Discourse Markers

Discourse represents an extremely complex phenomenon. The complexity of this topic

has caused little agreement as to what aspects should be included in a study of linguistics,
exclusively, or pragmatics.

The term “discourse” is normally used to refer to such extended, or multi-sentence,

utterances. Given that there is an interpretation question for single sentence utterances as well
as for multi-sentence utterances (=discourse), one should ask whether there is any connection
between the answers to these questions.

Discourse markers are linguistic elements that signal relations between units of talk,
relations at the exchange, action, ideational, and participation framework levels of the
discourse (Schiffrin,1987)1.

Discourse markers can be defined as a pragmatic class, lexical expressions drawn from
the syntactic classes of conjunctions, adverbials, and prepositional phrases. With certain
exceptions, they signal a relationship between the segment they introduce, and the prior
segment. They have a core meaning which is procedural, not conceptual, and their more
specific interpretation is “negotiated” by the context, both linguistic and conceptual.

One of the most important aspects of discourse markers is that they represent dynamic
interactive processes in which speech acts affect the situations in which they take place, and
how the situation affects the way the speech acts are understood.

In the same time, discourse in pragmatics appears as a complex argument in support of

the claim that all speech should be considered as action and, more specifically, that speech
can be described as the performing of actions similar to those performed by means of
performative utterances.

Pragmatic markers (also often called discourse markers or particles) in a broad sense
comprise the classes of speech act and politeness markers (pragmatics), text-sequence
markers of cohesion and coherence, and the signals of verbal interaction in discourse. They
are realized as words, phrases, or even syntactic constructions (e.g. question tags) and are
commonly characterized as short, recurrent (with high frequency), and optional verbal
elements, which are typically used in informal contexts (because of their stylistic

1
Schiffrin, Deborah, 1987. Discourse markers. Cambridge: Cambridge University Press.
1
stigmatization and negative evaluation) and display an extensive multi-functionality.
Pragmatic markers in a narrow sense refer to a set of linguistic items which have only little or
no lexical content and therefore do not contribute to the inference of the propositional part of
speech acts meanings. But they serve significant (non-propositional) functions of speech acts:
i.e. signaling speaker’s attitude, feeling, evaluation, or perspective on the things and matters,
empathy, ideology, etc. They also include modal modifiers like hedges, downtoners,
upgraders, etc., which are frequently used as politeness markers.

The central question concerns the relationship between linguistic meaning, its ways of
semantic and pragmatic modification of content, and the contextual parameters. The object of
this course thus touches the interface of semantics and pragmatics. Discourse markers in a
narrow sense are interactional and sequential signals and thus apply on the conversational or
discourse level. They serve “traffic” functions like connectivity of texts and talk
(“connectives”), turn taking activities and are of no relevance for the speech act meaning,
speaker’s attitudes, etc. Among these markers we also find invariant tag questions like: eh ?
Right ? Okay ? which may be subsumed under both classes of pragmatic/discourse markers.
The course will review the concepts and the system of markers, their use and functions,
contrasts between English and German, and then focus on theoretical and methodological
consequences for teaching pragmatic markers (in other words: the ways of expressing
speaker’s attitudes and evaluation) in the foreign language classroom.

According to http://dictionary.cambridge.org/grammar/british-grammar/discourse-
markers, discourse markers are words or phrases like anyway, right, okay, as I say, to begin
with. They are used to connect, organize and manage what speaker say or write or to express
attitude:

[Friends are talking]

A: So, I’ve decided I’m going to go to the bank and ask for a car loan.

B: That sounds like a good idea.

C: Well, you need a car.

B: Right.

A: Anyway, I was wondering if either of you would teach me how to drive.

The discourse markers in this extract have a number of uses:

- so marks the beginning of a new part of the conversation.

2
 - well marks a change in the focus (from getting a car loan to needing a car).

- right marks a response (B is agreeing with C).

- anyway marks a shift in topic (from buying a new car to having driving lessons).

We use different discourse markers in speaking and writing. In speaking, the

following discourse markers are very common:

- anyway, fine, great, good

- like, now, oh, ok

- right, well, so, mind you

- you know, I mean, for a start, as I say.

The following discourse markers are common in writing:

- firstly, secondly, thirdly

- in addition, in sum, in conclusion

- moreover, on the one hand

- on the other hand, to begin with.

It is known that discourse markers do not always have meanings that we will find in
our dictionary. However, they do have certain functions, and some discourse markers, such as
well, can have a number of functions. Some discourse markers are used to start and to end
conversations. Some are used to start new topics or to change topics.

