EESE 5/97

Speaking of Speech Acts

Friedrich Lenz (Passau)

1. Introduction

The term speech acts in the title should not suggest that this article is another contribution to what has come to be known as speech act theory. It does not deal with the abstract notion of speech acts but the occurrence and function of those utterances where the participants speak of what they are doing in their discourse. Explicit reference to the ongoing discourse, i.e. speaking of speech acts is regarded as a means of discourse organisation. In this way the article is rather a contribution to the empirically based investigation of discourse. However, unlike most discourse analytical studies,1 it not only attempts to explain interactive notions but also tries to show how these notions are linked to the lexico-grammatical inventory of English.

2. Speech Acts in Linguistics and Everyday Discourse

Speech acts are a firmly established topic in pragmatics. They have primarily been treated from a philosophical perspective. Philosophers like Austin (1962) and Searle (1969) have introduced the idea into linguistics that we perform actions when we speak. In fact, it is more appropriate to say that they re-introduced it as this thought is by no means new and can in principle be traced back to the scholars of Ancient Greece. But also in modern, i.e. post-Saussurean linguistics the concept had been worked with before it turned into a commonplace when pragmatic research became fashionable in the seventies.

In modern German linguistics it has at least been around since the psycho-linguist Karl Büuuml;hler wrote his Sprachtheorie. It is hardly known in this country however that in England roughly at the same time, namely in 1932, Alan Gardiner published his Theory of Speech and Language which runs along very similar lines. He also emphasises the functions of language and develops something like an early speech act theory. When Gardiner (1932, 62) for instance claims that "the act of speech is a highly complex, purposeful mode of human action [...which] arises in the intention of some member of the community to influence one or more of his fellows in reference to some particular thing", one could think that he enumerates what will later be called the illocutionary, the perlocutionary and the reference act. We shall return to some of his work in a while. For the moment the quotation from Gardiner can count as sufficient proof of the fact that speech acts had been the object of scholarly discussion in the humanities long before the term 'speech act theory' was coined.

The speaking of speech acts is, however, not confined to theory and research at all. In fact, the speaking of speech acts is and always has been an everyday routine that we regularly employ in our conversations - whether scientific or not. It is this explicit reference to speech acts within the ongoing discourse I will be discussing in this article, and I will do so not from a primarily philosophical, but from a discourse analytical point of view.

3. Reflexive language use

Philosophers, linguists and laymen use different terminology when they refer to speech acts. However, when philosophers, linguists and laymen speak of speech acts, we are not only confronted with different kinds of metalanguage, but with totally different types of reflexive language use altogether. When a philosopher like Searle speaks of speech acts, he is primarily interested in the conventions which govern action patterns. He differentiates between their "constitutive and regulative rules" neither of which in his view are language specific. Consequently, natural conversations play no role in his theory of speech acts and actual language forms occur only as far as he needs them for illustration. In any case, his speaking of speech acts has nothing to do with real segments of discourse that happen in real time.

Unlike speech act theory, discourse analysis is concerned with language or rather language use as it emerges in naturally occurring conversations. When Sinclair & Coulthard (1975), the pioneers of discourse analysis, speak of speech acts, it is in the sense of real communicative acts, but they still interpret these discourse units according to illocutionarily based categories.

A further essential difference between their reflexive language use and that of the participants is that discourse analysts speak about communication without being part of the communicative situation they analyse. This is the normal perspective of the empirical linguist and applies to a discourse analyst as well as to a phonetician for instance, the difference being that the former analyses discourse units, the latter phonetic units.

