Publishing Partner: Cambridge University Press CUP Extra Publisher Login

New from Cambridge University Press!


Revitalizing Endangered Languages

Edited by Justyna Olko & Julia Sallabank

Revitalizing Endangered Languages "This guidebook provides ideas and strategies, as well as some background, to help with the effective revitalization of endangered languages. It covers a broad scope of themes including effective planning, benefits, wellbeing, economic aspects, attitudes and ideologies."

New from Wiley!


We Have a New Site!

With the help of your donations we have been making good progress on designing and launching our new website! Check it out at!
***We are still in our beta stages for the new site--if you have any feedback, be sure to let us know at***

Review of  Evaluation in Text: Authorial Stance and the Construction of Discourse

Reviewer: Diana M. Lewis
Book Title: Evaluation in Text: Authorial Stance and the Construction of Discourse
Book Author: Susan Hunston Geoffrey Thompson
Publisher: Oxford University Press
Linguistic Field(s): Discourse Analysis
Text/Corpus Linguistics
Issue Number: 12.1637

Discuss this Review
Help on Posting

Hunston, Susan and Thompson, Geoff (eds.) (2000)
OF DISCOURSE. Oxford: Oxford University Press. Paperback.
225 pages.

Reviewed by Diana Lewis, University of Oxford

EVALUATION IN TEXT is a collection of nine papers on
evaluation in language, conceived as an introduction for
students wishing to research in this area. It also aims to
promote evaluation to a more prominent place in
descriptive linguistics. Eight of the nine papers deal
only with the English language.

Evaluation is defined as "the broad cover term for the
expression of the speaker or writer's attitude or stance
towards, viewpoint on, or feelings about the entities or
propositions that he or she is talking about" (p. 5) It
includes speaker assessment of both desirability and
likelihood; that is, both value-indicating comment and
epistemic/evidential comment. Most of the papers take a
qualitative approach, though several are based on data
from corpora.

The selection of papers is based on the editors' aim to
"represent as wide a range of approaches as possible,
while allowing our writers the luxury of comparatively
long contributions" (p. v). The approaches represented
are: systemic functional linguistic theory, narrative
discourse type, corpus linguistic methodology and
language-and-ideology studies (p. v).

The nine chapters are:

1. Geoff Thompson and Susan Hunston. 'Evaluation: an
introduction'. Defines and maps out evaluation.
2. Michael Hoey. 'Persuasive rhetoric in linguistics: a
stylistic study of some features of the language of Noam
Chomsky' Looks at how aspects of information structure
convey evaluation.
3. Joanna Channell. 'Corpus-based analysis of evaluative
lexis'. Analyses the 'evaluative polarity' of seven
4. Susan Conrad and Douglas Biber. 'Adverbial marking of
stance in speech and writing'. A corpus-based account of
evaluative adverbials in three registers of English.
5. Susan Hunston and John Sinclair. 'A local grammar of
evaluation'. Focuses on some English evaluative adjectives
and the contextual patterns associated with them.
6. Martin Cortazzi and Lixian Jin. 'Evaluating evaluation
in narrative'. Takes a wide view of evaluation and
considers how narratives are interpreted and assessed.
7. Geoff Thompson and Jianglin Zhou. 'Evaluation and
organization in text: the structuring role of evaluative
disjuncts'. Suggests that many evaluative disjuncts
function as indicators of coherence relations.
8. J.R. Martin. 'Beyond exchange: APPRAISAL systems in
English'. Proposes a model for the description of
evaluation in spoken and written English.
9. Susan Hunston. 'Evaluation and the planes of discourse:
status and value in persuasive texts'. Proposes a
multidimensional model of evaluation in text.

In their introduction, Thompson and Hunston give an
overview of evaluation, its functions in discourse and how
it can be recognized.