Starting a conversation or talk

A: Right, let’s get started. We need to get the suitcases into the car.

B: Okay. I’ll do that. Katie, will you help me?

[at the start of a radio interview]

Now, we have with us in the studio today someone you will all know from television.
John Rice, welcome to the show.

Ending a conversation

3
 [A mother (A) and daughter (B) on the telephone]

A: So we’ll see you Sunday, Liz.

B: Right, okay Mum.

A: Okay, see you then, love.

B: Bye, Mum. Thanks for calling.

A: Bye, Liz.

[At the end of a meeting]

A: Anyway, is that it? Has anyone got any questions?

B: No. I think we’re done.

A: Right, fine, thanks everyone for coming. We’ll circulate the documents tomorrow
and make some follow-up calls about the project.

As we talk, we monitor (or listen to) what we are saying and how our listener is
responding to what they hear. We often rephrase or change what we say depending on how
our listener is responding. We use words and phrases such as well, I mean, in other words, the
thing is, you know, you know what I mean, you see, what I mean is.

Saying something in another way

Sometimes, as we talk, we add phrases to show our listener that we are going to
rephrase, repeat or change what we are saying. These discourse markers help to make what
we say clearer for the listener:

I just had to leave early. What I mean is I hated the show. It just wasn’t funny.

You exercise regularly, you have a good diet and you don’t have too much stress. In
other words, I think you have nothing to worry about. Your health seems very good.

I think I’ve found a house I’d like to buy. Well it’s an apartment actually. It’s ideal for
me.

Shared knowledge

4
 When we talk, we think about how much knowledge we share with our listener. We
often mark what we think is old, shared or expected knowledge with you know and we mark
new knowledge that we see as not shared with the listener with phrases like see, you see, the
thing is:

You know, hiring a car was a great idea. (The speaker and the listener know about
hiring the car.)

A: Why don’t you come and stay with me when you’re in Lisbon?

B: It’d be difficult. I have to be back in Dublin by Friday. You see, my sister is getting
married on Saturday so I won’t have time to visit. (B assumes that A doesn’t know about her
sister’s wedding. This is new information)

As we listen to someone speaking, we usually show our response to what we hear

either by gesture (head nod) or by a short response (Mm, yeah, really, that’s a shame). This
shows that we are listening to and interested in what is being said.

In other words, we call these short responses “response tokens”: absolutely, fine, okay,
wow, (all) right, good, quite (more formal), yeah, certainly, great, really, yes, definitely, I see,
sure, exactly, no, wonderful, that’s great/interesting/amazing/awful.

Therefore, the speaker uses response tokens for a number of functions:

To show interest and to show that we want the speaker to continue

A: So he opened the door.

B: Yeah.

A: And he went in very quietly without waking her.

B: Right.

A: He opened her bag and…

To show surprise

A: We’ve decided to go to Africa for a month next year.

B: Oh really!

To show sympathy

A: He can’t play soccer for at least six months. He’s broken his leg.

5
 B: That’s terrible.

Subclassification of Discourse Markers

DMs may be seen as syntactically optional connective expressions which do not

affect the truth-conditional content of utterances (Schourup, 1998) 2. The particular sense
in which DMs may be said to connect, and what they connect are controversial. On any
view of DM function, however, each individual DM marks a particular type of
interpretation, and it is therefore to the point to ask what kinds of intepretative
relationships DMs may signal.

A convenient starting point is the classification of DMs into four principal types,
suggested by Fraser (1996: 187-188) 3. We will discuss Fraser’s categories individually,
noting where exception might be taken to the groupings and labels he proposes, and noting
alternative categorizations proposed by Quirk et al. (1985) 4 and Halliday and Hasan
(1976)5.

Topic markers

Fraser sees one set of DMs (e.g. before I forget, incidentally, speaking of X) as
indicating that the speaker sees the following utterance as a departure from the cur rent
topic. A smaller subgroup of topic markers is seen as refocusing attention on a part of the
current topic (e.g. in fact, indeed, now). Markers of both kinds are referred to by Quirk et
al. (1985: 639) as transitional.