3.1 Extracommunication and Metacommunication

When discourse analysts speak of speech acts, it is descriptive-explicative communication about other people's communication. It is - to use Wunderlich's term - extracommunication.2 Extracommunication is not confined to research because everyone can talk about other people's discourse. This everyday type of extracommunication is the favourite preoccupation of gossips, but will not interest us here any further. We are interested in metacommunication which is related to extra-communication, since in metacommunication what is being talked about is also a piece of empirical discourse. The difference between extracommunication and metacommunication lies in the fact that in metacommunication the participants of a discourse speak of their own speech acts. They are speaking of the units of the discourse they are currently engaged in. Whereas extracommunication, no matter whether it is descriptive-explicative or gossipy, refers to outside, usually past events, metacommunication is reflexive language use within an ongoing situation. It is this type of reflexive language use that we are interested in here. Discourse analysis is, as mentioned above, extracommunication and when it is concerned with metacommunication as in our case here, it is in fact extra-metacommunication.

3.2 The Organising Function of Metacommunication

Unfortunately, the term metacommunication has given rise to a great deal of confusion. First, it has not been distinguished from extracommunication. Second, it has not been kept apart from the general phenomenon of metalanguage.3 Third, metacommunication has been associated with communication about the socio-emotional relationships between the participants. Psychologists in particular follow Watzlawick's usage and assume that the so-called 'relationship aspect' is dealt with in meta-communication (cf. Watzlawick et al. 1967, 53ff.).

In speech act theory of all shades as well as in discourse analysis it is of course accepted that every speech act is a social act so that social relationships are necessarily involved. From this it follows that for Watzlawick every speech act is at the same time a metacommunicative act. But Watzlawick is somewhat inconsistent because he also describes metacommunication as "explicit communication about communication." This view, which is closer to our understanding, underlies his discussion of communication breakdowns, where he seems to think that metacommunication takes place - or should take place - where normal communication, the management of so-called 'content aspect', does not work properly. This remedial function of metacommunication has often been referred to and not only in psychological literature.4 It is of course possible that metacommunication in the sense of communication about communication is meant to 'heal' disturbed communication, but this is only one type of metacommunication. From a discourse analytical point of view there is no reason to narrow our focus on the remedial function of metacommunication. Whenever we speak of our own speech acts, this is metacommunication where some of the structure of the discourse is revealed.5 Regardless of the specific function or even the illocutionary point the metacommunicative utterance itself represents, the speaker refers to a part or unit of the discourse and by doing so identifies this unit for the addressees.

By speaking of what they have been doing, are doing and will be doing, the participants let each other know how they assume the discourse is organised. This general organising function is common to all metacommunication.

4. Reference to Speech Acts in Discourse

4.1 Temporality and Discourse Situations

Discourse is a temporal entity and it is temporally structured. This is particularly obvious with regard to spoken discourse the physical existence of which depends on and coincides with the acts of its production. Temporal entities have a place on the time line. They happen in time like any other event or more broadly speaking like any other situation. Discourses are situations and so are speech acts. They also happen in time. In fact they happen within the time of another situation, namely that of the discourse. It is here that Gardiner comes into our discussion again. Bühler (1934, 23) praised Gardiner's work as a "Situationstheorie der Sprache", because Gardiner recognises the act of speech as a situation which is itself embedded in other situations. Besides the fact that any speech act is purposeful, rule governed behaviour, Gardiner (1932, 71) stresses the temporal, situational character of speech acts. He writes: "An act of speech is not a mere set of words capable of being repeated on a number of separate occasions, but a particular, transient occurrence involving definite individuals and tied down to a special time and place."

It is this situational character that we must have in mind in order to find out what happens when conversationalists speak of their speech acts thereby organising their discourse. What they bring about during the time of the discourse are discourse situations and the structure of a discourse can be conceived of as a combination of different discourse situations. These discourse situations are unique - there is only one at a time - but they can be categorised according to situation types and that is part of what the participants do when they speak of their speech acts. They identify a token of a type by categorising the situation and saying that it happened at a certain time. The type is usually given by a verb and the token is singled out by temporal expressions.

In other words: Reference to a unique situation is made through a verb denoting a situation type combined with temporal expressions that locate a token of that type - the situation - on the time line. The view that verbal expressions can be considered referring expressions would certainly not be supported by all linguists and philosophers including Searle, whose reference act only draws on nominals. However, the idea is by no means new. Davidson (1967) had published it two years before Searle's speech act theory appeared. He claims that at least every action sentence not only presupposes the existence of the referents that are introduced by its nominals, but also an event. This event is introduced by an action verb,6 a verb that denotes a specific type of action.

