Second Language Acquisition - Theory and Pedagogy: Proceedings of the 6th Annual JALT Pan-SIG Conference.
May. 12 - 13, 2007. Sendai, Japan: Tohoku Bunka Gakuen University. (pp. 56 - 64)
The change-of-state token a
by Keiko Ikeda (Nagoya University)
in Japanese language proficiency interviews
This paper explores English and Japanese change-of-state tokens oh and a used in oral proficiency interviews.
Using a conversation analytic approach with twenty sets of OPI-simulated interview data for each language,
the study finds that the English testers withhold the use of oh in their listener responses whereas the Japanese
testers in fact frequently use a in various environments. The Japanese a, along with frequently co-occurring listener
responses expressions, denotes a rather neutral stance as an interviewer, which differs from the interpersonal stance
denoted by oh in English. The cross-linguistic approach to this study discourse markers may facilitate a deeper
understanding of intercultural communication processes.
change-of-state tokens, language proficiency interviews, conversation analysis, cross-linguistic research
This study employs conversation analysis (CA) to examine a set of Japanese and English language proficiency interviews(s) focusing on
"change-of-state" tokens by the interviewers. A language proficiency interview (henceforth LPI) 1 consists of at least one interviewer
(who is also usually the language rater) and at least one interviewee, who is often known as a candidate. The interaction begins with
the interviewer initiating a series of question-answer sequences. The interviewer's interactional management in a language proficiency
interview has attracted a great deal of recent interest among conversation analysts. For instance, Kasper and Ross (in press) examined the
multiple-question style employed in the English OPI, and Kasper (2006) explored interviewers' interactional efforts to keep the talk on task
(interactional management is accomplished throughout the talk, without awkward pauses or breaks).
In this study, the term language proficiency interviews (LPI) will be used to refer to various testing
batteries which take an interview format to test learners' speaking skills. The OPI (Oral Proficiency Interview)
developed by ACTFL (American Council on the Teaching of Foreign Languages) is one of the best known such interviews.
"The present study . . . seek[s] a deeper understanding of the challenges candidates face in a second-language interview."
Some studies on language proficiency interviews embrace a cross-linguistic perspective (e.g., Kim & Suh, 1998). Language-specific interactional
styles embedded in interviews have not been extensively examined (but see Ross, 1996). This, however, is certainly a critical issue since language
proficiency interviews always takes place in cross-linguistic contexts. The present study adopts such a comparative stance to seek a deeper
understanding of the challenges candidates face in a second-language interview.
Focus of the study: The change-of-state token
The English oh and Japanese a
In language proficiency interviews, the interviewer's main interactional role is to elicit as much oral performance from the candidate as possible. He or she
mostly listens to the candidate's talk, at times providing appropriate listener responses to further develop the topic of the conversation, or to
shift gears from one type of talk to another within the interview. This study focused on the use of what has been called a "change-of-state" token,
which is commonly used as part of interviewer's listener response.2
The use of this listener token has been well studied in relation to interview
talk (e.g., Clayman, 1992; Heritage, 1998). Exploring the ways such tokens are used by interviewers in LPIs may provide us with a better
understanding of the dynamics involved in the talk.
The change-of-state token is deployed to register an immediately proceeding utterance as informative; that is, the speaker makes use of it to
indicate that he or she has undergone a change from non-knowing to now-knowing (Schegloff, 2007, p. 118). In English conversation, the interjection
"oh" is generally recognized as a change-of-state display and it is one of the most frequently used discourse markers in conversations. Heritage
(1984; 1998; 2002) has examined the use of "oh" extensively using conversation analysis. For the purposes of this study, the most relevant part of
his analysis of the use of "oh" is in the [Question by Speaker A]-[Answer by B]-[Receipt of Answer by A] sequence. Heritage (1984) shows that the
questioner-exhibited "oh" proposes to treat the provided answer as complete, so that he or she is ready to bring that particular topic to a close.
Schegloff (2007) calls "oh" the "sequence-closing third," which occurs in the post-expansion turn after a basic adjacency pair. In the case of an
interview, this pair usually has a Question-Answer format. When the questioner wishes to probe further, he or she avoids producing "oh" (310).
The Japanese change-of-state token a3 is also extensively discussed in the literature, and scholars seem to agree that it functions very similarly to the English oh in the same environments (Saft 2001; Nishizaka, 2001). However, the use a in interview settings has not be examined closely.
Japanese communication and listening behavior have been a well-investigated topic in the literature (e.g., White,
1989; Mizutani, 1988; Maynard, 1989). While they facilitated our understanding of a general characteristics of Japanese listener behavior in mundane
conversations, there has not been much investigation when the listener is engaged in a specific interactional context such as an OPI.
