- 10 OO 90 I 1 I 2 I kHz 3 . I 4 5 ‘BARN’ 80 FIG. 3. LPC spectra of the prevoicing characteristics of two utterances, “dot” and “barn.” A 25.6-msec full Hamming window was placed in the period preceding the stop release. to the lower frequencies. Dental stop consonants also show a diffuse spread of energy, however, the distribution of peaks is fairly flat across the spectrum (Lahiri et al., 1984). To determine whether the patient produced alveolar or dental stops, LPC spectra were derived for 15 [t]initial utterances (see Table 3), using a 25.6-msec half-Hamming window placed at the burst release. Fig. 4 shows the results for three of these utterances. The spectra are clearly diffuse-rising, with greater energy in the higher frequencies than in the lower frequencies. These results are compatible with the production of alveolar stop consonants, typical of an English speaker. Manner of Articulation : Full Stops vs. Flaps Measurements of medial [t d] produced by English speakers indicate differences in the duration of the closure and release depending upon the phonetic environment (Zue & Laferriere, 1979). As indicated earlier, medial [t d] are realized phonetically as flaps when occurring after a FOREIGN ACCENT 227 SYNDROME 60 ! 40 FIG. 4. LPC spectra for three [t]-initial utterances, “tot,” “top,” 25.6-msec half-Hamming window placed at the burst release. and “tart,” using a stressed vowel, i.e. [VW], “butter” + [bArrI. However, they are realized as full stops medially when they occur in prestressed position, i.e., [VCV], “return” + [ritrn]. These differences are manifested acoustically in the duration of the closure and release interval. Flaps have durations of about 26 msec; full stops have durations of about 129 msec for [t] and 75 msec for [d] (Zue & Laferriere, 1979). The duration of this interval was measured for a total of 15 voiceless productions and 2 voiced productions of our patient. The stimulus utterances are phonetically of the type [VCV], and are realized as flaps by speakers of English. The results of this analysis are shown in Table 4 compared to those obtained by Zue and Laferriere (1979) for normal English speakers. The duration values of the closure and release interval produced by the patient are comparable to those of full stops and not those of flaps. Thus, consistent 228 BLUMSTEIN ET AL. TABLE 4 MEAN DURATION IN MILLISECONDS (AND STANDARD DEVIATION) Flapped [\iCV] Normals (From Zue & LaFerriere, 1979) Foreign accent patient FOR MEDIAL [t d] Prestressed [VC’?] [tl 26 (10) It] 129 (25) Ldl 27 (12) [t] 107 (34) [dl 75 (20) [dl 55 (5) with clinical impressions, our patient produces full stops rather than flaps for [t dl in medial position. ACOUSTIC ANALYSIS OF VOWELS In conversation, the patient’s vowels sounded different from normal English speakers’-shorter, more equal in duration, tenser, and less diphthongized. To better define the changes in vowel quality, acoustic analyses were conducted of the duration and formant frequencies of a series of vowels produced by the patient. These results were compared to those obtained with normals and aphasic patients (Ryalls, 1984; 1986). Stimuli and Procedures Stimuli consisted of the words heed, head, hid, hade, had, hod, hud, and who’d. They were printed in large type and were presented to the patient for reading. She read this series of words five times each, giving a total of 50 stimuli for analysis. All of these utterances were then transcribed. With the exception of the third and fourth repetition of the word had as [o] and [a], all were judged to be phonemically correct productions of the target word. These two utterances were eliminated from further analyses. The remaining 48 utterances were digitized and sampled on to the computer for acoustic analysis. hood, how’d, Duration Analysis Duration of the vowels was measured from the onset of periodicity in the waveform to the change in the waveform associated with the [d] closure. The averaged duration values obtained are presented in Table 5 along with the duration values for normal English speakers (Ryalls, 1984). There are no consistent differences between the values obtained for our patient and English speakers. In a manner similar to normal English speakers, she produces tense vowels [i] and [u] that are longer than the lax vowels [I] and [U]: c.p. 178 vs. 139 msec and 257 vs. 141 msec. Thus, there are no differences between our patient and normals in the intrinsic duration of vowels produced in stressed monosyllables. This analysis did, however, reveal an interesting