4.2 Speech act verbs

In the case of metacommunication the action verb has to be a speech act verb. Unlike a speech act's illocutionary point, which is only rarely indicated by a speech act verb, the metacommunicative character is mostly indicated by a speech act verb. One has to take into account that speech act verb here does not necessarily mean performative verb as is customary in speech act theory. Participants in a discourse do not refer to tokens of speech act patterns as they are defined in speech act theory, but to situations the types of which are represented in their (tacit) lexical knowledge.

This does not mean that our lexical knowledge of speech act types does not contain some of the conventions explained by Searle. Some of Searle's critics even claim that any speech act typology is ultimately nothing but the explanation of the lexical content of speech act verbs which questions Searle's fundamental assumption that speech acts are language independent (cf. Burkhardt 1986). It is not necessary to answer the question of the universality of speech acts here because as far as discourse situations are concerned it is enough to recognise that Searle's and similar typologies are too narrowly defined. They are primarily fixed on the illocution a verb in a performative formula expresses. However, there are verbs that cannot be used performatively but can still be used to refer to a discourse situation in the sense of a speech act which coincides with an utterance or sentence. One cannot say I insult you that you are an idiot, but when someone says you are an idiot this utterance usually counts as an insult and the participants can - metacommunicatively - refer to the act using the verb insult. The person insulted could, for instance ask why did you insult me?

Furthermore, there are complex speech act types which comprise more than one utterance or sentence (and also more than one illocution). When we explain or narrate something it will usually take more than one sentence. Some complex speech acts involve not only a number of utterances, but speaker change as well, for example when we negotiate or quarrel. In any case, there are verbs that denote such act types and neither the verbs nor the act types are covered in traditional speech act theory - though whenever we bring about such a discourse situation, we certainly do something by saying something.

The fact that we do say something seems to be the smallest common denominator of these verbs and that is exactly what Wierzbicka (1987) takes as the main criterion for her dictionary of English speech act verbs. She assumes that the semantic primitive say is common to all speech act verbs. As a consequence, she also lists verbs like talk, chat or discuss in her dictionary.

This approach proves to be useful for the analysis of metacommunication since it appears that particularly those verbs that would not figure in other lists of speech act verbs and which are illocutionarily rather vague are used most frequently by conversationalists when they want to refer to a part of their discourse. In a search through the Corpus of English Conversation the computer came up with talk as the most frequent speech act verb, and in context most examples looked like the following.7

  1. <425 d> ^I think 'this is :v\ery im_portant#
    <426 d> and I "^th\ink it _operates# .
    <427 d> [? @] ^in sur!prising :c\ircumstances#
    <428 d> I mean I ^think it`s "!tr\ue you 'see#
    <429 d> ^even 'in . the _kind of :graduate 'seminar 'work
    I /was 'talking about 'earlier :\on# -
    <430 d> where ^after all the :talk !\is a'bout# .
    <431 d> intel^lectual th=ings# .
    (LLC 5 7:Heading)

  2. <1191 B> +you ^call+ what you !s\ee a 'sign#
    <1192 A> ^y\es#
    <1193 B> ^y\es# - -
    <1194 B> *(laughs - - )*
    <1195 a> but I`m talking now about *[@] signs and symptoms*
    then and I I I`m putting them both under the heading symptom .
    <1196 A> ^y\es# -
    (LLC 2 9:Heading)

  3. <1012 d> there are ^r/\ecognised t/eachers# .
    <1013 d> ^/and !st\udents# -
    <1014 d> ^[?]at the S/enate# .
    <1015 d> there are ^r/\ecognised t/eachers and !st\udents#
    <1016 d> we are ^talking here [?]a^bout [?ei] committee of!!D\eans#- <1017 d> ^if !y\ou 'Sir#
    <1018 d> ^as !Chairman of this 'Academic :C\/ouncil#
    <1019 d> ^w/ere# .
    <1020 d> as I`m ^sure you d\o#
    <1021 d> from ^time to :t\ime# -
    <1022 d> to ^call . ^call to:g\ether# .
    (LLC 5 13:Heading)

