Introduction
Well, the little words are often the most interesting. This study investigates the discourse marker well in seven criminal jury trials in Auckland, New Zealand. As Aijmer Reference Aijmer2002 points out, a number of approaches have been used to consider well in the past, and each has something useful to offer, although more recent work has tended to choose one aspect rather than considering its variety. This article is based on the premise that a combination of approaches is likely to provide a more comprehensive and explanatory picture than would proceeding from a single viewpoint. The techniques of ethnography of speaking, Conversation Analysis (Sacks, Schegloff & Jefferson 1974), and interactional sociolinguistics (e.g. Drew & Heritage 1992) are applied, each considering a different contextual level (see Stubbe, Lane, Hilder, Vine, Vine, Marra, Holmes, & Weatherall 2003 for a discussion of the intersection of these). The combination facilitates a “functional-communicative perspective” (Mey Reference Mey2001:10), which is found to be more useful than focusing on social aspects such as powerlessness, gender, and ethnicity. However, an overall framework is also needed to answer the question “Why well?” Politeness theory (Brown & Levinson 1979) is considered and found to cover some but not all uses of well in these courtroom data. Relevance theory (e.g. Blakemore Reference Blakemore2002) provides a more useful framework in that it can account for all uses in an overarching way and in turn receives an application based in naturally occurring data.
It is well accepted that pragmatic markers, many of which are multifunctional, have interactional aspects. Such markers are “the linguistically encoded clues which signal the speaker’s potential communicative intentions” (Fraser Reference Fraser1996:68). One class of these, discourse markers, can be defined as “those natural language expressions whose primary function is to facilitate the process of interpreting the coherence relation(s) between a particular unit of discourse and other surrounding units and/or aspects of the communicative situation” (Risselada & Spooren Reference Risselada and Spooren1998).
Schiffrin (Reference Schiffrin1987:31) defines discourse markers, operationally, as “sequentially dependent elements which bracket units of talk.” By this she means that the markers “are not dependent on the smaller units of talk of which discourse is composed” (1987:37) and cannot be explained solely by looking at syntax. Her more theoretical definition says that markers are “members of a functional class of verbal (and non-verbal) devices which provide contextual co-ordinates for ongoing talk” (1987:41).
Redeker (Reference Redeker1991:1168) suggests a tighter definition and a change in terminology: “A discourse operator is a word or phrase – for instance, a conjunction, adverbial, comment clause, interjection – that is uttered with the primary function of bringing to the listener’s attention a particular kind of linkage of the upcoming utterance with the immediate discourse context.” Lenk notes that discourse markers “are used pragmatically, with a structuring and organising function” (1998:246). She points out that they can act globally as well as locally, showing relationships to other parts of the conversation. Others, such as Fischer Reference Fischer2000 and Blakemore Reference Blakemore2002, have continued to develop a picture of discourse markers but do not change the definitions markedly. It is in the theoretical framework and categorizations of functions that further advances have been made.
The body of work on language in legal settings is well known and will not be traversed here, other than noting the courtroom’s reliance on the adjacency pair structure (Atkinson & Drew 1979), the question of power/control (e.g. O’Barr Reference O’Barr1982, Danet 1984), and the New Zealand context (Lane Reference Lane1988). This article expands the picture by looking at all the uses of well by all participants in the examination phases of trials. It considers the different locations in which well occurs (in terms of both courtroom genre and conversation structure, and in both questions and answers) and the parts that social indicators, institutional roles and individual goals play. Finally, it considers how politeness and relevance can account for well at a more theoretical level.
Literature Review
Citing a range of previous research, Schiffrin Reference Schiffrin1987 notes that well can (i) be an interjection, filler, particle, hesitator, and initiator, (ii) begin turns, (iii) be a pre-closing device, offering an opportunity to return to an earlier topic or to open a new one, (iv) shift talk toward topics of mutual concern, (v) preface insufficient answers to questions, and (vi) preface disagreements and dispreferred moves. It appears when what is about to be said does not fit fully with the coherence options given immediately beforehand. She suggests that, unlike oh, well has no cognitive (referential) function, its purpose being solely interactive. Further, it “locates a speaker as a respondent to one level of discourse and allows a temporary release from attention to others” (Schiffrin Reference Schiffrin1987:127). She also notes that well “focuses on both speaker and hearer – for the one who uses ‘well’ is being defined as a respondent (a type of hearer) in relation to a prior speaker’s expectations who must also alter his or her expectations about the course of upcoming talk” (1987:323). Thus well focuses on both prior and upcoming utterances.
Schiffrin finds that in answers well appears more often (i) when answer options are “relatively” (1987:104) limited by the preceding question (more often before answers to wh-questions than before yes/no, disjunctive, and tag questions), and (ii) when the answer departs from the options provided by the question, for example through lack of knowledge. It can occur when expanding a minimal answer, in answer deferral, in repair, and after requests couched in forms other than questions. In contingent requests, Schiffrin suggests that it signals a response to both the preceding question and its answer; that is, it attends to more than one part of the exchange. Further, it has a role when expected appreciation is not offered, and in cooperative turn-taking. Her data, however, come from sociolinguistic interviews, and her conclusions seem particularly related to that situation.
In discussing commentaries on snooker games, Greasley Reference Greasley1994 too finds that well marks response, signaling that its speakers do not accept some aspect of a previous move or situation. He suggests that well’s function is to “take-up” a problem, consistent with Schiffrin, although it is stretching the point to imply that expressing surprise over a good snooker shot is a “problem.”
Brief discussion of well appears in Sai-hua Kuo’s (1994) article on agreement and disagreement in phone-in radio talk. She finds that speakers use well when delaying dispreferred responses and argues that the politeness principles are fundamental in speakers’ choices, at least for the “less powerful speaker” (1994:111). It is somewhat limiting to suggest that politeness is a tool used only by the powerless: it can also be used by those with power for a variety of reasons.