Since the notion topic can be applied to more than a single utterance, a topic
change marker can presumably serve to relate a single utterance to a property of a longer
stretch of discourse. Fraser’s inclusion of topic change markers in the DM category thus
opens the possibility that DMs, when they connect elements of text, mark not only local
pairwise relations between utterances, but also more “global” relations (Bell, 1994: 165-
166) 6. Schiffrin also allows for this possibility, but chooses not to discuss DMs of this
kind (Schiffrin, 1987:24). Lenk, however, focuses entirely on DMs she refers to as

2
Schourup, Lawrence, 1998, Discourse Markers – A tutorial overview – Lingua 107, 227-265, Elsevier Science B.V.
3
Fraser, Bruce – Pragmatic markers, Pragmatics 6 (2), 1996, Boston University, 1996
4
Quirk, R., S. Greenbaum, G. Leech and J. Svartvik, 1985. A comprehensive grammar of the English language. London:
Longman.
5
Halliday, M.A.K. and R. Hasan, 1976. Cohesion in English. London: Longman.
6
Bell, D.M., 1994. Cancelative discourse markers. Ph.D. dissertation, Boston University.
6
“globally oriented”, such as anyway, actually, and incidentally, which she sees as
performing 'topical actions” such as closing digressions, returning to prior topic, changing
topics, introducing a new topic, and inserting a subjective aside. Most of the functions
Lenk discusses are encompassed by Fraser's notion of topic change.

The idea that DMs may mark “topical actions” presupposes that the notion topic is
useful in describing discourse comprehension. The term “topic” is, however, notori ously
difficult to define, and as Fraser notes, many have avoided using the term. Fraser uses
“topic” loosely to refer to “what the discourse participants are ‘talking about’ at any
given time” (1988:27). Lenk 7 defines the notion “conversational topic” informally in the
same way (1998:25), but it is clear that, like many coherence theo rists (e.g., Giora, 1997) 8,
Lenk takes the notion of topic to be important theoretically and sees it as central to the
description of certain DMs. She stops short of rigorously defining the notion, but subscribes
to the view of Bublitz that “determining the topic for the text at hand is a matter of a
comprehending, interpretative ascription” (1988: 18f.) 9. Some writers, on the other hand,
have claimed that the notion “topic” is of no theoretical importance whatsoever (e.g.
Blass, 1990: 76)10.

Sperber and Wilson argue that since the function of discourse topics is simply to
provide contextual information required for comprehension, topic relevance is a derivative
notion and can be dispensed with on a relevance-theoretic account (1995: 216-217; cf.
Wilson, 1998: 68; for an opposing view see Giora, 1997). If topic relevance is a derivative
notion which has no role in utterance interpretation, topic change markers might more appro-
priately be classified as “change of context markers” (cf., Wilson, 1998: 72) 11. Unger
argues, more generally, that paragraphs and all other discourse segments have no sta tus as
primitive theoretical elements and that when particles, pauses, and other stylis tic means
appear superficially to indicate breaks between such units, such devices should be viewed
instead simply as “facilitators of context change” (1996: 431) 12.

Contrastive markers
7
Lenk, U., 1998. Marking discourse coherence. Tijbingen: Narr.
8
Giora, R., 1997. Discourse coherence and theory of relevance: Stumbling blocks in search of a unified theory. Journal of
Pragmatics 27, 17-34.
9
Bublitz, W., 1988. Supportive fellow-speakers and cooperative conversations. Amsterdam: Benjamins
10
Blass, R., 1990. Relevance relations in discourse: A study with special reference to Sissala. Cambridge: Cambridge
University Press.
11
Wilson, D., 1998. Discourse, coherence and relevance: A reply to Rachel Giora. Journal of Pragmatics 29, 57-74.
7
 Fraser sees contrastive markers, such as hut, conversely, and nevertheless, as sig-
naling that the following utterance is “either a denial or a contrast of some proposi tion
associated with the preceding discourse” (1996: 187). Fraser distinguishes between
contrastives which signal a sharp contrast in message content (e.g. conversely, in
contrast, on the contrary), those which signal a sharp but unexpected contrast (e.g. all the
same, still, instead), and those signaling a contrast between “a previous claim or like
message [...] and the claim in the current message” (e.g. I may be wrong hut, that said).
Quirk et al. (1985: 634-636), who refer to a set of conjuncts as contrastive, recognize
four subdivisions: reformulatory (rather, more accurately, alternatively), replacive (again,
on the other hand, better), antithetic (conversely, on the contrary, in comparison), and
concessive (anyway, yet, all the same). Halliday and Hasan refer to conjunctives of these
kinds as adversative, and suggest a general distinction between internal and external uses
(roughly, those expressing relations inherent in the phenomena language talks about and
those inherent in the communication process), with further subdivisions (1976: 240-244).