finding in her vowel FOREIGN ACCENT TABLE AVERAGE VOWH DURATION 5 (IN MILLISECONDS) FOR FOREIGN ACCENT PATIENT ENGLISH SPEAKERS Duration Foreign accent patient Normals 229 SYNDROME AND NORMAL. (in milliseconds) Ii1 III [El 1x1 [AI la1 IUI lul heed hid head had hud hod hood who’d 178 206 139 145 127 152 239 223 179 148 236 220 141 144 257 197 productions. In monosyllable test words, she released the final stop [dl and produced a reduced centralized vowel, i.e., “hid” + [hIda]. This coda changed the canonical syllable structure of the test words from CVC to CVCV. This change also probably contributed to the perception of a difference in the rhythm of her speech, and consequently to the perception of a foreign accent. Formant Frequency Analysis The first and second formant frequencies (Fl and F2) of the vowel were determined by LPC analysis, using a 25.6-msecfull-Hamming window placed at the midpoint of the vocalic portion of the utterance. Figure 5 shows a plot of the Fl/F2 distribution of our patient compared to the averages for normal female speakers (Peterson & Barney, 1952). The overall distribution of the subject’s vowels is similar to that of normals, although there is a tendency for Fl to be lower in frequency for the patient than for normal speakers. The lower frequency suggests that she is producing vowels with a more closed position of the vocal tract. Thus, these vowels are likely to be perceived as more tense and less diphthongized, even though, to repeat, they are, overall, very near normal. In addition, there is a high degree of variability in her productions of vowels from one repetition to the next. Table 6 shows the standard deviations for the Fl, F2, and duration values obtained for each vowel compared to measures taken in a study of anterior and posterior aphasics and normals (Ryalls, 1984, 1986). Our patient shows variability comparable to that found for aphasic patients. FUNDAMENTAL FREQUENCY ANALYSIS All listeners agreed that the patient’s prosody was disturbed. Fundamental frequency of phonation is the most important cue for voice pitch and accordingly for sentence level intonation (Bolinger, 1958; Lieberman, 1960, 1965; Crystal, 1969). One might, therefore, expect to find unusual fundamental frequency (F,) patterns present in her speech to 230 BLUMSTEIN ET AL. 1000 200 500 500 1000 1000 1500 1500 2000 2000 2500 2500 3000 F2 FIG. 5. Frequency plot of the first formant (Fl) and the second formant (F2) for vowel stimuli produced by the patient FDS compared to average values for normal female speakers PB (taken from Peterson & Barney, 1952). TABLE 6 MEAN STANDARD DEVIATION FOR Fl, F2 (IN HERTZ) AND DURATION FOR VOWELS (IN LISECONDS) PRODUCED BY FOREIGN ACCENT PATIENT COMPARED TO MEANS OF ANTERIOR POSTERIOR APHASICS AND NORMALS (TAKEN FROM RYALLS, 1984) MILAND Standard deviations Fl F2 Duration Foreign accent patient Normals Anterior aphasics Posterior aphasics 64 118 43.3 24 58 20.8 45 91 48.4 42 89 44.7 FOREIGN ACCENT SYNDROME 231 confirm the impression of abnormal prosody. To test this hypothesis, we conducted a number of analyses aimed at examining the F, characteristics of her speech, and compared these characteristics with patterns of a normal English speaker. We also compared the F, data of our subject with data gathered from a number of normal subjects tested in a previous experiment (Lieberman, 1984). Three speaking conditions were investigated: spontaneous speech, read sentences, and sentences which were repeated after a spoken model. Spontaneous speech provided a baseline of the patient’s typical speech pattern. Read sentences provided a speech sample with varying intonation patterns. Sentences repeated after a model provided a measure of the patient’s capacity to “match” normal F, speech patterns. Analysis of Spontaneous Speech Stimuli and Procedures A spontaneous speech interview was tape-recorded. The interview consisted of an informal “chat” between the patient and one of the investigators. She was told that we wished to obtain a sample of her speech on tape, and she was asked a number of questions concerning her job, family, hobbies, etc. For comparison, a control subject was also interviewed on similar topics. The control subject (CS) was a 62-yearold woman, a native speaker of English residing in Providence, Rhode Island. In addition to the spontaneous speech interview with CS, we also attempted to elicit in the spontaneous speech of CS several of the same lexical items produced by the patient. First, a transcript was prepared of the patient’s