In their review of accounts, Morris, White & Iltis 1994 mention that well shows that there will be some delay before the account itself. They define an account as “a description that reports trouble accomplishing what is expected ordinarily and, therefore, is understood and credited by its recipient as an explanation for a divergence from assumptions about what ordinarily will or should happen” (1994:130). They make the useful point that “people design accounts for particular recipients, on the fly, in light of what those persons have just said or might say … and … social relations and for their own interests, values and preferences” (1994:142).
Well is a “hesitating preface” and a positive politeness strategy, according to Holtgraves Reference Holtgraves1997. He reminds us that Conversation Analysis does not refer to psychological states and would therefore look at well only in terms of preference organization. His study is based on a small number of artificially contrived conversations in which the participants could not see each other. He concludes that, despite the findings of Conversation Analysis, linguistic devices used for positive politeness in arguments “must be largely motivated by the interpersonal motives of the interactants” (1997:236).
O’Barr & Atkins Reference O’Barr, Atkins, McConnell-Ginet, Borker and Furman1980, who include well in their list of powerless language features in the courtroom, call it a “meaningless particle” (1980:101) within the category of hesitation forms. This well-known study confines itself to the language of witnesses and therefore mainly to answers. It does not address well’s more powerful uses (e.g. in lawyers’ questions), nor does it consider its range of functions.
In her consideration of discourse markers during courtroom testimony involving Spanish interpreters in New South Wales, Hale Reference Hale1999 follows Schiffrin’s (Reference Schiffrin1987) definition. She claims that in cross-examination questioners use well to reject an answer or to provoke a disagreement by the witness or defendant; it is “a sign of contradiction and confrontation, expecting disagreement” (Hale Reference Hale1999:60). In examination-in-chief, where she finds well to occur rarely in the questions, she claims it is used as “a sign of frustration when the witness is not providing the desired answers” (1999:67). Following Hudson Reference Hudson1975, she categorizes the two uses as negative and positive conducive, respectively (i.e., expecting disagreement or agreement). However, she provides few examples and also looks only at questions.
Finally, in the courtroom context, Heffer Reference Heffer2005 discusses well briefly in his corpus-based analysis of jury trial language, although in the context of commenting on the skewing potential of idiolects rather than providing an analysis of well in his data. He characterizes it as an interpersonal discourse marker and notes that its use for adversarial purposes means that it is found more often in cross-examination than in direct examination.
Fischer’s (Reference Fischer2000) study of discourse markers is concerned with their use in automatic processing systems. Whereas others have acknowledged the markers’ sometimes fuzzy boundaries, this orientation requires Fischer to categorize them in a clear-cut way that also allows for their polysemy. Her position appears to be that discourse markers have lexical meaning, but that the particular application relates to conversation management as well. She uses data from modern plays, tidier perhaps than naturally occurring data, including courtroom data. She bases her analysis partly on questionnaire data about the functions of certain markers and partly on computational analysis; the latter appears to be confined to the turn in which the discourse marker occurs and the last word(s) of the previous turn, if she includes that in the “left context of the particle” (2000:100), although later she also mentions the dialogue acts of the preceding utterance. She arrives at eight functional categories: take-up, back-channel, framer, repair marker, answer, action, check, and modal. Neither these nor her inventory of dialogue acts include some of the functions found with well in the courtroom.
Aijmer & Vandenbergen 2003 also canvass the work on well, using translations of fictional works from English into Swedish and Dutch to provide the basis for categorizng its functions. While they acknowledge the problem of using translations, including that well is translated by many different particles, they argue that this approach allows a comparison of function that illuminates the item’s use in the original. They discuss well in terms of a core meaning (positive appraisal, which seems to include any subjective judgement) and a core function (signaling counter-expectation), noting that its interpretation always relates to the context. This approach additionally affords them some interesting insights, for instance that male translators into Swedish appeared not to include the interpersonal (negotiating common ground) function of well in the translations. They comment also on the use of well in conjunction with other lexico-grammatical items (e.g. “Well! Just fancy! Isn’t that nice dear”; 2003: 39), which they term “double functioning”; they do not suggest a separate role for this kind of well. Cuenca Reference Cuenca2008 also uses a contrastive analysis (English/Catalan), and concludes that continuity and downtoning are the core features (which presumably means functions) for well.
Brown & Levinson’s (Reference Brown, Levinson and Goody1978) politeness theory may seem a clear candidate for explaining the use of well. However, and importantly, Buck Reference Buck1997 questions their separation of politeness from other interactional features in conversation. She suggests extending their model to include face tending (i.e., not limiting it to face-threatening acts, or FTAs). This allows it to be used to look at larger sequences and to explain the “intricate dynamics at work, dynamics which go beyond the fact that speakers simply formulate linguistic strategies around politeness concerns” (Buck Reference Buck1997:103). Although the extent to which face tending occurs in the courtroom is debatable, the data will show that it is not precluded.
Relevance theory is the only framework that accounts for all uses of well, according to Jucker Reference Jucker1993 and McHoul Reference McHoul1997; that is, relevance is the motivating factor in speaker’s choice and hearer’s interpretation. Politeness becomes involved only if it is relevant to the situation in some way, rather than being necessarily communicated by certain features (interpretation involves hearer’s inferences, not merely decoding). Jucker says that well has four functions: delay, and the marking of frames, insufficiency, and face-threat mitigation. McHoul Reference McHoul1997 claims that well marks a shift in relevant context rather than the FTA directly. While his comments are reasonable, he looks at contexts that do not involve conflict (sociolinguistic interviews). He concludes that well relates to relevance, even when used to mitigate FTAs. It seems likely that relevance would explain its use in contexts involving conflict as well.