Since the analysis of these tokens in this study seem to indicate that a in some environments is not equivalent in its meaning with English oh.
Therefore, in this paper, Japanese change-of-state token a is not translated as English oh. Since the analysis of these tokens in this study seem
to indicate that a in some environments is not equivalent in its meaning with English oh.
Interviews and suppression of change-of-state displays
In the specialized interactional setting of an interview, change-of-state tokens are employed somewhat differently than in mundane conversation.
In news interviews in Britain and North America, for example, the interviewer generally attempts to maintain his or her neutrality vis-à-vis the
interviewee (Clayman, 1992; Clayman & Heritage, 2002). As a result, subjective listener responses, including change-of-state displays, are suppressed.
In an LPI, interviewers typically also seek to maintain a neutral stance. This is both because testers must avoid projecting potentially evaluative
responses, and because in OPI interactions they require multiple tasks simultaneously throughout the talk. One such task is eliciting as much
of a language sample as possible – in other words, they must encourage candidates to speak more than the interviewers themselves. For this reason
interviewers must avoid providing signals that might bring a question-answer sequence to a premature conclusion or pre-empt relevant elaboration.
With these background observations in mind, we are now in a position to pose two specific research questions:
- Are change-of-state tokens evident in LPI interviewers' listener responses? (If so, when and in which contexts are they used?)
- What are the similarities and differences between the English and Japanese LPIs existence in this regard? In other words, what does a comparative analysis tell us about frequent interview styles across both languages?
This study examined interviewers' use of change-of-state tokens in twenty Japanese a nd twenty English oral proficiency interviews. This paper will
focus just on the turn-initial placement of the tokens. They typically take on one of two patterns: the first is the placement of a4 or oh in a free-standing position as an independent turn by itself; the other is the insertion of a or oh + a listener response utterance. The prosodic
quality was marked as either a non-elongated token (e.g., a) or a token with a falling contour (e.g., a::) . For the Japanese OPI-simulated
interview data, the data samples include seven different interviewers (3 females, 4 males), and twenty different candidates (5 females, 15 males,
all in their 20's and 30's). For the English OPI-simulated interview data5, the corpus used indicated no record as to who the interviewers were; therefore we cannot know how many interviewers were involved (although the author detected several different speakers through the analysis). There
were twenty different candidates, and ten females and ten males (age range is not recorded) composed the sample.
Particularly for Japanese, some studies (e.g., Sudo, 2001) have shown that prosodic features may classify different functions mapped onto the tokens.
The format of the OPI-simulated interviews is called SST (Standard Speaking Test).
Distribution of oh and a in the interviews: A general overview
Tables 1 and 2 show the total count of tokens (a and oh) in five selected interviews in each language, with subtotals for the two specific targeted patterns6 :
Table 1. Occurrence of the change-of-state token "a" in the Japanese data
||a + listener responses
Table 2 illustrates the occurrence of oh in 20 English OPI-simulated interview sets provided by Izumi, Uchimoto, and Isahara (2004):
Table 2. Occurrence of the change-of-state token "oh" in the English data
||oh + listener responses
||1 (oh really?)
||1 (Oh, I see.)
The study examined all 20 sets in each language.
Owing to space limitations, however, I will discuss only five typical samples within each group.
Each interview lasted approximately 15 minutes. The analysis did not examine the role play parts of each interview.
The five sets shown in this paper were selected from the collection because they are most telling of the tendency
observed of each language sample.
From these two tables it is immediately apparent that interviewers in the English data do not utter oh in their listener responses as frequently as
interviewers utter a in the Japanese data. In the English setting, the use of oh seems to be withheld much as has been reported in news interviews.
Example 1 is an excerpt from the English data, in which the candidate (B) is supplying unknown information to the tester (A), yet we do not observe
any use of oh in A's responses.
In line 3, the interviewer inquires about the interviewee's personal information. Upon receiving a description of the area of Japan in
which the interviewee lives, A provides a minimum um in line 8 ("um" does not display receipt of a complete answer; Gardner, 1998).
The interviewee further provides an extended turn in line 9, to which the interviewer responds with a personal assessment, "that seems very nice place."
No change-of-state token was used, and the interviewer's personal input was delayed.
Example 1 (English example)
1 A: So where do you live?
2 B: Umm, I live in XXX8.
3 A: Hmm what kind of place is that?
7 B: Er, the nearest station is er XXX or XXX. It's very er er historical. Place in XXX.
And er (.) it's very comfortable to er live in and it's very convenient to er
for example to to er (.) to buy something or er to meet or to drink.
