Abstract
This article deals with a multiunit question format in which an open question is reformulated as a more specific yes–no question or a list of alternatives. The reformulation is considered a technique for preempting manifest or potential problems of formulating an appropriate answer. The reformulated question proposes one or more candidate answers to the original question and thereby guides the interlocutor in the direction of a relevant type of response. The question format is interpreted mainly as a practice for promoting understanding and participation by the nonnative speaker. In addition, it displays expectations of possible answers and may thus be considered a way of demonstrating common ground and thereby achieving affiliation. On the other hand, the alternatives proposed may on occasions restrain the freedom of the interlocutor to formulate an independent answer and may then be considered a dominating move.
Introduction
A striking feature found in conversations between native speakers (NS) and nonnative speakers (NNS) is that many questions by the NS are reformulated, that is, presented in a second version shortly after they have been formulated originally. Here are two forms this may take: (1) 1 S: hva- hva var det 2 wha- what was the first thing you did when you came here? 3 (.) 4 S: begynte du å 5 did you start working or, did you start taking courses or, (2) 1 S: [hvor] 2 where did you go then? 3 S: 4 Treider or something like that?
In the first example, the original question is followed by two questions, which specify the original question and present alternative types of answers to it. In the second, a request for a specific piece of information is followed by a tentative suggestion of an answer. The assumptions presented in the reformulated questions thus present candidate answers to the original question. The examples will be analyzed in more detail below.
The current article describes a set of practices involved in reformulating questions by adding candidate answers. It analyzes what sorts of contexts they tend to arise in and what is achieved by such reformulations. Common to them all is that the original question is an independent turn constructional unit (TCU), that is, a syntactically and prosodically complete contribution that is open to response by the interlocutor. The reformulated question is in some sense equivalent to it in that it does not alter the request for information significantly but presents tentative suggestions for possible answers to it.
The questions I seek to answer are in what ways the reformulation may help interlocutors produce an adequate answer, on the one hand, and to which degree it restrains their freedom of action, on the other. Furthermore, I analyze the interpersonal effects of reformulations, to see whether they contribute to a common ground and affiliation or rather stress the asymmetric relationship between the parties.
The data for the study come from video recordings of consultations at various social welfare offices in Norway, namely, an unemployment office, a job-qualifying center for immigrants, and a municipal office assisting immigrants and refugees during their first 2 years of residence. The corpus consists of 12 consultations with a total recording time of 5 h 6 min. In all cases, the clients are NNSs of Norwegian, and the social workers are NSs. Most of the clients come from Pakistan, but there are also informants from Iraq, Bosnia, Sri Lanka, Ghana, and Vietnam. A collection of 40 instances of reformulated questions were gathered and analyzed in depth using conversation analytic methodology.
Previous studies of reformulated questions in institutional interaction and second-language conversation
In the conversation analysis (CA) studies of institutional talk, several researchers have noted the existence of multiunit questioning turns. For instance, in the field of broadcast interviews, questions may be produced in several complex formats, including prefatory statements providing background for the question, multiple questions in so-called question cascades, turn increments, and so on (Clayman & Heritage, 2002; Heritage & Roth, 1995). It may seem that complex question formats are typical of institutional interaction, and Linell, Hofvendahl, and Lindholm (2003) find them in a range of settings, such as courtroom trials, police interrogations, and consultations in health care and social welfare offices. As for the subset of multiple questions, they note that subsequent versions may constitute either a specification of the previous question, a synonymous reformulation of it, or a generalizing invitation to respond.
Of special relevance to the current topic is a series of studies of questionnaire-based interviews (Antaki, 2002; Houtkoop-Steenstra, 1997; Houtkoop-Steenstra & Antaki, 1997). These studies show that interviewers frequently reformulate or revise questions in cases where they observe that interviewees have problems with responding in accordance with the given response alternatives. In such cases, they initiate third turn repair and redo the initial question in the form of a yes–no question that preempts or minimizes troubles in the interviewees’ responses or encourages “positive and optimistic” responses (Houtkoop-Steenstra & Antaki, 1997).