Blakemore Reference Blakemore2002, also discussing discourse markers in terms of relevance theory, describes well as encoding a procedure rather than a concept. Well tells the hearer how to interpret an utterance (as opposed to encoding a particular cognitive concept), and optimal relevance guides that interpretation. She notes that “it is conceivable that the speaker’s intention in such utterances might not include the construction of an assumption at all, in which case the discourse connective simply serves as a means of activating the right kind of cognitive effects” (Blakemore Reference Blakemore2002:87). She comments that “one must draw a distinction between what justifies the use of well and what it encodes” (2002:144). It encodes “that the utterance is relevant”; but its justification is “the speaker’s belief that certain assumptions are not manifest to the hearer” (2002:144). Blakemore writes that she sees one aspect of well as being a “green light” (2002:138) for using inferential processes, thus carrying optimal relevance, whereas Jucker sees it in terms of deviating from optimal relevance. Blakemore confirms Mey’s comment that this kind of approach is “disconnected from everyday communication and its problems” (Mey 1993:82). Despite that, it allows us to account, at an overarching level, for all the different functions found in the data for this study. As support for this approach, in looking at the use of well among Xhosa English speakers and, incidentally, comparing it with New Zealand English speakers, de Klerk 2005 also suggests it has a core meaning (“the search for relevance in an imperfect communicative context”; 2005:5), within which there are “certain loose categories of procedural meanings” (2005:18). In doing so, she notes both its lower use among Xhosa English speakers and its similar use in interpersonal functions by both groups.
To summarize, then, Schiffrin’s approach provides a useful starting point for discussing well in the courtroom. Other studies cited add useful points but have limitations either in the size and variety of the data pool or in its applicability to the courtroom. It is widely accepted that well is used as a response marker, indicating aspects of dispreferredness and structural coherence. Further, it has a role in politeness, although this has not received a fully realized analysis. At a more global level, relevance theory, as explicated by Jucker and Blakemore, appears to provide an umbrella wide enough to cover this variety of roles while also recognizing their different aspects. In particular, Blakemore’s notion of procedural meaning provides a unifying framework.
Methodology
The data for this study were obtained from audiotape recordings and observation of seven criminal jury hearings in the Auckland District Court. A combination of quantitative and qualitative approaches was used in order to gain the overall view made possible by the former as well as the fine-grained analysis available with the latter.
There were 50 active participants in the study (excluding court registrars, stenographers, interpreters and juries): 9 women and 41 men; 17 legal professionals (judges, lawyers) and 33 lay people (defendants, police witnesses, witnesses); and a mix of ethnic groups (Maori, Samoan, Niuean, Indian, and 37 New Zealanders of European descent, hereafter called NZE). Seven were L2 speakers of English.
The data contain 309 instances of well used as a discourse marker in a corpus of 90,528 words that made up the examination phases (examination-in-chief, cross-examination and re-examination) of the seven trials recorded.Footnote 1 A rate was calculated (per 1,000 words) for each participant and role group. For comparison purposes, a similar process was applied to five ordinary conversations taken from a series of recordings made as part of undergraduate linguistics coursework at the University of Auckland.
Who Uses Well?
All kinds of participants in these hearings use well as a discourse marker, rather than its being used predominantly by one group, however defined (e.g., by gender, ethnicity, power, or role). The highest-frequency users, who come from several ethnic groups and both genders, are four witnesses and two defendants. However, 18 participants (more than one-third) do not use well at all in this way, and they too come from all groups in the hearings.
TABLE 1. Rates for well by participant groups.
Table 1 shows that well broadly follows the patterns of power in the courtroom, in that the highest-using groups are the witnesses and the defendants. However, police witnesses, who fall near the middle in terms of courtroom power, use well the least. Perhaps unexpectedly, given that they hold the greatest power in the courtroom, the judges appear in the middle. Factors other than power will therefore provide a better explanation.
As Table 2 shows, women use well at a higher average rate than do men during these hearings. Five of the nine women do not use it at all, of whom three are witnesses. Two use an interpreter for questions but answer in English; we have no way of knowing whether their non-use of well is a consequence of the translation (the interpreters’ utterances to the witnesses were inaudible; as noted, interpreters have not been included in this analysis). Speaker 5DC uses it much more often than the other professional women (she does not appear to be significantly different from them, other than this being her first trial as a defense lawyer); however, she uses it less than the averages for men and all defense counsel. Function is more likely than gender to be relevant; however, the number of women in the study is too small to allow generalization.
TABLE 2. Rates of well by female participants.
Similarly, Table 3 reveals that, although the rate for non-NZE participants is higher than for NZE (perhaps reflecting Müller’s (2004) finding that non-native speakers use well more often than native speakers), being non-NZE cannot predict the use of well; of the eleven non-NZEs, six do not use it as a discourse marker. Here too, three (Niueans) use an interpreter (see comment above). Interestingly, other studies (Berk-Seligsen Reference Berk-Seligsen1999, Hale Reference Hale1999) have found that discourse markers are often not translated accurately. Given that the highest non-NZE users in the current study are both L2 speakers and neither uses an interpreter, we could speculate that being an L2 speaker may be relevant, but there are not enough non-NZE participants to go further.
TABLE 3. Rates of well for non-NZE participants.
Functions
Where does well appear in the courtroom?
Table 4 shows the location of well in terms of both Conversation Analysis and trial structure. Of the 309 tokens, 41% (127) appear in FPPs (first pair parts of adjacency pairs, describable as initiations that require certain responses; Sacks, Schegloff & Jefferson, 1974), usually questions in this context. Of these, 90% follow declarative turns, usually answers (unsurprising given the courtroom question-answer pattern). The remainder follow yes/no questions, wh-questions, and objections. Most of the wells in FPPs are addressed by professional participants to lay addressees, again unsurprising given the courtroom interaction pattern; the others involve only professionals and generally concern legal objections.