A major problem in this area, as Thompson and Hunston
point out, is the plethora of terminology and the lack of
a consensus among linguists on how best to delimit,
subcategorize and identify evaluation in text. The first
task is therefore to clarify terms and this is undertaken
in the introductory chapter. The editors distinguish
affective (good-bad) opinion, which relates primarily to
entities, from epistemic (probability) opinion, which
tends to relate to propositions. They then explain their
decision to view these two types of speaker/writer opinion
as subtypes of the category 'evaluation'. One reason given
for this is the overlap in the structural means of
expressing the two types, as in 'it is gratifying/fairly
certain ...'. Thompson and Hunston subsequently add two
further types of evaluation: expectedness and importance.
They suggest that the three parameters of probability,
expectedness and importance can be related to the 'basic'
good-bad parameter.

Identifying evaluation, claim Thompson and Hunston, "is a
question of identifying signals of comparison,
subjectivity and social value ... evaluation consists of
anything which is compared to or contrasts with the norm"
(p. 13). The problem is how to constrain this notion of
evaluation. "Most readers of a text agree about what
counts as evaluation in it" Thompson and Hunston assure us
(p. 13). But do they? Thompson and Hunston's definition is
wide: it encompasses lexical, grammatical and textual
structures; attitudinal, interpersonal and discourse-
organizational functions; pragmatic inferences as well as
conventional, coded meanings. Do readers, including
linguists, not constantly experience what Lyons refers to
as ".. the difficulty of deciding, in the case of
individual utterances and more generally, exactly what
meaning is encoded in the lexemes, particles and
grammatical categories of particular languages" (1995:
276-7)? Thompson and Hunston suggest that "the advantage
of looking at evaluation conceptually is that it does not
restrict what can be counted as evaluation", while "the
disadvantage .. is that the argument for what constitutes
evaluation becomes circular" (p. 14). That is, the
identified evaluation clarifies the yardstick of value,
which in turn facilitates the identification of
evaluation. This first chapter nevertheless admirably
brings together many and varied aspects of subjectivity
and puts some perspective on them.

Michael Hoey's paper is a reprint of his 1984 article in
Forum Linguisticum. It examines two short passages from
Chomsky's early writings for embedded evaluation, i.e.
evaluations that are presented as given, by use of factive
predicates, attributive rather than predicative adjectives
and so on. This chapter is a useful reminder of the
importance of the role that information structure plays in
expressing speaker/writer viewpoint.

Chapters 3 and 4 are corpus-based studies. In chapter. 3,
Channell is concerned with evaluation of the affective
(good-bad) type. She examines corpus data on seven
expressions - including 'fat', 'regime', 'par for the
course', 'roam' - to identify their positive or negative
'evaluation polarity'. Negative evaluative polarities seem
more frequent than positive ones. Channell rightly notes
that for this approach "researchers must have in front of
them a large number of examples" (p. 41). It is therefore
surprising not to find quantitative data on the
expressions discussed in this chapter.

Channell lucidly makes three important points. First, that
the 'evaluative polarity' of an expression as revealed by
analysis of corpus data is often not accessible to
introspection. Second, that evaluation is largely context-
dependent. She thus tackles the central issue of the
semantic/pragmatic boundary and the quasi-conventional
meanings that many lexical items carry in particular
contexts but not others. These two points may well be
related. Third, she draws attention to the need to link
use of evaluative language to 'facework' (in the sense of
Brown and Levinson 1987). She thus points the way towards
much valuable further research to be done along these

The second corpus-based paper, by Conrad and Biber, looks
at adverbial marking of stance (evaluation). The authors
calculate the frequency and distribution of a range of
stance-marking adverbials in conversation, in academic
prose and in news reportage.

The editors' introduction to this contribution is a little
confusing. To say, for example, that "in conversation
stance is most frequently indicated by an adverb, followed
by clauses, followed by prepositional phrases" (p.56)
seems to be a shorthand way of saying that a set of
adverbial expressions of stance is identified by the
authors, and that of the tokens of the set members found
in the conversation corpus, the largest number are
adverbs. In the paper itself, we several times find
'stance markers' used as shorthand for 'adverbials marking
stance', e.g. "these four adverbials account for about 70
per cent of all epistemic stance markers in conversation"
(p. 65). Such shorthand could confuse the inattentive

The chapter proceeds in a clear and well organized manner.
It describes the frequency and distribution across the
three registers - first of the three semantic categories
of stance identified: epistemic (including epistemic,
evidential and domain marking), attitudinal (including
value judgment and counter-expectation marking) and style
(speech-act comment); second, of the three grammatical
realizations into which the authors categorize stance
adverbials: 'subordinate finite clause', 'prepositional
phrase' (PP) and 'single adverb'; third, of the clause
position of the stance adverbials.