8 > A: Um.
9 B: It's very comfortable place.
10> A: Mm that seems very nice place.
11 B: Um.
From this finding, we can surmise that displaying a change of state has a different function in Japanese oral interviews than in English.
While the display of change-of-state (e.g., oh really? oh) indicates a clear personal footing (Goffman, 1981) in English, in Japanese it does not.
Example 2 shows how the [Question-Answer-Change-of-state token a prefacing Response] pattern occurs repeatedly in the Japanese data.
Example 2 (Japanese example)
1 B: Uh: Tookyoo de ano keimei gakuen to yuu (.) ano:
[Uh: in Tokyo, uhm a school called Keimei (.) uh:]
2 international no kookoo (.) no kookoo (.) soko de
[an international high school there I]
3 nihongo o benkyoo shimashita.
4 > A: A:: soo desu ka:: Nihon wa doo deshita ka.
[Ah soo desu ka. How was Japan?]
6 B: Un=tanoshikatta. Omoshirokatta. Ma (.) daisuki datta. Huhu.
[Yeah=it was enjoyable. Interesting. Uh (.) I loved it. Huhu.]
Mo:: mo:: nihon ni (.) ikitai. Hu.
[Once more I want to go to Japan.]
7 > A: A soo desu ka. ano: nani ga tanoshikatta desu ka.
[Ah, uhm: what was enjoyable?]
9 B: Etto: gakkoo wa tanoshikata:: ano:: (.) hoomusutei no kazoku (.)
[Well: school was fun. Uhm: (.) homestay family]
Uh: sugoi omoshirokatta ato wa:: arubaito mo shimashita.
[Uh: very interesting and additionally I also did part-time job]
11 tto:: hmmm (.) eigo o oshieta to:: un (.) un (.) club de hataraita
[uh:: hmmm (.) I taught English and:: I worked in a club.]
(.) iroiro arubaito o shimashita. Sore wa tanoshikatta
[I did various part-time work. That was enjoyable.]
13> A: A soo desu ka. ja (.) hosuto famirii no kazoku ni tuiste
[IAh soo desu ka. Then (.) about the host family]
(.) moo sukoshi oshiete kudasai.
[Please tell me more about them.]
XXX indicates a proper noun.
This excerpt reveals a fairly neutral stance by the tester on the content of the candidate's talk, despite the use of a in the turn-initial
position of his listener response. Indeed, the Japanese data showed that the interviewers employ the change-of-state token throughout the
interview. The change-of-state token a is used by the interviewer to indicate receipt of the prior turn, not to index receipt of newsworthy information.
Functions played by the Japanese change-of-state tokens
Change-of-state tokens and probing
In the remainder of this paper, I will narrow the discussion to the Japanese change-of-state token a and its functions in the LPIs. One typical
use of the token by the IR is to indicate to the IE that something just heard is interesting or worth expanding in the talk, initiating what
is called a "probing" phase in the LPI. In the probing phase, the tester pushed the IE to perform an extended or more complex task in their
target language; for example, providing a substantive elaboration rather than a simple yes or no response. In Example 3, speaker 1 has just
stated that the literary genres "stories" and "novels" are different:
This example illustrates how the interviewer utilized the change-of-state token a: to probe further within the same topic.
A: displays the interviewer's attention to certain news and warrants further questioning, as shown in line 5: "how are (stories and novels)
different?" The question asks the candidate to compare and contrast the two genres – a more complex task.
1 A: a (..) soo na n desu ka. monogatari to shoosetsu
[A(..) is that right. stories and novels]
tte no wa chigau n desu ka.
[are different, then.]
2 B: Hai,
3 A: a::
4 B: totemo
5 A: a:: do- dooyuu huu ni chigau n desu ka ne.
[A:: ho- how are they different, I wonder.]
6 B: ano (.) monogatari wa
[Well (.) stories are]
7 A: hmm
[hmm [The interviewee keeps answering]
The function of the Japanese a with combined listener responses
Among the frequent uses of a in combination with other listener responses, two patterns were particularly salient. The first is a soo desu ka, and
the other is a + the speaker's rephrase of the answer.
A soo desu ka
The most frequent constellation observed was [a] plus [soo desu ka]. Soo desu ka literally means 'is that so,' and often in the literature,
a soo desu ka is translated as "oh really?" in English. Now that I have examined change-of-state tokens in both languages, I will argue against this
habitual translation of a soo desu ka, and I will illustrate the reasons below.