    Most of these metacommunicative utterances can be regarded as what Bublitz (1988) calls topical actions in his analysis of everyday conversation. Topical actions of this type can also be found in more formal and particularly pre-structured discourses, for example business meetings (cf. Lenz 1989) or academic lectures. Parts of this article, for instance were originally held as a lecture and one of the first sentences in the lecture notes (in front of me) reads: "It is the explicit reference to speech acts within the ongoing discourse I will be talking about today."

    One always talks about something and that is presumably why talk about occurs so often when we deal with the topical organisation of a discourse - be it metacommunicatively or extracommunicatively. This does not mean other speech act verbs cannot contribute to the topical structure as well. One can also be more precise and use discuss, explain or clarify to introduce a topic.8 These speech act verbs denote different types of discourse situations which might be relevant for other levels of discourse organisation, but with regard to the topical structure they would have made the same contribution to the discourse.9

    4.3 Temporal expressions

    In connection with verb semantics it is important to remember that metacommunication is not only based on the lexical content of a speech act verb, but also on the temporal localisation of the speech act within the discourse. What we refer to are not the abstract speech act types, but the tokens of temporal entities as they are realised in the course of a discourse. Types are independent of time. The discourse structure develops in time and so do its parts that we can refer to. The identification of these entities is based on temporal relations and the usual ways in which they are expressed. We can see in our examples that the participants use different tenses, aspects and adverbials - the traditional temporal expressions as they are discussed in grammars.

    For the analysis of metacommunication, it is particularly important to see that the temporal localisation of events also depends on verbs, in our case: speech act verbs. They exhibit different internal temporal schemata according to which we can distinguish between different situation classes. Vendler (1967) introduced the idea of inherent temporal stages of events into modern linguistics, and in the relevant literature we mostly find his four situation or aspect classes: states, activities, achievements and accomplishments. As discourse situations are brought about by the participants, they must be dynamic and intentional so that we can exclude states from our discussion. The remaining three classes can be differentiated by a combination of two features: durativity and telicity. Telic situations are events with some sort of climax or culmination point that result in a change of state. Accomplishments are durative and telic, whereas achievements are telic, but not durative. Activities are durative, but non-telic situations. As mentioned above, these situation classes are reflected by the lexical content of verbs.10

    The following table exemplifies how some speech act verbs fit into the three relevant classes:

    classes: durative telic
    activities + - talk, converse, chat, gossip, lecture
    discuss, debate, negotiate, bargain, argue,
    quarrel... about something (and/or) with
    accomplishments + + explain, describe, narrate, relate, report
    comment on, prove, reply, summarize...
    achievements - + agree, apologise, regret, promise, order,
    command, recommend, affirm, claim,
    approve, name, appoint, say...

    As we can see, the verbs listed are speech act verbs in Wierzbicka's (1987) broad sense. Their temporal schemata give us an idea of the course or internal structure of the situation; that is, for example, whether the situation referred to comprises others, as is the case with durative discourse situations. They correspond to what is described in classical discourse analysis as higher order units: exchanges, transactions or the interactive unit as a whole (cf. Coulthard 1985). They also give us an idea of whether the situation has a certain outcome. When I stop in the middle of explaining something, I have not explained it, but when I stop talking about something, I have talked about it.

    Though the situation classes are lexically based, they set constraints on tense and aspect use. If I were interrupted in the middle of explaining something, I could not afterwards say I explained it, but I could say I was explaining it. The interaction between situation classes and tense and aspect can also be seen with the speech act verbs in the narrow sense. They belong to the achievement verbs. Achievements are conceptualised as practically instantaneous, as non-durative situations. That is why they cannot comprise other situations11 and that is also why they can be referred to in the present simple. I cannot say I explain something or I talk about something, but I can say I apologise, I agree, I approve (of something).

