Introduction

Field-coded questions are open questions from the respondents' point of view. While closed questions consist of both the question and the response options, field-coded questions consist of the question only. From the perspective of the interviewers, however, a field-coded question consists of more than just the question. The interviewers have a set of response options in front of them, and they must record the respondents' answers by checking the corresponding box. As Fowler and Mangione point out (1990: 88), field-coded questions require interviewers to be coders, because they have to classify the respondents' answers in order to be able to check one of the answer boxes.

Because the respondents are not informed of the responses they can choose from, their answers are often unformatted, that is, they do not match the response categories. Unformatted answers occur not only in response to field-coded questions. We also find them when the response options have been read to respondents. When answers are not fit for recording, in whatever context they are produced, interviewers are faced with the task of probing for a recordable answer. How do they do this? What conversational devices do interviewers employ to probe for a recordable answer?

Textbooks on survey interviewing do not offer much information on how to probe for answers, except to say that the interviewer should not probe in a directive manner.

Introduction

Survey methodology considers the standardized survey interview to be interaction involving two participants: the interviewer and the respondent. Survey methodology then analyzes the utterances of these participants in terms of their functions in a question-answer sequence. We see this approach at work when we study coding schemes for the analysis of verbal utterances in survey interviews, for example Dijkstra's elaborate and elegant coding scheme (forthcoming). According to Dijkstra: “Such an interaction sequence generally starts with the posing of a question from the questionnaire by the interviewer and ends at the moment interviewer indicates that the question is sufficiently answered by respondent, for example, by posing another question from the questionnaire.”Dijkstra's coding scheme contains three possible actors, namely, the interviewer, the respondent, and a possible third party who actively participates in the interview, usually the respondent's spouse or child. The second variable in Dijkstra's model is the type of information exchange that occurs:

• A “question” is a request to the other person for information related to the questionnaire, for example, a request for information, a probe, repeat of the question, or an elucidation.

• An “answer” is the requested information.

• A “repeat” is the repetition of an answer by the same speaker.

• “Perceived” is a remark indicating that an utterance is received and/

• understood. It includes utterances like “hm mm,” “fine,” “OK,” or “yes.”

• A “request” is a request for repetition.

• A “comment” is a remark on former talk, for example, “Difficult question!”

Introduction

Social science research is primarily devoted to describing and analyzing peoples' actions and attitudes. In modern states, governments need this information in order to make well-founded policy interventions and evaluate their effectiveness. For example, governments may want to know parents' attitudes towards the schooling of their children and whether or not proposed changes will meet with approval. Also, political parties need to know what issues their prospective voters consider important in order to adapt the election strategies accordingly. In addition, industries may want to know what the trends in fashion will be in two years time, in order to purchase the appropriate materials and dyes.

To gather such information, researchers may send people a questionnaire to fill out and return. Another increasingly popular procedure is to interview people, either in person or, far more economically, by telephone. In the 1980s, personal computers led to the introduction and rapid development of computer-assisted telephone interviewing (CATI).

Among the different types of research interviews, the standardized survey interview is the most prevalent one in use in western societies. Large numbers of people earn an income in the survey research industry, enormous amounts of public and private money are spent in this field, and many important decisions made by governments and companies are based on the results of survey research. In fact our lives are ruled by survey research to a considerable extent.

Introduction

This chapter examines how talk in standardized survey interviews differs from talk in ordinary conversation, and how this departure may confuse respondents. Section 4 provides an analysis of an extensive American interview fragment that illustrates how the interviewer strictly follows the rules of standardized survey interviewing and thus simultaneously disregards the principles of ordinary conversation. Section 5 discusses some interview fragments that show how interviewers try to follow the principles of ordinary conversation, thereby seemingly departing from the rules of standardized interviewing. The two preceding sections provide the theoretical basis for the analyses.

Although this book is written from a conversation analysis perspective, in this chapter I also make use of the conversational maxims of the language philosopher Paul Grice. I chose to use Grice's maxims in this chapter partly because these maxims are more specific than the related CA concept of recipient design. Also, although recent survey methodology literature has come to refer to Grice, especially in its discussion of context effect, I chose to make use of his work because I believe that he has more to offer the field than is usually acknowledged.

In his Lecture Notes of 1971 (published in 1992), Harvey Sacks discusses the concept of “recipient design.” This concept refers to the fact that participants in ordinary conversation design their talk for its specific recipients (Sacks 1992b: 438). Speakers are expected to orient toward what they know their co-participants know (564).

Introduction

I would like to begin this concluding chapter by quoting De Sola Pool. Although this quote is from 1957, it could well serve as a general conclusion for the present study. In his paper De Sola Pool criticizes the assumption that we can:

In this chapter I draw conclusions from the analyses of the “interpersonal character” of the interviews I have presented in this book. The first three sections briefly point out what I believe are the benefits of a CA approach to the study of standardized interviews. Section 4 presents some possible research questions based upon the findings of conversation analysis. Section 5 explains why interviewers should be allowed more flexibility than is currently granted to them by the rules of standardized interviewing. In the final section it will become clear that the costs of flexible interviewing are considerable, but whether or not the price is too high is a question that cannot be answered by survey methodology.

Introduction

In survey methodology questions are primarily seen as semantic entities, that is, linguistic units that have a certain meaning. Taking a conversation analysis viewpoint, we may look at survey questions as turns-at-talk. In addition to questions being semantic units, they are also interactional units to be used in an interview. Turns have an internal organizational structure. As we will see in this chapter, many survey questions are designed in such a way that their organizational structure is unfit for adequate use in the interview. This holds especially true for closed questions that are followed by a set of response options, as well as questions that contain a term or concept that is further explained or specified. The turn organizational character of these two types of questions makes them vulnerable to interruption by the respondent.