spontaneous speech, and a series of “target” items were selected from the transcript. “Probe” questions were then constructed for CS designed to elicit the target items in an informal conversation. For example, knowing that CS is a secretary, the lexical items “secretary,” “secretarial work,” and “stenography” were sought for by asking questions, e.g., “Can you tell us about your job history?” “Did you take any business courses prior to working here?” “What kind of classes did you take?” In this manner, we were able to compare identical lexical items produced in the spontaneous speech of the two subjects. Analysis of Results and Discussion From approximately 6 min of the patient’s spontaneous speech, we selected 11 sentences consisting of both NP VP structures and short answers containing only an NP for F, analysis by computer. These were the first 11 impressionistically “foreign” sounding sentences. These sentences are shown in Table 7. The pitch plots of these sentences were then compared with the ob- 232 BLUMSTEIN ET AL. TABLE 7 SPONTANEOUS SPEECH UTTERANCES PRODUCED BY THE FOREIGN ACCENT PATIENT USED IN THE F, ANALYSIS 1. To D . . R who was in Connecticut. 2. Who sort of makes me think of that voice I . . I dunno though. 3. Yeah . just an old, old friend, you know. 4. Well, she’s from Quincy. 5. She has been to Chicago and all kinds of places and married a marine. 6. We have been friends ever since we were little girls in the carriage together. 7. secretarial work 8. . in various terms of secretary. 9. Yes, I worked for Anodyne . 10. Uh yeah that was just a year. 11. And I had a gentleman who I did the housework for. servations made for normal speakers of English in a conversational setting (Lieberman et al., 1985). We were particularly interested in the “shape” of the contours, especially the last 150 msec or so of the “terminal” segment. The terminal segment is defined as a final rapid rise or fall in F,. It is an important linguistic cue which helps mark out an utterance’s phrase structure (Trager & Smith, 1951; Lieberman, 1967; Vanderslice & Ladefoged, 1972; Atkinson, 1973). A simple unemphatic declarative sentence will usually have an intonation contour with a sharply declining terminal segment (Lieberman, 1967). Other sentence types such as yes/no questions may display a terminal segment with a rising F,. Of the 11 pitch plots for the patient, 4 were judged to be “normal” in shape, and 7 had unusual attributes. The abnormal pitch plots are displayed in Figs. 6 and 7. The plots included in Fig. 6 represent those with relatively large F, excursions. Sentences 3 (top left panel) and 5 (bottom left panel) have very large peaks occurring over a relatively short period of time. The ‘F, range values for these sentences were also quite high (200 and 160 Hz, respectively). Sentence 4 (top right panel) has a very sharp series of peaks which resemble the contours found in “motherese” rather than those found in normal or even emphatic speech (cf. Fernald, 1982). All of the sentences contain “inappropriate” terminal segments. Sentence 4, “Well, she’s from Quincy,” for example, is a declarative sentence which is eventually conjoined by the patient to a number of other short declarative phrases. One would therefore expect to find a falling F, terminal segment. Instead, it has a sharply rising terminal, generally associated with a yes/no question. Another anomaly is found in the phrase “secretarial work” (Sentence 7) used as an unemphatic short answer to the question, “Can you tell me what kind of work you’ve done for a living?” This utterance was also marked by an inappropriate sudden sharp rise in the slope of the terminal element. 233 FOREIGN ACCENT SYNDROME “Yesh...lust a” old. old friend, “Well. you know” she’s from MSECS MSECS “She has..uh..been to Chkago and a1 kfnds of place8 and matid a Marhe” ‘SecretarId 4007 100, 1 0 , 2000 Oulncy...” work’ ,OO! / 4000 MSECS , 6000 j 6000 loo$vT-r--r~ 0 600 1000 MSECS 1500 2 00 FIG. 6. Fundamental frequency plots of four test sentences containing large F, excursions spoken by the patient. The F, contours shown in Fig. 7 include all of the segments with numerous relatively small F, excursions. The contour of Sentence 6 (top panel) is comparatively “flat” with little in the way of local prominence peaks (F, range is 92 Hz). The other two sentences also have relatively “flat” contours. Five target lexical items were successfully elicited from the control subject using the methodology described above. The items were analyzed for comparison to the same productions of the patient. The comparison is displayed in Fig. 8 as a series of overlapping plots. The contours produced by the subjects for the lexical item “secretarial” appear to be roughly similar in shape. For the other four lexical items, the contours produced by the two subjects are qualitatively different. For example, for CS the contour of the word generally is roughly level with a brief rise at the end of the word; for the patient, the contour has two peaks which seem to correspond to the first and last syllables of the word. Unlike CS, the F, of the words secretury and stenography spoken by the patient was rising on the last syllable. 