    Performative formulae represent present achievements, i.e. discourse situations which are simultaneous with the time of utterance but do not exceed it. Activities and accomplishments do exceed it and therefore we need the progressive. This does not mean that performative verbs could not occur in the present progressive. I can, for instance say I am claiming that metacommunication is interesting to study. This can be interpreted as a so-called multi-event activity (cf. Smith 1991). The activity of claiming would then last longer than one sentence and consist of more than one speech act. In the Corpus of English Conversation the progressive occurs very often with say. Say is the second most frequent speech act word which supports Vendler's (1972) claim that it should be taken as the general performative. In a sense it can take up any illocutionary point.

    The preceding remarks show that speech act verbs together with temporal expressions give the analysts as well as the participants substantial information on the course and structure of a discourse. The tense and aspect system is too complex to give a systematic account of how temporality is expressed. It also interacts with temporal adverbs, which are completely left out here as it should in principle be clear that they can also be used and understood as a means to localise a discourse situation within the time of the discourse (cf. Lenz 1997). However, there are some local expressions which seem to serve the same purpose. Because they apparently do not fit into our temporal framework, some brief comment seems to be appropriate here.

    4.4 Local Expressions

    Given the temporal character of the referents, it is hardly surprising that we primarily find temporal adverbs in metacommunication as example (1) and (2) above show. But particularly in writing we also find local adverbs. That again is no wonder: books or volumes have three dimensions. However, even there we transform the three-dimensional space into a one-dimensional one and in this way we recreate the same linear order in which we conceive of time. The three spatial dimensions are usually reduced to the vertical. We refer to the speech acts of a written text as to the parts above or below.12 As far as metacommunication is concerned, we somehow still seem to conceptualise our writings in terms of paper rolls, which in an era of electronic media seems to have become adequate once more: On the computer screen we continually move up and down again. It is important to notice that the one-dimensionality of written text only applies to the speech acts because beyond the metacommunicative line we have a three-dimensional space: we can metaphorically put text referents aside or place them in the foreground or background of the textual world or space (cf. Nöth 1996).

    In spoken metacommunication, and that is what I am discussing here, temporal adverbs are almost exclusively used. One of the few exceptions is the adverb here, which I just used in the last sentence: spoken metacommunication and that is what I am discussing here... Imagine the adverb had been uttered in a conversation. In a spoken discourse here can simply be interpreted locally (cf. Harweg 1990, 182), and the above utterance could be one such borderline case. The interpretation could be based on situational information that speaker and hearer share. In this interpretation, here would be considerered a local autodeictic expression. It would be a three-dimensional local expression referring to the space around the standpoint of the speaker.

    The more adequate interpretation, however, is that the adverb locates a temporal entity, namely a portion of the discourse, on the time line, and the actual local space where the discourse takes place is irrelevant. Here is still used as an auto-deictic expression that expresses an identity relation with the speaker's co-ordinates, but it is the temporal standpoint of the speaker that is identified, i.e. here is used in the same sense as now. This temporal interpretation is more likely, and in other cases, as in example (3), the local interpretation seems hardly possible.

    In written texts the standpoint of the speaker is mostly unknown. It is the reader's standpoint that matters (except in some letters). His or her local co-ordinates, however, are irrelevant again. The here in the above metacommunicative utterance refers to the time of reading a certain text segment. As this segment is also represented on paper (or screen), it has in fact a place as well. But the place identified by here is neither the place where the reader is physically located nor the place where the letters appear on the paper (or screen), but the place where the reader is in his or her reading process. Any process is a temporal entity and therefore the place is again a temporal one. Here refers to the present stage of the receptive process and that is why it ultimately corresponds to now in the same way as in spoken discourse.

    The use of local expressions to convey temporal relations is not as unusual as it may at first seem. Time is conceptualised in local terms. We cannot talk about time without reverting to time points, time spans or the time line as localist approaches in cognitive semantics amply show. In any case, the use of local expressions does not contradict our claim that speaking of speech acts gives us, analysts and conversationalists, information on the temporal structure of a discourse.