In the field of second-language communication, already Hatch (1978) noted that NSs tended to repair WH-questions by YES/NO questions or by so-called OR CHOICE questions when the original question seemed to cause a problem. She explained this practice as a device for making it easier for the learner to participate in the conversation: “He need only agree or disagree with the YES/NO questions, take one of the choices offered him by OR CHOICE questions, or take one of the answers suggested by the NS” (Hatch, 1978, p. 419). A quantitative comparison of NS–NNS conversation with NS–NS conversation carried out by Long (1984) gives evidence that this feature is characteristic of NS–NNS conversation. He found that NSs used considerably more questions when talking to NNSs (66% of all turns) than when talking to fellow NSs (16%) and, furthermore, that the percentage of OR CHOICE questions was also much higher (21% vs. 11%).
Two recent CA studies investigate multiple questions in second-language conversations. Gardner (2004) investigated “extended question sequences” in informal conversations and found two distinct practices, which included second versions of questions. One of them is produced in pursuit of a missing answer: “First there is a question, which is followed by a silence beyond the normal transition space, which in turn is followed by the pursuit of an answer, by some form of redoing the original question” (p. 256). The other practice consists in immediate question expansions, that is, with no intervening silence between the parts. These “redoings” are observed to clarify the question and make it more explicit. Gardner’s conclusion is that a major focus of the expansions was the avoidance of disagreement and more generally the avoidance of dispreferred response, and the general paucity in the talk of the SL-speakers of markers of dispreference appears to have been a factor in this. (p. 265)
Kasper and Ross (2007) analyzed multiple questions in oral proficiency interviews. They also make a similar distinction as Gardner (2004) in that they distinguish between subsequent versions that are “reactive” and “proactive.” The former are responsive to some indication that the interlocutor has trouble answering the original question, namely, repair initiations, silences, and problematic answers. The “proactive” multiple questions differ in that they occur in the same turn without any gap or problem indication. They are described as preemptive techniques used to enable relevant responses in interactional environments that are especially vulnerable to trouble, such as “topic changes and shifts, requests for extended verbal action beyond ‘answering questions,’ hypothetical circumstances or events, and delicate matters” (Kasper & Ross, 2007, p. 2066).
Delimitation of the phenomenon
The previous studies make a highly relevant distinction between reformulations that occur as self-repair in a third turn (after some indication of trouble by the interlocutor) and same-turn, integrated reformulations. They have also identified contexts of occurrence and potential types of trouble they are used to preempt. However, a problem with several of the previous studies is that they treat multiunit question turns as a single and homogenous category and ascribe functions to the category as such. I will argue that interactional functions are rather associated with a range of different and more specific practices that all may be labeled “multiunit questions” or “extended question sequences.” If we take only the case of reformulations, we may distinguish at least three different practices that have very little functional similarity. The first class consists of reformulations that present candidate answers to the original question (such as in (1) and (2) above), the second consists of synonymous paraphrases, and the third consists of reformulations that change the preference structure of the question. Before going into the analysis of the first type, which is my main concern in this article, I will briefly outline some characteristics of the other two types, in order to distinguish them from my object of inquiry.
A synonymous paraphrase merely poses the “same” question in other words. In second- language conversations, this usually implies simplifying the question and making it more explicit, such as here: (3) 1 nei men du er ikke redd for å 2 no but you aren’t afraid of trying in that case, 3 hvis du kan: m: (.) 4 if you can m .. get a job you want to try?
In this case, the (conventionally) metaphorical expression “afraid of trying” is substituted with a more congruent one (“want to try”), and some more contextual background is provided (“If you can get a job”). In paraphrases such as this one, there is no candidate answer offered, and the fundamental question format is not altered, only the specific formulation of it.
