Sunteți pe pagina 1din 6

FRSM-2014

A cross linguistic comparison of AOR & CBTC using ultrasound imaging


1

A CROSS LINGUISTIC COMPARISON OF ANGLE OF
RETROFLEXION (AOR) AND CAVITY BENEATH THE
TONGUE CURL (CBTC) USING ULTRASOUND
IMAGING
Sindhusha Chandran
1
, Irfana M
2
and N. Sreedevi
3
Abstract
Ultrasound imaging of the tongue is a state of the art technology for studying speech
production. The present study compared the AOR and cavity beneath the tongue curl (CBTC) of
tongue tip during the production of unvoiced and voiced retroflex sounds /TT/ and / DD/ in 10
female participants of 20-30 years in native Kannada and Malayalam. The tongue movements
were imaged using the Mindray ultrasound instrument 6600 and analyzed using Articulate
Assistance Advanced (AAA) software. Results showed significant difference between the two
languages with respect to AOR. The tongue contour in Malayalam speakers showed a more
retroflection curvature. That is there was difference in degree of retroflection within the
Dravidian language family itself. The study classifies Kannada and Malayalam retroflexes
using AOR and CBTC and this augments our understanding of tongue dynamics during
retroflexion using the two independent and innovative parameters which can be applied to
clinical population.

Keywords: Ultrasound tongue contours, Angle of retroflection (AOR), cavity beneath the
tongue curl (CBTC), Retroflex, Dravidian language, Articulate Assistance Advanced (AAA).

Introduction
The retroflex is a Latin-based term (Dixit 1990), refers articulation with the tongue tip bent
backwards and having contact on the post-alveolar area. Retroflexes are present in all the
Dravidian languages, Indo-Aryan languages, in few Tibeto-Burman languages and Austro-
Asiatic languages. They are interest to researchers due to its articulatory complexity. Most
studies researched on constriction location (Ladefoged, & Bhaskararao, 1983) and tongue
contours shape (Svarny & Zvelebil, 1955).
Hamann (2003) discussed four characteristics of retroflex production: apicality, posteriority,
presence of sublingual cavity and retraction. Keating (1991) studied changes in sublingual
cavity in different retroflexes. Keating (1991) and Hall (1997) described an apical and a
subapical post-alveolar retroflex pattern. Apical retroflection is the contact of tongue tip to
post-alveolar region while subapical retroflection is contact of area below tongue tip involving
an increased retroflex gesture than the former. These variations in retroflection are examined
across and within language families in phonetic and phonological literature. Figure 1 illustrates
the two types, with a retroflex stop from Hindi on the left, and one from Tamil on the right.

Figure 1: Hindi retroflex [D] (left) and Tamil retroflex [D] (right), based on sagittal x-ray
tracings from Ladefoged and Maddieson (1996).



1
Research Officer, Department of Speech and Language Sciences, Email: sinaiish@gmail.com,
2
Junior Research
Fellow, Email: fanairfana@gmail.com, Department of Speech Language Sciences,
3
Reader & Head, Department of
Speech Language Sciences, Email: sreedeviaiish@gmail.com, All India Institute of Speech and Hearing, Mysore,
India


Catford (1977) distinguished
and Indo-Aryan. He proposed that the retroflex
articulation with the underneath of the tongue against the prepalatal area termed as sublamino
prepalatal articulation and
tongue (subapical area) against the back of the alveolar ridge termed as sublaminopost
alveolar. Svarny and Zvelebil (1955)
retroflexed segments in Indo
language. The above studies i
patterns of speech sound articulation
languages had retroflex sounds
alveolar ridge, and certain others
pattern of retroflection in different la
Although individual languages or language families show a preference for a particular pattern
of retroflection, the variation has also
families. Dixits (1990) stud
narrower constriction than the
Dravidian languages spoken
both the languages belong to the sam
phonotactics of these languages are different.
significant articulatory pattern like sub
and laminal denti-alveolar /t/
failed to conclude if there was similar or distinct production of the same retroflex sounds within
the same language family especially in terms of the
The investigations to study the tongue patt
palatography, MRI scans
imaging techniques to study
state of the art technique to analyze tongue movements during speech production.
convenient, safe, non-invasive,
Kochetov, Sreedevi, Manjula and Kasim, (2012)
sound /TT/ using ultrasound with respect to dentals /t/ and velars /k
showed that retroflexes were predominantly subapical but
an obligatory property of retroflection.
Figure 2: The
The cavity beneath the tongue
Dravidian languages. The
the researcher to describe
study is to understand the nature of retroflex sounds in
and CBTC.

