Do Japanese ESL Learners Pronunciation Errors Com

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/228421264
Do Japanese ESL learners' pronunciation errors come from inability to articulate

or misconceptions about the target sounds?
Article · January 2010
CITATIONS READS
9 2,531
1 author:
Akitsugu Nogita
University of Victoria
24 PUBLICATIONS 24 CITATIONS
SEE PROFILE
All content following this page was uploaded by Akitsugu Nogita on 19 August 2016.
The user has requested enhancement of the downloaded file.

Do Japanese ESL learners’ pronunciation errors
come from inability to articulate
or misconceptions about target sounds?
Akitsugu Nogita
MA Student, University of Victoria Linguistics
[email protected]
This paper aims to examine whether Japanese English-as-a-Second-

Language (ESL) learners’ pronunciation errors are due to their inability
to articulate, or to misunderstandings of target phonemes and the
English phonological system. Four Japanese ESL learners read an
English passage and some particular segments were analyzed for errors.
After the analysis, each participant was interviewed about the errors.
Results showed that the participants often purposely pronounced the
same phoneme written with the same alphabet letter differently. For
example, <v> in “gave” and that in “traveler” were misunderstood to be
different phonemes. On the other hand, different phonemes spelled with
different alphabet letters were often purposely pronounced the same.
For example, <ar> in “hard” and <ir> in “first” were misunderstood to
be the same phoneme. In addition, participants did not understand the
whole picture of the English phonological inventory. I argue that the
participants’ mispronunciations are often due to the fact that they
reportedly had not yet been taught basic English symbol–sound
correspondence rules, not necessarily due to their inability to produce
particular sounds. Since letter knowledge precedes phoneme awareness,
the participants were not quite aware of English phonemes. If Japanese
ESL learners in general adopt the same behaviour, pronunciation
lessons need to pay more attention to Japanese ESL learners’
understanding of the basic English phonological system, not only to
what learners actually produce.
1 Introduction
This paper examines second language (L2) learners’ understandings of the

phonological system of a target language at the segmental level. The motivation
of this study comes from the author’s speculation that pronunciation errors by L2
learners are not necessarily errors, but instead may result from learners’ own
interpretations of L2 phonology due to learners’ language backgrounds in which
they have not yet been taught the phonological system of their target language.
For example, if a Japanese English-as-a-Second-Language (ESL) learner
83
mispronounces “change” as [tʃɛndʒ]1 when it is supposed to be pronounced as

[tʃendʒ], it would be careless to immediately conclude that the learner has
difficulty in the distinction between the tense vowel [e], as in “pain,” and the lax
vowel [ɛ], as in “pen,” because the learner’s first language (L1) does not have
this tense and lax distinction. There is a possibility that the learner has not yet
been taught the basics of the English vowel system, and the learner
misunderstands that “change” is supposed to be pronounced as [tʃɛndʒ]. If this is
the case, the learner has yet to attempt to pronounce “change” as [tʃendʒ], and
there is no way to know whether the learner has difficulty with [e]. Once the
learner has been taught to pronounce “change” as [tʃendʒ], the learner might
produce it without difficulty. Apparently, in the area of L2 pronunciation,
analyses of learners’ understandings of L2 phonology are less common than
phonetic or phonological analyses of what learners actually pronounce. To
address this gap in the research, this paper examines whether Japanese ESL
learners have proper knowledge of English sounds when they commit
pronunciation errors.
2 Background
2.1 Literature review
L2 learners’ pronunciation errors are caused by factors other than difficulty in

production. One possible factor is cognitive skills. For example, according to a
finding of Fraser’s (in press) study of /l/ and /ɹ/ distinction by Asian ESL
learners, the participants produced these sounds without much difficulty, despite
the fact that participants were unaware that /l/ and /ɹ/ were two different
phonemes, which change lexical meaning in English.
Another possibility is what Richards (1971) called “false concepts” and
what Stenson (1974) termed “induced errors” (as cited in Brown, 2007). These
are errors caused by misleading teaching. In fact, Suarez and Tanaka’s study
(2001) with 88 Japanese college students found that 40% of the students claimed
that their pronunciation problems came from a lack of pronunciation instruction
in their six years of English curriculum in junior and senior high school. Another
24% felt psychological barriers had hampered correct pronunciation: when
students try to pronounce English accurately, they are afraid of being teased or
they feel embarrassed. From a teacher’s perspective, according to Muroi’s (2005)
observation in Summer Teacher Training, about 30% of the Japanese teachers of
English answered that they had never taught pronunciation to their students.
1
Phonemic transcriptions of English vowels in this paper are based on “American
English “R-Colored” Vowels as Complex Segments” Green (2001).
Working Papers of the Linguistics Circle of the University of Victoria 20, 82–116
© 2010 Akitsugu Nogita
84
More specifically, Avery & Ehrlich (2003) explained an example of induced

errors. They suggest that many mispronunciations by Portuguese speakers come
from the influence of the Portuguese spelling system rather than from difficulty
producing particular sounds; teachers with Portuguese students often familiarize
themselves with the Portuguese spelling system. According to Bayraktaroğlu
(2008), in terms of L2 learners’ pronunciation errors, L1 orthographic
interference and L1 phonological interference are completely different; the
former is differences of one-to-one letter–sound correspondence between L1 and
L2, while the latter is differences in the sound systems. The Japanese writing
system has Romaji, Japanese Romanization, in which the symbol–sound
correspondence rules are quite different from those in English in many respects.
If Japanese learners of English familiarize themselves with the sound–spelling
correspondences of Japanese Romanization similar to Portuguese learners of
English, then their pronunciation errors may be a result of orthographic influence.
Moreover, L2 learners may need orthographic knowledge of the target
language in order to understand its phonological system. Siegel and Wade-
Woolley (1997) stated that phonological processing and literacy are strongly
related. According to Carroll’s (2004) study about first language (L1) acquisition,
letter knowledge precedes phoneme awareness; letter learning helps children
learn to separate phonemes from phonetic contexts and identify the same
phoneme in different words. As well, according to Cook (2004), English speakers
may understand that an alphabet character corresponds to an individual phoneme.
When the number of alphabet characters and the number of phonemes do not
match, for example “month” (five characters) and /mʌnθ/ (four phonemes), adult
English speakers try to reconcile the contradictions in information, which results
in difficulties in identifying the number of phonemes in the words (C. Pytlyk, in-
class presentation). Furthermore, Goble (2002) revealed that his participants,
Japanese college students, astonishingly lacked awareness that English loanwords
in Japanese and their English counterparts are different entities, and the students’
pronunciation and spelling errors showed an inordinate amount of loanword
influence. Since Japanese has many loanwords from English, Japanese ESL
learners’ mispronunciations may come from loanword interference, rather than an
inability to articulate. If Japanese ESL learners are not taught the English spelling
system, they might be deficient in phonological awareness in English as well. It
is worth examining this possibility.
85
2.2 Research questions
The present study is designed to address the following two questions:

1) When Japanese ESL learners make pronunciation errors, do they try to
pronounce target sounds but fail to articulate the targets, or do they
misunderstand or not know the target sounds to begin with? I specifically
examine some segmental errors, namely the consonants /θ/, /ð/, /v/, /l/, /ɹ/, /f/, /h/,
the distinctions between /s/ and /ʃ/, /t/ and /tʃ/, and /d/ and /dʒ/ before high front
vowels, and the vowels /æ/, /ɛ/ and /e/.2
2) If Japanese ESL learners do not have knowledge of target sounds, what
lessons do they need to understand target sounds properly?
3 Methods
3.1 Participants
There were four participants, labelled as P1, P2, P3 and P4. They were all
Japanese ESL learners in British Columbia, Canada. Like the majority of
Japanese people, all of the participants had studied English in junior and senior
high school for six years. All of them claimed that they were not confident with
their pronunciation, nor could they read the International Phonetic Alphabet
(IPA), except P4, who was trying to learn the IPA on her own. P1 had been
taught English by her mother who spoke British English, so P1’s pronunciation
might have been influenced by this exposure to the British accent. P1 was
working, P2 was in a lower-intermediate class, and P3 and P4 were in an
intermediate English class in an ESL school. Table 1 summarizes a number of the
participants’ traits.
Table 1. Summary of participant characteristics. Nb: Kansai= Osaka and Kyoto area.
Kanto= the area around Tokyo.
P1 P2 P3 P4
Age: 33 28 20 19
Gender: F F F F
Length of residence in Canada: 3 years 5 months 5 months 5 months
Home region in Japan: Kansai Kanto Kanto Kansai
2
These segments are often considered problematic sounds for Japanese ESL learners
(Avery et al, 2003; Ohata, 2004; Taniguchi, 2009).
86
3.2 Stimulus and procedure
The experiment took place over two days. On the first day, the participants read
an English passage and Japanese nonsense words, and were asked to complete
four phonemic contrast identification tasks. After their recordings were analyzed,
I later talked to each participant individually about the results of the analysis. On
the first day, the participants read the English passage, “The North Wind and the
Sun” from the Handbook of the International Phonetic Association (p.44)
(reproduced in the Appendix).
This reading task is designed to examine how participants interpret the
segments mentioned in the research question 1. This story is familiar to many
Japanese speakers and I expected that the participants would feel more
comfortable with a familiar story than an unknown story. As well, this passage is
commonly used in phonetic demonstrations. I handed the participants a sheet of
paper with the passage on it a few minutes before recording, so they did not have
time to ask native speakers about pronunciation or to check a dictionary.
However, I taught them the sounds and meanings of presumably new words, such
as “oblige” and “cloak.” After practicing a couple of times, they were recorded.
Recording was done in the Phonetics Lab at University of Victoria with a Luna
1.1 inch large diaphram condenser microphone, M-Audio Firewire 410, with
PRAAT set to 44100 Hz.
Participants also recorded 10 nonsense Japanese words written with the
Katakana syllabary. Table 2 shows the stimuli words presented to each
participant. Some segments mentioned in Research Question 1, such as /l/ and
/ɹ/, are obviously not distinctive phonemes in Japanese. Conversely, the contrasts
in Table 2 are sometimes considered problematic (Avery & Ehrlich, 2003; Ohata,
2004) although these contrasts are also sometimes considered to exist in Japanese
(Matsuzki, 1993; Inozuka, 2009). This task was designed to ascertain whether
Japanese ESL learners have to articulatorily practice the distinctions between /s/
and /ʃ/, /t/ and /tʃ/, and /d/ and /dʒ/ before high front vowels, and the distinction
between /ɛ/ and /e/, or if Japanese ESL learners can economically utilize L1
distinctions for these L2 distinctions.
87
Table 2. The 10 nonsense Japanese stimulus words. Nb: In Japanese [ti] and [tʃi] are
distinctive, and so are [di] and [(d)ʒi] (Matsuzaki, 1993). Whether [ʃi] and [si] are
distinctive is debatable; I argue that it depends on lexical classes.
ティー [tiː] チー [tʃiː]3

