Professional Documents
Culture Documents
An Acceptable Alternative
Articulation to Remediate
Mispronunciation of the
English /l/ Sound: Can
Production Precede Perception?
GREG RAVER-LAMPMAN
CORINNE WILSON
Old Dominion University
Instrument
Based on procedures employed by Kusumoto (2012), a list of 20
English /r/-/l/ minimal pairs with an initial contrast was created
including words such as lip, lap, lamp, light, and lot (see
Appendix A). The list contained mirror images of the pairs (e.g.,
rip, rap, ramp, right, rot). The selected words were derived from the
minimal pair list developed by Higgins (2015) with the criteria that
they be differentiated solely by the initial consonant and that the
meanings be clearly distinct and familiar to advanced speakers of
English as measured by the Common European Framework
(CEFR). These are randomized with an equal number of /r/ and
/l/ exemplars in each column. A rating sheet listed these words
side by side. This study recruited participants from an intensive
English program at a southeastern U.S. university program.
Method
Participants. Four L1 speakers of Korean and two L1 speakers
of Japanese were recruited. Four were female and two were male.
They self-identified as having problems both perceiving and
producing the English /r/ and /l/ sounds. These participants had
studied English for 9 to 13 years, and the total length of residence
in the United States ranged from 3 to 9 months. All spoke English
at an advanced level.
Procedures. The intervention took place in a soundproof room
in a university library. To assess perception, an L1 speaker of
American English recorded a list of the /r/ and /l/ minimal pairs.
RESULTS
Although this study did not focus on the relation between
production and perception, the data showed little correlation
between the ability to perceive the distinction and the ability to
produce it (see Figure 1). In this preliminary study, Participant 5
tied with Participant 3 for the highest score on the perception test,
correctly identifying 19 of 20 (95%) of the /r/ and /l/ minimal
pairs, yet had the lowest score on the preintervention production
test, reading only 4 of 20 (20%). Participant 2, on the other hand,
had the lowest score on the perception test, 14 of 20 (70%), but the
highest score on the preintervention production test, 20 of 20
(100%). This corroborated the research findings that L1 speakers of
Japanese can master the production of the /l/ and /r/ sounds
even if they cannot perceive it (Goto, 1971; Sheldon & Strange,
1982).
The small number of participants and the varied
preintervention production of the /l/ sound by the participants
made it impossible to determine statistical significance, but the
ability to pronounce the distinction improved markedly in the two
participants who had the most trouble producing the distinction.
Participant 5’s correct differentiation rose from 4 of 20 (20%)
correctly produced preintervention to 18 of 20 (90%) after 20
EXPANDED STUDY
The results of the pilot study suggested that an expanded study
using more participants and testing for more variables, including
the articulation of the /l/ sounding the word initial, medial, and
final position, and in blends, was warranted. An important change
was the development of an instrument that included the /l/
sound in the initial, medial, and final positions, including blends
such as blush. Although the length of a single instruction session
remained the same, participants were asked to attend three instead
212 TESOL Journal
of one. The hypothesis of the expanded study was that
participants who had failed to master the production of the /l/
sound in all positions (initial, medial, final, and blends) would
increase their production after three 30-minute teaching sessions.
As part of the design, analysis would exclude participants who
had already mastered pronunciation of the /l/ sound in the
pretest.
Instrument
A list of 25 minimal pairs was compiled that included 13 minimal
pairs with a word-initial contrast, six with a word final contrast,
and six with a word-medial contrast, including blends (Higgins,
2015). To evaluate the accuracy of the rater, the minimal pairs
were recorded by an L1 English speaker.
Method
Participants. Thirty-seven L1 speakers of Japanese, Korean,
Mandarin Chinese, Vietnamese, and Thai agreed to participate in
three 30-minute sessions spaced 1 week apart. Other instructors at
the intensive English program were aware of the alternative
articulation, but they did not incorporate it into their classes until
after the study was complete.
Procedures. As in the pilot study, participants were given a
sheet containing all the minimal pairs, this time with the /l/ and
/r/ sound in all positions. They listened a recording of a native
speaker of American English reading the words. They placed a
check in front of the word they heard—for example, light or right—
or marked unsure if they could not hear a difference. They then
read from a list of the minimal pairs, which were recorded on
iPads using Voice Recorder Pro.