Method

Participants: 10 female
including 5 native Kannada (
M4 and M5) speakers. All the subjects were
Instrument setup: A desktop
ultrasound software Version 2.14
The stabilization headset was
and avoid movements during the ultrasound imaging
A cross linguistic comparison of AOR & CBTC
distinguished the patterns of retroflection in two language families:
proposed that the retroflex productions on Dravidian languages are
articulation with the underneath of the tongue against the prepalatal area termed as sublamino
and on Indo-Aryan languages are articulated with the underside of the
tongue (subapical area) against the back of the alveolar ridge termed as sublaminopost
Zvelebil (1955) and Ladefoged and Maddieson (1996)
retroflexed segments in Indo-Aryan languages and extremely retroflexed segments in Dravidian
e above studies indicated that languages within a language family share
patterns of speech sound articulation. Whereas Butcher (1992) indicated
retroflex sounds which were articulated with the tongue tip just behind the
and certain others with subapical pattern. Thus there are different views on the
pattern of retroflection in different languages.
Although individual languages or language families show a preference for a particular pattern
variation has also been seen within individual languages of
study indicated that the voiced retroflex sounds (and dental) had a
narrower constriction than their voiceless counterparts in Hindi. Kannada and Malayalam are
spoken approximately by 49 and 38 million speakers respectively
both the languages belong to the same language family, the number of phonemes and the
phonotactics of these languages are different. Hamann (2003) reported that Malayalam has
significant articulatory pattern like sub-apical palatal // which contrasts with apical alveolar /t/
alveolar /t/. Limited study in this particular area on language families have
failed to conclude if there was similar or distinct production of the same retroflex sounds within
the same language family especially in terms of the angle of retroflection (AOR)
investigations to study the tongue patterns are predominantly using x
palatography, MRI scans, EMMA and Ultrasound. Ultrasound is one of the most recent
imaging techniques to study the overall tongue dynamics. The present study make
art technique to analyze tongue movements during speech production.
invasive, and captures the fast tongue movements
Kochetov, Sreedevi, Manjula and Kasim, (2012) compared the tongue
/ using ultrasound with respect to dentals /t/ and velars /k/ in Kannada
s were predominantly subapical but the tongue body retraction was not
an obligatory property of retroflection. But AOR was not explored in their study
: The AOR and CBTC during production of a retroflex sound
The cavity beneath the tongue curl (CBTC) as seen in Figure 2 is minimally researched in
Dravidian languages. The AOR and CBTC are relevant parameters to be studied
researcher to describe the retroflex sounds in a novel way. Thus, the a
understand the nature of retroflex sounds in Kannada and Malayalam using AOR
female subjects of the age range 18 to 25 years participated in the study
Kannada (K1, K2, K3, K4 and K5) and 5 native Malayalam (M1, M2, M3,
. All the subjects were and proficient in their respective
desktop PC computer loaded with Articulate Assistant Advanced (AAA)
Version 2.14 was connected to the Mindray ultrasound instrument 6600
set was placed comfortably on the participants head
and avoid movements during the ultrasound imaging. The long-handled microconvex
FRSM-2014
using ultrasound imaging
2
language families: Dravidian
productions on Dravidian languages are
articulation with the underneath of the tongue against the prepalatal area termed as sublamino-
Aryan languages are articulated with the underside of the
tongue (subapical area) against the back of the alveolar ridge termed as sublaminopost-
Ladefoged and Maddieson (1996) found moderately
Aryan languages and extremely retroflexed segments in Dravidian
language family share distinct
indicated certain Australian
re articulated with the tongue tip just behind the
different views on the
Although individual languages or language families show a preference for a particular pattern
within individual languages of language
retroflex sounds (and dental) had a
Kannada and Malayalam are
respectively. Though
e language family, the number of phonemes and the
Hamann (2003) reported that Malayalam has
/ which contrasts with apical alveolar /t/
Limited study in this particular area on language families have
failed to conclude if there was similar or distinct production of the same retroflex sounds within
AOR).
predominantly using x-rays, static
Ultrasound is one of the most recent
The present study makes use of this
art technique to analyze tongue movements during speech production. It is
the fast tongue movements during speech.
pared the tongue contour of retroflex
in Kannada. The findings
tongue body retraction was not
in their study.