A few clarifications regarding terminology and method
would have been helpful. For instance, it is not clear why
adverbial phrases, be they adverbs, clauses or PPs, are
described (p.58) as "grammaticalized expression of stance"
by contrast with "lexical expressions of stance"; nor
quite what is meant by the category of 'actuality'. It is
not clear to what extent adverbial modifiers of noun
phrases (NPs) and adjectival phrases (APs) were included
as tokens of stance markers nor whether syntactic position
was used as a criterion for the inclusion of an expression
in the set of adverbial stance markers. A list of the
stance adverbials included in the study would have been
most useful.

Interesting quantitative data is provided on the
frequencies with which adverbials marking the three
categories of stance occur in the three registers. The
authors highlight the following findings as surprising:
"academic prose writers use stance markers almost twice as
often as newspaper writers" (p. 64); "style adverbials are
.. moderately common in news reportage" (p. 67); "attitude
stance adverbials are moderately common in news reportage
and academic prose, but relatively rare in conversation"
(p. 68); "finite subordinate clauses are by far the most
common as stance adverbials in conversation" (p. 70);
"stance adverbials in final position are particularly
common in conversation" (p. 72). The authors' analysis of
the position of stance adverbials in the clause does not
distinguish among simple sentence adverbs, prepositional
phrases, subordinate clauses, parentheticals, transparent
predicates (e.g. 'I think') and NP-modifiers (e.g. 'about'
in 'about the fourth time'). It is therefore difficult to
assess the significance of the differences in position
that the authors discover across the three registers.

Corpus-based studies are essential to our understanding of
language use, language variation and language change.
Interested students can be referred to the excellent
studies of stance in English by Biber and Finegan (1988
and 1989).

In chapter 5, Hunston and Sinclair look at how local
grammars of evaluation might be built. They consider the
degree to which it might be possible to automate the
identification of evaluation. Focusing on adjectival
expression of evaluation, and using examples from the Bank
of English, they identify a number of grammatical patterns
which typically select an evaluative adjective. An example
is 'it + link verb + adjectival group + clause', as in 'It
was certain that he was much to blame'. The notion of
using these types of regular patterns to build 'local
grammars' with relatively fine-grained categories has many
attractions. There are obvious parallels here with recent
work in Construction Grammar. It would be interesting to
see some quantitative data on the corpus frequency and
distribution of these various evaluative adjectival

Cortazzi and Jin's chapter on 'Evaluating evaluation in
narrative' defines evaluation rather differently from the
other contributions to the volume. Whereas for Thompson
and Hunston evaluation seems to be defined in a way that
makes it identifiable in text, for Cortazzi and Jin
evaluation is interpretation and assessment. In addition
to the evaluative signals given by the narrator, Cortazzi
and Jin consider the evaluation OF narrative, and how
people and situations are evaluated THROUGH narrative.
This is the only paper that makes reference to a language
other than English. The authors provide a valuable
discussion of the role of contextual factors (in the
widest sense) in narrative interpretation, with examples
from professional settings and from Chinese and Jain

In chapter 7, Thompson and Zhou pick up the third of
Thompson and Hunston's functions of evaluation (that of
organizing the discourse) and relate it to the other two
(expressing opinion and maintaining interpersonal
relations). Their claim is that expressions of writer
opinion can have cohesive functions; that disjuncts can
signal coherence relations. Further, they claim that there
are relations that can ONLY be signalled by disjuncts.
They examine a number of disjuncts according to which
coherence relation they signal: 'admittedly', 'plainly',
etc. (to signal concession), 'surprisingly',
'unfortunately' etc. (to signal expectancy), 'ostensibly'
etc. (to signal hypothetical-real contradiction) and
'maybe .. maybe ..' etc. (to signal the alternative

The distinction drawn by Thompson and Zhou between
'propositional coherence' and 'evaluative coherence'
provides us with yet another pair of terms relating to the
propositional/non-propositional divide to add to the
'subject-matter' vs 'presentational' relations of
rhetorical structure theory, the 'informational' vs
'intentional' relations of Moore and Paris (1992), the
'semantic' vs 'pragmatic' sources of coherence of Sanders
et al (1993) and so on.