ディー [diː] ジー [(d)ʒiː] ズィー [(d)ziː]4
シー [ʃiː] スィー [siː]
ベタ [beta] ベータ [beːta] ベイタ [beita]
The words were aligned in this order on the sheet from which the participants
read. They were asked to pronounce the words in a natural Japanese way. Since
Japanese has phonemic pitch patterns,5 which are not shown with regular
orthography, most of the participants asked me about what pitch pattern they
should use. Then, I answered that they could use whichever they felt was natural.
Participants also completed four phonemic contrast identification tasks.
They were shown the homophones and minimal pairs in Table 3 and were asked
to identify whether the words in each pair were the same or different in
pronunciation. They were also asked to identify the difference between any two
words they felt were not homophones. For example, I asked, “Do you know
whether ‘meat’ and ‘meet’ are the same or different in pronunciation?” If
participants confidently answered, “Yes, they are the same,” I gave them a credit.
If they showed uncertainty, “Um, I’m not sure. Maybe the same?” I did not give
them a credit even if the answer was right as this may have been accidentally
correct. The purpose of these tasks was to examine their L2 phonological and
orthographic awareness in general. Since this task did not involve production, I
could focus my investigation on the participants’ understanding. The reason I
chose these pairs is that each pair of words would likely be pronounced in the
same way by Japanese speakers as a result of loanword adaptation processes. I
tried to examine whether the participants could identify the phonemic structure of
each word without being distracted by Japanese loanword adaptation.
3
The transcription of the Japanese voiceless lamino-alveolo-palatal fricatives vary
between [ʃ] and [ɕ] for the voiceless one, and between [ʒ] and [ʑ] for the voiced one (Pan,
Utsugi and Yamazaki 2004). In this paper, I use [ʃ] and [ʒ] in order to be consistent with
the English counterparts.
4
In Japanese [dʒ] and [ʒ] are allophonic variations of one phoneme, and so are [dz] and
[z] (Inozuka & Inozuka, 2009).
5
Japanese is known as a pitch accent language in which pitch is the primary indicator of
accent (stress) (Avery & Ehrlich, 2003).
88
Table 3. Phonemic contrast identification tasks
“meat” versus “meet”

“ear” versus “year”
“bone” versus “born”
“who’d” versus “food”
On the second day, I conducted semi-structured interviews with each

participant in their L1, Japanese. Each interview lasted 30 to 40 minutes. I mainly
asked participants what sound they actually tried to pronounce. This interview
was designed to examine whether the participants’ pronunciation errors were due
to mis-articulation of intended target or misunderstanding of target. I used this
method in order to best ascertain each participant’s understanding in a
straightforward manner. Below are examples of the questions:
• “Did you try to pronounce [θ] in ‘north’?”
• “Did you try to pronounce [v] in ‘gave,’ ‘of,’ and ‘traveler’?””
• “You pronounced [v] in ‘gave’ but pronounced [b] in ‘of.’ Can you
come up with any reason?”
• “You said you did not try to distinguish /l/ and /ɹ/, and in fact you mostly
did not. However, you pronounced native-like [ɹ] in ‘wrap.’ Can you
come up with any reason?”
Below are examples of participants’ answers:
• “Yes, I tried to pronounce [θ] to distinguish it from [s].”
• “No, I didn’t put any extra effort into /θ/ and just pronounced [s] just as
Japanese speakers commonly do.”
When a participant mispronounced /θ/ as [s] and answered that she tried to
pronounce [θ], her error was analyzed as mis-articulation of the intended target.
When a participant mispronounced /θ/ as [s] and answered that she intended to
pronounce [s], her error was analyzed as misunderstanding of target.
3.3 Analysis of the recordings
The transcribers were a phonetically trained native speaker of North American

English and myself, a native Japanese speaker and trained phonetician. For the
English data, the transcription was based on whether or not they produced the
target phoneme; if a segment produced by a participant was obviously accented
but was still easy to understand, it was considered correct. If a participant’s
production sounded like a different phoneme or was difficult to be categorized as
any English phoneme, we transcribed what they actually pronounced. For the
Japanese data, the transcribers judged what English phonemes could be
represented by their productions.
89
4 Results and discussion
Overall, in terms of English phonemes that are generally considered problematic

for Japanese ESL learners, participants quite often intended to pronounce
different sounds. In other words, their mispronunciations often came from their
understanding of what phonemes they were supposed to produce. More
interestingly, the participants’ own interpretations of English phonology varied
much more than I expected from sound to sound, from word to word, and from
individual to individual. Below are the details of each phoneme.
4.1 Results for /θ/
There are four occurrences of /θ/ in the passage, all of which are in the word
“north.” All the participants realized /θ/ as [s]. However, what is important in
this paper is not the productions themselves, but whether the participants knew
that the target was /θ/. For example, if P1 misunderstood that the target was /s/,
instead of /θ/, she actually did not attempt to pronounce [θ]. In this case, I would
conclude that she tried to pronounce [s] four times and successfully produced it
four times. Consequently, what a teacher would then want to consider is teaching
the proper target phoneme as it is identified orthographically, rather than
articulation of [θ]. Therefore, I asked the participants if they knew that the target
was /θ/, or if they intended to pronounce the dental fricative [θ].
• P1 reported she had never tried to pronounce [θ] although she knew that
/θ/ and /s/ should be different. Therefore, she actually intended to
pronounce [s] and she successfully produced what she was aiming to
produce. In this case, there is no way to know if she was able to produce
[θ] at the time of the recording because she had not attempted it.
• P2 reported she knew that the target was /θ/ and tried to pronounce it.
• P3 reported she knew that the letters <th> sounded different from the
letter <s>, but she was more influenced by the English loanword in
Japanese “ノース” [noːsɯ], which means “north.” Since her underlying
representation was /noːsɯ/, but not /nɔɹθ/, there is no way to examine if
she was able to produce [θ] in this experiment.
• P4: Like P2, P4 reportedly tried to pronounce [θ].
The table below shows how many times the participants intended to produce the
target phoneme and how many times they did so. Since P1 and P3 did not intend
to pronounce [θ], the number of “Intended” is 0.
90
Table 4. Results for /θ/. Nb: “Occurrences: T”= The total occurrences in the passage;
“Occurrences: Int”= How many times the participants intended to pronounce the target
sound; “Correct: Int”= The number of correct productions when the participants intended
to pronounce the target; “Correct: Acc”= The number of accidentally correct productions.
Target: /θ/
Occurrences Correct Incorrectly pronunciations
T= Int Int Acc (sound: occurrences)
P1 4 0 0 0 [s]: 4
P2 4 4 0 0 [s]: 4
P3 4 0 0 0 [s]: 4
P4 4 4 0 0 [s]: 4
4.2 Results for /ð/
There are 23 occurrences of /ð/ in the passage. The following summarizes

participants’ comments.
• P1: Just like [θ], she had never tried to pronounce [ð] and always realized
it as the Japanese /z/ although she knew that /ð/ was not the Japanese /z/.
In fact, she pronounced both [dz] and [z], which are allophonic variations
of the Japanese /z/.
• P2 reported that targets were different from /z/ because the target was
spelled as <th>, but not <z>. She said that she tried to distinguish
between /ð/ and /z/ although she was not quite sure of the sound quality
of /ð/. In fact, all of her /z/s were consistently pronounced as [z] or a
somewhat devoiced [z]; she pronounced [ð] five times. She also
pronounced the dental stop [d̪ ] nine times, which was close to the target
/ð/ but was categorized as /d/. In fact, since the combination of
phonological features [+continuant] and [−sonorant] is relatively
difficult, stopping or affricating a fricative such as /ð/ to [d] is common
in child language (Bernhardt & Stemberger, 1998).
• P3 reported definitely trying to distinguish between /ð/ and /z/.
Moreover, she was aware that when function words, such as “the,” “that”
and “than” were followed by a difficult word in pronunciation, such as
“traveler,” she ignored the correct pronunciation of [ð] in order to
concentrate her effort on the next word. Therefore, like P2, she
understood the target correctly and she was aware of her errors.
• P4: Like P3, P4 reported trying to distinguish between /ð/ and /z/, which
means she understood the target correctly. The big difference from P3 is
91
that she did not quite realize her production was not [ð], but its
characteristic was much like the Japanese /z/.
Incidentally, all the participants did not know the cross-linguistic phonetic
difference of /z/ between English and Japanese; the default form of J/z/ is the
affricate [dz] (Grenon, 2008). Nevertheless, P1 produced the pure fricative [z]
more often than [dz]. I will mention this phonetic issue in §4.13.
The table below shows how many times the target actually occurs in the
passage, how many times the participants intended to produce the proper target,
how many times the participants correctly pronounced the target when intending
to do so, and how many times the participants accidentally pronounced the target.
Also summarized are the incorrectly pronounced sounds of each participant.
Since P4 missed the word “the” in the passage on one occasion, her occurrences
were counted as 22. The question mark beside the number in the column of
“Occurrences: Intended” means that the participant was not sure if she really
intended to pronounce the target.
Table 5. Results for /ð/; Int= intended, Acc= accidental. Nb: In Japanese, [dz] and [z] are
allophonic variations of the phoneme /z/ (J/z/) (Tsuzuki, 1996; Grenon, 2008; Vance,
2008; Inozuka and Inozuka, 2009). [d̪ ] is more dental than [d]. “?” indicates that the
participants themselves were not really sure if they were aware of the target sounds.
Target: /ð/
P1 23 0 0 0 [z]: 20 [dz]: 3
P2 23 23? 5 0 [d̪ ]: 9 [dz]: 7 [z]: 2
P3 23 23 4 0 [d̪ ]: 8 [dz]: 8 [z]: 3
P4 22 22 0 0 [dz]: 11 [z]: 11 [ts]: 1
4.3 Results for /v/
There are six occurrences of /v/ in the passage: “gave,” “of,” and four occurences
of “traveler.”
• P1 reported she did not distinguish between /v/ and /b/, and substituted
/v/ with the Japanese /b/. In fact, she pronounced both [β] and [b], which
are allophonic variations of the Japanese /b/. Her /v/ in “of” sounded like
[v], but it was actually the weakened version of /b/,6 suggesting she
accidentally hit the target.
6
[β], [b] and [v] are all allophonic variations of /b/ in Japanese (Inozuka et al., 2009).
92
• P2: Due to having learned “gave” and “of” in Japan, P2 incorrectly