A different type of reformulation is found in cases where the preference is reversed in the subsequent version of the question, as described by Sacks (1987): A sort of thing you get is: the person asks the question exhibiting a preference, and there is a pause, no answer, and they then revise that question to exhibit the reverse preference, and then (with no further delay) they get an answer in accord with it. (Sacks, 1987, p. 64)
This is what we find in the following two examples: (4) 1 S: i- i Pakistan a var du 2 i- in Pakistan where you at home there too or what? 3 hhh 4 S: 5 did you work or go to school in Pakistan? 6 C: ja 7 yes a bit job.
Here, the original question is designed to prefer one type of answer (that the interlocutor was at home) and is subsequently replaced by a reformulated question, which prefers another (that she worked or studied). As we see, the first question does not immediately generate a response (as can be seen in the prolonged outbreath in line 3 during which the client could have started speaking), and this delay may indicate to the speaker that a dispreferred response is forthcoming. He thus produces a revised version, displaying an opposite expectation and this time receives a (preferred) response.
Reformulations that reverse the preference structure may take the form of alternative questions in the same way as candidate answers (cf. (1) above). However, in contrast to the latter, they present an opposite alternative to the original question rather than a set of alternatives as potential answers to it. This is shown in the following example, where they are talking about the client’s wish to become a mother-tongue teacher for bilingual children (referred to as “bilingual” below): (5) 1 S: eh: (.) jeg bare tenker 2 eh … I’m just thinking that eh do you have any alternatives? 3 no andre: 4 some other job wishes than bilingual? 5 (.) 6 S: eller er det 7 or is that what you would like to try first. 8 G: ja 9 yes I will try first eh 10 S: ja. 11 yes.
Here, the first question does not receive a response, even after having been further specified in line 3. The speaker then reformulates the question by inquiring about the opposite alternative (line 6), and this question does get a response.
In contrast to most previous studies of multiunit question turns, I believe that these different types of reformulation constitute different practices with distinct functions. I therefore do not believe that it is possible to assign functional characteristics to the whole class of multiunit questions but that this will have to be done at the level of more specific practices. The current study is thus an attempt to do just that, by investigating one single type of multiunit question turn, namely, reformulations with candidate answers.
Candidate answers
In asking questions, people have the choice of asking an open question that does not display an expectation of what the answer will be (such as “Where were you yesterday?”) or offering for confirmation of one’s assumption about the state of affairs (such as “Were you at the movies yesterday?”). In the following excerpt, the speaker initiates an open question but cuts it off midcourse and replaces it with a more specific question, which displays his assumption about the situation: (6) 1 S: da får du bare: (1.0) gå på 2 then you just ought to … (1.0) go to the unemployment office and talk to them, 3 hvilket arb- 4 which unem- 5 er det her 6 is it down here you use or? 7 K: mhm.
Since the speaker here aborts one way of asking the question and replaces it with the other, the two ways of asking are oriented to as having different functions or conditions of use.
Pomerantz (1988) analyzes questions that present the speaker’s assumptions (such as the latter in (6)) in terms of “offering a candidate answer.” She notes that this way of formulating questions (in contrast to open questions) gives the interlocutor “models” of the types of answers that would satisfy their purpose for asking. This, she notes, is a useful practice when speakers have a reason to guide their interlocutor to respond in a particular way, either because they are interested in some particular piece of information or because the interlocutor has difficulty in giving a satisfactory answer (p. 367). Furthermore, speakers display their knowledge of, and familiarity with, the situation. Choosing an open or a closed question thus also has consequences for the degree of common ground established between the interlocutors.
This understanding of candidate answers is the point of departure for this investigation. In contrast to the questions studied by Pomerantz, however, the questions in this study are delivered in two rounds, first as open questions and subsequently as questions that offer a candidate answer to that question. The analysis begins with reformulations that come in response to some indication of trouble with producing an answer. The next part then analyzes same-turn reformulations that occur without any such indication. After that, the patterns of response to reformulated questions are outlined, and the use of the question format is discussed in relation to the institutional activity type and the second-language context.