234 BLUMSTEIN ET AL. ‘We have been friends ever elnce we were We ~lrle In the carriage together” 4oo$ 00 MSECS “...ln variow terms of secretary.” 4oo< 100 1 1 1 1 ! ( 0 1000 0 1 8 , 3 ’ 1 I ’ 9 2000 3000 “Yes. I worked 100 0 FIG. 7. Fundamental excursions. ““,““[““,““,“” 1000 frequency 2000 1 4000 for Anodyne-” 3000 MSECS plots of sentence 4000 I 61 00 segments with relatively small F, 235 FOREIGN ACCENT SYNDROME “Secretary’ 400 100’ 30007 I ( 200 0 1 , 400 MSECS I , 800 r .a* 800 “Connectkut” Y3ecretariar 0 100 200 300 400 500 MSECS “Stenooraphy” 400 300 j 1 4 H 200z -0 . . . . . ..I................ *-.A.. 0 100 200 300 400 MSECS 500 800 70 IOO- o-,, I,,, 0 100 ,,,, 200 ,,,, ,,,, ,,,, ,,,, 300 400 500 600 700 MSECS FIG. 8. Comparison of fundamental frequency of individual lexical items as produced by the foreign accent patient, represented by inverted V’S,and the control subject, represented by the filled triangles. The results of the analysis of spontaneously produced sentence level and word level fundamental frequency suggest that F, differences contribute to the “foreign” quality of our patient’s speech. Inappropriate terminal segments, unusually large and frequent pitch excursions, occasional “flat” regions as well as changes in word level contours yield a percept of nonEnglish sentence intonation. 236 BLUMSTEIN ET AL. TABLE 8 SENTENCES USED IN F, ANALYSIS OF READ AND REPEATED SPEECH 1. The boy is here. 2. The girl is very pretty. 3. The rug is colored blue, red, and yellow. 4. He tried to reach him by the telephone. 5. Go ahead and do it if possible. 6. Where is he? 7. Is the boy here? 8. Come in and sit down. 9. Sing along with me. 10. When did he arrive? 11. Why did you do that? 12. After he received the mail, he sat down and wrote a letter. 13. Who told you to do that? 14. The pretty girl saw the baby sleeping in the carriage. 15. The fat man ate a juicy, big, T-bone steak. 16. When he arrived home last night, he went right to sleep. 17. The boy told the man that he had found the missing ring. 18. She saw something lying in the road. 19. 1 told him to go there without me. 20. This is a funny story. 21. When he heard this joke, he laughed right out loud. Analysis of Read and Repeated Speech Stimuli and Procedures In the initial clinical examination, there seemed to be less abnormality in intonation when reading or repeating speech. In order to test this quantitatively, a series of test sentences were developed to represent a variety of sentence level intonation patterns (e.g., simple declarative sentences, yes/no questions, Wh-questions, and emphatic sentences). The sentences also included varying proportions of open- and closedclass words. These sentences are listed in Table 8. The patient read the list of sentences immediately following the informal discussion in which the spontaneous speech was gathered. The control subject read the same list of sentences. The sample of repeated speech was obtained from the patient several days after the spontaneous speech/reading test session. In this session, she was told that she would hear a list of sentences read individually by the examiner; her task was to repeat each sentence. In order to have control subject CS repeat the identical sentences, a test tape was constructed which consisted of the original sentences recorded by the examiner and taped during the testing of the patient. FOREIGN ACCENT SYNDROME 237 Analysis of Results and Discussion Two sets of pitch plots were obtained for each of the 21 sentences. The first set allowed comparision of the “repeated” productions of the patient and the control to the “model” read by the examiner. The second set allowed comparison of each subjects’s read and repeated versions of the sentences. As in the case of the analysis of spontaneous speech, the overall shapes of the contours were examined, with particular attention paid to the terminal