This work is based on data culled from a 20-million word
genre-diversified corpus. Yet very few quantitative
arguments are put forward. It would be interesting to see
more figures, and normalised figures, for the uses
discussed. The tack taken by Thompson and Zhou implies
that they are dealing with the pragmatic functions of
disjuncts. Quantitative data are therefore needed to
estimate the extent of these cohesive functions and their
defeasibility. The authors write "perhaps ... disjuncts
invoke negotiation and the reader's co-operation in
constructing discourse, whereas conjuncts reflect a more
dominant role for the writer" (p. 141). Here, Channell's
recommendation to consider the role of face might usefully
be taken up.

Thompson and Zhou are right to point out that the
traditional distinction between English conjuncts and
disjuncts, or indeed between conjunctions and sentence
adverbs, is unsatisfactory. Part of the problem is a
tendency to confuse types and tokens and to ignore the
polyfunctionality of many expressions described as
disjuncts (Kolar 1975). Corpus studies such as Thompson
and Zhou's are ideally suited to clarifying these issues.

Martin's chapter focuses on evaluative lexis. It describes
a proposed multidimensional model of the conceptual space
of appraisal (evaluation). Appraisal is characterized by
three 'evaluative resources': affect, judgment and
appreciation, together with systems for marking speaker
commitment and gradation of evaluation. This model is
applied to several English texts, to illustrate how the
lexical resources available for the expression of
appraisal can profitably be mapped out. For affect
(emotive response), a series of parameters is proposed for
a further classification, including that of realis vs
irrealis, i.e. response to real vs unrealized stimuli.
This promises to be an interesting line of enquiry in
future research.

The model of appraisal is suggestive, thoughtful and well
illustrated. It naturally leaves a number of questions
unanswered. One challenging issue is where to draw the
boundary of appraisal. With respect to affect, Martin
distinguishes 'inscribed' (explicit) affect from 'evoked'
(implicit) affect (p.154), a distinction that could be
seen in terms of the use of semanticized vs contextual
means of expressing speaker evaluation. Martin warns that
"we need to be cautious about reading position when
analysing ideational meaning as tokens of affect" (p.
154). This approach raises questions of intentionality
and of whether there is much in text at all that cannot be
construed as appraisal. Explicitness may be a matter of
degree rather than a binary parameter. A related issue is
that of the mapping between linguistic expressions and the
corresponding appraisal values: Martin's aim in coding
text for affect has been to "work with the smallest
domains that can be associated with a particular affect

Overall, Martin's model promises to provide a valuable,
flexible tool for both English language and cross-
linguistic research into the expression of attitude.

In the final paper in the collection, Hunston explores
complexity in evaluation and proposes a model of
evaluation in text. Following Sinclair, she posits two
'planes of discourse' - the 'autonomous plane' and the
'interactive plane' - to distinguish between two aspects
of meaning that correspond roughly to state-of-affairs, or
that which is interpreted in informational terms, and
textual argument, or that which is interpreted as
speaker/writer comment. (Readers will see parallels here
with other frameworks, such as Sweetser's (1990) content,
epistemic and speech-act domains, or even with Schiffrin's
(1987) five 'planes of discourse', though the latter are
rather different.) In Hunston's model, to oversimplify
rather, each statement has status, and optionally has
value. Status and value operate on both the interactive
and the autonomous planes. Status is epistemic/evidential
on the interactive plane, but can also be judgmental on
the autonomous plane. Modal adverbs such as 'undoubtedly',
'certainly' are treated as indicators of source (the
source being 'implied consensus'), as well as concessives.
Status "reifies statements, making each discourse-entity
into a thing that can be given a value" (p. 193). Value
attributed may be positive or negative, but is
interpretable only in the light of the status of what is
evaluated. Statements can thus be characterized by their
status and their value, on both the interactive and the
autonomous planes. Moreover, the model can be applied to
textual organization, in that the rhetorical relations
between two clauses can often be seen in terms of one
statement giving value to the other. For example, the
evidential relation in 'assessment + evidence', can be
seen in terms of the evidence giving value to the
assessment by supporting or undermining it.