Bell (1994) argues that the defining notion for most such markers should be one
not of contrast or denial of expectations but of cancelation. Cancelation refers to the way
in which aspects of information the hearer considers derivable from the prior utterance are
considered to be non-operative with respect to the speaker's intended meaning. An
example involving the prototypical cancelative marker but is:

“I went to the liquor store last Sunday but it was closed.” (Bell, 1994: 152)

Bell sees but here as canceling an aspect of information contextually inferrable

from I went to the liquor store last Sunday. What is canceled in particular is the inference
that the speaker expected the store to be open. The notion of cancelation is not, of course,
far removed from that of contrast: canceling information results in a contrast between the
information or assumptions entertained before cancelation and those entertained
afterward. Nevertheless, the notion of cancelation might be seen as giv ing more specific
content to the somewhat vague notion of “contrast”.

Elaborative markers

An elaborative marker (e.g. above all, by the same token, moreover) is in Fraser's
view one which indicates that the following utterance “constitutes a refinement of some
sort on the preceding discourse” (1996: 188). A similar category for conjuncts, additive, is
12
Unger, C., 1996. The scope of discourse connectives: Implications for discourse organization. Journal of Linguistics 32,
403-438
8
proposed by Quirk et al., and divided into equative conjuncts (correspondingly, equally,
by the same token) and reinforcing conjuncts (further, moreover, in addition). Quirk et
al., however, place additive conjuncts within a more general category of listing conjuncts
which also includes enumerative items, such as in the first place, next, and finally. Fraser
considers the latter to be commentary markers with representational meaning (1996:
182).

Halliday and Hasan also use the term additive to refer to a class of conjunctive
elements within which they distinguish three subcategories: additive (and, and also,
furthermore), negative (nor, and ... not), and alternative (or, or else, alternatively). Also
included among additives are conjunctions indicating afterthought (incidentally, by the
way) which Fraser would classify as topic markers, and Quirk et al. as transitional
conjuncts.

Inferential markers

According to Fraser, inferential markers, including after all, so, and as a result,
have the function of indicating that "the force of the utterance is a conclusion which
follows from the preceding discourse" (1996: 188). It is difficult to see why Fraser has
restricted the function of inferential DMs to marking conclusions. This descrip tion clearly
does not apply to after all, which marks not a conclusion but a premise for the deduction
of the proposition expressed by a previous clause (as does you see; see Blakemore, 1987:
82, 89), but if after all and you see are excluded, it is not clear what other category might
accept them. The alternative is to broaden the inferential category to include both markers
of premises and conclusions, but this, too, is problematic, since if premise markers are
allowed, moreover could be included here rather than among elaborative markers
because it links two premises offered in support of the same conclusion (see Blakemore,
1987: 91).

Quirk et al. also posit a class of inferential markers (otherwise, in that case, in
other words) but distinguish these from resultive markers (so, accordingly, as a result) on
the grounds that only the former indicate a conclusion based on "logic and supposition"
(1985: 638). Not all writers would concur either in identifying the notion of result as
operative in the case of markers like so, or in the view that such markers do not indicate a
conclusion based on "logic and supposition". The term inferential marker as used by Quirk
et al. has a broader reference than the same term as used by Fraser and also from the use of

9
the term inferential in the RT literature, where “inferential connective” can substitute for
DC. Halliday and Hasan refer to a broad class of inferential conjunctive elements as
causal, a cover term for relations of result, reason, and purpose.

Discussion

The foregoing comments are sufficient to suggest that the subcategorization of

DMs is no more settled an issue than that of how the superordinate category DM should
be described. There are broad content similarities among the four systems of classification
considered here. All four systems include major categories which are, roughly, additive,
inferential, and contrastive, but beyond this superficial similarity lies a maze of
disagreements. First, there is disagreement about the classification of particular items. That
is, for example, is classed both by Quirk et al. and Halliday and Hasan as appositive, but
while Halliday and Hasan regard such appositives as a subclass of additive conjunctions,
Quirk et al. distinguish separate classes of appositive and additive conjuncts. Such
disagreements reflect more fundamental disagreements about the basis of co-classification.
Thus Halliday and Hasan see the additive relation as based on the logical notion 'and'
(1976: 244), while Quirk et al. consider this relation to effect the strong integration of the
conjoined elements (1985: 636). Fraser's similarly named category of elaborative DMs, on
the other hand, focuses on the notion that one utterance is a refinement of the preceding
one (1996: 188).

Finally, there are differences in the number of general categories. Fraser and
Halliday and Hasan require only four, while Quirk et al. posit seven. Differences of these
kinds are at least in part linked to the way the superordinate category DM/con -
junct/conjunctive is viewed. For example, it is the fact that Halliday and Hasan make no
use of the notion of non-truth-conditionality in defining conjunctives that makes it
possible for them to include a general class of 'temporal' items among conjunc tives and
to include in this class items such as an hour later and at once which would be roundly
excluded from the DM class based on their truth-conditionality.