TABLE 4. Well in examination phases.
* The turns in which these appear cannot be classified as either FPPs or SPPs.
** These occur purely as part of procedure (e.g. when the hearing was interrupted to expel a witness who should not have been present), or during the introduction to a voir dire, or the judge giving unsolicited advice.
One hundred seventy-four wells appear in SPPs (second pair parts, usually answers in this context). Showing a different pattern from those in FPPs, 25% follow declaratives, 40% follow wh-questions and another 25% follow yes/no questions. This is consistent with Schiffrin’s (Reference Schiffrin1987) results in that well in SPPs follows wh-questions rather more than other forms. However, she does not appear to find it following declarative FPPs; this may reflect differences between courtroom patterns and sociolinguistic interviews.
In the predictable reversal from FPPs, by far the most wells in SPPs (84%) are used by lay participants to professionals. However, a further 12% are addressed by professionals to professionals. The fact that 8 of these appear in FPPs and 20 in SPPs supports the notion that, while role may be important in well use, function is also a determining factor (see further below).
Also important is the kind of examination in which well occurs: co-examination (examination-in-chief, re-examination) or cross-examination. The difference stems from whether the questioner is questioning someone from his or her own side (prosecution or defense) of the proceedings or from the other side. If well is important in signaling speaker response, then it will appear differently in these two examination types, because of their different aims (more cooperative vs. more combative); and, indeed, 68% of the wells occur during cross-examination (comprising 54.76% of the corpus) and 28% during co-examination (more frequent in the latter than Hale Reference Hale1999 finds).
Well occurs in FPPs far more often during cross-examination (81%) than co-examination. On the other hand, 59% of those in SPPs occur in cross-examination, a difference statistically significant at the 0.05 level (even taking into account the low expected values for some cells). This suggests that well may have different functions according to the kinds of turns it appears in and when it is used.
An ethnographic approach suggests that, owing to the adversarial framework of the New Zealand court process, we would expect to find challenge, clarification, and confirmation in questions, and justification and hedging in answers in the cross-examination phases. If the Conversation Analytic approach shows well to be occurring at those times, we can surmise that its use depends both on the users’ roles (cf. Fuller Reference Fuller2003) and their goals for those turns.
In these data, well is clearly associated with declaratives in SPPs (unsurprising, since these are answers to questions). But in FPPs well appears in a variety of constructions, most often in declaratives, yes/no questions, and wh-questions. It occurs less often with alternative questions in these data.
What is well doing in court?
The functions found with well in these data are summarized in Table 5 and discussed further below (with relevant examples) when four categories are proposed. Many instances are associated with more than one function, and it is not possible to choose the primary function in a principled way.
TABLE 5. Functions of well.
* includes confirmation, admission
** all requests (including for clarification, justification, opinions, rulings)
*** includes denying, disclaiming
**** includes denying knowledge, refusing compliance, refusing a witness’s excuse
***** includes scene setting, quoting, pre-sequences, signalling incomprehension, criticising
Predictably, there is very little overlap between the functions in FPPs and SPPs. Previous research generally agrees that well occurs in dispreferred SPPs (Sacks, Schegloff & Jefferson 1974), but less is said about how often this is so. Dispreferred SPPs are defined as not meeting the structural requirements of the previous speaker’s FPP and often exhibit mitigating signals. In the present study, 120 wells appear in dispreferred SPPs, leaving 46 that do not. Those 46 occur with all the forms and functions found among the dispreferred SPPs (i.e., no different pattern emerges here).
Well also occurs in FPPs, where by definition dispreferredness does not apply. However, a similar awareness of preference is at work here, even if it does not have a structural element as dispreferredness does. This may be best described as FTAs (Brown & Levinson 1987), as has been done effectively by others (e.g. Holmes & Stubbe Reference Holmes and Stubbe2003, Cameron Reference Cameron2000). It seems to me that dispreferredness implicitly involves an element of face threat. Therefore, it may be more economical to categorize both uses of well under the FTA umbrella, rather than having to categorize them differently for FPPs and SPPs.
Let us now consider wells in non-dispreferred SPPs with the 17 wells in non-face-threatening FPPs. Although the numbers are small, there is a pattern here. First, they occur more often during co-examination than cross-examination, and this applies for both genders and all ethnic groups. However, in addition, the women use them in a nonthreatening way more often than do the men. This is consistent with research by Holmes and others who suggest that women use facilitative language more often than men do (e.g. Holmes Reference Holmes, Bell and Holmes1990). Given the aims of co-examination, I suggest that well may be used as a facilitator by lawyers or as a mitigation device by witnesses.
Professional men use well in non-face-threatening FPPs in both co- and cross-examination, but professional women do so more often in co-examination; the lay participants do not show a strong gender difference. Again the numbers are small, but the fact that the users of well are not all powerless in court allows us to surmise that it is being used for different functions in different examination phases.
Before considering functions further, we need to address delay, often assumed as a given with well. Many instances (19%) occur in turns with explicit delay (see examples 1 and 2 below), mainly pauses and hesitations. Thus well is often associated with delay.
(1) 1 Defense counsel: (3) how would you describe him.
2 Defendant: (1/3) oh well (3) he’s a violent man?
(2) 1 Defense counsel: (2/) did you see what sort of party it was that was being held?
2 Witness: (0.5) well (0.5) yeah xxx were ah (0.5) people who were well
dressed.
However, there are many examples where well occurs without any other delay indicators.
(3) 1 Prosecutor: (1/) well why didn’t you tell the police that Mr D2 (0.5) had thrown (0.5)
2 the bin at the window.