Reformulations in pursuit of a response
Reformulations regularly occur in contexts where an open question does not generate a response or only generates an inadequate one. In the following example (an expanded version of extract (1)), the coparticipant displays having problems answering the original question: (7) 1 S: eh (.) så kom du til 2 eh then you came to Norway. what was the first thing you did after you came here? 3 (1.2) 4 K: her? 5 here? 6 S: ((kremt)) ja, 7 ((throat)) yes, 8 hva- hva var det 9 wha- what was the first thing you did when you came here? 10 (.) 11 S: begynte du å 12 did you start working or, did you start taking courses or, 13 K: [ eh ] ne:i 14 eh no never been to school or courses.
First, the recipient of the question initiates repair. The repair initiator is a category-specific request for information about location (“here?”). However, the original speaker (S) treats this repair initiation as possibly indicating a more encompassing problem and proceeds to repeat the question (cf. Svennevig, 2008). When this second question also meets with silence, he starts reformulating it, and the subsequent questions take the form of a series of suggestions of potential activities that he might have engaged in. This solves the problem, and the recipient produces an answer that disconfirms one of these candidate answers.
Another case of problematic recipiency is when there is indeed an answer to the question, but it is treated as inadequate. This is the case here: (8) 1 S: Kan du fortelle litt om hva du (.) har gjort 2 Can you tell me a bit about what you have done after you came to Norway? 3 C: Ingenting. 4 Nothing. 5 S: Ingenting? 6 Nothing? 7 C: Ja ha ha ha [ha ha] 8 Yeah ha ha ha ha ha 9 S: [ha ha] ha 10 .h Du kan vel ikke ha gjort 11 (you can well not have done nothing) 12 You can’t have done nothing can you? 13 C: m:hm:, 14 S: Du må vel ha gjort 15 You must have done something? 16 C: ha ha ha 17 S: Hæ? 18 Huh? 19 C: .hhh 20 S: Har du gått på no 21 Have you attended any Norwegian courses? 22 C: eh ja 23 uh yes
In this extract, the social worker starts with a very open and wide question in the form of an invitation to tell about what she has done in Norway. The answer “nothing” is treated by the social worker as inadequate in that he initiates repair in all three times (lines 5, 10, and 14). However, none of these initiations present any candidate solutions to the problem, and they do not result in a modified answer. He then reformulates the question as a candidate answer—a specific yes–no question about one single activity, namely, language courses (line 20)—and then receives an answer.
In the cases analyzed here, the reformulations seem to help the interlocutors produce an answer in that they provide them with models of an answer. The problem they address is the question of relevance: what sort of activities is being asked for? In the last example above (8), the client displays an interpretation of the question in her answer (“nothing”), implying that what she has actually done in Norway is not relevant as an answer to the question. This assumption is what is questioned by the social worker in his repair initiations, and when the client still does not provide any relevant activities, the reformulated question proposes a candidate. The reformulations thus address potential problems of relevance associated with the original question and propose a solution by formulating it in a more specific and explicit way.
In the examples above, the reformulations are new full questions, formulated as main clauses with interrogative syntax. A second class of reformulations is “guesses” in the form of concrete suggestions for answers to the original question. These rather take the form of phrasal TCUs, produced with rising intonation. We find an example in (9), where the client has just told the social worker that she reads newspapers: (9) 1 S: hva er det du 2 what is it you read then? 3 A: he he he [he] 4 S: [an]nonser? 5 the classifieds? 6 A: HA HA HA HA HA HA 7 ja: noen: jeg 8 yes some I don’t know but I read
The question is responded to by laughter, and the social worker deals with this lack of response by suggesting “the classifieds” as a possible answer. This “guess” is a display of the speaker’s expectations and knowledge of the interlocutor’s situation in that he relates to the fact that she is looking for a job.
In all these examples, we see open questions (a wh-question in (7) and (9) and a request for an extended account in (8)) reformulated as requests for confirmation of specific alternatives. This amounts to a certain “closure” of the question, both semantically and interactionally. Semantically, it restrains the thematic range of the question from a broad to a narrow focus, and interactionally, it reduces the projected response from an extended account to a simple confirmation (more about responses to such questions below).
