Content uploaded by Tae-Jin Yoon
Author content
All content in this area was uploaded by Tae-Jin Yoon on May 03, 2025
Content may be subject to copyright.
Speech Communication / Communication par la parole
Pho n et ic an d P h on o lo g ic al In fl ue nc e o f a Sp e e c h -Im p a ir e d Sp e a k er o n
Rhy t h m
Gurnikaita Chhina1, Tae-Jin Yoon2, and Karin Humphreys3
'Dept. of Life Science, McMaster University, 1280 Main St. W., Hamilton, Canada, L8S 4M2
2Dept. of Linguistics, McMaster University, 1280 Main St. W., Hamilton, Canada, L8S 4M2
3Dept. of Psychology, Neuroscience, and Behaviour, McMaster University, 1280 Main St. W., Hamilton, Canada, L8S 4M2
1. in t r o d u c t io n
We present detailed phonetic analyses of a 61 year-old
monolingual female English speaker. The speaker from
Halifax, Nova Scotia, acquired a speech disorder from a
motor vehicle accident. The disordered speech was detected
three years after the brain injury she suffered from the
accident, and perceived by her family and friends as being
similar to Southern U.S. dialect or Scottish English. It is
reported that her disordered speech is a rare but possible
case of Foreign Accent Syndrome (FAS; Louch 2009). FAS
may lead to long-term muscular adjustments of the vocal
apparatus that lead to changes in articulatory, phonatory,
prosodic settings (Moen, 2000). There is considerable
variability among reported cases of FAS in terms of
phonetic characteristics and impairments (Varley,
Whiteside, Hammill & Cooper, 2006). We hypothesize that
one of the factors that make the accent change in her speech
is due to a change in her rhythmic characteristics. In order to
test the hypothesis, we calculated %V (proportion of vocalic
duration over the total duration of an utterance) as a
measure of rhythm. In order to find out factors that may
affect the %V value, we further conducted detailed phonetic
analyses to understand the impact of the speaker’s speech
characteristics on the %V that deviates from that of
Canadian English. The results indicate that both phonetic
and phonological factors contribute to her non-canonical
rhythmic pattern.
2. DATA
The speaker, LA, is a monolingual English-speaking
Canadian Woman and she was 61-year old when the data
were collected. She was born in St. Catherines, Ontario,
moved out east around age 6, and has resided in Nova
Scotia since the fifth grade. She has lived in Eastern Atlantic
Canada for most of her life. She was a retired special
education teacher. The speaker acquired a speech disorder
three years after suffering from a brain injury from a motor
vehicle accident. One day three years after the accident, her
family member observed noticeable changes in her speech
such as word searching, stuttering, and a robotic style of
speech. These changes were salient when the speaker was
extremely tired or anxious. According to the family
members, she woke up one morning speaking with slow and
broken speech sounding like a foreign accent. Her speech is
regarded as often shifting from her Atlantic Canadian
English to what was perceived as Scottish English at times
or as Southern United States English at times (Louch 2010).
Upon request by the family members, the third author
visited the speaker and collected the speech recordings
through sessions with the patient. The collected recordings
range from simple read sentences to passages, and to
spontaneous description of pictures. In this study, we report
our preliminary analysis of a subset of the recordings of
simple sentences.
In the subset of the collected data, the speaker pronounced
11 monophthongal vowels that are inserted in a /hVd/
context: heed, hid, hade, head, had, who’d, hood, hoed, hud,
hawed, and hod. Each target was embedded in one of two
carrier phrases, “the next word is /hVd/.” and “I said /hVd/
again.” In order to conduct phonetic analyses including %V,
we segmented the utterances phonetically and then
collapsed the segments into consonant intervals and vocalic
intervals, using a custom-made script in Praat (Boersma &
Weenink, 2010).
3. RESULTS
Typical stress-timed languages such as English have
complex syllable structure and tend to have reduced vowels
in unstressed positions. Therefore, the average vocalic
duration is expected to be relatively shorter than the average
consonantal duration. Indeed, Ramus et al. (1999) reports
that the %V is about 40.1 in English. Given the complex
syllable structure and reduced vowels in unstressed
positions, relatively lower value of %V is expected. But
measurement can be affected by different phonetic
characteristics even within the same language group, or
even between different dialects of the same language. For
example, Grenon (2010) reports that the %V value is 47 in
Canadian English, and she attributes the higher %V than the
one reported in Ramus et al. (1999) to the relatively longer
aspiration in Canadian stops. We found the %V of the
speaker is 52, even higher than the average for Canadian
English. At least two possibilities can be entertained: (1)
disordered speech characteristics of our subject prolonged
vocalic duration more than consonantal duration, or (2) the
rhythmic characteristic of her speech may resemble syllable-
timed languages (with simpler syllable structure).