memorized that “gave” and “of” were [geib(ɯ)] and [ob(ɯ)]
respectively. Therefore, she misunderstood that the target sound was /b/,
instead of /v/. As for “traveler,” she reportedly knew that there was [v],
but she ignored the correct pronunciation of [v] in order to concentrate
her effort on [l] and [ɹ].
• P3: As for “traveler” and “of”, she misunderstood that /v/ in these words
was /b/. She explained that since she quite frequently used the word
“traveler” in her conversation in Canada, she made up her own way to
pronounce it. As for “of,” just like P2, she was not aware that “of” has a
[v]. She pronounced [v] in one of the occasions of “traveler” but it was
actually the weakened version of /b/, similar to P1. As for “gave,” she
tried to pronounce [v] and successfully pronounced it. Therefore, she got
one correct [v] out of one attempt.
• P4 reported possibly trying to pronounce [v] in “gave.” However, she
actually pronounced this with a [b]. As for “traveler” and “of,” she
consciously used the Japanese /b/. She mentioned that she concentrated
too much on [l] and [ɹ] in “traveler” and could not afford to think about
[v], just like P2.
Most of the time, the errors regarding /v/ were due to inappropriate intentions,
rather than production problems. Only one instance, by P3, was appropriately
intended and correctly produced.
Table 6. Results for /v/; Int= intended, Acc= accidental. Nb: “?” indicates that the
participants were not really sure if they were aware of the target sounds.
Target: /v/
P1 6 0 0 1 [β]: 3 [b]: 2
P2 6 0 0 0 [β]: 5 [b]: 1
P3 6 1 1 1 [β]: 4
P4 6 1? 0 0 [β]: 5 [b]: 1
4.4 Results for /l/ and /ɹ/ in onset position
Japanese ESL learners often spend much time attempting to acquire the contrast
between the North American English /l/ and /ɹ/ (E/l/ and E/ɹ/) because Japanese
has only one liquid, /ɾ/, that can appear as [l] and even [ɹ] allophonically or in
quasi-free variation (Magnuson, 2008). This distinction is so extensively studied
that I put /l/ and /ɹ/ in the same section. From the point of view of this study, I
93
will give more credit to the participants who tried to distinguish between them
but did not quite hit the target than to those who did not try to distinguish
between them but accidentally hit the target. In examining whether the
participants’ attempts to distinguish /l/ and /ɹ/ affected the sound quality, I
categorized their productions along two parameters following Magnuson (2008):
rhoticity versus laterality and central oral stricture. 7 In this scheme, [l] is lateral
and narrow while [ɹ] is rhotic and open.
There are 17 occurrences of /l/ in onset position, including consonant
clusters, and 10 occurrences of /ɹ/ in onset position, including consonant clusters.
• P1 reported not trying to distinguish between /l/ and /ɹ/ in onset position
at all. In fact, she almost consistently used rhotic liquid for both /l/ and
/ɹ/. Interestingly, according to Magnuson (2008), J/ɾ/ is most commonly
realized as a raised alveolar flap, but P1 pronounced [ɹ] much more often
than a flap. The study by Akahane-Yamada, Aoyama, Fledge, Guion and
Yamada (2004) showed that Japanese ESL learners more successfully
acquire E/ɹ/ than E/l/ because the difference between E/ɹ/ and J/ɾ/ is
perceptually more salient than the difference between E/l/ and J/ɾ/. In this
way, P1 acquired E/ɹ/ and over-generalized it for /l/.
• P2 reported trying to distinguish between E/l/ and /ɹ/. In fact, she quite
consistently pronounced more rhotic and open sounds, namely [ɹ] and [ɾ̞],
for /ɹ/ while pronouncing lateral and narrow sounds, namely [l] and [ɺ],
for /l/.
• P3 reported knowing that /l/ and /ɹ/ were supposed to be distinguished;
however, she abandoned this contrast in her inter-language due to her
low self confidence. In fact, she almost uniformly used rhotic and open
sounds, namely [ɹ] and [ɾ̞], and the flap [ɾ] for both /l/ and /ɹ/.
Interestingly, she pronounced an accurate [ɹ] in “wrap”. However, she
mentioned that the <w> in the spelling of “wrap” encouraged her to
round her lips, which accidentally resulted in quite native-like [ɹ]. She
was sure that she would pronounce the homophone “rap” with flap [ɾ].
Therefore, her [ɹ] was actually an accidental production caused by her
misconception of the spelling and English phonotactics8 where the
sequence of /*wɹ/ at word-initial is not allowed.
• P4: Like P2, P4 said that she tried to distinguish between E/l/ and /ɹ/. In
fact, she more successfully distinguished /l/ from /ɹ/ than P1 and P3. She
pronounced [ɹ] better in “traveler” and “agree” than other words. She
mentioned that she more frequently used the words “travel” and “agree”
7
“Rhoticity” is “[ɹ]-like quality,” while “laterality” is “[l]-like quality.” “Central oral
stricture” is how narrow or wide the space in the oral cavity is (Magnuson, 2008).
8
Phonotactics deals with restrictions in a particular language on the permissible
combinations of phonemes.
94
than “strong” or “around,” and she felt more comfortable pronouncing [ɹ]
in familiar words.
Interestingly, P1 and P3, who did not try to distinguish /l/ from /ɹ/, performed
notably worse with /l/ than P2 and P4, who tried to distinguish these. Conversely,
P1 and P3 performed quite well with /ɹ/. Akahane-Yamada et al.’s finding that
Japanese ESL learners acquire [ɹ] earlier than [l] might apply only to those who
do not intend to distinguish /l/ from /ɹ/. Once they try, they might acquire [l]
earlier than [ɹ] because [l] is less marked than [ɹ]. Moreover, English speaking
children typically acquire [l] earlier than [ɹ] (Vihman, 1996).
Table 7. Results for /l/ in onset position; Int= intended, Acc= accidental. Nb: Ø= no
consonant; [l] = the alveolar lateral; [ɺ] = the alveolar lateral flap; [ɾ] = the alveolar flap;
[r] = the alveolar trill; [ɾ̞] = the lowered flap (the tongue does not quite touch the roof of
the mouth); [ɹ] = the alveolar rhotic approximant.9
Target: /l/
P1 17 0 0 2 [ɹ]: 8 [ɾ]: 2 [ɾ̞]: 2 [ɺ]: 1 Ø: 2
P2 17 17 13 0 [ɺ]: 3 [ɾ̞ ]: 1
P3 17 0 0 1 [ɾ̞ ]: 6 [ɹ]: 5 [ɾ]: 5
P4 17 17 12 0 [ɾ]: 3 [r]: 2
Table 8. Results for /ɹ/ in onset position; Int= intended, Acc= accidental.
Target: /ɹ/
P1 10 0? 0 9 [ɾ]: 1
P2 10 10 6 0 [ɾ]: 3 [ɾ̞ ]: 1
P3 10 0 0 4 [ɾ]: 4 [ɾ̞ ]: 2
P4 10 10 5 0 [ɾ]: 5
9
According to the studies by Tsuzuki (1996), Magnuson (2008), and Inozuka et al.
(2009), all of these sounds are possible allophones of J/ɾ/.
95
4.5 Results for velarized /l/ in coda position
• P1 reported not thinking about the difference between /l/ and /ɹ/ at all, so
her correct production is considered to be accidental.
• P2 reported attempting to pronounce [l] but did not pronounce it
successfully.
• P3 reported not attempting to distinguish between /l/ and /ɹ/. Just like
onset position, she used the rhotic sound [ɹ] for /l/.
• P4 tried to pronounce [l] but actually produced [ɾ], which is neither rhotic
nor lateral (Magnuson, 2008).
All in all, their realizations of /l/ in coda position seems the same as those in
onset position; however, both P2 and P4 failed to pronounce [l].
Table 9. Results for velarized /l/ in coda position/; Int= intended, Acc= accidental, Ø= no
consonant.
Target: /l/
P1 1 0 0 1
P2 1 1 0 0 Ø: 1
P3 1 0 0 0 [ɹ]: 1
P4 1 1 0 0 [ɾ]: 1
4.6 Results for /ɹ/ in coda position
/ɹ/ occurs 20 times in coda position. I divided them into three smaller groups
based on the preceding vowel: /ɔɹ/ as in “north,” /ɑɹ/ as in “hard,” and /ɝ/ in both
a stressed syllable (as in “first”) and an unstressed syllable (as in “stronger”). /ɔɹ/
occurs 8 times, /ɑɹ/ occurs once, and /ɝ/ occurs 11 times.
We will first examine /ɔɹ/ separately from /ɑɹ/ and /ɝ/ because the
participants behaved interestingly. /ɔɹ/ occurs in three different morphemes,
“north,” “warm,” and “more” in the passage, and all the participants consistently
pronounced /ɔɹ/s in three different ways depending on the morpheme as shown in
Table 10 below.
• P1 reported not really being aware that she pronounced “or” in “north”
and <ar> in “warm” differently. However, she mentioned that she
consciously pronounced <or> in “north” differently from the others
because she was influenced by the pronunciation of Japanese teachers of
English.
96
• P2 reported not knowing that <or> in “north,” <ar> in “warm” and <or>
in “more” were phonemically the same, and she was influenced by
English loanwords. However, she was taught the pronunciation of
“warm” by a native speaker in an ESL school, so she pronounced only
“warm” and “warmly” correctly. Therefore, she was actually able to
produce [ɔɹ] but misunderstood that <or> in “north” and <or> in “more”
were not [ɔɹ]. In other words, she could not generalize the skill of
pronouncing [ɔɹ] to words other than “warm.”
• P3: As mentioned in §3.1, she pronounced “north” in the same way as
the loanword [noːsɯ]. As for “warm”, she misunderstood that <ar> in
“warm” might be more like <ar> in “hard.” As for “more,” she
pronounced it acceptably. However, she mentioned that, when facing
<r>, she became intimidated and sometimes pronounced it strangely.
Therefore, [ɔɹ] in “more” was counted as an accidentally correct
production.
• P4 reported intending to pronounce the /ɔɹ/ in three different ways,
consistent with what she had been taught in junior high school.
According to the participants’ feedback, the three different realizations of /ɔɹ/ are
not caused by phonetic environments. Rather, they are misconceptions that the
/ɔɹ/ in all three instances was supposed to be different.
Table 10. Results for /ɔɹ/; Int= intended, Acc= accidental. Nb: “?” indicates that the
participants themselves were not really sure if they were aware of the target sounds.
Target: /ɔɹ/
P1 8 0 0 0 [oʊ] in “north” [oː] in “warm”
[oa], [oə] in “more”
P2 8 2 2 0 [oː] in “north” [oa] in “more”
P3 8 0 0 2 [oː] in “north” [ʌː] in “warm”
P4 8 0 0 0 [oː] in “north” [aː] in “warm”
[oə] in “more”
Note that the quality of the Japanese /a/ (J/a/) is between the cardinal vowels [a]
and [ɑ], and it has a wider range of allophonic variations than the other Japanese
vowels; [a], [ə], [ʌ] and [ɑ] can all be allophones of J/a/ (Tsuzuki, 1996). The
participants produce J/a/ as [a], [ə], and [ʌ]. In addition, English loanwords in
Japanese, “north,” “warm” and “more” are typically adapted into [noːsɯ],
[woːmɯ] and [moa] respectively.
Next, I will examine /ɑɹ/ and /ɝ/. Although /ɑɹ/ occurred only once in the
97
passage, it is interesting to compare /ɑɹ/ and /ɝ/.