Summary

The term DM typically refers to a more or less open class of syntactically optional, non-
truth-conditional connective expressions. There is, however, wide disagreement about
the nature of the connection DMs express, the nature and extent of the elements
connected, and the grammatical status of the DM category. An inconvenient result of

10
such disagreements is that the items (and uses of items) designated by the term DM on
one definition sometimes overlap only minimally with those designated on another
definition. Equally inconvenient is the fact that the referential overlap between the term DM
and other similar terms, such as pragmatic marker and pragmatic particle, can in some
cases be as great as that between variant definitions of the term DM itself.

The overabundance of terms and definitions in this area cannot be described to

claim-staking or fashion. In general, the terms and definitions used in each framework
are chosen to reflect theoretical preoccupations, to avoid unwanted associations, or to
rule in or out particular linguistic items or functions. Such variation is to be expected in
an area that has only recently become a focus of intensive study and which bears on
many different areas of discourse research, cognitive, social, textual, and linguistic. On
the other hand, so long as such uncertainties exist, DM must remain a term with
theoretical aspirations, but whose precise reference remains at issue.

“Context” is an all-pervasive term in pragmatics. For some authors “context” is the

defining concept of pragmatics. But too many different concepts are included under this term.

In Linguistics, “context” commonly means the previous and subsequent linguistic

material in a given text. In Kaplan's scheme, the context is objective, it comprises the actual
basic facts about an utterance: the speaker, time, place, and possible world in which it occurs.
In Stalnaker's scheme, context is basically subjective: a matter of common ground: that is,
shared beliefs that serve as common presuppositions for the interpretation of assertions. Often
the term is used for anything in the indefinitely large surrounding of an utterance, from the
intentions of the speaker to the previous topics of conversation to the objects discernible in the
environment.

Here are a number of distinctions that have been made with respect to the concept(s)
of context that the reader may find helpful. The list is not intended to be exhaustive, mutually
exclusive, or to represent a single coherent view of pragmatic phenomena.

Linguistic context versus extralinguistic context

Considering the context of an utterance, one of the most intuitive distinctions is

between the context as consisting of its previous and subsequent utterances — the linguistic
contexts — and any other extra-linguistic circumstance surrounding the utterance. In the case
of ellipsis and anaphoric (and cataphoric) pronouns the designation is determined, or at least

11
constrained, by the linguistic context of the utterance, while the designation of deictic
demonstratives is fixed by contextual extralinguistic facts.

Narrow versus broad

Narrow context is usually understood as the list of parameters for basic indexicals,
parameters that correspond to basic facts about the utterance. Speaker, place and time are on
almost everyone’s list, as required for the interpretation of “I”, “here”, “now” and tense.
Kaplan adds the possible world, for the interpretation of “actual”. Arguably, every utterance
occurs in a world, at a time, in a place, and with a speaker.

In contrast, wide or broad context is understood as all other kinds of information, in

particular, information relative to the speaker's communicative intention, used for the
interpretation of ‘pragmatic aspects’ of the utterance. In Bach's words:

Wide context concerns any contextual information relevant to determining the

speaker's intention and to the successful and felicitous performance of the speech act…
Narrow context concerns information specifically relevant to determining the semantic values
of [indexicals]… (Bach 1999a)

Bach goes on to say that narrow context is semantic, wide context pragmatic.

But on this there would be some disagreement, as many would hold that information
about the speaker's intentions, and perhaps also about causal and informational chains,
ongoing topics of conversations, and much else, are needed for semantics.

Epistemic versus doxastic

It is sometimes assumed, particularly by writers with a psychological orientation, that,

together with speaker's intentions, it is speaker's beliefs what determine the content of
expressions in her utterance, with the issue of the truth (or falsity) of her beliefs having no
relevance. So, belief rather than knowledge is the relevant concept to characterize context.
According to this view, there would not be a significant difference between ‘intending to say’
and ‘saying,’ which goes about some truths of intentions in general: one usual way of failing
to perform an action comes from the falsity of some agent's belief. This is related to another
possible distinction between objective and subjective contexts.

Objective versus subjective

There are a number of cases, however, in which the speaker's beliefs, even if shared by
everyone in the conversation, do not seem to determine content. Suppose, for example, that
12
Elwood's hero worship of John Searle has reached such a point that he now takes himself to
be John Searle. He introduces himself to the new class of Stanford graduate students by
saying, “I'm John Searle, from across the Bay.” It seems that even if he and everyone in the
conversation believes he is Searle, what he has said is the falsehood, that Elwood is John
Searle.

13