Halliday FunctionalDiversityLanguage

Functional Diversity in Language as Seen from a Consideration of Modality and Mood in
English
Author(s): M. A. K. Halliday
Source: Foundations of Language , Aug., 1970, Vol. 6, No. 3 (Aug., 1970), pp. 322-361
Published by: Springer
Stable URL: https://www.jstor.org/stable/25000463
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/25000463?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Foundations of
Language
This content downloaded from

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
All use subject to https://about.jstor.org/terms
M. A. K. HALLIDAY
FUNCTIONAL DIVERSITY IN LANGUAGE

AS SEEN FROM A CONSIDERATION OF
MODALITY AND MOOD IN ENGLISH1
Recent work in a number of different areas within linguistics has tended to

bring about a renewed interest in functional approaches to language. It is
enough to mention here Hymes' work in sociolinguistic theory, Labov's
studies of urban dialectology and of social and situational variation, and
Bernstein's research into language and educational failure.2 Also in the
field of language acquisition it has become clear that it is necessary to operate
with basic concepts of a functional kind; what the child learns he learns in
situations of use, and the structures he builds up reflect the functions which
he himself internalizes.3 We are able to understand the structures underlying
the utterances of the child to the extent that we understand the purposes he
is using language for. The traditional interest of linguists, particularly those
of or influenced by the Prague school, in a functional analysis of language is
very directly relevant to current problems.4
The link between function and structure is seen most clearly in the language
of the child. One of the earliest uses of language that the child explores is an
instrumental one: language used for the satisfaction of his material and emo
tional needs. When the small child who has dropped his teddy-bear over
the side of his cot says teddy!, perhaps accompanied by some appropriate
paralinguistic signal such as a sob, we can recognise this utterance as the
exponent of some structure, perhaps a structure consisting of one element
which we might label 'Object of desire'; and this structural function is
directly derivable from the linguistic function just mentioned.5 In this respect,
1 Paper read to the Yale Linguistics Club, 13 January 1970. My thanks are due to those
who took part in the ensuing discussion for valuable comments and criticisms.
2 Cf. the discussion in Dell H. Hymes, 'Linguistic Theory and the Functions of Speech'
(paper prepared for the conference 'International Days of Sociolinguistics', Rome, Septem
ber 1969, and to be published in its proceedings).
3 Cf. Daniel I. Slobin, 'Universals of Grammatical Development in Young Children', in
Proceedings of the Conference on Psycholinguistics (ed. by G. Flores d'Arcais and W.
Levelt), North-Holland, Amsterdam (forthcoming). Compare also Supplement 1 (by
Bronislaw Malinowski) to C. K. Ogden and I. A. Richards, The Meaning ofMeaning, Kegan
Paul, London (International Library of Psychology, Philosophy and Scientific Method),
1923; e.g. p. 321 'In all the child's experience, words mean, in so far as they act ...'.
4 Roman Jakobson, 'Efforts Towards a Means-Ends Model of Language in Interwar
Continental Linguistics', in Trends in Modern Linguistics (ed. by Christine Mohrmann
et al.), Spectrum, Utrecht, 1963. See also reference in n. 10 below.
5 The unified concept of 'function' underlying these two apparently somewhat different
senses of the term - structural function, e.g. 'agent', 'process', 'subject', 'theme', and
Foundations of Language 6 (1970) 322-361. All rights reserved.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
FUNCTIONAL DIVERSITY IN LANGUAGE 323
however, it is no different in kind from the elements of structure of the adult