• P1 reported not knowing that <ar> in “hard” and <ir> in “first” were
different. In addition, she did not know that there was [ɹ] in those words
(in a rhotic dialect). Although P1’s mother, who has a British English
background, taught her English when she was young, she did not know
<ir> in “first” and <er> in “consider” or “other” were different in British
English.
• P2: Like P1, P2 reported not knowing that <ar> in “hard” and <ir> in
“first” were different. She said that she could produce the [ɹ]-like sound
if she tried, as shown in §4.4 and §4.6, but she also insisted that [ɹ] in
onset and coda sounded like completely different entities for her. This
can be explained in terms of Brown’s (2000) finding that Japanese
subjects perceived E/l/ and E/ɹ/ in onset with only 31% accuracy while
they did so in coda with nearly 100% accuracy (cited in Archibald,
2005). P2 also misconceived that the English letter <r> in coda position
was the same as the Japanese symbol <ー>, which phonemically
lengthens the preceding vowel. In fact, she quite consistently pronounced
the long vowel [aː] for both /ɑɹ/ and /ɝ/. She misinterpreted the English
orthographic information and did not know that English lacks the
contrast of vowel length, unlike her L1.
• P3 reported not knowing <ar> in “hard” and <ir> in “first” were
different. In fact, she pronounced [ɝ] in “hard” as she intended.
However, as for the word “were,” she said that she ignored [ɹ]. She
pronounced “were” as [waː], as she intended. My interpretation of this is
that she attempted /ɝ/ ten times out of 11 occurrences and succeeded four
times. The problem is that she was not sure if /ɝ/ in each word in the
passage was phonologically the same.
• P4: Like P2, she reported misconceiving that the letter <r> played the
role of phonemically lengthening its preceding vowel. She also did not
know that English lacks the contrast in vowel length present in her L1. In
fact, she pronounced “hard” as [haːd] where [a] was lengthened, just as
P1 and P2 did, while she pronounced <ir> “first” as [ɑɹ]. However, she
believed that Japanese speakers can pronounce [ɹ], so that she was not
intimidated by [ɹ]. She might have been aware that [ɹ] can appear as an
allophone of J/ɾ/. The problem is that she did not know that /ɝ/ in each
word in the passage was phonologically the same. In fact, the phonetic
quality of her /ɝ/ varied from occasion to occasion even when her
production was within the phonologically acceptable range. She
mentioned that she pronounced /ɝ/ instinctually. Therefore, it may not be
appropriate to consider that she attempted to pronounce /ɝ/.
All in all, the errors regarding post-vocalic /ɹ/ were mostly due to inappropriate
intentions.
98
Table 11. Results for /ɑɹ/; Int= intended, Acc= accidental.
Target: /ɑɹ/
P1 1 0 0 0 [aː]: 1
P2 1 0 0 0 [aː]: 1
P3 1 0 0 0 [ɝ]: 1
P4 1 0 0 0 [aː]: 1
Table 12. Results for /ɝ/; Int= intended, Acc= accidental.
Target: /ɝ/
P1 11 0 0 0 [aː]: 9 [ɑː]: 1 [a]: 1
P2 11 0 0 0 [aː]: 9 [ə]: 2
P3 11 10? 4 0 [ə]: 4 [aː]: 1 [a]: 1 [oʊ]: 1
P4 11 0? 0 4 [aː]: 6 [ɑɹ]: 1
Again, [a], [ə] and [ɑ] are possible allophonic variants of J/a/. Note also that
vowel length is phonemic in Japanese: e.g. /soɾi/ (sled) versus /soːɾi/ (Prime
Minister).
4.7 Results for /f/ and /h/
It is important to note that Japanese has phonemic contrast between /f/ and /h/
which is neutralized before the vowel /ɯ/. Also, phonetically J/f/10 is the bilabial
fricative [ɸ] (Vance, 2008; Inozuka et al., 2009). In a questionnaire administered
to 13 experienced ESL teachers in British Columbia, Canada, one respondent
(and advanced ESL level instructor) pointed out the /f/ and /h/ distinction as one
of Japanese learners’ problems. As well, Berman, Lambacher, Martens, & Nelson
(2001) found that Japanese learners perceptually confused /f/ and /h/ before [u].
Therefore, it is worth examining it. /f/ occurs five times and /h/ occurs eight
times in the passage. The results show that the contrast between E/f/ and E/h/
does not seem problematic, except for “fold” and “who.”
10
Vance (2008) phonemicized the Japanese bilabial fricative as /f/. In this paper, I follow
Vance’s method. (cf. Akamatsu, 2000)
99
•
P1 reported being aware of the cross-linguistic phonetic difference
between E/f/ and J/f/. She pronounced the labio-dental [f], except <f> in
“first” was [ɸ]. [ɸ] is more marked than [f] (Maddieson, 1984, 2005).
Based on Eckman’s Markedness Differential Hypothesis (2003), if one’s
L1 has a more marked sound, the less marked counterpart in L2 is not
difficult to acquire. Therefore, her acquisition of [f] is not surprising
although [ɸ] still appeared. She pronounced <f> in “fold” as [h], and she
said that it was a slip of tongue. E/f/ is sometimes adapted as both J/f/
and J/h/ in loanwords: e.g. “telephone” can be pronounced and written as
either /teɾefoN/ or /teɾehoN/11 (Matsuzaki, 1992, 1993). She might have
been influenced by that. As for “who,” she did not know that <wh> in
“who” and <f> in “food” were different. Therefore, she simply
transferred L1 phonetics and phonotactics, namely neutralization of /h/
and /f/ before /ɯ/, and ended up with [ɸ] in “who.”
• P2 reported not being aware of the phonetic difference between E/f/ and
J/f/. Therefore, she was going to pronounce the bilabial [ɸ] and
consistently did so. However, it was still within the acceptable range of
E/f/. As for “who,” she did not know that <wh> in “who” and <f> in
“food” were different, like P1. J/f/ (or /h/) before a high back vowel was
typically pronounced as [ɸ], similar to P1, but she happened to produce
[h], or weakened [ɸ], in “who.” Therefore, I consider it accidental.
• P3: Like P1, P3 reported being aware of the phonetic difference between
E/f/ and J/f/. However, she pronounced “fold” as “hold,” just as P1 did.
The difference from P1 is that P3 more consistently produced [f] than P1,
but she simply misread “fold” as “hold” and intended to pronounce
“hold.” As for <wh> in “who,” she did not know that it was different
from <f> in “food,” like P1 and P2. Therefore, she simply transferred L1
phonetics and phonotactics, like P1.
• P4: Like P1 and P3, P4 reported being aware that E/f/ was not [ɸ], and
she pronounced [f] in some words. However, she pronounced [ɸ] in
“off.” She said that she pronounced easy words, like “off,” in the
Japanese way, whereas she was careful with relatively difficult words.
However, her [ɸ] was still phonologically within E/f/. The problem is
that she purposely pronounced <f> in “off” and <f> in the other words
differently, when English does not have this contrast. As for <wh> in
“who,” she did not know that it was different from <f> in “food,” like all
the other participants. P4 misconceived that <wh> in “who” was [f], and
clearly pronounced “who” as [fu].
In some dialects, <wh> is categorized as /hw/ which is distinct from /w/. “Who”
pronounced by P1 and P3 were phonologically within the acceptable range of
11
/N/ stands for placeless moraic nasal.
100
such dialects. However, P1 and P3 did not intend to pronounce <wh> this way. In
fact, they pronounced “when” as [wɛn]. Therefore, in this case, their L1 transfer
happened to be within the acceptable range.
Table 13. Results for /f/; Int= intended, Acc= accidental. Nb: “?” for P4 is due to her
purposely distinguished [f] and [ɸ] although English does not have this contrast.
Target: /f/
P1 5 5 4 0 [h]: 1
P2 5 5 5 0
P3 5 4 4 0 [h]: 1
P4 5 5? 5 0
Table 14. Results for /h/; Int= intended, Acc= accidental.
Target: /h/
P1 8 7 7 0 [ɸ]: 1
P2 8 7 7 1
P3 8 7 7 0 [ɸ]: 1
P4 8 7 7 0 [f]: 1
4.8 Results for /t/, /d/ and /s/ before high front vowels
According to Ohata (2004), Japanese ESL learners may pronounce “seat” and
“tip,” for example, as like “sheet” and “chip” because they transfer the Japanese
allophonic alternation of /t/, /d/, and /s/ which become [tʃ], [dʒ] and [ʃ]
respectively before high front vowels. Such allophonic alternations occur in some
classes of lexicon in Japanese; for example, the inflectional variations of the verb
“win,” /kata/ and /kato/(irrealis), /kati/(adverbial), /katɯ/(conclusive), and
/kate/(imperative), in which the stem is /kat/, are pronounced as [kata], [kato],
[katʃi], [katsɯ], and [kate] respectively. My focus is on whether this L1 transfer
occurs at the level of their understanding or at the level of their production
ability. I examined /s/, /t/ and /d/ before either /i/ or /ɪ/, namely “succeeded,”
“consider,” “disputing,” “did” and “immediately.” Based on those allophonic
101
alternations, these words are expected to be pronounced as [səkʃidɛd] (or