In the last two examples, the question is replaced by a single, closed question or suggestion. However, in many cases reformulated questions take the form of open-ended lists of alternatives, as in (7) above. Typical of these list constructions is the final particle “eller” (or) and the rising intonation on each of the alternatives, which indicates that the list is open-ended (cf. Jefferson, 1990; Selting, 2007).
1
This marks the suggested alternatives as only some of a wider range of possible answers and thus counteracts somewhat the narrowing down of response alternatives (Linell et al., 2003, p. 552). In some cases, the open-ended character of the list is further stressed by adding a new reformulation, this time returning to an open question: (10) 1 S: og 2 and now you have enrolled for a test here. and registered at the unemployment office 3 S: hva betyr 4 what does that mean, does it mean that you’re looking for a job or? eh 5 (1.5) 6 S: 7 courses or? 8 (.) 9 S: hvilke planer 10 which plans do you have. 11 K: mh: det er 12 mh it’s both. 13 S: begge 14 both? 15 K: ja. 16 yes
As the candidate answers do not receive a confirmation or a disconfirmation by the interlocutor, the speaker produces a new reformulation, which opens up to a wider range of response options: “which plans do you have?”
Immediate reformulations
As noted by Gardner (2004) and Kasper and Ross (2007), many question expansions are produced immediately after the first TCU is completed and are thus not reactively oriented to a manifest problem of answering the question. Here is an example: (11) 1 S: har du 2 have you talked med someone at the unemployment office (or)? 3 C: ja. 4 yes. 5 S: (.h)ja 6 yes 7 hva 8 what did talk about then? 9 S: om 10 about job or courses or, 11 C: .. eh: jeg 12 .. eh I apply first eh for a- a job, 13 S: ja, 14 yes, 15 C: men dessverre ikke, ((shakes head)); 16 but regretfully not,
After having brought the first question to completion, both syntactically and prosodically, the speaker here immediately starts reformulating the question by offering candidate answers in an open-ended list construction. This instance resembles the examples investigated in the previous section in that the reformulation presents models for potential answers and thus helps the interlocutor in producing an appropriate and adequate answer. However, here the help is preemptive rather than reactive and thus seeks to avoid a potential problem rather than a manifest one. And in some cases, the problem may not be obtaining an answer but rather what sort of an answer one is seeking. Consider the following example: (12) (IFF 4) 1 S: men e::m (.) hvis du nå 2 now if you don’t get a taxi course from the unemployment office, do you have any 3 al 4 alternatives? 5 D: [ .h nei ] 6 .h no 7 S: [tenker du å gå på] noe 8 [do you plan on taking] any Norwegian training, 9 D: [ nei da ] 10 no then 11 S: [eller tenker du] på å søke noen 12 or do you plan to apply for any other jobs or? 13 D: ne:i egentlig så: for å få jobb på butikken, ↑ 14 no actually cause getting a job at the store, that isn’t hard. 15 S: nei så 16 no so that you can manage yourself. 17 D: det kan jeg [greie sjøl.] 18 that I can manage myself. 19 S: [ ja ja ] 20 yeah yeah
This reformulation is clearly not oriented to problems of getting an answer, since the interlocutor tries to take the floor twice, in line 5 and 9, but is driven to relinquish it by the social worker’s continued talk. Rather than a technique for generating or pursuing a response, this form of reformulation must be considered as oriented toward guiding the response in a certain direction by narrowing the response options. In the current case, the social worker signals that he is not just interested in any alternative but only a more restricted subset of alternatives. In this way, he displays that the question is not response ready at the first completion point, and that, consequently, he is not yet ready to give the interlocutor the floor, even if the latter shows readiness to start responding. In this sense, the current type of reformulation is indeed forward oriented (or “proactive”) in that it prepares the ground and specifies the conditions for the forthcoming response. In the current case, it may also be considered a preemptive technique, oriented to avoiding inappropriate or inadequate responses. As we see, the “alternatives” that are specified in the reformulation are courses of action that potentially involve services provided by the institution in question (assistance with job-seeking and job-qualifying courses). The reformulation thus guides the interlocutor toward a specific type of answers that are relevant to the conversational activity at hand, namely, professional counseling, and desirable from the perspective of the institution’s goals and societal role.