To students unfamiliar with this type of terminology, the
chapter may seem a little daunting. But again, this is a
suggestive and interesting model, embracing both overtly
evaluative language and highly implicit evaluations.

This collection of studies of evaluation in text is very
welcome. The expression of speaker/writer attitude is an
area that has tended to be neglected by linguists, thanks
to what Lyons calls "the intellectualist - and objectivist
- prejudice that language is essentially an instrument for
the expression of propositional thought" (1995: 336). A
book on evaluation as it is defined by Thompson and
Hunston is, of course, an ambitious undertaking. The
problems of deciding where and how to draw a line between
representational/descriptive meaning and
attitudinal/evaluative meaning are well known. There is a
multitude of differing approaches and overlapping
categories, as the range of terms, categories and models
to be found in this volume testifies.

Thompson and Hunston present 'evaluation' as both the
expression of speaker/writer opinion and the opinion
itself. They define 'evaluation' in their introduction as
"the expression of the speaker's or writer's attitude or
stance towards, viewpoint on, or feelings about the
entities or propositions that he or she is talking about"
(p. 5). They note that "there are three functions that
evaluation is used to perform" (p. 6). All this suggests
evaluation as a linguistic category. Later, however, they
gloss evaluation as "what the writer thinks about what he
or she is writing" (p. 121) and write about exploring "the
different types of evaluation that are expressed" (p.
176). It might have been useful to distinguish these two
meanings of 'evaluation' more clearly.

Evaluation as attitude is heterogeneous and is expressed
through lexical, grammatical and prosodic choices. A book
on evaluation in text must therefore be extremely
selective in its coverage. Most of the papers in this
volume focus on the lexical expression of evaluation. The
relationship to evaluation of word order, prosody, mood,
or specialized grammatical constructions receives less
attention. Thompson and Hunston's volume is almost
exclusively concerned with the English language (this
might usefully have been signalled in the subtitle).
Evaluation has tended to be particularly neglected in
lingustic studies of English, perhaps because
subjectivity is comparatively little grammaticalized in
English (v. Palmer 1986). As Lyons points out, "it is
much easier to objectify and propositionalize the
inherently expressive and subjective, non-propositional,
components of the meaning of utterances in English than
it is in many languages" (Lyons 1995: 180). The focus on
English is therefore welcome.

Overall, EVAULATION IN TEXT is a timely reminder of the
extent to which expressions of evaluation pervade


Biber, D. and Finegan, E. (1988). Adverbial stance types
in English. Discourse Processes 11, 1-34.
Biber, D. and Finegan, E. (1989). Styles of stance in
English: lexical and grammatical marking of evidentiality
and affect. Text 9 (1), 93-124.
Brown, P. and Levinson, S.C. (1978). Politeness. Some
universals in language usage. Cambridge: Cambridge
University Press.
Kolar, S. (1975). Aspects of polyfunctionality in English
sentence adverbs. Philologica Pragensia 18 (4), 222-227.
Lyons, John (1995). Linguistic semantics. Cambridge:
Cambridge University Press.
Moore, J.D. and Paris, C.L. (1992). Planning text for
advisory dialogues: capturing intentional and rhetorical
information. Marina del Rey, CA: Information Sciences
Institute Technical Report no. 92-22. Also published in
Computational Linguistics 19(4), 1993, pp 651-695.
Palmer, F.R. (1986). Mood and modality. Cambridge:
Cambridge University Press.
Sanders, T.J.M., Spooren, W.P.M. and Noordman, L.G.M.
(1993). Coherence relations in a cognitive theory of
discourse representation. Cognitive Linguistics 4, 93-133.
Schiffrin, Deborah (1987). Discourse markers. Cambridge:
Cambridge University Press.
Sweetser, E.E. (1990). From etymology to pragmatics.
Metaphorical and cultural aspects of semantic structure.
Cambridge: Cambridge University Press.


Diana Lewis has research interests in lexical semantics
and pragmatics, language change and variation, corpus
linguistics and contrastive linguistics.