Paradoxical facilitation: the resolution of foreign accent syndrome after cerebellar stroke

Appendix e-1

METHODS

Approximately three and a half years after the patient’s second stroke, samples of her speech were recorded at the Beth Israel Deaconess Medical Center (BIDMC) and then compared to short speech samples provided by a family member from voice mail messages left by the patient at various points including: nine days prior to the first stroke (CVA1), labeled (“pre-CVA1 -9”); one day prior to CVA1 (“pre-CVA1 -1”); four days post CVA1 (“CVA1 +4”); and three weeks post CVA1 (“CVA1 +20”). Shortly after her second stroke (CVA2), the patient made one recording of her own speech, which included both a spontaneous commentary on her losing her foreign accent and a reading sample. Word counts from these spontaneous speech samples are as follows: 89 words (pre-CVA1 -9); 76 words (pre-CVA1 -1); 48 words (CVA1 +4); 82 words (CVA1 +20) and 192 words (CVA2). These speech samples provided not only data for acoustical analysis of her spontaneous speech at several points in her FAS history but also scripts which the patient was asked to read at the BIDMC recording session three and a half years later (“BIDMC CVA2”).

All recordings were transferred to a Dell computer and were sampled at 20 kHz with a 9.0 kHz low-pass filter (Butterworth 24 dB/octave) and a 12-bit quantization for acoustic analysis. Following descriptive analysis of the characteristics of speech, acoustic analyses included: 1) pitch plots to track the fundamental frequency contours at word, phrase, and sentence levels; 2) vowel formant frequency and duration analysis using a 25.6 msec full-Hamming window centered on the vowel midpoint to perform Linear Predictive Coding (LPC) and to track changes in the first (F1) and second (F2) formants; 3) stop consonant analysis including voice-onset time (VOT) and spectral analysis for place of articulation.

RESULTS

DESCRIPTIVE CHARACTERISTICS OF SPEECH

At CVA1 +4, the most salient changes in her speech contributing to the perception of a foreign accent were aberrant clausal and sentential prosody due to inappropriate rising inflection, as well as excessive labialization of high back vowels. Other vowel changes included reduction of diphthongs to singleton vowels (i.e. [miu  mu] in ‘museum’) and a tendency for some front vowels to become more central ([æ] [a]; [E]  [a]). With respect to consonants, the voiced inter-dental fricative [ð] occurred as the voiced stop [d] word-initially. These anomalous descriptive characteristics of the patient’s speech fell within the constellation of speech deficits reported in a number of other FAS cases1. At CVA1 +20, all of these deficits persisted. A tendency for high front and high back vowel laxing was also noted in this sample. After the second stroke occurred (CVA2), both local and global prosody appeared to be normal. Excessive lip rounding and diphthong reduction were absent. The only vowel anomaly perceived in the patient’s speech sample was sporadic laxing of the high back vowel.

ACOUSTIC ANALYSIS OF SPEECH VARIABLES

Prosody

No examples of dysprosody were found in the patient’s spontaneous speech samples prior to her stroke as shown by normal small pitch rises at the end of phrases and normal falling pitch contours at the end of declarative sentences. In contrast, the pitch plots for sentences at CVA1 +4 showed prosodic deficits on the phrase, clause, and sentence level with excessive, inappropriate rising intonation at phrasal, clausal and sentence boundaries. Moreover, there was very little recovery in these prosodic patterns by CVA1 +20. Subsequent to CVA2, the FAS prosodic patterns disappeared and her prosody returned to normal (Figure e-1).

Vowels - Formant Frequency

Formant frequencies for high and low vowels [i, u, æ, a], which typically form the endpoints of a standard vowel chart, are plotted in Figure e-2. The vowel space at CVA1 +4 and CVA1 +20 both demonstrate typical findings of FAS speech: at CVA1 +4 the vowel space is constricted due to a centralizing tendency that affected three out of the four point vowels; at CVA1 +20 there is evidence of excessive lip rounding as evidenced by a lowering of F1 and/or F2. Comparisons between the vowel space at these two time points suggest that the clinical manifestation of FAS in a single patient may not be static. After CVA2, the vowel space shifted and approximated the BIDMC session. There was increased variability of F1 and F2 measures after CVA1, which normalized after CVA2. Standard deviations of F1 and F2 in Hz at various time points compared to normal controls2 are as follows: F1,normal control (24); F1,CVA1 +4 (56.5); F1,CVA1 +20 (59.2); F1,CVA2 (38.9); F1,BIDMC CVA2 (36.5); F2,normal control (58); F2,CVA1 +4 (147.7); F2,CVA1 +20 (104.2); F2,CVA2 (63.9); F2,BIDMC CVA2 (59.2).