[səkʃidʒɪd]), [kənʃɪdɝ], [dʒɪspjutʃɪŋ], [dʒɪd] and [ɪmidʒiətli].
Contrary to the prediction, as shown in Tables 15 through 17, all the
participants correctly pronounced these phonemes. In addition, participants said
that the aforementioned predicted sounds were highly unlikely even in Japanese
accented English, except [kənʃɪdɝ] for “consider” and possibly [səkʃid] for
“succeed” were acceptable.
According to 15 scholars’ interpretations of Japanese phonology in
Matuzaki’s (1993) paper, [ti] and [tʃi] are unanimously considered contrastive in
Japanese except in some lexical classes mentioned above, and so are [di] and
[dʒi]. Whether [si] and [ʃi] are contrastive in Japanese is debatable. Nogita (2010)
argues that [si] and [ʃi] are not contrastive in core lexical classes, but they are in
peripheral lexical classes such as technical terms and social dialects. As well, in
Nogita’s experiment, 93 monolingual standard Japanese speakers all
distinguished [si] and [ʃi] regardless of their age. Thus, there is no reason that
Japanese ESL learners have difficulty in pronouncing [t], [d], and [s] before high
front vowels.
Additionally, the participants also recorded Japanese nonsense words
written in Japanese orthography, and all of them distinguished “ティー” [tiː]
from “チー” [tʃiː], “ディー” [diː] from “ジー” [(d)ʒiː], and “シー” [ʃiː] from
“スィー” [siː] (as mentioned in §4.13). Therefore, if Japanese speakers
pronounce /t/, /d/, and /s/, before high front vowels as [tʃ], [dʒ], and [ʃ], it makes
more sense to consider that such errors are caused by other factors, such as
loanword interference. In fact, there is variation in loanword adaptation. For
example, [tɪ] in “tip” and [di] in “radio” were adapted as [tʃi] and [(d)ʒi]
respectively, while “tea” [ti] and [dɪ] in “Disney” were adapted to [tiː] and [di]
respectively.12 The L1 transfer regarding [t], [d], and [s] is confined to loanword
interference, but the transfer is not likely to occur in words that are not a part of
Japanese vocabulary, such as “succeeded,” “consider,” “disputing,” “did,” and
“immediately.”
12
Before the government in Japan standardized the writing system in 1991, the Agency
for Cultural Affairs stipulated that in loanwords, [ti]/[tɪ] and [di]/[dɪ] in the original words
should be written as “チ” [tʃi] and “ジ” [(d)ʒi] respectively as much as possible (with a
few exceptions) (Matsuzaki, 1992).
102
Table 15. Results for /t/ before /i/ or /ɪ/; Int= intended, Acc= accidental.
Target: /t/
P1 1 1 1 0
P2 1 1 1 0
P3 1 1 1 0
P4 1 1 1 0
Table 16. Results for /d/ before /i/ or /ɪ/; Int= intended, Acc= accidental.
Target: /d/
P1 3 3 3 0
P2 3 3 3 0
P3 3 3 3 0
P4 3 3 3 0
Table 17. Results for /s/ before /i/ or /ɪ/; Int= intended, Acc= accidental.
Target: /s/
T= Int Int Accl (sound: occurrences)
P1 2 2 2 0
P2 2 2 2 0
P3 2 2 2 0
P4 2 2 2 0
4.9 Results for /æ/
As mentioned in §4.6, the Japanese /a/ is situated between the cardinal vowels [a]
and [ɑ], and Japanese lacks a vowel in the low front region. I observed six
occurrences of /æ/ in content words, namely “wrap,” “last,” and four occurrences
of “traveller”. Since a vowel in a function word is often reduced to schwa, I did
not include function words, such as “and” and “that.”
103
• P1 reported having never tried to pronounce [æ]. Despite her British

background, she did not know <a> in “wrapped” and <a> in “last” were
often pronounced differently in British English. However, she mentioned
that she purposely pronounced “can’t” as [kɑnt], instead of [kænt], even
when talking to Canadian people because of her preference of British
accent. At the same time, she realized that <a> in “can’t” in British
English was the same as J/a/, and in fact, the quality of her production
had the characteristics of J/a/. Her production of /æ/ was a mixture of
Japanese interference and over-generalization of British accent.
• P2 reported misconceiving that /æ/ in the passage was supposed to be the
same as J/a/. Because of Japanese Romanization rule in which the letter
<a> corresponds to the vowel J/a/, she had been habituated to this L1
symbol–sound correspondence.
• P3: Like P2, P3 also pronounced the letter <a> as J/a/ even in the English
contexts.
• P4: She did the same as P2 and P3.
The results indicate that all of the participants did not try the low front [æ], but
used J/a/.
Table 18. Results for /æ/; Int= intended, Acc= accidental. Nb: [a], [ʌ] and [ɑ] can be
allophones of J/a/.
Target: /æ/
P1 6 0 0 0 [ʌ]: 4 [a]: 2
P2 6 0 0 0 [a]: 3 [ʌ]: 2 [ɑ]: 1
P3 6 0 0 0 [ʌ]: 5 [a]: 1
P4 6 0 0 0 [a]: 5 [ʌ]: 1
4.10 Results for /e/ and /ɛ/
Ohata (2004) pointed out that Japanese ESL learners may make errors between
the tense vowel /e/ and the lax vowel /ɛ/ because the Japanese vowel system does
not have the tense-lax distinction. However, Ladefoged (2006) mentioned that
the terms “tense” and “lax” are really just labels, as opposed to simply a matter of
phonetic tenseness versus laxness. I will examine whether such errors come from
Japanese ESL learners’ misconceptions or their inability of production. /e/ occurs
in “came,” “they,” “make,” “take,” and “gave,” while /ɛ/ occurs in “when,”
104
“attempt,” “confess,” and two occurrences of “then.” Since P1 misread “they” as