    5. Conclusion

    Speaking of speech acts is reflexive language use. We were concerned with one of its types: metacommunication which is brought about by the conversationalists when they refer to parts of their discourse. These discourse situations are identified by lexico-grammatical elements which express their temporality and their corresponding speech act types. In this way, speaking of speech acts gives information on the temporal structure of the discourse and can be regarded as a means of discourse organisation. It is of course only one means among others. It obviously operates on the level of topic organisation as was just demonstrated, but on other levels, for example turn-taking it is relatively rare.

    The last remarks concern the way speaking of speech acts has been treated by the professional analysts. It is somewhat surprising that classical discourse analysis has not studied it more thoroughly. The analysts more or less treat metacommunicative utterances as any other speech act according to their illocutionary points. They mostly work bottom-up, i.e. from the lowest and smallest units to larger structures thereby concentrating on the flow of speech acts as they interpret and categorise them. The ethnomethodological school of conversation analysis, on the other hand, attempts to analyse what they call 'the practices' the participants employ to organise their conversations, and in fact the study of metacommunication is rather central in their analyses. They try to avoid philosophical and linguistic concepts and metacommunication appears under the label of formulations. Sacks and Garfinkel (1970, 351) state: "We shall speak of conversationalists' practices of saying-in-so-many-words-what-we-are-doing as formulating...".

    In trying to follow the participants practices they ignored linguistic terms to the point that one might think they wanted to re-invent linguistics again. We have seen that formulating is done by lexico-grammatical means. The participants systematically draw on their grammatical and lexical knowledge. It is the task of the linguist to describe this process.


    Austin, John. 1962. How to Do Things with Words. Oxford: Clarendon Press.

    Bublitz, Wolfram. 1988. Supportive Fellow-Speakers and Cooperative Conversations. Discourse Topics and Topical Actions, Participant Roles and 'Recipient Roles' in a Particular Type of Everyday Conversation. Amsterdam: Benjamins.

    Bühler, Karl. 1934/1965. Sprachtheorie. Die Darstellungsfunktion der Sprache. Stuttgart: Fischer.

    Burkhardt, Armin. 1986. Soziale Akte, Sprechakte und Textillokutionen. A. Reihnachs Rechtsphilosophie und die moderne Linguistik. Tübingen: Niemeyer.

    Coulthard, Malcolm. 1985. An Introduction to Discourse Analysis. London: Longman.

    Davidson, Donald. 1967. "The Logical Form of Action Sentences". In: Rescher, N. (ed). The Logic of Decision and Action. Pittsburgh, 81-95.

    Gardiner, Alan. 1932. The Theory of Speech and Language. Oxford: Clarendon.

    Garfinkel, Harold/Harvey Sacks. 1970. "On Formal Structures of Practical Actions". In: McKinney, J./E. Tiryakian (eds). Theoretical Sociology. N.Y., 337-366.

    Harweg, Roland. 1990. Studien zur Deixis. Bochum: Brockmeyer.

    Kamp, Hans/Uwe Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer.

    Lenz, Friedrich. 1989. Organisationsprinzipien in mündlicher Fachkommunikation. Zur Gesprächsorganisation von 'Technical Meetings'. Frankfurt: Lang.

    Lenz, Friedrich. 1993. "Ansätze zur Analyse mündlicher Fachkommunikation". In: Bungarten, T. (ed). Fachsprachentheorie Bd.1, Tostedt, 341-367.

    Lenz, Friedrich. 1997. Diskursdeixis im Englischen. Sprachtheoretische Überlegungen und lexikogrammatische Analsyen. Tübingen: Niemeyer.

    Nöth, Winfried. 1996. "The (Meta-) Textual Space". In: Pütz, M./R. Dirven (eds). The Construal of Space in Language and Thought, Berlin, 599-612.

    Searle, John. 1969. Speech Acts. An Essay in the Philosophy of Language. Cambridge: CUP.