Canadian Acoustics / Acoustique canadienne Vol. 39 No. 3 (2011) - 148
4. DISCUSSIONS AND CONCLUSION
We conducted detailed phonetic analyses to understand the
impact of the speaker’s speech characteristics on the
relatively higher %V than that of Canadian English. Speech
characteristics of the speaker include slow, enunciated, and
prolonged realization of segments, frequent insertion of
pauses, and release of word-final stop consonants, as well as
the following non-canonical characteristics: (1) Occasional
modification of syllable structure via a schwa-like vowel
insertion, as in Figure 1, (2) fully aspirated stop in the
intervocalic context, as in Figure 2, and (3) a variant
realization of [aj] to [a] (monophthongization), as in Figure
4 compared to Figure 3. The perception of her speech
sounding as though Southern US English can be attributed
to the occasional monophthongization o f a diphthong [aj] to
[a].
Figure. 1. Insertion of two schwa vowels in words of the CVC
structure. The words ‘word’ and ‘hid’ have modified syllable
structure (i.e., CVCV) due to the vowel insertion.
into account possible modifications o f phonetic properties
and phonological structure. Further analyses and tests are
needed to see whether these %V has indeed perceptual
consequences.
Figure. 3. The canonical realization of the word ‘I’ as a
diphthong is shown by two diverging F1 and F2. The release of
final consonant is indicated by the stop burst on the final
consonant in the word ‘said.’ The release of final consonant is
optional in English, but the speaker tends to release most of the
word final consonants.
mtn.**..-
.....
•
..........
•• be";
270 rm
i200 ms 205 m *
said
a: s I 0 I d
Figure. 2. Fully aspirated stop in the intervocalic context. The
intervocalic ‘t’ is realized with long closure duration (180ms)
and aspiration (40ms).
Some of these non-canonical phonetic properties work
toward reducing the %V value. The vocalic duration is
asymmetrically longer than the consonantal duration. In
addition, the perception of her speech sounding as though
Southern U.S. English can be attributed to the occasional
monophthongization of a diphthong [aj] to [a]. These
phonetic differences make the speaker’s post-stroke speech
so distinctively different from her pre-stroke speech that her
family and friend think she speaks a foreign accent. These
characteristics o f disordered speech that deviates from
canonical speech properties of mainstream English poses a
challenge on models of speech production that does not take
Figure. 4. The non-canonical realization of the word ‘I’ as
diphthong is shown by low two stable formants (F1 and F2).
The final consonant is released and then followed by non-
canonical schwa-like vowels as shown by voiced waveform and
spectrogram at the end of the word ‘said.’ The figure is in
contrast to Figure 3.
REFERENCES
Boersma, P. & Weenink D. (2010). Pr aa t: d oin g P ho ne tic s by
C om put er (ve rsio n 5.2) [downloadable from http://praat.org].
Grenon, I. (2010). A n a coust ic analy sis o f sp eec h r hyth m i n
Ca nadian En gl ish a nd Ja pan ese, (ms.) University of Victoria.
Louch, M. (2009). A C ase o f For eig n Ac ce nt S ynd rom e. B.Sc.
Thesis. McMaster University.
Moen, I. (2000). “Foreign accent syndrome: A review of
contemporary explanations.” Asph asiolo gy, 14, 5-15.
Ramus, F., M. Nespor & J. Mehler (1999) Correlates of linguistic
rhythm in the speech signal. Cognition 73, 265-292.
Varley, R., Whiteside, S., Hammill, C., & Cooper, K. (2006).
Phases in speech encoding and foreign accent syndrome. Jou r n al
o f Ne ur olingu ist ics , 19, 356-369.
AUTHOR NOTES
The work was conducted while the first author, Gurnikaita
Chhina, was a student at McMaster University.
149 - Vol. 39 No. 3 (2011) Canadian Acoustics / Acoustique canadienne