Also “guesses” presenting specific answers may be produced immediately after the question, such as here: (13) 1 D: men 2 but afterwards when I was starting to take this business education, 3 S: ja: hva var 4 yes what was that? 5 D: det var sånn 6 it was a business diploma with reduced portfolio. 7 S: så du har tatt 8 so you’ve taken that? 9 D: ja. 10 yes. 11 (. ) 12 D: [( )] 13 S: [hvor] 14 where did you go then? 15 S: 16 Treider or something like that? 17 K: sånn eh: eh: 18 this eh eh no it was eh N NKI. 19 S: NKI ja. 20 ok NKI. 21 S: ettårig 22 a one-year correspondence course or something like that then? 23 D: nei jeg tok eh: første to fag i: søtti 24 no I took eh first two classes in seventy-eight, 25 S: ja. 26 yes.
Immediately upon completion of the question, the speaker proposes the name of a school as a candidate answer to his own question (line 15). That it is “just a guess” is indicated by the tag “or something like that?,” which opens for other alternatives as well. Clearly, “guesses” such as this one contrast with the former example in that they do not narrow the response options. Rather, they seem to be involved in displaying the speaker’s own knowledge of or familiarity with the situation (cf. Pomerantz, 1988). In the current example, displaying his familiarity with schools offering the specific type of education mentioned provides evidence for the interlocutor that they may have some common background knowledge. This may be useful both for establishing successful reference in the following talk and maybe even for creating affiliation between the parties. The social worker’s involvement in creating common ground is further evidenced in that he makes another guess in line 21, this time about what sort of course he took. However, this extract also illustrates the risks one runs in making guesses. Both suggestions are rejected by the interlocutor, and thus, the opposite is achieved, namely, a lack of affiliation and common ground.
Many of the previous studies of multiple question sequences have only focused on their role in dealing with problems in conversation and have not mentioned this type of reformulation. For instance, Gardner (2004) claims that a common factor of all extended question sequences in his data is that there is some “perceived inadequacy with the first version of the question” (p. 257) and that “a major focus of the expansions was the avoidance of disagreement” (p. 265). As we can see, none of this is involved in this type of question sequence.
Responses to reformulated questions
As noted, reformulated questions most often take the form of requests for confirmation or various forms of alternative questions, usually in the form of list constructions. Requests for confirmation may be answered by a simple confirmation, as in (6) and (8) above. However, a disconfirmation of a proposed candidate answer will in general not be an adequate answer in itself. It will systematically be expanded by providing the “correct” answer. For instance in (13) above, the guess about which school the interlocutor went to (Treider) is disconfirmed, and the negative answer is followed by the name of the school in question (NKI). This practice of answering “more than the question” in cases of disconfirmation has been described by Sacks (1992). He noted that certain questions propose a certain class member as a candidate answer and thereby can be seen as eliciting another class member in cases of disconfirmation (Sacks, 1992, vol. 1, p. 21f on the “correction invitation device”).
The answer presented in addition to the disconfirmation relates back to the original question and thus shows that it has not been canceled by the reformulation. For instance, in (12) above, the rejection of the alternatives suggested by the social worker is followed by an account of what the client sees as his alternative to taking a taxi course, namely, returning to his old job at the store. The original question thus remains relevant and has consequences for how the answer is produced.