(the first pause is necessitated by waiting for the typist)
(4) 1 Witness: (0.5) he did not give it to me
2 Defense counsel: (0.5) xxx or even to (0.5) C =
3 Witness: = well he gave it to C (0.5) not to me.
While it is possible that well is adding delay here, it is difficult to be sure because there is no corroborating evidence. This particularly applies in example (4), where well begins a turn latching the previous turn. We can justifiably attribute several functions to well in such examples, and it is unwise to rely on delay as the only explanation. Further, we can often attribute other functions to well in utterances where there is also explicit delay.
Categorizing well
While others have provided categorizations of well, they have found that its polysemy requires a number of classes. It may be useful instead to group the functions of the utterances in which well appears. I initially used three broad categories: neutral, agreement, and disagreement. These categories relate logically with the notion of preferredness so fundamental in Conversation Analysis but also give a wider sense of function.
While I considered following Hale Reference Hale1999 in using Hudson’s (Reference Hudson1975) “negative” and “positive conducive” categories, I do not do so for two reasons. First, they were proposed for questions, and answers do not fall so easily into “conducive” categories. Second, the categories I propose allow for two further qualities clearly present in the data but different in character from expecting agreement and disagreement.
“Neutral” relates to management of the process and includes quoting, procedural matters, rulings, continuation, raising new issues, beginning stories, and rephrasing questions. This is consistent with the notion of conformity to a norm, favored, for example, by Aijmer Reference Aijmer2002 and Aijmer & Simon-Vanderbergen 2003.
(5) 1 Defense counsel: (1) and as he was walking towards (0.5) his car (1) did you see
2 something happen?
3 Defendant: (2/3) yeah I I. (0.5) I walked up to him (0.5) and I just told him to
4 leave (0.5) as I as I was talking to him (2/1) ah (0.5) person came
5 up beside me or behind him (0.5) and punched him on the side of
6 the head (0.5) or face (0.5) up here
5 Defense counsel: (2/1) well can you tell us what that person looked like
“Agreement” involves the speaker’s providing something and includes agreements, acknowledgments, admissions, confirmations, or giving and requesting straightforward and undisputed information (names, addresses, locations). This too fits with notions of acceptance and positive appraisal suggested by others (reviewed in Aijmer & Simon-Vandenbergen Reference Aijmer and Simon-Vandenbergen2003).
(6) 1 Defense counsel: (1/1) why did you leave the area.
2 Defendant: (3/1) well (2) S ask (0.5) ask us to walk down to the shops (0.5)
3 and (0.5) the shop was close
“Disagreement” involves a lack of acceptance in some way and includes denials, hedges, requests for clarification, refusals to comply, challenges, posing unpalatable questions, giving unacceptable answers, and legal objections.
(7) 1 Prosecutor: (3) so how many (0.5) hours did you spend (0.5) in Queenstown.
2 Defendant: (1/1) well (2) it’s hard to say (0.5) exactly (0.5) it
I include reinvoking earlier topics or statements as disagreements because it generally involves challenging something the witness has already said, but I do not include reinvoking something said by another witness. I follow Muntigl & Turnbull’s definition of a challenge as “any negative thought, attitude, or action that a speaker attributes to an addressee” (1998:230; after Labov & Fanshel 1977).
However, this still leaves some tokens unaccounted for. I therefore propose a fourth category: “evaluative.” This also involves the speaker’s taking a position, although not necessarily either of acceptance or nonacceptance. It includes qualifying statements as well as giving opinions, explanations, and justifications. Aijmer & Simon-Vandenbergen Reference Aijmer and Simon-Vandenbergen2003 seem to include this in the term “positive appraisal.” I prefer the more neutral “evaluative,” which in turn allows separation from the category I have called “agreement.”
(8) 1 Defense counsel: (3/1) well if you had had a suspect (0.5) identified (0.5) at a party
2 (0.5) you might well include a number of the other party goers in
3 the identification parade wouldn’t you
4 Police witness: (0.5) well we get whoever we can to ask (0.5) willing to ah
5 undergo (1) the parade
Table 6 shows how the uses of well in the hearings fall into the four categories. That neutral instances occur nearly twice as often in FPPs as in SPPs can be explained by the fact that often they are concerned with the progress of a hearing, which is controlled by the lawyers and the judges. A similar comment can be made about the agreement category: witnesses and defendants are required to provide information, which appears in SPPs by virtue of the nature of the interaction.
TABLE 6. Categorizing well functions.
However, this does not apply with the disagreement category, where well over half of the instances fall. Disagreement occurs more or less equally in FPPs and SPPs, although slightly more in FPPs. This supports previous findings that well is often associated with dispreferred responses. That FPPs are not responses, by definition, supports the suggestion that we should consider well in a wider sense, as often indicating a speaker’s awareness that he or she is about to say something that may be unpalatable to the hearer, regardless of whether it is in an FPP or an SPP (see discussion of FTAs above).
Marking speaker attitude
This study repeats Schiffrin’s and others’ findings that well looks back to what has already been said.
(9) 1 Defense counsel: (2/) so (0.5) there no direct evidence (1) that this xxx that that
2 laptop computer was (0.5) at his house.
3 Police witness: (1) I didn’t say that.
4 Defense counsel: (2/1) well all I’m trying to say is you can’t have it both ways it was
5 either (1) there at the time of the search (0.5) which is one of your
6 own witnesses (1) or it (1) now your next comment (5/1) that
7 doesn’t appear to follow.
Here the lawyer’s (2DC2) second turn has two distinct parts. It begins with a challenging comment in which well signals that the lawyer does not accept the witness’s (2PW1) previous answers. However, we can also argue that it signals that the lawyer is about to say something unpalatable. Thus well also looks forward, acting as a signal. Having expressed his challenge, the lawyer then turns to the next issue he wishes to raise, another challenge. Similarly, well looks both back and forward in the following example.