and CBTC during production of a retroflex sound
is minimally researched in
to be studied which enable
Thus, the aim of the present
annada and Malayalam using AOR
participated in the study
Malayalam (M1, M2, M3,
respective languages.
computer loaded with Articulate Assistant Advanced (AAA)
ultrasound instrument 6600.
placed comfortably on the participants head to stabilise the head
handled microconvex
FRSM-2014
A cross linguistic comparison of AOR & CBTC using ultrasound imaging
3

transducer probe operating at 6.5 MHz was placed beneath the chin after smearing the
conduction gel Aquasonic 100. Microphone i ball 333 recorded the audio stimuli in
synchronization with the ultrasound image signal.
Procedure
Stimuli and instructions: The stimuli considered were geminate unvoiced and voiced retroflex
sounds /TT/ and /DD/ in the vowel environment of /a/ i,e; /aTTa/ and /aDDa/. The participants
were instructed to repeat the target words four times each. Thus, 80 recordings (10
participants*2 stimuli*4 times) in total were made.
Measurement of AOR: The target frame showing the tongue retroflexing was identified and
subjected to spline fitting offline. It was subjected to angle function of measurement tool
program loaded in MATLAB software. Three points were identified: point on the blade of the
tongue, point at the initiation of the tongue curl, and the point on the tongue tip. These points
were marked on the frame and the angle function of measurement tool in MATLAB software
provided the AOR. The steps of measurement are shown in the figure 3.

Figure 3: Demonstrates the steps for the measurement of AOR: selection of the target frame,
fitting 2D tongue spline, identifying the three points for measurement of AOR and obtaining
AOR using angle function of measurement tool in MATLAB software.

Data Analysis: The recordings were identified, annotated and fitted with 2D spline to plot the
tongue contour. The AOR values obtained was subjected to statistical analysis SPSS-16.

RESULTS AND DISCUSSION
The present study aimed to study the AOR and CBTC of tongue tip during the production of
unvoiced and voiced retroflex sounds in two Dravidian languages Kannada and Malayalam
using ultrasound imaging. Table 1 shows the mean AOR for /TT/ and /DD/ in Kannada (K) and
Malayalam (M) speaking individual participants.
Table 1: Mean angle of retroflection (AOR) for /TT/ and /DD/ in Kannada (K) and Malayalam
(M) speaking participants
Participants K1 K2 K3 K4 K5 M1 M2 M3 M4 M5
/TT/ in /aTTa/ 108.35 89.6 92.44 121.15 151.23 85.43 61.2 68.13 57.31 60.33
/ DD/ in /aDDa/ 132.71 90.22 100.16 106.08 156.04 102.27 85.52 65.14 66.84 68.37

The results of the study are discussed under 2 segments:
1. Comparison of AOR and CBTC of /TT/ and /DD/ across Kannada and Malayalam.
2. Comparison of AOR and CBTC of /TT/ and /DD/ within Kannada and Malayalam.

1. Comparison of AOR and CBTC of /TT/ and /DD/ across Kannada and Malayalam

a) AOR
Table 1 shows the mean AOR of retroflex /TT/ in all Kannada participants as >90 except in K2
(89.6) and all Malayalam participants had AOR of <90. Similarly, AOR of retroflex /DD/ in
all Kannada participants was >90 for /DD/ production and all Malayalam participants had <90
except in M1 (102.27). An acute AOR (<90) implies greater degree of tongue curl than obtuse
angle (>90). Thus, the unvoiced and voiced retroflex sounds in Malayalam were found to have
a larger tongue curl than in Kannada.
Non parametric Mann Whitney U test revealed a significant difference in AOR of retroflex /TT/
and /DD/ in Kannada and Malayalam at p<0.05. Even though the participants produced same
retroflex sounds, the AOR was different in both the languages. The present study showed that
the two retroflex sounds had a distinct AOR for both languages, despite the fact that both
FRSM-2014
A cross linguistic comparison of AOR & CBTC using ultrasound imaging
4

belong to the Dravidian family. The study finds support from Butcher (1992) that different
languages of Australian language family produced retroflex segments distinctly. The present
study is not in consonance with Catford (1977), Svarny and Zvelebil (1955), and Ladefoged
and Bhaskararao (1983). This study suggests that languages within a language family can have
divergent AOR during the production of the same retroflex sounds.