Whereas reformulations in pursuit of an answer constitute an emergent phenomenon, due to the emergence of a problem, immediate reformulations are designed to present two (or more) questions in a single turn. There may be certain advantages in this particular design, in which one question is open and projects an extended answer, and the other is narrow and projects a minimal answer. The reformulated question makes it easy to produce a local answer and thus fulfill the conditional relevance of the first pair part. The original question, however, gives the opportunity (but not the obligation) to expand the answer in certain directions. In this way, same-turn reformulations seem to be a practice for inviting but not requiring, extended answers, while simultaneously giving the opportunity of merely responding minimally. This may be useful in contexts where the interlocutor has limited oral proficiency in the language spoken. In (9), for instance, the client accepts the suggested candidate answer (“classifieds?”) but seems to try in what follows to expand on the matter (although she subsequently gives up and abandons the turn initiated).
Immediate reformulations involve a systematic violation of the preference for contiguity between first and second pair parts of adjacency pairs (Sacks, 1987). Having produced a complete question as a recognizable first pair part, the speaker has in principle brought the turn to a transition relevance place (TRP) and allocated the next turn to the coparticipant. In example (12), we see that the interlocutor orients to this expectation by starting to respond after the original question has been produced. In this situation, then, speakers will have to employ special techniques for holding the floor beyond the TRP. This involves not leaving a pause before starting to reformulate and not producing conventional paralinguistic signals of TCU completion, such as reduction in tempo and loudness, lengthening of final vowels, and so on. Instead, they perform a quick transition into the next TCU, in the form of a so-called “rush-through” (Local & Walker, 2002; Schegloff, 1982).
The reformulated question may thus be considered a special type of speaker-initiated insertion, delaying, but not canceling, the conditional relevance of a second pair part. As forward-oriented insertions laying the ground for a particular type of response, they can be characterized as “pre-seconds” in the terminology of Schegloff (1990, 2007). The insertions that pursue a missing response may also be characterized as a form of insertion, but they are mainly oriented backward, toward a problem with the prior question, and may thus rather be characterized as “post-firsts.”
The role of reformulated questions in different types of interaction
As noted above, multiunit questions have mainly been observed and described in institutional interaction, but Gardner’s data are informal conversations between first and second-language speakers. The question then arises whether reformulated questions are associated with specific activity types or with native/nonnative interaction. As noted, I do not think this question can be answered for the whole class of multiunit questions as such but has to be addressed according to the various practices involved in different sorts of reformulations and other question extensions. I therefore inspected a corpus of informal NS interaction involving unacquainted interlocutors (cf. Svennevig, 1999). In this corpus, amounting to more than 4 h of conversation, I did not find a single instance of reformulated questions that presented candidate answers in the form of new, more specific questions of the type found in (1) above. 2 However, I did find certain instances of “guessing” (as in (2) above), and I also found a number of reformulations that reversed the preference structure of the question (as in (4) and (5) above). This seems to suggest that offering candidate answers in the form of new questions is a practice primarily associated with second-language conversation or institutional interaction, or both.
For someone with a limited proficiency in a language, it is easier to confirm a candidate answer presented by the interlocutor than to author and formulate an answer oneself. In addition, reformulations make the question easier to understand in that they make it more explicit and more specific and contextualize it more. This may indicate that reformulated questions have a special role in second-language conversations. Furthermore, my impression from analyzing the instances in the corpus is that they are more frequent in conversations involving second-language speakers with low proficiency than in conversations with more fluent speakers (see for instance excerpts (9) and (11) above).
In addition, there are obvious reasons for why this question format is especially frequent in institutional interaction. Many researchers have noted that a recurring source of trouble for clients in institutional interaction is the fact that questions are based on predefined questionnaires or lists and are introduced into the conversation in a topically disjunctive manner (Drew & Heritage, 1992). The clients thus have few contextual cues in the previous conversation as to the purpose for asking and, consequently, what may be a relevant answer. One solution to this problem may be to introduce contextual cues into the question itself, either by presenting background information in question prefaces or by reformulating the question in more specific and explicit candidate answers. This finds support in many previous studies which observe that multiunit questions frequently occur in topic shifts (Kasper & Ross, 2007; Linell et al., 2003).