language, such as Agent, or Process, or Goal. The adult 'translation' of this
particular utterance, say give me my teddy!, embodying some such structural
functions as Goal and Beneficiary, reveals very clearly the systematic
relation between the structure of the child language at that early stage and
the structure of the adult model. The latter is also related, although less
directly, to the functions that language is made to serve.
The example from language acquisition provides a useful point of entry
to a consideration of the functions of language. Just as the child builds up
his linguistic structures in a way which reflects his acquisition of the uses
of language, so the structure of language as a whole has been built up in
such a way that it reflects the demands that are made on language and the
functions it is required to serve. The phenomena of linguistic ontogeny are
therefore functionally complex: the child's developing language system
represents a progressive adjustment between structures derived from his
own uses of language, that is, from those functions which he has already
acquired, and the structures of the model to which he has access - structures
which are themselves functional in origin. Thus the adult model provides
him with instances which are from his point of view not nearly as fragmentary
and disordered as we are sometimes invited to think. Most speech is function
ally highly coherent, in the sense that its structure reflects what it is being
made to do; and this is what (given just the assumption that speech is
purposeful activity) makes it accessible to the learner.
(It could perhaps be added that it is also coherent structurally. The view
that the speech to which a young child is exposed is a rather impoverished
array of formless semi-sentences is very far from the truth, and appears to
rest on a number of doubtful assumptions. One is that a set of very disparate
phenomena (the list often runs something like 'hesitations, repetitions, false
starts, slips of the tongue, coughs and sneezes') can be lumped together as
instances of formlessness, whereas some of these are in fact structured and
are made to look formless either by the way they are expressed ('false starts'
instead of 'restructurings') or by being collocated with the others. Another
is that the only structure in language is 'cognitive' (i.e. ideational) structure,
whereas in fact the pragmatic and expressive utterances which make up much
of the linguistic experience of the young child are also fully structured, and
are less liable to continuous replanning than the very intellectuated discourse
that is usually used to exemplify structural disorder. The young child is,
linguistic function (i.e. the functions of language), e.g. Biihler's 'representational', 'cona
tive', 'expressive' (and cf. n. 37 below) - is embodied in Firth's notion of 'meaning as
function in context'. See J. R. Firth, 'The Technique of Semantics', TPhS 1935, reprinted
in Papers in Linguistics 1934-1951, Oxford University Press, London, 1957, pp. 7-33.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
324 M. A. K. HALLIDAY
typically, exposed to a large quantity of highly structu

types where the functions of the speech act tend to be
the structure directly relatable to them. Thus his exper
'means' is not restricted to its ideational meaning or 'co
broader in this respect than the view of meaning th
of the poverty of his linguistic environment.)6
The present paper suggests a functional treatmen
English syntax. The orientation is 'functional' in the
tulating that in order to explain the structure of la
sider its use: language is as it is because of the fun
serve. The classical functional theories such as thos
Biihler are extrinsic in origin: there the functions
investigated from an ethnographic or a psychologica
categories that are set up are validated on other than li
will vary according to the particular psychological
Here we shall be concerned with questions that are in
language. How is the functional diversity of language r
system? what idea of linguistic function do we arriv
consideration of the nature of language itself? and i
work which will satisfy both requirements, that of t
matical structure and that of providing an insight into
It seems possible to recognize a simple but very ba
tion in language, one which is inherent in the gramma
mines the form taken by grammatical structure. This n
linguistic theory derives mainly from the work of
has been being developed by Danes, Firbas and others w
to the structure of the sentence and the clause;7 but it
the grammar as a whole. We are concerned, theref
theory which is both general and linguistic. The p
however, an exposition of a functional theory. Such
to become largely terminological; and while this would
since it may be necessary to consider the functions of
standpoint, involving new categories and new inter
would be vague and unconvincing. A very brief sket
functional framework, in order to provide the necessar
6 Cf. also the discussion in John Morton, 'What Could Possibly

to Second International Congress of Applied Linguistics, C
7 Frantisek Danes, 'A Three-Level Approach to Syntax', TLP
'The Third Syntactic Plan', and other papers in the same vo
Thematic Subjects in Contemporary English', TLP 2, 1968; Al
of Communicative Units and Fields as Illustrated by English
Brno Studies in English 7, 1968, and numerous references therein