    Sinclair, John/Malcolm Coulthard. 1975. Towards an Analysis of Discourse. Oxford: OUP

    Smith, Carlota. 1991. The Parameter of Aspect. Dordrecht: Kluwer.

    Svartvik, Jan/Randolph Quirk. 1980. A Corpus of English Conversation. Lund: Glerup.

    Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca: Cornell University Press.

    Vendler, Zeno. 1972. Res cogitans. Ithaca: Cornell University Press.

    Watzlawick, Paul et al. 1967. Pragmatics of Human Communication. A Study of Interactional Patterns, Pathologies, and Paradoxes. New York: Norton.

    Wierzbicka, Anna. 1987. English Speech Act Verbs. A Semantic Dictionary. Sydney: Academic Press.

    Wunderlich, Dieter. 1970. "Die Rolle der Pragmatik in der Linguistik". In: Der Deutschunterricht 22, 5-41.

    1 Here (and in the following) no distinction is made between the pragma-linguistic tradition of discourse analysis and the ethnomethodological school of conversation analysis. For a detailed discussion of the theoretical and methodological assumptions of discourse analysis and conversation analysis see Lenz (1993).

    2 The term in fact goes back to Ungeheuer, but became more widely known because of an article by Wunderlich (1970).

    3 A metalanguage is the natural or formal language we need to describe a certain object. In linguistics this object is again a language, namely the language under consideration. Even in everyday communication we sometimes talk about language as such and the metalanguage is then (part of) our natural language. In this article, we are not interested in the general phenomena of metalanguage, but in the way we use English for metacommunication.

    4 Even in linguistic studies metacommunication is sometimes understood along the lines of power, symmetry, complementarity or other relationships. The characteristics of socio-emotional relationships are entities which are not limited to the time of a specific discourse. It is problematic to find their manifestations in the language data of a specific speech event and therefore their examination is a particularly problematic area in the framework of discourse analysis.

    5 In this context, it is quite surprising that the linguists concerned with discourse analysis more or less have ignored metacommunication and concentrated on the flow of speech acts as they interpreted them. On the other hand, the sociologists concerned with ethnomethodological conversation analysis have concentrated on those parts of conversation or discourse where the participants organise their gatherings. They have consequently looked at what they call 'formulations', but more or less ignored the linguistic means on which these are based. See below.

    6 In the framework of Discourse Representation Theory, which has nothing to do with discourse analysis but is an exercise in model theory and logic (cf. Kamp/Reyle 1993 for example), Davidson's idea has been more comprehensively developed, and it has recently become quite familiar in semantics to include verbs when considering reference. There is no reason discourse analysis should not follow this line.

    7 The transcript conventions used on the CD-Rom version are retained here. For a printed explanation see Svartvik/Quirk (1980, pp. 21ff.).

    8 T he more formal a discourse is, the less frequent is the use of talk. In written discourse we hardly find it at all. In fact, in this written and modified version of my former talk I used discuss instead of talk (cf. the last sentence of chapter 2). In principle, the same regularities apply to spoken and written discourse, but I will exclude written discourse from the discussion here since in a written (usually printed) text additional structuring techniques are used. It is often more explicit, and not all graphic devices have prosodic counterparts and vice versa.

    9 It is not necessary to discuss the semantics of speech act verbs in detail here. It is sufficient to see that whenever we use them to refer to parts of the discourse, this is a means of discourse organisation. In the examples given above, it contributed to the topical structure of the discourse; in other cases it contributes to the organisation of turn-taking or the sequencing.

    10 This in fact is a simplification. One has to take the arguments into account as well (cf. Lenz 1997). It is this verb constellation, I borrow the term from Smith (1991), which should ultimately be classified.

    11 Interestingly enough, we do not conceptualise Searle's simultaneous (sub-) acts as situations; we do not have verbs for utterance acts or propositional acts.

    12 The essential point is not the question which of the three spatial dimensions is selected, but the fact that it is only one. In Hausa, for instance the sagittal dimension can be used for discourse-deictic reference. I would like to thank Clifford Hill for drawing my attention to this.