“then,” I counted four for the occurrences of /e/.
• P1 reported knowing that /e/ and /ɛ/ in the passage were different, but she
also mentioned that she pronounced them “by instinct.” She
diphthongized /e/ and made it longer than /ɛ/. However, as long as she
pronounced it “by instinct,” the consistency may not be guaranteed.
• P2: She pronounced E/ɛ/ in the passage as the Japanese short
monophthong /e/, likely because of the Japanese Romanization where it
corresponds to the letter <e>. As for E/e/, she pronounced the target
words as she was taught in junior high school. However, she was not sure
that E/e/ in “came,” “make,” “take,” and “gave” were the same as E/e/ in
“they” because the spelling looked different.
• P3: She was sure that E/e/ in the passage was relatively diphthongized,
while E/ɛ/ was relatively monophthongized. In fact, she distinguished
them clearly in production.
• P4: She thought that E/e/ and E/ɛ/ in the passage were different, but it
was because she had memorized those words with the Japanese
pronunciation. She did not have connection to E/e/ and E/ɛ/ in the
English phonological system. A lack of knowledge may be the reason of
her mispronunciation.
Importantly, as I will mention in §4.13, the participants pronounced the nonsense
Japanese words written in Japanese orthography: [beta], [beːta] and [beita].13
The vowel part of the first syllable in each word was categorized as E/ɛ/, E/ɛ/ and
E/e/ respectively, by the North American judge (see §3.3). The distinction of
length did not change the English categories. Therefore, the fact that Japanese
does not use the label of “tenseness” for grouping vowels does not mean
Japanese speakers cannot distinguish between E/e/ and E/ɛ/. Since E/ɛ/ is one-
mora and E/e/ is two-mora (Duran, 2005), Japanese speakers can pronounce these
two by efficiently deploying the Japanese one-mora /e/ and the two-mora vowel
sequence /ei/. However, E/e/ is usually adapted to three different Japanese
categories depending on the words, namely J/eː/, J/ei/, and J/e/ – although J/e/ is
not as common as the other two (Okada, 2004). For example, the English words
“game,” “paint,” and “change” are adapted to /geːmɯ/, /peiNto/, and /tʃeNdʒi/.
This inconsistent loanword adaptation may confuse Japanese learners of English.
It makes more sense to consider that Japanese ESL learners’ errors regarding E/e/
are because of loanword interference, not their inability to articulate.
13
J/e/ is between the cardinal vowels [e] and [ɛ], so it can be transcribed as [e̞ ].
105
Table 19. Results for /e/; Int= intended, Acc= accidental.
Target: /e/
P1 4 4? 4 0
P2 5 5? 5 0
P3 5 5 5 0
P4 5 0? 0 0 [ɛ]: 5
Table 20. Results for /ɛ/; Int= intended, Acc= accidental.
Target: /ɛ/
P1 5 4? 4 0 [iː]: 1
P2 5 5? 5 0
P3 5 5 5 0
P4 5 5? 5 0
4.11 Summary of segmental errors
Table 21 summarizes the numbers of errors due to production problems, errors

due to inappropriate intentions, and accidentally correct productions. To see
tendencies, I divided the errors into two types: “Consonants” (/θ/, /ð/, /v/, /l/, pre-
vocalic /ɹ/, /f/, /h/, /s/, /ʃ/, /t/, /tʃ/, /d/, and /dʒ/) and “Vowels (Rhymes)” (/æ/, /ɛ/,
/e/, /ɔɹ/, /ɑɹ/, and /ɝ/). As mentioned above, when the participants themselves
were not sure whether or not they intended to pronounce appropriate targets, I
marked this with a question mark in the corresponding results tables. In this
summary I ignore those question marks.
Of the total 281 pronunciation errors, 186 (66.2%) were due to inappropriate
intentions while 95 (33.8%) were due to production problems. Twenty-six
productions that appeared to be correct were actually accidental. In detail, among
98 errors regarding vowels or rhymes, 92 (93.9%) were due to inappropriate
intentions. This large number should not be ignored. In the errors regarding
consonants, there are individual differences; P2 and P4 exhibited fewer
inappropriate intentions than the others, and P2 and P4’s total pronunciation
errors were also fewer than the others’. Interestingly, in spite of P2’s proficiency
106
in English being the lowest and that of P1 the highest, while P2 made the fewest
errors in both. Based on these limited data, there seems to be no correlation
between English proficiency and understanding of pronunciation.
Table 21. Summary of the results for the segmental errors: inappropriate intentions vs.
production problems. Nb: Pro= the number of “production problems”; Int= the total
number of “inappropriate intentions;” Acc= the number of “accidentally correct”
productions.
Consonants Vowels (Rhymes) Total

Errors Errors Errors
Pro Int Acc Pro Int Acc Pro Int Acc
50 27 77
P1 13 0 13
1 49 0 27 1 76
37 24 61
P2 1 0 1
31 6 0 24 31 30
52 20 72
P3 6 2 8
19 33 6 14 25 47
44 27 71
P4 0 4 4
38 6 0 27 38 33
183 98 281
T= 20 6 26
89 94 6 92 95 186
4.12 Results for the phonemic contrast identification tasks
According to the participants’ comments and the summary of their errors, the
participants seem to lack phonological awareness in English in many cases. In
order to examine their phonological awareness more deeply, I asked the
participants whether the homophone pair and minimal pairs were the same in
pronunciation or not: meat/meet, ear/year, bone/born, and who’d/food. Since this
task does not involve production, I could focus on the participants’
understanding.
The result was that none of the participants were certain whether the words
in each pair were the same or different in pronunciation. What is intriguing is that
their production of both “meat” and “meet” sounded (almost) the same.14
Nevertheless, the participants were still not certain that these words were
14
Both “meat” and “meet” as loanwords in Japanese are also homophones: [mi:to].
107
homophones. Another interesting point is that between “ear” and “year,” the
pronunciation difference is obviously shown in the spelling, namely presence or
absence of <y>, but none of them paid attention to it and became perplexed. The
comments from each participant listed below are intriguing with respect to the
participants’ own interpretations of English phonology.
• P1: Between “who’d” and “food,” she guessed that the tongue position
might be different. (She did not mention for what sound the tongue
position might be different.) What is interesting here is she paid attention
to tongue position, rather than phonological categorization.
• P2 reported not knowing what the difference was, but she misunderstood
that “meat” and “meet” were different because the spellings were
different. Meanwhile, “bone” and “born” were the same because
Japanese EFL learners typically pronounced these words in the same
way, [boːN]. She inconsistently referred to either spelling or Japanese
EFL learners’ pronunciation or loanwords.
• P3 said that she had been pronouncing the two words in each pair
probably in the same way, except she was taught that “ear” and “year”
were different in junior high school although she was not sure what the
difference was.
• P4 claimed that she had no awareness of the connection between spelling
and sounds in English, or no knowledge about the English pronunciation
system. In contrast, in Japanese she had the clear connection between
orthography and sounds, and had the whole picture of the Japanese
phonological system. Therefore, she had no idea about these English
homophones and minimal pairs.
The participants’ comments indicated that they do not really have a clear picture
of the English sound system. Moreover, although they often referred to English
loanwords in Japanese or the rules of Japanese Romanization, they did not fully
depend on the Japanese phonology. Hocket (1960) defined linguistic sounds as
discrete, whereas non-linguistic sounds form a continuum. More specifically,
according to D. McKercher (personal communication, November, 2009),
linguistic sounds must be categorized as phonemes in particular languages, while
non-linguistic sounds cannot be categorized as phonemes. Since the participants
often could not categorize sounds in the stimuli as particular English phonemes,
they often might have pronounced English words with non-linguistic continuum
sounds. In this way, the participants’ vowel and rhotacized vowel qualities 15
varied substantially. It will be worth examining whether their vowel qualities will
be more consistent after they learn the structure of English vowel inventory.
15
“Rhotacized vowels” = /ɔɹ/, /ɑɹ/, and /ɝ/.
108
4.13 Orthographic pairs reading results
As mentioned earlier, when participants read Japanese stimuli written in Japanese

orthography, they distinguished between [si] and [ʃi], [ti] and [tʃi], and [di] and
[(d)ʒi]. As for their [e], [eː] and [ei], the first two can be categorised as E/ɛ/ while
[ei] can be categorized as E/e/.
Interestingly, P1 (the participant with the longest residency in Canada at
three years) showed different phonetic characteristics than P2, P3 and P4, whom
had all lived Canada for five months in. P1 aspirated [t] in both a word-initial and
word-medial position, and did not show pre-voicing for [d], [dʒ], or [b], and did
not affricate [z]. In contrast, P2, P3, and P4 did not quite aspirate [t], (except P2
aspirated word-initial [t]) and often showed pre-voicing for [d], [dʒ], [b], and
even [dz], and also pronounced the affricate [dz], instead of [z]. Japanese /z/ is
typically the affricate [dz] (Grenon, 2008), as mentioned in 4.2 above. Vance
(2008) mentioned that, according to some descriptions, Japanese /p/, /t/, and /k/
are typically weakly aspirated in word-initial position or in an accented syllable,
and unaspirated elsewhere. According to Takada (2008), in Tokyo and Kansai
region, voiced stops typically have negative voice onset time (VOT) values,16 in
other words “pre-voicing”. Recall that P2 and P3 are from near Tokyo and P1 and
P4 are from Kansai region. Therefore, P2, P3, and P4 showed typical phonetic
characteristics in the Japanese stimuli, whereas P1 showed different
characteristics. Since P1 had been in an English environment much longer than
the others, her L1 may have been influenced by her L2. In fact, in Haraguchi’s
(2003) study, advanced Japanese ESL learners acquire English aspiration patterns
without special endeavour. However, as shown in §4.1 to §4.10, P1’s
phonological realization was similar to that of the other participants. This implies
that longer length of residence may help Japanese adult ESL learners acquire
phonetic characteristics, but may not help them construct L2 phonological
categorization. Incidentally, according to Hirayama (1994), Kyoto dialect
speakers do not affricate /z/. Since P1 is from near Kyoto, her true fricative [z] is
possibly not from L2 influence, but rather a characteristic of her L1 dialect.
Another interesting phonetic characteristic is that P1 and P4 added a glide in
the Japanese [(d)ziː] and [siː] data, like [(d)zwiː] and [swiː], or unrounded
[(d)zɰiː] and [sɰiː]. Conversely, they did not show such glide insertion in the
English data. For example, they did not pronounce “succeeded” as [səksɰidɪd].
Their purposely differentiated productions may be due to orthographical
interference. In the Japanese syllabary system, when a new syllable comes into
use, it is written with the combination of two existing symbols (a big symbol and
a small symbol), instead of creating a new symbol (Inozuka, 2009). For example,
16
In the Tokyo area, younger speakers more often show positive VOT values in voiced
stops than older speakers do (Takada, 2008).
109
the new syllables [(d)zi] and [si] are written with <ズィ> and <スィ>
respectively. This two-symbol structure may cause some Japanese speakers to
add an extra sound.
5 Discussion
5.1 Pedagogical implications
The pronunciation error patterns of these Japanese ESL learners can be divided
into the following four types, summarized in Table 22.
Table 22. Summary of four types of the errors committed by the participants.
A: A lack of phonological knowledge or misunderstanding of target sounds