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
however, the discussion will take the form of illustration, centring around two
areas in the grammar of English which are crucial to the point at issue
because each is bound up with two separate functions. The two are the
system of modality (probable, possible &c.), and the system of mood (de
clarative, interrogative &c.). A study of either of these systems raises certain
questions, of a different kind in each case, which can be answered most
effectively when it is shown that the system in question involves the in
terrelation of different components in the grammar, and that these com
ponents are functional in origin.8
The functions of language, as they are to be understood here, are not
different 'levels', in the traditional linguistic sense of the term; nor have they
to do with a distinction such as that between a formal system, its inter
pretation and its use. Rather, the functions are to be understood as gen
eralized uses of language, which, since they seem to determine the nature
of the language system, require to be incorporated into our account of that
system. Here is an illustration in microcosm. Consider the sentence Smith
died. It expresses a content, relatable to the speaker's experience of the real
world; this is one aspect of the use of language - language in its experiential
or ideational function. We could change the content in a systematic way if
we replaced it by Jones killed Smith: there are now two participants in the
process instead of one.
Secondly, Smith died also expresses a role relationship between speaker
and hearer. In making a statement, the speaker is taking upon himself a
particular communication role, that of (let us say) 'declarer', and is inviting
the hearer to take on the complementary role.9 Here the contrast is with
did Smith die?, where the speaker is casting both himself and the hearer in a
different role relationship. This is an illustration of language in its interper
sonal function. The expression of speech roles in a communication situation
is one instance of the wider function whereby the speaker enters into the
communication process in its social and personal aspects.
Finally, Smith died also expresses texture. It takes on a particular form,
as a message, that is operational in the given context. If instead we had
8 For a fuller discussion see M. A. K. Halliday, 'The Functions of Language' and 'The
Components of a Grammar', 1969 (mimeographed).
9 The system of mood is the expression of the speaker's choice of role in the communicative
situation. A role such as that of 'declarer' is merely a general label; the importance lies in
the more specific categories such as those of giving information and expressing opinion
(cf. the various types of 'performative' role). In taking on one such role the speaker also
defines the range of options for the hearer: assent to or contradiction of the opinion;
acknowledgment, dismissal, or claiming prior knowledge of the information; and so on.
Such responses are also positive role selections, and have their own linguistic expressions
such as I see, oh, I know, yes (note that yes is considerably more frequent in conversation
as response to statements than as response to questions).

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
the one who died was Smith this would be a differen

different presuppositions. This is an illustration of l
function. The textual function is internal to language an
the other two; it is only because we can select the d
message that we can also use language effectively
experience and to interact with those around us.
Each of these three pairs of sentences contrasts in
function, the 'ideational' or content function, the 'in
role function and the 'textual' or discourse function.
sentence Smith died thus includes components derive
tions. All these components of meaning are in fact n
adult utterance. An utterance usually embodies an elemen
what I have to say'; an element of speaker's involvem
come in'; and a third element, 'this is the kind of m
gives the utterance the status of a text. In a typical
simple sentence as the one we have been discussing, th
all these three elements simultaneously.10
Naturally this example does no more than hint at t
cannot display the full generality of the functions it is be
What needs to be made clear, therefore, is that it is b
systematic contrasts in the grammar derive from o
functions. A suggested categorization of the grammatica
according to function is set out in Table I. In the table, e
as having just one function as its origin. In fact, howeve
one way or other exemplify the intersection of a pair
shed further light on the nature of the functional c
instances of this kind that we shall be concerned in wha
The first topic that we shall examine from a functi

that of modality. Modality in English poses one very
may be stated by means of an example: what is the rel
meanings of must in (1.1) and (1.2)?
(1.1) you must be very careful