segments. Figure 9 shows representative pitch plots of the patients’ and CS’s repeated sentences compared to the model. Both closely matched the exemplars set by the model. Sample plots of the F, contours from selected read and repeated sentences produced by the patient and CS are given in Figs. 10 and 1I. Figure 10 shows examples for each subject of read vs. repeated productions which were closely matched. Note that the contours for the control subject were all “normally shaped,” i.e., had no unusual prominences or inappropriate terminal segments, and there was generally a close match between her read and repeated productions. For the patient, however, there are some “unusual” patterns. Five of the 21 read sentences had contours which differed greatly from their repeated counterparts. Two of these are given in Fig. 11. Sentences I and 11 show read vs. repeated differences in their terminal portions. The terminal or Read Sentence 1 rises, but the terminal in the repeated sentence falls. A rising terminal is unexpected in a simple declarative sentence as “The boy is here.” In Sentence 11, the terminal rises in the read version, but falls in the repeated version. The rising terminal segment in the read version is unexpected for the Wh-question, “Why did you do that?” In Sentences 8, 11, 13, and 19, there are large local “peaks” (regions of local prominence) in the read sentences. These peaks are flatter or nonexistent in the repeated sentences. In all cases, the repeated sentences are more similar to the read “model” sentences. From these observations, it seems that the patient did reasonably well in “matching” sentence level prosody to that of a model. In read speech, her sentence level prosody patterns were generally similar to those of a normal speaker, but there were a number of instances in which the prosody differed markedly from what would typically be expected. These differences were found particularly in the terminal portion of the utterance (as was also found in the prosodic patterns of her spontaneous speech), or at points of local prominence within the sentence. GENERAL DISCUSSION It was the object of this study to determine the acoustic characteristics of the speech of a patient presenting with the foreign accent syndrome. 238 BLUMSTEIN ET AL. “The boy I8 here” MSECS “Where is he’?” Y”” 300 : ,^ r z” 200 100 200 300 400 500 MSECS “I6 the boy here?” 800 7 FIG. 9. Representative fundamental frequency plots of the foreign accent patient’s (represented by the inverted v’s) and control subject’s (represented by the diamonds) repeated sentences compared to a model (represented by the empty circles and filled thick lines). FOREIGN ACCENT SYNDROME “The boy I I 239 is here” L..-- 100 II 0 200 400 III1 600 600 I’ 1000 1 00 MSECS “When did he arrive?” H Z 200 10 MSECS FIG. 10. Fundamental frequency plots of the read and repeated versions of a sentence which were closely matched in F,. The top panel is an example of a sentence spoken by the control subject and the bottom panel is one spoken by the foreign accent patient. The read sentences are represented by the inverted V’S and the repeated sentences by the triangles. The acoustic analyses investigated particular phonetic dimensions which clinical impressions had suggested might contribute to the “foreign accent” of the patient. These analyses allowed assessment of the extent to which the patient’s speech differed from normal English. Any phonetic differences might potentially reflect characteristics of other languages, and thus be the “source” of the “foreign accent.” The phonetic attributes investigated included segmental and prosodic dimensions. For segmental dimensions, we explored consonantal voicing, place, and manner of articulation and 240 BLUMSTEIN ET AL. ’ The boy ia here” MSECS “Why did you do that?” ‘OOi 100 1”“,1’“,1’1’,1111,1111,1111,,,~ 0 250 500 750 1000 12501600 l- 171 MSECS “Who 400 0 500 1000 MSECS -1 told him to go there 1600 without 2000 me” loo),,,,,,,,, 0 toob”“““““‘. 