B: Abandonment of particular phonemes in learners’ inter-language
C: Difficulty in articulation or a lack of knowledge of the sound quality of a target
phoneme
D: Accidentally correct productions
Only C is a phonetic error, but the others are caused by misunderstanding. In fact,
in many of the cases, the participants did not intend to pronounce the proper
target phonemes. If native Japanese-speaking learners of English adopt the same
behaviour, articulatory training often does not help them improve their
pronunciation. The findings of this research suggest that pronunciation lessons
need to stress learners’ understanding of target sounds and the phonological
system of the target language, and not only what learners actually produce. Each
type of error is discussed in more detail below.
5.1.1 Type A: Lack of phonological knowledge/ misunderstanding target sounds
Learners do not know what they are supposed to pronounce. Learners often do
not consider the target sound as a discrete phonological category, but as a non-
linguistic sound. For example, the participants were not sure whether /e/’s in
“gave” and “they” were the same. Therefore, their productions phonetically
varied over a wide range. Another example is the participants’ misunderstanding
that /ɔɹ/s in “north,” “warm,” and “more” were supposed to be pronounced
differently. The source of this type of error is that learners have not been taught
the English symbol–sound correspondence rules. As Carroll (2004) stated, letter
knowledge precedes phoneme awareness, as mentioned in 2.1. Learners need to
know the concept of discrete phonological categorization with the visual cue, the
110
orthography. As well, as Makino (2008) mentioned, learners need to be shown all

the English consonants and vowels to grasp the whole picture of the phonological
inventory.
5.1.2 Type B: Abandonment of particular phonemes in learners’ inter-language
Learners know what the target sound is, but they have abandoned the particular
phoneme in their inter-language. For example, P3 knew that the English /l/ and /ɹ/
were different from the Japanese /ɾ/, but she gave up trying to acquire E/l/ and
E/ɹ/, and substituted both with J /ɾ/. According to the participants’ comments,
they did not know why some particular phonemic contrasts, such as E/l/ and E/ɹ/,
must be distinguished, and so they were not motivated to practice the contrasts.
In order to help them understand the concept of contrasts, other ESL learners’
errors or JSL (Japanese as a second language) learners’ errors seemed effective.
For example, naming the /p/ and /f/ confusion by Korean speakers and the /p/ and
/b/ confusion by Arabic speakers, which are not problematic for Japanese
speakers, helped the participants understand what the confusions like /l/ and /ɹ/
sound like.
5.1.3 Type C: Difficulty in articulation or a lack of knowledge of the sound

quality of a target phoneme
Learners know what the target phoneme is and attempt to pronounce it, but fail to
meet the target in terms of articulation, or can meet the target in isolation or
careful speech but cannot afford it in a practical situation. Alternatively, they
misunderstand the sound quality of the target phoneme. For example, P3 tried to
pronounce /ð/ in the right place, but sometimes she affricated it. This type of
error is a purely phonetic issue. Learners need some phonetic tips or need some
practice on their own.
5.1.4 Type D: Accidentally correct productions
Learners accidentally met the target, but their production of target sounds was not
intentional. In other words, it can be called a covert error. For example, P3
pronounced a target-like [ɹ] in “wrap”, but actually she intended to pronounce
/wɾ/, which happened to be realized as [ɹ]. Although this type of error may be
difficult to find, the source of the error is the same as either A or B. Therefore, if
learners understand the concept of phonological categorization and symbol–
sound correspondence rules, this type of error can be avoided.
111
5.2 Limitations of the analysis
Firstly, this study’s data were limited as some sounds appeared in only one word.
For example, /θ/ appeared only in the word “North”, and /ɑɹ/ appeared only once
in the whole passage. There is no way to know how the participants pronounce
these sounds in other words. However, this study was still able to identify the
participants’ understandings of these sounds, as mentioned in §4.1 and §4.6.
Secondly, the participants themselves were often not sure of what they
intended to pronounce, which made it difficult to judge whether they attempted
the proper target sounds or not. Moreover, they often used their L1 sound without
understanding the L2 phonological inventory, and the L1 sound happened to be
within the acceptable L2 target phoneme. For example, when they consistently
used J/e/ for E/ɛ/, it sounded correct, but if they did not picture E/ɛ/, or so called
“Short E”, it is questionable whether they intended to pronounce the proper target
or not. In these ambiguous cases, I put a question mark. At the same time, these
ambiguous cases indicate that the participants did not understand the target
sounds.
5.3 Future research
Firstly, this study found that the participants lack phonological and
orthographical awareness in English in many cases. I should examine whether the
finding from this qualitative study are generalizable by performing more
quantitative research.
Secondly, P1 with three years of residence in Canada had better sound
qualities of [ɹ] and aspiration than the other participants, while in terms of
phonological and orthographical awareness, P1 was similar to the others. For
example, P2 and P4 (with only five months of residence in Canada) performed
much better than P1 in the /l/ and /ɹ/ distinction. Longer residence may help
learners improve phonetic accuracy but may not help learners naturally acquire
L2 phonological mapping and spelling rules. Since the participants had been
exposed to English loanwords or strongly Japanese accented English for a long
time, this exposure may have prevented them from constructing the L2 rules.
This has to be studied with more participants in the future.
112
6 Conclusion
When Japanese ESL learners mispronounce English, they often intend to

pronounce different sounds due to their misconception about target sounds, or
due to their own interpretations of English phonology, as opposed to current
ideas about Japanese learners’ articulatory inability to produce particular sounds.
Especially, in this experiment, the participants’ errors regarding vowels or
vowels followed by /ɹ/ were due to misguided intentions 93.9% of the time.
Japanese ESL learners’ misconceptions are likely due to their often not having
been taught the basics of English phonological and orthographical systems.
Therefore, the same phoneme spelled with the same alphabet letter(s) is often
purposely pronounced differently when it appears in different words, For
example, <v> in “gave” and that in “traveler” are misunderstood to be different
phonemes. On the other hand, different phonemes spelled with different alphabet
letters are often purposely pronounced the same: <ar> in “hard” and <ir> in
“first” are in this way misunderstood to be the same phoneme(s). Moreover, even
in the production of the notorious /l/ and /ɹ/, there was a clear difference in their
productions between those who tried to distinguish them and those who did not.
Insofar as Japanese ESL learners have not yet been taught the basic English
symbol–sound correspondence rules, they would be limited to pronouncing
according to their own interpretations of English phonology. If Japanese ESL
learners have not yet pronounced their L2 sounds according to true English
phonology, there is no way to know whether they actually have difficulty in
producing particular sounds. It would be safe to avoid immediately concluding
that pronunciation errors by Japanese ESL learners come from their articulatory
inability to produce.
113
References
Agency for Cultural Affairs, Japan. (2002). Kokugo-hyoki no kijun. [Online

resource]. Retrieved 30 Mar., 2009, from: http://www.bunka.go.jp/kokugo/.
Akahane-Yamada, R., Aoyama, K., Fledge, J.E., Guion, G. & Yamada, T. (2004).
Perceived phonetic dissimilarity and L2 speech learning: The case of
Japanese /r/ and English /l/ and /r/. Journal of Phonetics 32, 233–255.
Akamatsu, T. (2000). Japanese phonology. Munich: Lincom Europa.
Akinaga, K. & Kindaichi, H. (Eds.). (2008). Shinmêkai nihongo accent jiten.
Tokyo, Japan: Sansêdo.
Akiyama, K. (2009). Sensê seminar: Hatsuon sassê. NHK CTI Nihongo Centre.
Archibald, J. (2005). Second language phonology as redeployment of
phonological knowledge. Canadian Journal of Linguistics 50, 285–314.
Avery, P. & Ehrlich, S. (2003). Teaching American English Pronunciation. New
York: Oxford University Press.
Bayraktaroğlu, S. (2008). Orthographic interference and the teaching of British
pronunciation to Turkish learners. Journal of language and Linguistics
Studies 4(2), 107–143.
Beckman, M. E., Edwards, J. & Li, F. (2009). Contrast and covert contrast: The
phonetic development of voiceless sibilant fricatives in English and
Japanese toddlers. Journals of Phonetics 37, 111–124.
Berman, J., Lambacher, S., Martens, W. & Nelson, B. (2001). Identification of
English voiceless fricatives by Japanese listeners: The influence of vowel
context on sensitivity and response bias. Acoustical Science and Technology
22(5), 334–343.
Bernhardt, B.H. & Stemberger, J.P. (1998). Handbook of phonological
development from the perspective of constraint-based nonlinear phonology.
California: Academic Press.
Brown, H.D. (2007). Principles of language learning and teaching. New York:
Pearson Education.
Carroll, J.M. (2004). Letter knowledge precipitates phoneme segmentation, but
not phoneme invariance. Journal of Research in Reading 27(3), 212–225.
Davenport, M. & Hannahs, S.J. (2005). Introducing phonetics & phonology.
London: Hodder Education.
Duran, J. (2005). Tense/Lax, The Vowel System of English and Phonological
Theory. In: P. Carr, J. Durand & C.J. Ewen (Eds.), Headhood, elements,
specification and contrastivity: Phonological papers in honor of John
Anderson. Philadelphia, PA: John Benjamins, 77–98.
Eckman, F.R. (2008). Typological markedness and second language phonology.
In: J.G. Hansen Edwards & M.L. Zampini (Eds.), Phonology and second
language acquisition. Philadelphia, PA: John Benjamins, 95–116.
114
Fraser, H. (in press). Pronunciation as categorization: The role of contrast in