(10) 1 Defense counsel: (2) did (0.5) D know that the car had been stolen at that stage.
2 Police witness: (2) I couldn’t be sure
3 Defense counsel: (0.5) couldn’t be sure (2) well wouldn’t you tell him that you were
4 making enquiries as to whether (0.5) the car was (0.5) in his
5 possession legitimately?
Here the lawyer does not appear to accept the police witness’s apparent lack of certainty or knowledge. The well looks back, signaling that response, but it also looks forward, signaling the upcoming challenge, expressed in a negative yes/no question (a coercive strategy). The pause here is explicit delay and lends weight to the suggestion that the forward-looking aspect of this well may be primary, but does not rule out its backward glance.
Despite not meeting the structural requirements of the FPP, and possibly not meeting the FPP’s speaker’s goals, a dispreferred answer may well suit its own speaker’s goals. For instance, a lawyer may ask a question for which the preferred answer (preferred structurally, but also desired by the lawyer) is one thing, but the hearer desires to give a different answer (and in court this often revolves around the fraught question of truth). Well seems particularly useful here.
(11) 1 Defense counsel: what did you (0.5) observe (0.5) A and (0.5) W1 (0.5) to do (1/)
2 after you’d received these (1/1) instructions from (1/1) Detective S.
3 Defendant: (1/) um (0.5) well (0.5) on a couple of occasions (2/) I saw them
4 driving (0.5) late model vehicles (0.5) around the streets of Te
5 Atatu.
Here 2D is apparently giving a preferred answer if we look from 2DC2’s point of view and at the structure of his question. This is a typical environment for well to occur: the wh-question asks for information, and 2D provides it. However, it is delayed, and only speculation can suggest why. Perhaps 2D knows that 2DC2 requires a more damning answer, or perhaps 2D is uncomfortable with his role as police informer. This exchange continues:
(11, continuation)
6 Defense counsel: (2/14) why (1/1) did (0.5) or do you know (0.5) the reason (2/1)
7 you were asked to (0.5) do this work for Detective M.
8 Defendant: (1/) yes um (0.5) they’d been (0.5) trying to catch (0.5) C and W1
9 for (0.5) quite some time (2) especially after (0.5) C had stolen (1)
10a high ranking (0.5) police officer’s car out of his house.
The lawyer continues trying to establish why 2D was acting as an informer, eventually getting him to admit that it was for self-serving reasons. In such situations well marks not only the speaker’s response to what has already been said, but also the speaker’s attitude toward what he or she is about to say.
Well is used in a similar way with sarcasm. Jorgensen (Reference Jorgensen1996:629) suggests that the use of irony is a face-saving measure because it “guarantees distance between the speaker and the mentioned proposition, and thus retains ambiguity about the seriousness of intent.” However, Jorgensen’s experiments involved friends, which allowed him to conclude that sarcasm is used to promote solidarity. This cannot not apply with the following examples from the current study. Nonetheless, his suggestion that sarcasm involves an element of complaint does apply here and ties in with the notion of face threat.
(12) 1 Prosecutor: (0.5) how do you know they were smoking heavily.
2 Defendant: (1) well (2) because (0.5) you see (0.5) um (0.5) white (0.5) um
3 (2) paper (1) with smoke coming out of it (2) being dragged into their
4 mouth and then passed to the next person.
The question here relates to cannabis. 4D gives a preferred response, structurally speaking, in that she answers the wh-question explicitly. However, she violates Grice’s maxims of quantity and relevance, using well to signal that upcoming violation.
(13) 1 Prosecutor: (1/) all right I’ll (0.5) try and be a bit more specific did you sell any (0.5)
2 in (1) Queenstown.
3 Witness: (3) well (0.5) we sort of (1) when we got to (1/1) Bluff (0.5) we had no
4 money so apparently I can’t have (0.5) sold any in (0.5) Queenstown I
5 sort of had a (0.5) pretty heavy session in (2) Queenstown (3) and
6 anything coulda happened.
Clearly delay is also part of both these examples. However, it is arguable that there is more to it than that. Although these answers are preferred in that they answer the questions, the speakers are relying, indeed playing, on everyone’s knowledge that they are not the desired answers. Well may be used here to signal that awareness on the part of the speaker and perhaps even to warn the respondent: it looks both ways.
Well can look forward in FPPs as well as in SPPs. Thus, it can project something about the speaker’s orientation or stance toward the upcoming FPP and can look back and forward and between sequences at the same time. This is clearly illustrated with challenges, of which there are a great many in these data.
(14) 1 Defense counsel:( 2) what can you tell us about S?
2 Witness: (4) he became a s-source of ah annoyance (0.5) as the weeks went
3 by.
4 Defense counsel: (2) well let’s put let’s let’s (0.5) lets put it a little more strongly
5 than that you’d barred him from the shop hadn’t you.
6 Witness: (1) I barred him from the shop yes?
This sequence occurs in cross-examination between a defense lawyer and a prosecution witness. On the face of it, the witness’s utterance fulfills the requirements of the question and is not a strongly dispreferred answer (although the delays in s-source and the hesitation ah suggest some discomfort). However, clearly it is not satisfactory for the lawyer’s purpose – discrediting S in order to shift blame from the defendant. Well appears at the beginning of a question that implies that the answer just given is not satisfactory, and then the lawyer poses the expected answer with a tag question severely constraining the witness’s reply. There is some delay in the lawyer’s challenge, suggesting some unpalatability about the question. It is also possible that the well is used to soften the challenge: this witness is the victim of the crime.
We saw earlier that well functions as a linking device between FPPs and SPPs; now we see that it can also link adjacency pairs (i.e., between SPPs and FPPs or between sequences). Its function is more than that, however, as the linkage often also carries a projection of the speaker’s stance. This, then, brings the two aspects of well being discussed here into the same framework: dispreferred answers and FPPs that are unpalatable in some way.
