Metadiscourse: Theoretical background
In the last several years, few elements of
language have elicited more study from
scholars in various related academic fields
than have elements that can be classified as
metadiscourse. Scholars involved in the
study of metadiscourse represent fields such
as discourse analysis, linguistics, applied
linguistics, pragmatics, rhetoric, and second-language theory and pedagogy. And the
languages in which they have focused on
metadiscourse are numerous; in his 2005
book Metadiscourse: Exploring Interaction
in Writing, Ken Hyland (2005) refers to
studies of metadiscourse in at least eleven
different languages, ranging from Arabic
through Iranian to Vietnamese.
The term metadiscourse is closely related to
terms such as metatalk (cf. metalanguage
and metacommunication), which some
researchers in conversational analysis use to
name the language people employ to talk
about language (cf. Schiffrin, 1980). In my
work, I have focused mainly on written
language, and I have used metadiscourse to
designate elements of texts that convey
meanings other than those that are primarily
referential.
In describing such meanings, I follow
Halliday (1973), who has shown that when
people use language, they usually work
toward fulfilling three macro-functions.
They try to give expression to their
experience, to interact with their audience,
and to organize their expressions into
cohesive discourses that their addressees can
make coherent sense of. In other words,
Halliday (1973) asserts that people
communicate with messages that are
integrated expressions of three different
kinds of meaning, which he calls ideational,
interpersonal, and textual.
Linguistic elements that convey ideational
meaning “are concerned with the content of
language, its function as a means of the
expression of our experience, both of the
external world and of the inner world of our
own consciousness” (Halliday, 1973, p. 58).
Some examples of these elements in clauses
are those that express transitivity. The most
accessible labels for these are identical to
those found in many case grammars, such as
“agent,” “process,” and “goal.”
Elements that convey interpersonal meaning
are concerned with “language as the
mediator of role, including all that may be
understood by the expression of our own
personalities and personal feelings on the
one hand, and forms of interaction and
social interplay with other participants in the
communication situation on the other hand”
(Halliday, 1973, p. 58). These elements
show how a person steps into the rhetorical
situation and tries to affect others; these
elements carry essentially social meanings.
In clauses, some of these elements indicate
choices of mood; others are some of the
modal verbs.
Finally, elements within the textual set have
“an enabling function, that of creating text,
which is language in operation as distinct
from strings of words or isolated sentences
and clauses. It is this component that
enables the speaker to organize what he [or
she] is saying in such a way that it makes
sense in context and fulfills its function as a
message” (Halliday, 1973, p. 58).
What I up to this point have labeled
referential meaning is equivalent to what
Halliday (1973) calls ideational meaning.
And I suggest that kinds of metadiscourse
convey interpersonal or textual meanings.
Interpersonal metadiscourse helps writers
express their personalities, reveal their
evaluations of and attitudes toward
ideational material, show what role in the
communication situation they are choosing,
and indicate how they hope readers will
respond to the ideational material. Textual
metadiscourse helps writers show how they
relate bits of ideational material within a text
and how that text makes sense in a particular
situation or situations.
What this analysis suggests is that as we
write, we generally proceed on more than
one level (cf. Williams, 1981, p. 47). On one
level, we expand ideational material. On the
levels of metadiscourse, we do not expand
ideational material but help our readers
connect, organize, interpret, evaluate, and
develop attitudes toward that material.
Thus, although a well-formed text is, strictly
speaking, an integrated expression of three
kinds of meaning, there is a sense in which
what I have called metadiscourse is
discourse about discourse.
Kinds of metadiscourse
In 2002, I offered a taxonomy of
metadiscourse that included six main
categories (Vande Kopple, 2002):
1. Text Connectives: These show
readers how the parts of texts are
connected to one another and
how texts are organized.
Specific examples of these are
elements that indicate sequences
(first, next, in the third place) as
well as those that indicate logical
or temporal relationships
(consequently, at the same time).
Also included with the
connectives are reminders about
material presented earlier in texts
(as we saw in Chapter One) and
statements about forthcoming
material (as we shall see in the
next chapter). Finally,
sometimes writers use what
Williams calls topicalizers (as
for, with regard to, in connection
with). These are words that
“focus attention on a particular
phrase as the main topic of a
sentence, paragraph, or whole
section . . .” (p. 50).
2. Code glosses: These “help
readers grasp the appropriate
meanings of elements in texts”
(Vande Kopple, 1985, p. 84).
Sometimes we judge that we
should define a word or phrase
for our readers. Or sometimes
we signal that there is a problem
with the ordinary interpretation
of a word; we use expressions
such as so-called or what some
people call (cf. Stubbs, 1986, p.
13). At other times, we signal
how strictly or loosely we wish
readers to take our words—we
use expressions like strictly
speaking or technically to signal
strict or technical interpretations,
and expressions like sort of (cf.
Aijmer, 1984) and roughly
speaking to signal loose
interpretations. At still other
times, we predict that readers
might be having trouble
interpreting passages, and we
signal that we will re-phrase: I’ll
put it this way or What I mean to
say is.