500 1000 (500 MSECS 2000 told YQ” to do that?” , 2500 500 ,,,, too0 MSECS ,,,, 1500 J 213130 3000 FIG. 11. Fundamental frequency plots of the read and repeated versions of five sentences produced by the patient which were not closely matched in F,. The read sentences are represented by the inverted v’s and the repeated sentences by the triangles. vowel quality. For prosodic dimensions, we explored fundamental frequency. The results indicated that many of the attributes which might have contributed to the foreign quality of our patient’s speech were, in fact, similar to those of normal English speakers. The voiceless stop consonants [p t k] were aspirated, the stops [d] and [t] were produced as alveolars, and vowel quality and intrinsic vowel duration were characteristic of native English speakers. Thus, while our patient’s speech sounded as FOREIGN ACCENT SYNDROME 241 though she were a speaker of a Romance or Slavic language, many of the phonetic attributes corresponding to these languages-unaspirated voiceless stops, [d] and [t] produced as dentals, and shortened vowel durations-were not present in her speech. Nevertheless, there was a constellation of acoustic features which were not characteristic of English. First, initial voiced stops were produced with unusually long prevoicing and, at times, had an epenthetic neutral vowel produced prior to the release of the stop closure. Second, monosyllabic CVC utterances were produced with an epenthetic neutral vowel introduced at the end of the utterance resulting in a CVCV canonical sequence. Third, prestressed alveolar stops were produced as full stops rather than flaps. Fourth, vowels were more tense than normal. Finally, the patient’s prosody showed large F, excursions. There were uncharacteristic rising pitch contours at the terminal elements even for simple declarative sentences and Wh-questions. These anomalous acoustic features in the speech of our patient may be linked to a common underlying deficit relating to speech prosody. Speech prosody involves both the melody and rhythm of language. The melody of speech relates to the fundamental frequency characteristics of words, phrases, and sentences. The rhythm of speech relates to the stress pattern and timing of syllable structure. The F, impairment of our patient clearly implicates a prosodic disorder. What is less clear is how the disturbances found in the analyses of particular characteristics of segment production also reflect a prosodic disorder. It will be recalled that our patient’s speech was described as syllable-timed, rather than stress-timed, as is the case in English, A change in speech rhythm could result in this perception. Changes in speech rhythm could affect a number of segmental properties, while leaving others intact. in particular, it could result in alterations of syllable structure, specific characteristics of segmental features, and vowel quality. Thus, the addition of epenthetic vowels in the environment of two contiguous consonants, e.g. ‘this top’ + [dIsatap], and in the environment of utterance final consonants, e.g., “heed” * [hida], changes the syllable structure of words from CVC to CVCV. More equal stress and timing of contiguous syllables could result in the emergence of full stops instead of flaps for alveolar consonants, perhaps because the phonetic environment for the occurrence of flaps in English is no longer realized. Rhythm changes could also affect the production of vowels, such that they would be produced in a more clipped and rhythmically uniform manner and thus be more tense and less diphthongized. It is, of course, just these segmental characteristics which were impaired in our patient’s speech. Nevertheless, many elements of segmental production were absolutely normal. In particular, voicing and place of articulation for consonants and vowel quality were all unaffected. The implication is that there is no deficit in the production of the intrinsic segmental properties of speech, but that instead, there are a few highly 242 BLUMSTEIN ET AL. specific segmental production anomalies that are secondary to a basic disturbance in prosody. It is the combination of abnormalities in fundamental frequency characteristics and particular segmental properties which suggests that a prosodic disturbance underlies the speech output disorder of our patient. While the preceding analyses delineate the patient’s prosodic deficits, they do not explain why most listeners hear her as a “normal” speaker talking with a foreign accent instead of as an impaired speaker of native English. The question is why this constellation of features takes on the quality of a foreign accent. We suggest two factors. First, at the lexical and grammatical levels, her language was very near normal, and while articulation was impaired, it was always easily comprehensible. There were minimal obvious phonemic, semantic, lexical, or grammatical markers of impaired language in conversation. Second, unlike the speech disturbance found in Broca’s aphasia or in most dysarthrias, this patient’s speech contains only features