Vowel Duration

Tense (i.e. [i e u]) and lax [I E U] vowel durations vary in English with tense vowels consistently longer than lax vowels. This tense/lax distinction was maintained in earlier FAS studies3, 4, 5and in this patient as well. For example, the [i] vs. [I] distinction at various time points were: pre-CVA -9: 86 vs. 47; CVA1 +4: 159 vs. 121; CVA1 +20: 136 vs. 91; CVA2: 111 vs. 78; BIDMC CVA2: 88 vs. 70.A number of additional observations are noteworthy: The tense-lax contrast was maintained in all vowel comparisons; after her first stroke resulting in the FAS, vowel durations were substantially longer than normal. These durations shortened after her second stroke when the FAS resolved and approximated normal values at the BIDMC recording. The vowel durations produced pre-morbidly and at the BIDMC recording were similar, indicating that any pathological productions as a consequence of her strokes had resolved. There was also increased variability of vowel duration after CVA1, which normalized after CVA2. Standard deviations of vowel duration in milliseconds at various time points compared to normal controls are as follows: normal control (21); CVA1 +4 (71); CVA1 +20 (49); CVA2 (28); BIDMC CVA2 (29).

Voice Onset Time (VOT)

Figure e-3 displays the frequency distribution of the patient’s VOT values by place of articulation (labial, alveolar, velar). As can be seen, VOT productions of this patient from BIDMC CVA2 are similar to those of normal controls, showing a clear-cut differentiation of voiced from voiceless tokens for each place of articulation with no overlap between the voiced and voiceless phonetic categories. In addition, her distribution of VOT durations patterns are like that of an American English speaker: (1) voiceless stop consonants have long-lag VOTs, while voiced stop consonants have short-lag VOTs; and (2) voiceless VOT values increase as place of articulation moves from front to back (i.e. from labial to alveolar to velar). Finally, these normal VOT results are similar to those found in earlier FAS patients3, 4, 5.

Stop Consonants - Place of articulation

Results of the spectral analysis of place of articulationshow that, similar to other FAS patients, the spectral patterning for each place of articulation in this patient’s stop consonants generally matches that of a typical speaker of English6. Figure e-4 shows the LPC and discrete Fourier transfer (DFT) spectral envelopes for [t] in the words “too” [from pre-CVA1] and “to” [from CVA1 +4 (resulting in FAS) and CVA2 and BIDMC CVA2 (resulting in the resolution of FAS)]. All four time periods show the same normal diffuse-rising alveolar pattern for [t, d], indicating that the production of place of articulation in stop consonants remained normal throughout her course.

REFERENCES

1. Blumstein SE, Kurowski K. The foreign accent syndrome: a perspective. J Neurolinguistics2006; 19: 346-355.

2. Ryall JH. An acoustic investigation of vowel production in aphasia. Unpublished doctoral dissertation, BrownUniversity 1984.

3. Kurowski KM, Blumstein SE, Alexander M. The foreign accent syndrome: a reconsideration.Brain Lang 1996; 54: 1-25.

4. Blumstein SE, Alexander MP, Ryalls JH, Katz W, Dworetzky B. On the nature of the foreign accent syndrome: a case study. Brain Lang 1987; 31: 215-244.

5. Moonis M, Swearer JM, Blumstein SE, et al. Foreign accent syndrome following a closed head injury: perfusion deficit on SPECT with normal MRI. Neuropsychiatry Neuropsychol Behav Neurol 1996; 9 (4): 272-9.

6. Stevens KN, Blumstein, SE. Invariant cues for place of articulation in stop consonants. J Acoust Soc Am 1978; 64: 1358-1368.

ACKNOWLEDGEMENTS

This research was supported in part by NIH Grant RO1 DC00314 from the National Institute on Deafness and Other Communication Disorders. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Deafness and Other Communication Disorders or the National Institutes of Health.