3. Illocution Markers: With these
we “make explicit to our readers
what speech or discourse act we
are performing at certain points
in texts” (Vande Kopple, 1985, p.
84). For example, we can use
elements such as I hypothesize
that, to sum up, we claim that, I
promise to, and for example.
Further, we can modify the
amount of force that many
illocution markers and the acts
they signal have. For instance,
we might add a modal verb to a
direct request: I must ask that
you. All elements that attenuate
the force of speech acts can be
called mitigators (cf. Fraser,
1980, p. 342). On the other
hand, we can increase the force
of certain speech acts with
boosters such as enthusiastically
and most sincerely.
4. Epistemology Markers: Several
kinds of metadiscourse are linked
in the overarching function of
indicating some stance on our
part toward the epistemological
status of the ideational material
we convey. One stance has to do
with how committed we are to
the truth of ideational material.
Sometimes we are cautious, and
we signal that caution with what
I call shields (such as it is
possible that and perhaps).
Sometimes we as writers
“underscore what we really
believe—or would like our
reader to think we believe”
(Williams 49) by using what are
called emphatics (such as without
a doubt and most certainly). A
second kind of stance that we can
take with regard to the
epistemological status of
ideational material has to do with
the “kinds of evidence”
(Anderson, 1986, p. 273) or
bases we have for that material.
Chafe (1986) notes that there are
several different bases we might
have for ideational material, from
our personal beliefs (I believe
that) through sensory experience
(it feels like) to what we hear
from others (the professor told
me that).
5. Attitude Markers: The function
of the fifth kind of metadiscourse
is to help us reveal what attitude
we have toward ideational
material. To express such
attitudes, for example, we can
use adverbs such as fortunately,
parenthetical expressions such as
I regret and I rejoice, and clauses
such as I am grateful that.
6. Commentary: The final kind of
metadiscourse in my 2002
taxonomy is commentary, with
which we address readers
directly, often appearing to draw
them into an implicit dialogue.
For example, we can comment
on their probable moods, views,
or reactions to our ideational
material (some of you will be
amazed that) or even recommend
a mode of reading (You might
wish to skip to the next chapter).
In Metadiscourse: Exploring Interaction in
Writing, which is probably the best place
now for those interested in metadiscourse to
become oriented to it and its study, Ken
Hyland offers a somewhat different
taxonomy from mine. Here it is not my goal
to try to reconcile these two taxonomies or
expand upon them. Rather, I would like to
devote a few pages to some thoughts about
why the study of metadiscourse is so
interesting and important. And I am hoping
that these thoughts will stimulate others to
add to them in the future.
The study of metadiscourse: Aspects of its
interest and importance
One of the reasons the study of
metadiscourse is so interesting and
important is that it shows how intricately
structured language is and how attentive to
detail one must be in the study of language
and its effects. Consider some examples
directly related to metadiscourse: In some
sentences readers can find several different
kinds of metadiscourse. For instance, in
Finally, I am sorry to proclaim that you are
guilty, readers find a text connective, an
attitude marker, and then an illocution
marker before they get to the ideational
material.
Further, in other sentences some linguistic
forms appear to fulfill more than one
metadiscoursal function at a certain point in
a text. As Barton (1995) suggests, some
kinds of metadiscourse may fulfill functions
in both the interpersonal and textual
domains. She suggests that text connectives,
which clearly have textual functions, can in
academic argumentation also serve
“complex interpersonal purposes” (235).
Similarly, I hypothesize that probably
functions in some texts as both an illocution
marker and a shield. And phrases like to
conclude this section probably often
function both as text connectives and
illocution markers. Perhaps the kind of
categorization that will emerge in future
research will show overlaps between
Halliday’s (1973) macro-functions of
language.
Finally, it is important to note that some
linguistic forms can function as
metadiscourse in some contexts but as
conveyors of ideational information in
others. The clause I guess in “He is, I guess,
at work” functions as a shield. The same
clause, albeit now lacking the commas that
set it off parenthetically, conveys ideational
content in “I guess a lot on standardized
tests.” Similarly, may (along with be) in
“There may be a correlation between the two
findings” functions as a shield; however,
may in “Teachers in that school may never
reprimand students” functions as part of the
ideational content.
Just these few examples show how finely
nuanced meanings conveyed by
metadiscourse can be and how carefully
researchers must examine linguistic
elements, meanings, and probable effects of
those meanings within particular contexts.
Another reason why research on
metadiscourse is interesting and important is
that it opens up intriguing questions about
ethics. A good study to examine in this
connection is Simpson’s (1990) “Modality
in Literary-Critical Discourse.” Simpson
examined F. R. Leavis’s use of shields. He
notes that in “The Great Tradition,” Leavis
shields relatively uncontroversial statements
(for example, about influences of Dickens
on Conrad) and then leaves truly
controversial statements unshielded (for
example, “D. H. Lawrence . . . was the great
genius of our time”; cited in Simpson, 1990,
p. 89). Simpson points out that this tactic
helps Leavis nudge his readers into asking
the little questions but skipping the big ones
about issues in the history of English
literature.