b) CBTC
The result of CBTC is described and discussed in terms of visual inspection of the tongue
curvature. As discussed earlier, the retroflex /TT/ and /DD/ in Malayalam had an acute AOR
than Kannada; implying that the Malayalam speakers made larger degree of tongue curl than
Kannada. This can be prominently observed in Figure 4.

Figure 4: Shows the tongue contours of unvoiced retroflex sound /TT/ in a. Kannada and b.
Malayalam

The Figure 4 shows the tongue tip contact with hard palate in unvoiced retroflex sound /TT/ in
both languages. It can be seen that there is not evident CBTC in Kannada unvoiced retroflex
while an evident CBTC for unvoiced retroflex production in Malayalam.

Figure 5: Shows the tongue contours of voiced retroflex sound /DD/ in a. Kannada and b.
Malayalam
Figure 5 depicts the tongue imaging of voiced retroflex sound in Kannada and Malayalam. The
tongue contours showed the presence of CBTC in both the languages; but it was more marked
in Malayalam speakers than in Kannada. Thus a greater CBTC was found in Malayalam.
Thus, findings from AOR and visual examination of CBTC of tongue images across languages
showed a greater CBTC in Malayalam than Kannada for both unvoiced and voiced retroflex
production. Keatings (1991) and Halls (1997) described two different patterns of retroflection
mainly an apical and a subapical post-alveolar retroflex pattern. The findings of the present
study also revealed a classification of retroflection pattern. The apical post-alveolar pattern of
retroflection was seen in the production of unvoiced retroflex in Kannada and subapical post-
alveolar pattern was seen in Malayalam unvoiced and voiced retroflex sounds. Kannada voiced
retroflex showed a combination of apical and subapical post-alveolar pattern of retroflection.
Thus the present study could classify retroflexion pattern in Kannada and Malayalam.
The dominant subapical pattern in Malayalam is possibly due to functional influence to
preserve the contrast of sub-apical palatal // with apical alveolar /t/ and laminal denti-alveolar
/t/ (Hamann, 2003). Thus, there are a distinct pattern of retroflex production in terms of AOR
and the variation in CBTC within the same language family.


FRSM-2014
A cross linguistic comparison of AOR & CBTC using ultrasound imaging
5

2. Comparison of AOR and CBTC of /TT/ and /DD/ within Kannada and Malayalam
a) AOR
Table 1 showed greater AOR in voiced compared to unvoiced retroflex in Kannada. Similar
pattern was found in Malayalam speaking subjects. Though independent group difference was
seen in Table 1, statistics indicated no significant difference between unvoiced and voiced
retroflexes with in both languages. This result is incongruent with previous report of Dixit
(1990) and Ladefoged (1964) where they found different patterns for voiced and unvoiced
retroflexes counterparts.

b) CBTC

Figure 6: Shows the tongue contours of a. unvoiced /TT/ and b. voiced /DD/ retroflex sounds in
Kannada

Figure 6 shows the tongue contours of unvoiced /TT/ and voiced /DD/ retroflex sounds
in Kannada. The tongue contour for unvoiced production demonstrated a tongue tip contact
with the palate and voiced /DD/ retroflex production had a rounded appearance. Even though
no significant difference in the AOR in unvoiced and voiced retroflex sound was obtained,
CBTC could be distinctly differentiated. Thus, there was a presence of CBTC in voiced
retroflex compared to unvoiced retroflex production in Kannada speakers.