which actually do occur in natural languages. Syllable-timed speech rhythm, CVCV syllable structure, epenthetic vowels, tense vowel systems, intonational patterns with rising final contours, and varied melodic structure are all common characteristics of natural language. Thus, while these features are inappropriate and in this sense are “pathological” for English, they are normal characteristics of language. In contrast, the speech production disturbances found in Broca’s aphasia and in dysarthrias often contain elements which violate the phonetic features of natural language. For example, in the production of voicing, Broca’s aphasics commonly produce voiced and voiceless stops with overlapping VOTs (Blumstein et al., 1980; Rosenbek, McNeil, & Aronson, 1984).Dysarthric patients commonly produce aspirated stops with unusually long VOTs (Blumstein et al., 1980). No language with a voiced-voiceless contrast shows an overlap between voiced and voiceless productions, nor does any language contain voiceless stops with such extreme degrees of aspiration. In the production of place of articulation, Broca’s aphasics commonly produce alveolar stops with reduced energy in the high frequencies (Shinn & Blumstein, 1983). No languages contain alveolar stops with such a feature. Finally, the intonational patterns of Broca’s aphasics are often flattened, containing a limited F, range (Ryalls, 1982; Cooper, Soares, Nicol, Michelow, & Goloskie, 1984). No language has an intonational system which is atonal and monotonic. Thus, the patterns of speech characteristic of the foreign accent syndrome are qualitatively different from those of Broca’s aphasia and dysarthria. For the patient with the foreign accent syndrome, the characteristic features of speech contain only potential attributes of natural language. For the patient with Broca’s aphasia or dysarthria, the characteristic features of speech are largely distortions of the properties of natural language. Perception studies have shown that normal listeners are indeed very adept at perceiving a foreign accent, and in fact can correctly perceive FOREIGN ACCENT SYNDROME 243 that someone is speaking with a foreign accent not only in the absence of lexical and syntactic cues, but having been presented with only 20odd msec of speech (Flege, 1984). Given that our patient’s speech patterns contain potential attributes of natural language, it is no surprise that listeners agree that she sounds foreign but disagree about what foreign accent she has. In essence, her speech has a “generic” foreign accent, not that of any particular language. The listener attempts to frame her speech in such a way as to categorize it in terms of a native speaker of a particular foreign language, but her speech is actually categorized on the basis of stereotypical features which are part of the universal properties found in natural language. This syndrome may well not have a single pathologic anatomy. There are CT scan data on three cases (Graff-Radford et al., 1986; Schiff et al., 1983; the present case). Two cases had lesions of sensorimotor cortex for face (Schiff et al., 1983; the present case). Our case had additional small lesions located in areas that could potentially affect prosody; we cannot be absolutely confident that the lesion in lower sensory-motor cortex &one produced the “foreign accent”. The lesion profile of limited sensory-motor cortex damage would be compatible with impairment in bulbar sensorimotor function disturbing rapid and smooth integration of bulbar movements. The lesions were small, and thus resulted in minimal aphasia. The third case had a lesion of premotor structures at the level of the middle frontal gyrus (Graff-Radford et al., 1986). As the authors acknowledge, this lesion usually results in transcortical motor aphasia without a “foreign accent.” One can only speculate that disruption to motor association cortex for bulbar/facial muscles carries some potential to disrupt the rapid integration of movement without producing frank dysarthria. Further cases will be required to confirm the underlying mechanisms and anatomy of the “foreign accent syndrome.” REFERENCES Atkinson, J. R. 1973. Aspects of intonation in speech: Implications from un experimental study of fitndumenfal frequency. Unpublished Ph.D. dissertation, University of Connect&, Storrs. Blumstein, S. E., Cooper, W., Goodglass, H.. Statlender, S., & Gottlieb, J. 1980. Production deficits in aphasia: A voice-onset time analysis. Bruin und Language, 9, 153-170. Blumstein, S. E. 1981. Phonological