One especially interesting thing about
Leavis’s tactic is that it raises important
questions about how to convey material that
is not accepted as certain. When a writer
uses an emphatic such as obviously (as in
“Obviously, Trollope was a great writer”),
or when a writer uses an attitude marker like
regrettably (as in “Regrettably, they stopped
doing meaningful research last year”), is the
writer in any way seeking to “sneakily
strengthen the force of the proposition by
presupposing its truth” (Holmes, 1984, p.
353)?
So which ways of using metadiscourse with
debatable material are fair and just? If some
ways are not fair and just, how serious is the
harm that those ways cause? In this
connection, I believe that beyond shields,
emphatics, and attitude markers, evidentials
and bits of commentary could reward further
study.
A third way in which the study of
metadiscourse is interesting and important
has to do with how metadiscourse is used in
similar texts in different languages.
Mauranen (1993) has found that native
speakers of Finnish use few text connectives
in economics texts in Finnish, whereas
native speakers of English, in similar texts in
The importance of studying metadiscourse 44
English, use a good many connectives. This
finding accords in part with research by
Clyne (1991), who found that texts in
linguistics and sociology produced “by
English speakers are far more likely to have
advance organizers than those [in German]
by Germans” (54).
Mauranen (1993) writes that the Finnish
school system teaches that using connectives
“is not only superfluous, but the sign of a
poor writer” (8). Many North American
schools, on the other hand, stress using
connectives, especially between paragraphs.
Mauranen comments that the different
Finnish and Anglo-American practices of
using textual metadiscourse probably reflect
different ideas of politeness and of what
should be expected of readers. She would
say that Finnish writers show respect for
their readers by leaving more of the textual
processing up to them.
These points lead me to wonder about how
much misunderstanding and possibly even
conflict might be associated with different
practices with and understandings of
metadiscourse. For example, imagine
people from a cultural-linguistic background
that values individual deference and group
identity encountering important messages
full of attitude markers from people from a
different cultural-linguistic background. Or
imagine people from a cultural-linguistic
background that leads them to be very
careful about the truth value and sources of
their ideation conveying such material to
people whose cultural-linguistic background
leads them to view messages containing
shields and evidentials as exhibiting a kind
of weakness. Or consider the following
example, one that students often bring to my
attention: Imagine how people from a
cultural-linguistic background that shuns all
direct references in writing to writers and
readers would react to texts full of
commentary including the pronouns I and
you.
One final example in this connection: I
remember hearing James Paul Gee giving a
talk at a convention, a talk that dealt with
different understandings of uses of
interpersonal metadiscourse, among other
things. He told about how a doctoral student
from a non-Western culture for some reason
lost her dissertation director at a university
in the United States and was nearly in panic
trying to line up a new one. She came to
Gee and said, “You must be my new
director.” Gee never revealed how he
ultimately responded to her, but he made it
clear that her approach, with its very strong
modal verb, shocked him in that it was not
the polite and deferential one that he had
become accustomed to in North American
graduate education.
The final way that I would like to discuss
about how the study of metadiscourse is
interesting and important is actually implicit
in some of the comments appearing directly
above. That is, metadiscourse deserves a
prominent place in second-language
instruction. We have seen that texts in one
language might contain more text
connectives than do similar texts in other
languages. It is easy and reasonable to
assume that certain kinds of texts in some
languages would pay more attention to
expressing precise degrees of certainty about
information than would texts in other
languages. Similarly, it is easy and
reasonable to assume that texts in some
languages would contain more instances of I
and you within bits of commentary than
would similar texts in other languages.
Expanding this list of possible differences in
uses of metadiscourse would be relatively
easy.
What is more important now, however, is
seeing how much detailed instruction in
metadiscourse would be needed to lead a
speaker and writer of one language to
approach fluency and facility in another.
Such teaching would probably have to
include at least the following several steps:
1. Looking closely at a variety of
texts in the L2 to discover
what elements of
metadiscourse appear.
2. With the help of a native
speaker of L2, discovering
whether the uses of
metadiscourse are natural and
successful or not.
3. Deciding which function or
functions the elements of
metadiscourse are meant to
fulfill.
4. Discussing whether or not
other specific elements of
metadiscourse could be
substituted for the elements
of metadiscourse that do
appear.
5. Discussing whether there is a
link between functions of
metadiscourse and aspects of
the culture that sustain and is
sustained by the particular
L2.
6. Discussing why there might
be a link between functions
of metadiscourse and aspects
of the culture that sustain and
is sustained by the particular
L2.
7. Working on analyses,
exercises, and real-world
tasks to help the students
learn appropriate uses of the
metadiscourse.
This list probably only hints at the
pedagogical steps necessary to help students
acquire skillful use of metadiscourse in an
L2. And even though the list is probably
incomplete, it shows how challenging the
task of teaching something like full
acquisition of an L2 is. But such teaching is
certainly worth pursuing, for it is with such
teaching that we move toward true cultural
and linguistic meeting of minds.