The phrase a soo desu ka, uttered with a falling intonation, commonly follows the receipt of new information from the candidate in response to a
question and registers the answer as appropriate and complete so far as it goes. It also commonly indicates that there is more that can be said
on the subject and hence serves to elicit elaboration. However, we sometimes observe it being used even when the candidate fails to provide an
adequate answer. Example 4 is an illustration. Here, the candidate was just asked to explain the plot of the movie The Usual Suspects that she has
Upon the candidate's indication of incapability to proceed with the task (line 6), the IR in line 7 utters a soo desu ka, then follows up by
saying hai ii desu yo: 'that is okay.' Then in line 8 the IR changes the topic completely asking what time the candidate wakes up in the morning.
1 A: chotto oshiete kudasai.
[Please tell me about it.]
[omitting line 3-4: the interviewee is struggling to answer]
5 A: n
6 B: huhuhu (.5) nanka (.) chotto sumimasen dekimasen.
[giggles (.5) well (.) I am sorry I cannot do it.]
A: a: soo desu ka. un. Hai. ii desu yo::.
[soo desu ka. hmm okay. That is okay.]
A: E:to jaa eeto (.)ichinichi asa (.) e: nanji ni okimasu ka.
[Well then well (.) all day morning (.) uh:]
[what time do you wake up?]
Thus it is apparent that the more basic interactional role if a soo desu ka is to mark the completion of a Q-A-R cycle regardless of whether
the answer to the original question was informative and regardless of whether the interviewer wishes to probe for further information on the
topic or to change the topic altogether.
The Japanese a+ reformulation of the prior turn
Another frequently employed pattern was a plus a reformulation of the prior turn, which is the candidate's answer. The sequence mapping is as follows:
The interviewer's turn 3, a with reformulation, almost always elicits turn 4, a post-post-expansion turn in addition to the three-turn cycle
previously mentioned. The fourth turn is produced by the candidate. Unlike the typical Q-A-R three part sequence, this pattern seems to be
in use to provide the candidate a space for self-repair of the prior turn in the interview talk. Example 5 below is an illustration.
Turn 1 Interviewer: Question
Turn 2 Interviewee: Response (Answer)
Turn 3 Interviewer: A + Reformulation of Turn 2
Turn 4 Interviewee: Confirmation of the Reformulation
The candidate in this example made a linguistic error confusing the pronunciations of karai 'hot' and kirai 'hateful' in Japanese.
He managed to describe what he intended to say (karai) by circumlocution in line 5 'I get hot when I eat', and the interviewer provided the
target word in the pattern a + reformulation. In line 7, the candidate displayed uptake and provided an affirmation by repetition:
karai desu. By means of this sequence, the candidate was cooperatively given a chance to overcome difficulties in the target language,
and hence to perform.
1 A: Kirai.
2 B: ki (.) ki (.) kirai ja nai n desu kedo (.) ano: (1)
[Dis(.) dis (.) it is not dislike but (.) uhm: (1)]
[some lines omitted: 2 is struggling to produce the next line]
3 B: Ano (.) a (.) tabe (.) tabetekara
[Uhm (.) oh (.) eat (.) after I eat]
4 A: Ee
5 B: a (.) tabete uh:: atsuku narimasu.
[A eat uh:: I become hot. ]
6 A: a:: (.) a:: karai desu ka.
[a spicy is it? ]
7 B: karai desu.
8 A: a:
9 B: sumimasen. Kirai wa suki ja nai desu ne::.
[I am sorry. I do not like spicy ( (hateful)) ]
10 A: soo desu ne: (.) hm: (.) a! karai desu ka. soo desu ka.
[Let me see: hm: a! spicy, is it. I see.]
hmm: etto ima Tai kara kita
[hmm: well you came from Thailand]
11 A: To iimashita ne:.
[You said, right?]
The analysis of Japanese a and English LPI interactional data yields several findings. While the change-of-state tokens a (in Japanese)
(in English) were both used by the interviewers, Japanese interviewers used the token frequently whereas English interviewers used it sparingly.
The Japanese token appeared in almost every response turn by interviewers in Q-A-R sequences. The English token appeared in a very limited context;
in this study, it was found only after the candidate completed his or her response and the tester indicated a desire to begin a different task
(e.g., to start a role play, or to initiate a whole new topic of discussion).
". . . the change-of-state token "a" in Japanese functions differently from its so-called equivalent marker in English."
The contrastive analysis of the LPI data suggests that the change-of-state token a in Japanese functions differently from its somewhat equivalent
marker in English. The data seem to confirm that the use of a does not inherently violate the neutrality of the interviewer's footing.