This is consistent with Schiffrin’s conclusion that well should be treated as a coherence device, but one that is not “pairwise.” It frequently relates adjacent statements, but these are not necessarily adjacency pairs. In addition, it can relate utterances that are not adjacent, as with Schiffrin’s “re-invoking” function, which concerns still earlier utterances. Further, it is arguable that it can project speaker stance further ahead than just the turn in which it appears (at the same time as marking response to the previous turn).
(15) 1 Prosecutor: (1/1) did you see any Polynesian men in the bar with sticking plasters on
2 their nose and chin (0.5) that day?
3 Defendant: yes
4 Prosecutor: (1/4) well tell us about this person that you saw with the (0.5) sticking
5 plaster on his face.
6 Defendant:no I can’t
7 Prosecutor: (1/1) because it’s not true is it Mr D.
We can suggest here that the well prefigures the challenge that will occur in line 7. The close relationship between these two turns by 6PC is pre-invoked. This would be worth investigating over a larger corpus, particularly in instances of conflict interaction.
Most wells appear in the initial position of turn construction units (TCUs); only 12% appear elsewhere. Most (82%) also appear initially in their turns. Fox, Hayashi & Jasperson Reference Fox, Hayashi, Jasperson and Ochs1996 point out that English speakers can use the beginnings of TCUs “to project a possible course for the utterance” (1996:213); clearly well can fulfil this function.
Well in repair
Almost half of the wells occurring in these data do so in repair, where it appears in FPPs almost as often as in SPPs. While space here does not permit a detailed discussion, there is a clear functional pattern: when well occurs in repair sequences, it is strongly associated with challenges. It also occurs in the responses to challenges (clarification, explanation, justification), often associated with delay and hedging. It appears that well signals speaker stance rather than the upcoming repair per se (many repairs occur without well).
Well as a softener
Speaker 4D often uses well when agreeing, explaining or justifying, often with concurrent use of hedging. Her wells could be softening her answers, seeking to mitigate her culpability. Speaker 5W4, also female, uses well often when agreeing and hedging. It is possible that their gender combines with their respective roles of defendant and defense witness to account for their use of well.
As stated earlier, 5DC uses it more often than the other professional women. Ten of her 22 tokens involve challenges, while 6 involve continuation and raising new issues. She too may use well to mitigate challenges, perhaps aiming to put witnesses at ease. However, 9 of her 10 challenges are addressed to prosecution witnesses, her adversaries in a sense. Her role suggests that she is heightening her challenges, whereas other factors, such as gender, might suggest mitigation and/or facilitation to be informing her choice of well. The fact that she is the only professional who uses someone’s first name (which she does only when addressing the defendant or a witness) supports the latter interpretation, but we cannot be sure which interpretation is correct here.
In keeping with their role in court, the judges use well in some ways that are different from the other participants’s uses. The most common single function for the judges is delay, although it always appears in conjunction with other functions. Other functions particularly associated with their role are giving rulings, resolving issues, clarifying, giving advice, and directing procedure. The judges may use well to soften or mitigate the potential threat of their intervention (a negative politeness device). One judge does not use well at all during the two trials in which she takes part; and she does not participate in a substantive way, thus not posing the threat of intervention, unlike the judges in the other five hearings.
For PCs and PWs, who use well the least, mitigation clearly cannot be a significant factor. Both groups could be said to be on the attack: PCs try to prove the defendants’ guilt, and PWs are intent on showing that the police have done their job appropriately. Fully 93% of PCs’ 82 tokens occur in FPPs (unsurprisingly); 63% of these involve challenges and requests for clarification or confirmation. Delay occurs in only 8 examples. I suggested above that lawyers may be using well to heighten their challenges. On the other hand, PCs could be using it to mitigate the force of their bald challenges.
The question then becomes one of telling the difference between well used to heighten challenge and well used to mitigate it. With 5DC we have seen that either argument applies, but perhaps she is unusual. Generally the lawyers tend to reserve their challenges, or at least those using well, for addressing witnesses giving evidence for “the other side.” Given New Zealand’s adversarial legal process, we would expect that well would be used to heighten the effect. Supporting this is the fact that all the DCs’ challenges to defendants are less bald, being requests for clarification, opinions, or justification, and all of them occur in co-examination. It does not appear likely that lawyers use well to mitigate challenges, unless their challenge is directed to someone appearing on “the same side” of the issue (which supports the interpretation that 5DC uses well to heighten her challenges to prosecution witnesses).
A Framework For Well?
The emergence of challenge and mitigation as operative factors in the use of well suggests that politeness theory may provide a useful framework. Some uses of well fall squarely into politeness theory: those involving face threat (e.g. challenges). However, other uses do not fit clearly here, including face tending (Buck, Reference Buck1997). Although the extent to which face tending occurs in the courtroom is debatable, clearly it is not precluded, as demonstrated by 5DC’s use of first names.
However, relevance theory suggests that the desire to communicate politeness is not sufficient on its own to explain language choices (cf. Buck Reference Buck1997). In discussing Brown & Levinson’s model, Jary (Reference Jary1998:2–3) adds that “even when the speaker is aiming to protect the hearer’s face, her ultimate motivation is to maintain or raise her own status within the group and/or to ensure her continued well-being, in both the short and the long term.” This comment seems appropriate for describing much courtroom interaction. The different participants have differing goals at different points in the hearings, and we cannot put them all down to the need to be polite. On the other hand, at all times all participants are seeking to maintain or raise their status, so that their version of events will be accepted by the court.