In addition, clients do not always know the institutional criteria behind the formulation of a question and may thus have problems in knowing what is relevant to report. Questions such as “what have you done since you came to Norway?” and “do you have any alternatives?” are wide and unspecific, and may be interpreted as a request for information about a range of different activities, only a part of which are institutionally relevant. As we see from example (8) above, at least one client does in fact have institutionally relevant activities to report (taking Norwegian courses) but does not think of it as relevant given the original formulation of the question. Reformulations are thus clearly functional in guiding the interpretation of the original question toward what may be a relevant answer.
Guiding the answer in a certain direction by means of a preemptive reformulation in the same turn may in many circumstances be an appropriate action, especially when the speaker anticipates that the interlocutor may have a problem giving a relevant answer or producing an answer at all. However, it may also constitute an unnecessarily dominating move. By suggesting answers on behalf of their interlocutors, speakers restrain their freedom of action and restrain their contribution to a choice between alternatives defined by them. Furthermore, the institutional representative may miss important information by not letting the client define what is relevant to the report. There is thus a real danger of “putting words into the mouths” of the clients and depriving them of the right to author their account of the situation (cf. Goffman, 1981).
The NSs in the corpus may on different occasions be considered as relating differently to this risk. In (8) above, the social worker gives the client several opportunities to give her own account of what she has done since she came to Norway by just producing open repair initiators and withholding any suggestions. Only after this approach has failed repeatedly does he offer a candidate answer. In (12) however, the opposite occurs—the client shows readiness to respond, but the social worker does not give him the opportunity. Reformulations of questions and answers thus involve an inherent tension between assisting interlocutors in finding words and restraining them in their freedom of action. Offering a candidate answer in an unwarranted situation may accentuate the asymmetry between the parties–in this case an already strong asymmetry both in terms of linguistic proficiency and institutional authority.
However, not all reformulations discussed here deal with problems. The class labeled “guesses” seems rather to present the speaker’s expectations and thereby a claim to be knowledgeable about the interlocutor’s situation. As noted above, this may contribute to establishing common ground between the interlocutors and to creating affiliation. Displaying one’s expectations may also be considered a sign of interest and involvement in the topic. In the current institutional context, this may then be considered a way of creating a more personal and informal style and thus of reducing the institutional character of the conversation.
A candidate answer expresses an assumption about the interlocutor to a larger extent than an open wh-question. These assumptions may portray the interlocutor in socially acceptable or unacceptable terms. In the current corpus, the candidate answers all express expectations that the clients wish to qualify themselves for a job and start working (as in (12) above). This is in line with the positive and “optimistic” assumptions found by Houtkoop-Steenstra and Antaki (1997) in their studies of reformulated questions in survey interviews. In the current context of professional counseling, these positive portraits of the clients are not just functional in creating affiliation but also in encouraging the sort of activities that the institution wants their clients to engage in. The candidate answers thus implicitly convey a model of the “ideal job seeker” that the interlocutors are expected to conform to and convey a preference for agreement with the suggestions offered.
Conclusion
Multiunit questions can take many forms and have many different functions in various activity contexts. I have analyzed one specific type, namely, reformulated questions that offer a candidate answer. Even this format turns out to involve two functionally rather distinct practices, namely, “guesses” (in the form of phrasal TCUs) and new, more specific questions (in the form of interrogatives). The analysis shows that the former are mainly used to show involvement in the topic, seek common ground with the interlocutor, and thereby invite affiliation. The latter are used to guide the interlocutor toward a certain type of answer. This occurs first and foremost in contexts when the interlocutors display problems in producing an answer and may thus be considered a practice for pursuing a response by helping the interlocutor find and formulate an appropriate answer. This practice may thus not only be associated with talk addressed to second-language speakers but also to other speakers who lack competence in the activity type in question. Reformulations also occur as an integrated question format within the same turn, and in such cases, the reformulation is used to invite certain types of answers and, consequently, to discourage others. This is shown to be primarily associated with the exigencies of institutional interaction, where the institution has defined in advance which types of answers are relevant to its categories and procedures.
Footnotes
Funding
This research has been partly funded by a grant from the Norwegian Research Council.