A with its collocated listener response expressions, particularly soo desu ka, merely acknowledges receipt of the prior turn,
indicating to the other: "I understood what you said." In English, the use of the change-of-state token oh in the same places would be evaluative.
In both interview settings, the candidates were given opportunities to perform in a wide range of oral tasks during the interviews. The interviewers
controlled the development of the talk in both settings, so that the tasks were successfully performed. Cross-linguistically, however, how the
interviewers went about to do so differed. In the English cases, the testers withheld the insertion of change-of-state tokens during the candidates'
deliveries of responses to questions so that the candidates would expand their talk on their own. In Japanese interviews, the testers employed
listener responses, particularly the change-of-state token a and collocated expressions, to elicit further language samples. The mean length of
the utterance in the candidate's turn may be shorter in the Japanese cases as a result of interviewer feedback, but by means of token insertion
the Japanese interviewers could pinpoint precisely what they expected to happen next.
For the purpose of evaluating the LPIs as a means of testing a learner's oral proficiency level, this study suggests that a cross-linguistic
perspective is required to better understand what goes on in their actual development. The different listening styles of Japanese and English
testers are likely to be salient and consequential: for instance, English-speaking learners of Japanese may expect testers to adopt a relatively
passive role, while Japanese-speaking learners of English may expect the opposite. Knowing that interview styles can differ across languages
is also important when seeking to establish standard criteria for LPIs.
Clayman, S. (1992). Footing in the achievement of neutrality: The case of news interview discourse.
In P. Drew and J. Heritage (Eds.) Talk at Work, (pp. 163-198). Cambridge: Cambridge University Press.
Clayman, S., & Heritage, J. (2002). The news interview: Journalists and public figures on the air. Cambridge: Cambridge University Press.
Heritage, J. (1984). A change-of-state token and aspects of its sequential placement. In J. M. Atkinson and J. Heritage (Eds.)
Structures of social action (pp. 299-345). Cambridge: Cambridge University Press.
Heritage, J. (1998). Oh-prefaced responses to inquiry. Language in Society, 27, 291-334.
Heritage, J. (2002). Oh-prefaced responses to assessments: A method of modifying agreement/disagreement. In C.E. Ford, B.A. Fox & S.A. Thompson
(Eds.) The language of turn and sequence (pp. 196-224). London: Oxford University Press.
Izumi, E. Uchimoto, K., & Isahara, H. (2004). Nihonjin no 1200 nin koopasu [a corpus of 1200 Japanese speakers of English]. Tokyo: ALC.
Kasper, G., & Ross, S. (in press). Multiple questions in oral proficiency interviews. Journal of Pragmatics.
Kasper, G. (2006). Staying on task in oral proficiency interviews.
Paper presented at the SLA Colloquium Series of the Language Institute of the University of Wisconsin, Madison, October, 26th.
Kim, K. H. & Suh, K. H. (1998). Confirmation sequences as interactional resources in Korean language proficiency interviews.
In R. Young and A. W. He. (Eds). Talking and testing: Discourse approaches to the assessment of oral proficiency (pp. 297-332).
Amsterdam/ Philadelphia: John Benjamins.
Maynard, S. K. (1989). Japanese conversation. Norwood, NJ: Ablex.
Mizutani, N. (1988). Aizuchiron [On Conversation Fillers]. Nihongo-gaku, 7 (12). 4-11.
Nishizaka, A. (2001). Kokoro to kooi: Esunomesodorojii no shiten [Mind and action: An ethnomethodological perspective] Tokyo: Iwanami Shoten.
Rose, K. (1996). Formulae and inter-interviewer variation in oral proficiency interview discourse. Prospect, 11 (3), 3-15.
Saft, S. (2001). Displays of concession in university faculty meetings: Culture and interaction in Japanese, Pragmatics 11, 223-62.
Schiffirin, D. (1987). Discourse markers. Cambridge: Cambridge University Press.
Schegloff, E. (2007). Sequence organization in interaction. Cambridge: Cambridge University Press.
Sudo, J, (2001). Kandooshi a no onseiteki tokuchoo to kaiwa sankasha-kan no shakaiteki kankei
[Social relationship among participants and the prosodic features of a in Japanese] Nihongo-Nihon Bunka Kenkyuu, 11, 117-128.
White, S. (1989). Backchannels across cultures: A study of Americans and Japanese. Language in Society, 18, 59-76.
Uemura, R. (Ed). (1998). CD-ROM Intabyuu keishiki ni yoru nihongo kaiwa deetabeesu [Japanase Spoken Corpus of Intereviews]. Database Vol. 1. Tsukuba, Japan: Tsukuba University.