aspects of aphasia. In M. T. Sarno (Ed.). Acquired uphusiu. Academic Press, New York. Blumstein, S. E., & Stevens, K. N. 1979. Acoustic invariance in speech production: Evidence from measurements of the spectral characteristics of stop consonants. Jownul of the Acoustical Society of America, 66, 1001-1017. Bolinger, D. L. 1958. A theory of pitch accent in English. Word. 14, 109-149. Cooper, W. E., Soares, C., Nicol, J., Michelow. D., & Goloskie. S. 1984. Clausal intonation after unilateral brain damage. Lunguagf, & Speech, 1, 17-24. Crystal. D. 1969. Prosodic systems and intonation in EngIish. Cambridge: Cambridge Univ. Press. Femald. A. 1982. Acoustic determinunts of infimt prqferrnce .fiw “Mothere.se.” Unpublished Ph.D. dissertation. University of Oregon. 244 BLUMSTEIN ET AL. Flege, J. E. 1984. The detection of French accent by American listeners. Journal of the Acoustical Society of America, 76, 692-707. Graff-Radford, N. R., Cooper, W. E., Colsher, P. L., & Damasio, A. R. 1986. An unlearned foreign “accent” in a patient with aphasia. Brain and Language, 28, 86-94. Lahiri, A., Gewirth, L., & Blumstein, S. E. 1984. A reconsideration of acoustic invariance for place of articulation in diffuse stop consonants: Evidence from a cross-language study. Journal of the Acoustical Society of America, 16, 391-404. Lecours, A. R., & Lhermitte, F. 1976. The “pure form” of the phonetic disintegration syndrome (pure anarthria): Anatomo-clinical report of a historical case. Brain and Language, 3, 88-113. Lieberman, P. 1960. Some acoustic correlates of word stress in American English. Journal of the Acoustical Society qf America, 33, 431-454. Lieberman, P. 1965. On the acoustic basis of the perception of intonation by linguists. Word, 21, 40-54. Lieberman, P. 1967. Intonation, perception and language. Cambridge, MA: MIT Press. Lieberman, P. 1984. The biology and evolution of language. Harvard Univ. Press: Cambridge, MA. Lieberman, P., Katz, W.. Jongman, A., Zimmerman, R.. & Miller, M. 1985. Measures of the sentence intonation of read and spontaneous speech in American English. Journal of the Acoustical Society of America, 77, 649-657. Monrad-Krohn, G. H. 1947. Dysprosody or altered “melody of language.” Brain, 70,405415. Peterson, G., & Barney, M. 1952. Control methods used in a study of the vowels. Journal of the Acoustical Society of America, 24, 175-184. Pick, A. 1919. Uber Anderungen des Sprachcarakters als Begleiterscheinung aphasicher Storungen. Zeitschrift fiir gesamte Neurologie und Psychiatric, 45, 230-241. Rosenbek, J. C., McNeil, M. R., & Aronson, A. E. (Eds.). 1984. Apraxia of speech: Physiology, acoustics, linguistics, management. San Diego, CA: College-Hill Press. Ross, E. D. 1981. The aprosodias: Functional-anatomic organization of the affective components of language in the right hemisphere. Archives of Neurology, 38, 561-569. Ryalls, J. H. 1982. Intonation in Broca’s aphasia. Nel4ropsycholoRiu, 20, 355-369. Ryalls, J. H. 1984. An ucoustic investigation of vowbe/ prodwtion in aphasia. Unpublished doctoral dissertation, Brown University. Ryalls, J. H. 1986. An acoustic study of vowel production in aphasia. Bruin and Lungrruge, 29, 48-67. Ryalls, J. H., & Reinvang, I. 1985. Some further notes on Monrad-Krohn’s case study of foreign accent syndrome. Folia Phoniutricu, 37, 160-162. Schiff, H. B., Alexander, M. P., Naeser, M. A., & Galaburda, A. M. 1983. Aphemia: Clinical-anatomic correlations. Archives of Neurology, 40, 720-727. Shinn, P., & Blumstein, S. E. 1983. Phonetic disintegration in aphasia: Acoustic analysis of spectral characteristics for place of articulation. Brain und Language, 20, 90-l 14. Stevens, K. N., & Blumstein, S. E. 1978. Invariant cues for place of articulation in stop consonants. Journul of the Acoustical Society of America, 64, 1358-1368. Trager, G. L., & Smith, H. L. 1951. Outline ofEnglish structure. Norman, OK: Battenburg Press. Vanderslice, R., & Ladefoged. P. 1972. Binary suprasegmental features and transformational word-accentuation rules. Language, 48, 819-838. Whitaker, H. A. 1982. Levels of impairment in disorders of speech. In R. N. Malatesha & L. C. Hartlage (Eds.), Neuropsychology and cognition (Vol. I, pp. 168-207). The Hague: Nijhoff (Nato Advanced Study Institutes Series D, No. 9). Whitty, C. W. M. 1964. Cortical dysarthria and dysprosody of speech. Journal ofNeurology, Neurosurgery und Psychiatry, 27, 507-510. Zue, V. W., & Laferriere, M. 1979. Acoustic study of medial /t,d/. JotrrnuI of the Acousticul Society of Americu. 66, 1039-1050.