If we accept this, we then have to conclude that, while politeness is certainly attractive as an explanation and covers many of the examples, it still does not account for all of the attitudinal functions found for well in these data. Nor is it the most powerful way of explaining its use in coherence. Relevance theory, in contrast, provides a comprehensive explanation covering both aspects. The notion of encoding procedure allows the myriad of functions found in these data to be accounted for in one economical frame: these data make it very clear that well is justified, in Blakemore’s sense, by the speaker’s belief that certain assumptions (relating, I suggest, to my proposed categories of neutrality, agreement, disagreement and evaluation) are “not manifest to the hearer” (Blakemore Reference Blakemore2002:144). This is not to say that the hearer may not have been able to predict or make these assumptions. This fits nicely with the individual goals of participants in jury trials: the lawyers, defendants and witnesses need to make their versions of the story manifest to the jury, and the judges need to ensure the fairness of the hearing, including making that fairness manifest to the jury, defendants, and witnesses. This also allows us to attribute a function to well in instances of the double functioning mentioned above (Aijmer 2003).
Comparison with Everyday Conversation
Five ordinary conversations have been analyzed for comparison purposes. All are informal conversations between family members or friends in Auckland, cover a variety of topics, and involve no conflict. They therefore provide a strong contrast with the courtroom data.
Well occurs 101 times in these conversations, a rate of 5 per 1,000 words (vs. 3.4 per 1,000 in the courtroom data). Apparently it is possible that well occurs more frequently in ordinary conversation than in the courtroom. Further, if what I have concluded is correct, we would not expect well to be used in the same ways in the two sets of data.
There are 20 participants in these conversations, 8 men and 12 women. One involves speakers who are Maori; the remainder involve NZEs. Five do not use well at all (2 male, 3 female). The highest-frequency user is a woman whose rate is 17.7 per 1,000 words. Interestingly, this is almost exactly the same rate as for the highest user in the courtroom data, also a woman (5W4). Among those who use well, the average rate for women is 6.4 and for men 4.6. As in the courtroom data, there is a clear association between well and women’s usage in these conversations. However, it is likely here too that a functional analysis would reveal differences between the men’s and women’s use. This line could be pursued usefully later.
The other point of interest here is that well’s pattern in the courtroom is not repeated in these conversations. Here only 4% appear in FPPs, with 17% in SPPs, markedly different from the 41% and 56% respectively in the courtroom data. All the rest occur in topic talk (discussion, as opposed to sequences of questions and answers), and their use is consistent with earlier research.
Predictably, given that more than 75% of the tokens appear in topic talk, well appears commonly in declarative utterances, again very different from the courtroom. Only 5 are otherwise: 4 occur with tag questions and 1 in a yes/no question. Only 2 occur in repair. Nineteen occur in FTAs (all mild forms), and 9 in dispreferred responses. Applying the four categories proposed above, well is used here for agreement and evaluation, again quite a different picture from the courtroom.
The difference between the two datasets may well be due to the completely different situations and goals, remembering that the conversations all take place between people who know one another reasonably well and involve no conflict, as opposed to the courtroom data. In addition, the questioning behavior found in court would not be appropriate in ordinary conversation; in fact, it would be seen as hostile.
Conclusions
This study has attempted an account of the functions of well in the courtroom and considers whether politeness theory or relevance theory provides the better explanation for its use. Several conclusions can be drawn.
First, well clearly operates as a discourse marker in the courtroom according to the definitions cited at the beginning of this article – that is, revealing contextual coordinates. It not only reveals simple coherence but also indicates participant responses and attitudes.
Second, and predictably, it is used pragmatically in a structural sense – as a link between utterances. It also signals both speaker response and awareness of the hearer’s likely response, thus looking back to what has been said and forward to what is about to be said.
Third, its broad function may be understood as procedural encoding: making manifest to the hearer the fact that, although the speaker intends a link to the previous turn, the upcoming utterance may not be exactly what the previous turns have projected, and the hearer will have to make inferences in order to interpret the utterance. Well is used in four distinct kinds of utterances in these data. I suggest that these can be understood as relating to that interpretive procedure encoded in well and as justifying its use (in Blakemore’s sense). In other words, those four kinds of utterances can be seen as justifying the choice to use well, and it is therefore carrying one of four pragmatic functions. It can carry a neutral function, which is very often procedural in nature. When it falls into the agreement category, it is facilitating or mitigating in nature. When it is in the disagreement category, it is often used for explicit challenges or for utterances that are unpalatable in some way, whether for the speaker or the hearer. Finally, when it is evaluative, it is signaling the speaker’s awareness that he or she is providing opinion but without its being either agreement or disagreement. Thus, we can conclude that the choice to use well may be accounted for (i) as a signal that the utterance may differ from expectation, and (ii) by perceived needs for challenge, facilitation, or mitigation. “Perceived” here refers to the speaker’s possible intentions as well as to the notion that the speaker may believe that the hearer is unaware of those intentions. The choice between those perceived needs is made by the hearer on the basis of optimal relevance. In turn, it may well be found in further research that the absence of such motivation can account for the non-selection of well. The results of this study therefore support Schiffrin’s conclusions and add a further dimension by looking in detail at well’s use in FPPs in addition to SPPs, and by looking at a different context. The results show well operating proactively rather than being confined to reaction. In addition, they show an application of relevance theory to naturally occurring data.
In conclusion, therefore, well certainly has a role in providing the contextual coordinates in talk. Further, while the power/powerlessness axis has some bearing on the use of well in these data, this relates to the participants’ roles and their goals rather than to social indicators. Some uses of well are clearly related to face and politeness, but not all of them. Discovering that well appears to be used differently in the adversarial courtroom and ordinary friendly conversation also provides support for describing it in a broader manner than under the rubrics of structural coherence, social indicators, or politeness. Relevance theory may be a useful framework in that it can account for all well’s different functions within the same account; that is, the structural (cohesion and delay) and the attitudinal (including face and politeness) can be gathered under the umbrella of procedural encoding and optimal relevance.