A speaker can perform several actions in one turn-at-talk. For example, when a speaker says, “Uh, John, can I have the sugar, please?” he first takes the turn and indicates that he is going to say something (“Uh”). Next he attracts somebody's attention (“John”), displaying that he selects John rather than Liza as his addressee. Next he requests the sugar (“can I have the sugar”), and finally he is being polite (“please”). Note that “being polite” is a separate action, and that this action can be performed by several phrases, such as “please” or “Do you mind?”

When people talk they not only produce meaningful utterances.

Listener roles

Accounts of listener roles

The standard account of the communicative situation, as I remarked in 1.6, envisages a single speaker addressing a single listener, where the role of the listener is seen as understanding what the speaker says, and the listener's full attention is given to what is being said (see Goffman 1981:129 for a characterisation of such a traditional view). We have seen, in the Map task data, that listeners may have intentions and goals in listening which are, to a greater or lesser degree, independent of those of the speaker. We have observed the same phenomenon in the Stolen letter task data, but here we noted as well that different listeners in a group listen in different ways.

Goffman draws our attention to three different types of listener: ‘those who overheat, whether or not their unratified participation is inadvertent and whether or not it has been encouraged; those (in the case of more than two-person talk) who are ratified participants but are not specifically addressed by the speaker; and those ratified participants who are addressed’ (1981:9—10). Addressees are particularly oriented to by the speaker and will be the persons designated to respond in an appropriate manner to what the speaker says. Later in the same volume, Goffman writes ‘the relation(s) among speaker, addressed recipient and unaddressed recipients) are complicated, significant, and not much explored’ (1981:133).

Locations and landscape features

In chapter 3, we observed a range of instances where the initial referring expression used by the speaker was insufficient to enable the listener to locate the landscape feature referred to, at least initially. In such cases, since all of the entities were represented in a spatially structured context, the standard response by B was to request information on the location of the entity, either in order to constrain the search field, or to insert a new feature on to the map. In the first section of this chapter, we shall note some of the issues raised by expressions of location in English, before going on in later sections to consider how some expressions of location can be used to create and structure the context if they are interpreted as terms of spatial deixis.

Lyons (1991: 142) draws a distinction between entity-referring and place-referring expressions, and between entities and places, suggesting that ‘it is … arguable that places (as distinct from spaces) are ontologically secondary, being identifiable as such by virtue of the entities that are located in or near them’. Such a characterisation seems appropriate to entities mentioned in prepositional phrases of a type such as behind the barn, on the shelf, over the sea, when such phrases are used to locate an object. Such an analysis is also consistent with the classic psychological distinction drawn between figure and ground (see the discussions in Clark 1976, Hanks 1987).

The nature of communication

How do people use language to communicate with each other? For centuries the commonsense view has been that articulated by Locke: ‘Unless a man's words excite the same ideas in the hearer which he makes them stand for in speaking he does not speak intelligibly’ ([1689] 1971:262). Locke was himself fully aware of the difficulties of achieving such an ideal, guaranteed, form of communication, but it still claims its adherents in the twentieth century. In the simplest version of this view, the speaker has a thought which is encoded into words and transmitted through the air by sound-waves so that it reaches the listener, who decodes the words and then has the speaker's thought. Such an account of communication would have little more to say than that it consists of speakers exchanging thoughts (see for instance Shannon & Weaver's account of signal-information in Information Theory, 1949).

Swift parodies such a simplistic view in Gulliver's Travels, where Gulliver describes an even more direct method of ensuring the passage of the same thought. The mathematics master in the Grand Academy of Lagado (Swift's splenetic version of the Royal Society) requires his students to eat each idea, so that it may progress directly to the brain without any distorting mediation arising from the student's own contemplation of the idea: ‘the proposition and demonstration were fairly written on a thin wafer, with ink composed of a cephalic tincture. This the student was to swallow upon a fasting stomach, and for three days following eat nothing but bread and water. As the wafer digested, the tincture mounted to his brain, bearing the proposition along with it’ ([1726] 1960:224).

The initials of members of the pairs of subjects who produced the dialogue from which each extract has been taken are indicated at the top of each dialogue taken from the Map task data. Similarly the number of the relevant quartet or of the undergraduate (ug) pair is indicated at the top of extracts from the Stolen letter task.

The nature of reference

The term reference is generally held to hold between a linguistic expression and a referent (for example a particular individual or entity in the world). However the term has been used in a variety of ways and in this short introductory discussion of the nature of reference I shall begin by making use of the three-way terminological distinction introduced by Lyons (1977: chapter 7) between sense, denotation and reference. His account relates to that of Frege, who called attention to the importance of the distinction between sense and reference, a distinction widely accepted in philosophical discussion. However, Lyons uses the term sense to describe meaning relations holding between linguistic units. His use of the term denotation is in some respects closer to Frege's use of sense. The distinction between these terms is not always easy to maintain (though see Lyons' discussion of the issues in Lyons 1977). Despite the difficulties of drawing a clear boundary between these categories on all occasions, we shall nevertheless find it useful to call upon the distinction in analysing certain types of misunderstanding.

Sense

The term sense, as used by Lyons, applies to words taken out of context, as they are when they appear in a dictionary. (Strictly speaking the term applies not to words, but to those meaningful elements which are common to a set of words related in meaning, to lexemes.