Figure 7: Shows the tongue contours of a. unvoiced /TT/ and b. voiced /DD/ retroflex sounds in
Malayalam

Figure 7 showed that both unvoiced and voiced retroflexes in Malayalam were
produced with tongue curling initiating from blade of the tongue and contact to the palate by
underside of the tongue tip making a rounded appearance of the anterior tongue similar to
subapical post-alveolar contact described by Keating (1991) and Hall (1997). The CBTC was
larger in production of voiced compared to unvoiced retroflexes in Malayalam even though
their AOR was found to be similar.
Thus the present study also revealed that AOR and CBTC were independent. AOR provided
information on the degree to which tongue is curled to make a contact with the palate, while
CBTC provided the measure of tongue cavity made beneath the AOR. The study also suggested
that apart from the voicing aspect, the CBTC also varied in voiced and unvoiced retroflex
counterparts in Kannada and Malayalam. Variation in CBTC could imply variation in the place
of articulation. Thus, a slight variation in the place of articulation is possibly present in voiced
and unvoiced retroflex counterparts, which needs to be further researched on large group of
subjects and in different languages. Thus, in Kannada and Malayalam, CBTC varies in the
production of unvoiced /TT/ and voiced /DD/ retroflex sounds.
To summarize the present study both the Dravidian languages studied had distinct AOR and
CBTC for retroflexes. Apical pattern of retroflection was seen in unvoiced retroflex and a
combination of apical and subapical pattern was seen in voiced retroflex in Kannada.
Malayalam unvoiced and voiced retroflexes had subapical pattern. AOR was similar and CBTC
was different between unvoiced and voiced retroflex cognates. The study enhances the
knowledge of two novel or innovative parameters angle of retroflection (AOR) and cavity
FRSM-2014
A cross linguistic comparison of AOR & CBTC using ultrasound imaging
6

beneath the tongue curl (CBTC) in the production of retroflexes. These observations can be
served as an input for development of speech production models. It should be noted that there
could be variation in these parameters across different languages. They can be explored in
normal subjects and applied in speech sound disordered population as a feedback tool for
clients with retroflex errors.

Acknowledgments
The authors are grateful to Dr. S. R. Savithri, the Director, and Dr. Y. V. Geetha, Head of the
Department of Speech Language Sciences of All India Institute of Speech and Hearing, Mysore
for permitting to conduct this study. We sincerely acknowledge Mr. Ajish K. Abraham, Head,
and Mr. Raghavendra G N, Software Engineer, Department of Electronics for programming
angle measurement tool in MATLAB software. We thank the participants for their cooperation
and valuable participation in the study.

References
1. Butcher, A. (1992). The phonetics of neutralisation: the case of Australian coronals. In: Jack
Windson (ed.) Studies in general and English phonetics. London: Routledge; 10-38.
2. Catford, J. C. (1977). Fundamental Problems in Phonetics.Edinburgh: University Press.
3. Dixit, R. P. (1990). Linguotectal contact patterns in the dental and retroflex stops of
Hindi. Journal of Phonetics, 18, 189-201.
4. Dixit, R. P., & Flege, J. E. (1991). Vowel context, rate and loudness effects on linguopalatal
contact patterns in Hindi retroflex //. Journal of Phonetics, 19, 213-229.
5. Hamann, S. (2003). The Phonetics and Phonology of Retroflexes. Utrecht: LOT.
6. Keating, P. A. (1991). Coronal Places of Articulation. In: C. Paradis & J.-F. Prunet (eds.),
29-48.
7. Kochetov, A., Sreedevi, N., & Kasim, M., & Manjula, R. (2012). Analysis of tongue shapes
during the production of Kannada consonants. AIISH Project.
8. Ladefoged, P., & Bhaskararao, P .(1983). Non-quantal Aspects of Consonant Production: A
Study of Retroflex Consonants. Journal of Phonetics, 11, 291-302.
9. Ladefoged, P. & Maddieson, I. (1996). The Sounds of the Worlds Languages. Oxford:
Blackwell.
10. Maddieson, I. (1984). Patterns of Sounds. Cambridge: Cambridge University Press.
11. Svarny, O., & Zvelebil, K. (1955). Some remarks on the articulation of the cerebral
consonants in Indian languages, especially in Tamil. Archiv Orientalni, 23, 374-434.

S-ar putea să vă placă și