teaching English /r/ AND /l/. Australian Review of Applied Linguistics.
Gass, S.M. & Selinker L. (2009). Second language acquisition. New York:
Routledge.
Goble, D. (2002). Loanword-induced interference in Japanese students’ foreign
language acquisition: Developing student awareness through experiential
learning. Journal of Chikushi Jyogakuen Junior College 37, 55–77.
Green, A.D. (2001). American English “r-colored” vowels as complex segments.
Linguistics in Potsdam 15, 70–78.
Grenon, I. (2008). The acquisition of English soun[dz] by native Japanese
speakers. A perceptual study. Saarbrücken, Germany: VDM Verlag.
Haraguchi, Y. (2003). The acquisition of aspiration of voiceless stops and
intonation patterns of English learners: Pilot study. Proceeding of the 8th
Conference of Pan-Pacific Association of Applied Linguistics, 83–91.
Hirata, Y. (2004). Effects of speaking rate on the vowel length distinction in
Japanese. Journal of Phonetics 32(4), 565–589.
Hirayama, T. (1994). Gendai nihongo hôgen daijiten. Tokyo, Japan: Mêjishoin.
Hockett, C.F. (1960). The origin of speech. Scientific American 203, 88–96.
Inozuka, E., Inozuka, H. & Machida, K. (Eds.). (2009). Nihongo onsêgaku no
shikumi. Tokyo, Japan: Kenkyûsha.
Ito, J. & Mester, A. (1995). Japanese phonology. In: J.A. Goldsmith (Ed.), The
handbook of phonological theory. Cambridge, MA: Blackwell, 817–838.
Koizumi, T. (2003). An introduction to phonetics. Tokyo: Daigaku-Shorin.
Kurath, H. & McDavid, R.I. (1961). The pronunciation of English in the Atlantic
states. Ann Arbor: University of Michigan Press.
Ladefoged, P. (2006). A course in phonetics. Boston: Thompson Wadsworth.
Maddieson, I. (1984). Patterns of sounds. New York: Cambridge Press.
Maddieson, I. (2005). Bilabial and labio-dental fricatives in Ewe. UC Berkeley
Phonology Lab Annual Report, 199–215.
Magnuson, T.J. (2008). What /r/ sounds like in Kansai Japanese: A phonetic
investigation of liquid variation in unscripted discourse. MA thesis,
University of Victoria.
Makino, T. (2008). English pronunciation teaching and phonetic study.
Proceedings of the 162nd meeting of Osaka linguistics society. Osaka,
Japan.
Matsuzaki, H. (1992). A typification of variation in loanword sounds in
dictionary items. Tsukuba Working Papers in Linguistics 10(11), 43–56.
Matsuzaki, H. (1993). A quantitative study of fluctuation in Japanese loanword-
phone notation. Journal of the Department of Japanese, Tohoku
University 3, 83–94.
Muroi, A. (2005). Hatsuon shido ni okeru kyôshi no yakuwari: Ayashii hatsuon
shido no shôtai. Êgokyôiku 2005. 12. Taishûnshoten.
115
NHK Hosobunkakenkyûjo (Ed.). (2008). Nihongo hatsuon accent jiten. Tokyo,

Japan: NHK Shuppan.
NHK Shuppan (Ed.). (2005). NHK announce jissen training. Tokyo, Japan: NHK
Shuppan.
Ohata, K. (2004). Phonological difference between Japanese and English: Several
potentially problematic areas of pronunciation for Japanese ESL/EFL
learners. Asian EFL Journal 6(4).
Okada, S. (2004). Language variation of the hiatus /ei/ in the Corpus of
Spontaneous Japanese: In the case of loanwords. IEICE Technical Report.
Speech 104(148), 35–40.
Ota, K., Riney, T.J., Takagi, N. & Uchida, Y. (2007). The intermediate degree of
VOT in Japanese initial voiceless stops. Journal of Phonetics 35(3), 439–
443.
Pan, H.Y., Utsugi, A. & Yamazaki, S. (2004). An acoustic phonetic study of
voiceless alveolo-palatal fricatives in Japanese, Korean and Chinese.
Journal of General Linguistics 7, 1–27.
Siegel, L.S. & Wade-Woolley, L. (1997). The spelling performance of ESL and
native speakers of English as a function of reading skill. Reading and
Writing 9(5), 387–406.
Suarez, A. & Tanaka, Y. (2001). Japanese learners’ attitudes towards English
pronunciation. Bulletin of Niigata Seiryo University 1, 99–111.
Takada, M. (2008). Geographical pattern of VOT in Japanese initial voiced stops.
Studies in the Japanese Language 4(4), 48–62.
Takayama, M. (2003). Issues in Japanese historical phonology. Journal of the
Phonetic Society of Japan 7(1), 35–46.
The International Phonetic Association. (2003). A guide to the use of the
international phonetic alphabet. Cambridge: Cambridge University Press.
Tsuzuki, M. (1996). An allophonic study of Japanese vowels and consonants. The
Journal of Aichi Gakuin University Humanities & Sciences 43(2), 41–66.
Vance, T.J. (2008). The sounds of Japanese. New York: Cambridge University
Press.
Taniguchi, M. (2009). Japanese learners’ weak points in English pronunciation,
rhythm and intonation. Sênangakuin daigaku êgo shidoryoku kaihatsu
wâkushoppu. [Online resource.] Retrieved 12 Aug., 2010, from
http://www.seinan-gu.ac.jp/syakai/school/pdf/090822taniguti.pdf.
Vihman, M.M. (1996). Phonological development: The origin of the language in
the child. Massachusetts: Blackwell Publishers Inc.
Yamaguchi, K. & Yamanaka, K. (2003). The confusion of alveolar and palato-
alveolar fricatives by Japanese learners of English. Bulletin of the Yamagata
University. Cultural Science 15(2), 207–220.
116
Appendix
Reading Task: The North Wind and the Sun
The north wind and the sun were disputing which was stronger, when a traveler
came along wrapped in a warm cloak. They agreed that the one who first
succeeded in making the traveler take his cloak off should be considered stronger
than the other. Then the north wind blew as hard as he could. But the more he
blew the more closely did the traveler fold his cloak around him; and at last the
north wind gave up the attempt. Then the sun shone out warmly, and immediately
the traveler took off his cloak. And so the north wind was obliged to confess that
the sun was the stronger of the two.
View publication stats

Do Japanese ESL Learners Pronunciation Errors Com

Uploaded by

Copyright:

Available Formats

Do Japanese ESL Learners Pronunciation Errors Com

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Do Japanese ESL Learners Pronunciation Errors Com

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Do Japanese ESL learners' pronunciation errors come from inability to articulate

Article · January 2010

The user has requested enhancement of the downloaded file.

This paper aims to examine whether Japanese English-as-a-Second-

This paper examines second language (L2) learners’ understandings of the

mispronounces “change” as [tʃɛndʒ]1 when it is supposed to be pronounced as

2.1 Literature review

L2 learners’ pronunciation errors are caused by factors other than difficulty in

More specifically, Avery & Ehrlich (2003) explained an example of induced

2.2 Research questions

The present study is designed to address the following two questions:

3.2 Stimulus and procedure

ティー [tiː] チー [tʃiː]3

Table 3. Phonemic contrast identification tasks

“meat” versus “meet”

On the second day, I conducted semi-structured interviews with each

3.3 Analysis of the recordings

The transcribers were a phonetically trained native speaker of North American

4 Results and discussion

Overall, in terms of English phonemes that are generally considered problematic

4.1 Results for /θ/

4.2 Results for /ð/

There are 23 occurrences of /ð/ in the passage. The following summarizes

4.3 Results for /v/

• P2: Due to having learned “gave” and “of” in Japan, P2 incorrectly

4.4 Results for /l/ and /ɹ/ in onset position

4.5 Results for velarized /l/ in coda position

4.6 Results for /ɹ/ in coda position

passage, it is interesting to compare /ɑɹ/ and /ɝ/.

Table 11. Results for /ɑɹ/; Int= intended, Acc= accidental.

Table 12. Results for /ɝ/; Int= intended, Acc= accidental.

4.7 Results for /f/ and /h/

Table 14. Results for /h/; Int= intended, Acc= accidental.

alternations, these words are expected to be pronounced as [səkʃidɛd] (or

4.9 Results for /æ/

• P1 reported having never tried to pronounce [æ]. Despite her British

4.10 Results for /e/ and /ɛ/

“attempt,” “confess,” and two occurrences of “then.” Since P1 misread “they” as

Table 19. Results for /e/; Int= intended, Acc= accidental.

Table 20. Results for /ɛ/; Int= intended, Acc= accidental.

4.11 Summary of segmental errors

Table 21 summarizes the numbers of errors due to production problems, errors

Consonants Vowels (Rhymes) Total

4.12 Results for the phonemic contrast identification tasks

4.13 Orthographic pairs reading results

As mentioned earlier, when participants read Japanese stimuli written in Japanese

5.1 Pedagogical implications

A: A lack of phonological knowledge or misunderstanding of target sounds

5.1.1 Type A: Lack of phonological knowledge/ misunderstanding target sounds

orthography. As well, as Makino (2008) mentioned, learners need to be shown all

5.1.2 Type B: Abandonment of particular phonemes in learners’ inter-language

5.1.3 Type C: Difficulty in articulation or a lack of knowledge of the sound

5.1.4 Type D: Accidentally correct productions

5.2 Limitations of the analysis

5.3 Future research

When Japanese ESL learners mispronounce English, they often intend to

Agency for Cultural Affairs, Japan. (2002). Kokugo-hyoki no kijun. [Online

Fraser, H. (in press). Pronunciation as categorization: The role of contrast in