To make certain all understood the correct articulation of the
alternative articulation, an explanation was translated into
Japanese, Korean, Chinese, Vietnamese, and Thai and projected on
a screen in front of the participants. Needless to say, participants
were made aware of the purpose of the intervention. During the
training sessions, the instructor conducted group activities but also
focused on each participant to ascertain that each had mastered
the correct articulation, with the tongue tip visible below the top
Acceptable Alternative Articulation of the English /l/ 213
front teeth. In addition to pronouncing a list of words containing
the /l/ sounds, participants again were asked to sing songs,
including the Christmas song that contains the fa, la, la, la, la, la, la,
la, la. Each session took less than 30 minutes.
At the end of all three teaching sessions, participants were
again presented with a sheet of paper that included only one
column from the minimal pair list. They then read and recorded
25 minimal pairs, 13 with the /l/ sound and 12 with the /r/
sound in word-initial, medial, and word final positions (see
Appendix B). These words included consonant blends such as
blush and brush. Half the participants were asked to read from one
column, the other half from the mirror image. One participant, for
instance, would read from the column with the word raw while
another read from the column with the word law. These were also
recorded onto iPads using Voice Recorder Pro. These pre- and
postintervention recordings were uploaded to a shared folder in a
random order so one could not tell which recording was made
first. A rater from the university’s applied linguistics master’s
program listened to and scored all the recordings on a minimal
pair sheet similar to that used by the participants. The minimal
pair list recorded by the L1 speaker of American English was
embedded among the recordings to assess the accuracy of the
rater; the L1 speaker of American English received a rating of
100% accurately pronounced.
Analysis
Of the 37 participants who started the study, 8 completed only one
teaching session and 12 completed only two teaching sessions.
That left 17 participants who completed all three sessions, 10 from
Japan, 2 from China, 4 from South Korea, and 1 from Thailand. Of
those, nine were female and seven were male. The mean time
studying English was 9 years, with a range from 7 years to 22
years.
Because data from the preliminary study found that people
who have mastered the alveolar articulation showed no
improvement, this study focused on those who had yet to master
the pronunciation. Because minimal-pair scoring contains just two
conditions, a perceived /l/ sound or a perceived /r/ sound, it is
214 TESOL Journal
conceivable that a random distribution could result in a score of
50%; for this reason, the study evaluated only those participants
whose score fell below that figure. Of the 17 participants who
completed all three teaching sessions, six met the criteria. Of these,
all showed an improvement in the ability to produce the /l/
sound, with no significant difference between the initial, medial,
and final positions or the blends (see Figure 3).
The data were analyzed using a Wilcoxon signed rank test, a
statistical hypothesis test that can be used for repeated
measurements in a sample from a small population. Because it is a
nonparametric test, it does not require divergence from a normal
distribution. The null hypothesis of this test was that the median
difference between the preintervention and postintervention
scores, expressed in percentages, is zero versus the alternative
hypothesis that the median difference is positive.
RESULTS
The results failed to confirm the assertions that perception and
production of the /r/ and /l/ contrast were unrelated (see
Figure 4). Nonetheless, although the results confirmed a
correlation between perception and production, the relationship
was not one-to-one. In other words, one could not assert from
these data that improved perception would necessarily result in
Figure 3. Pre- and post-intervention production of the /r/ and /l/ for
participants whose mastery of the articulation of the /r/ /l/ sounds fell
below 50 percent
Acceptable Alternative Articulation of the English /l/ 215
Figure 4. Perception and production of the English /l/ /r/ contrast
DISCUSSION
Although this intervention was decontextualized and performed
outside of traditional class time, the training sequence suggests
that the acquisition of the new articulatory could be incorporated
into ongoing classes using the model outlined by Levis and Grant
(2003), in which instructors target specific segmentals in one-on-
one sessions. The intervention has been successfully incorporated
into an intensive English program at the university where the
research was conducted. Anecdotally, several instructors have
commented on how quickly the students mastered a native-like
/l/ sound. It’s not unusual for Korean, Japanese, and Chinese
students to mention the mastery of this notoriously difficult
216 TESOL Journal
consonant contrast as among the most important skills they have
learned during in the program.
THE AUTHORS
Greg Raver-Lampman lectures at the English Language Center at
Old Dominion University in Norfolk, Virginia. He researches
methods to improve teaching both English as a second language
and English as a foreign language, and has presented in Japan and
published in China.
REFERENCES
Aoyama, K., Flege, J. E., Guion, S. G., Akahane-Yamada, R., &
Yamada, T. (2004). Perceived phonetic dissimilarity and L2
Please put a check by the word you hear. If you are not certain, mark
“unsure.”
APPENDIX B
# ___Name:___________________
Put a check by the word you hear. If you are not certain, mark
“unsure.”