(1.2) you must be very careless
The first means 'you are required to be', the second 'i
10 Cf. the formulation in Pavel Novak and Petr Sgall 'On the Pra
TLP 3, 1968, p. 294 '... the "functions" are recognised to be
language structure (system)'. I would prefer to say that the funct
the language system (and would question the opposition impli
remark (ibid. p. 292) that 'the traditional term "language func
since the functions concerned should be ascribed not to the lang

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
TABLE I
function IDEATIONAL INTERPERSONA
rank I Experiential Logical
TRANSITIVITY MOOD T
types of process condition types of speech functio
participants & circumstances addition modality (
CLAUSE (identity clauses) report (the WH-funct
(things, facts & reports) pr
g POLARITY sub
.o
___ _ ..__
Verbal GROUP TENSE catenation PERSO
(verb classes) secondary tense ('marked' op
o
MODIFICATION classification
epithet function . sub-modification attitudinal
enumeration intensifiers 'phor
Nominal GROUP (noun classes
(adjective classes) (definit
Adverbial 'MINOR PROCESSES' narrowing COMMENT

(incl. prepositional) prepositional relations . sub-modification (classes o
GROUP (classes of circum- u cour
stantial adjunct) __ ____
WORD LEXICAL 'CONTENT' compounding LEXICAL 'REG
(incl. lexical item) (taxonomic organization F derivation 5 (expres
~~~of vocabulary) (stylistic organization o
IJS^~X~ ? ~ voca
INFORMATION TONE INFOR
UNIT intonation systems distribu

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
are'; an example such as you must be very sympathet

preted in either sense and this brings out the ambiguity
ambiguity is it? Is it the same as that which is the so
Mary can't think that! Mary can't think (period!)? An
have done that yesterday ambiguous whereas you can't h
day is not? (The first may mean either (i) 'you were n
'it is impossible that you did it'; the second has only
put the last question in another form, why can we sa
Mexico but not, or only very restrictedly, I can have bee
As a startingpoint let us consider a small set of rel
possibility. In Greenwich, in southeast London, there
which is attributed to Sir Christopher Wren, architect o
It is a rather indistinguished structure which, if it was
been a task set for homework while he was at school. At
(2.1) possibly this gazebo was built by Sir Christop
We can express more or less the same content by saying
(2.2) this gazebo may have been built by Sir Chris
with may have been instead of possibly ... was; and

the two:
(2.3) possibly this gazebo may have been built by Sir Christopher Wren
To show that this is a general pattern, here is a similar triad; note that the
underlined elements are tonic (carry the intonation pattern associated with
primary stress) and have the same tone in each case:
(3.1) surely he'll stop talking soon

(3.2) he must be going to stop talking soon
(3.3) surely he must be going to stop talking soon
And there are a number of others. The principal system, as it is for a wide
spread variety of British English, is set out in Table II.
These forms represent the speaker's assessment of the probability of what
he is saying, or the extent to which he regards it as self-evident. They are thus
restricted to finite, declarative, independent clauses, and finite dependent
clauses such as conditionals; there is also a minor system in interrogative,
whereby the speaker invites the hearer to express HIS assessment, as in
(4) could they perhaps have left a note somewhere?
i.e. 'do you consider that there is a chance that ...?'

These meanings are what we understand by 'modalities'. A modality is
expressed by either or both of two elements, one verbal and the other non

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
TABLE II
MODALITY (interpersonal function)
POSS
PROBABLE
POSSIBLE VIRT
I_ I l _
probably possib
NEU|lTRAL
NEUTRAL POS will may, can* (could
NEG (i) won't may not ca
NEG (ii) won't [can't (could
UNDERTONE: presumably perhaps assured

tentative, POS would (will) might, could should,
deduced NEG (i) t ( might not shouldn't, oughtn't
NEG (ii) [ wouldn't (won't) [couldn't (can't)]
OVERTONE: predictably (tone 1) conceivably (tone 4) su

assertive, with POS would may, might, could should,
reservation (tonic) NEG (i) might not shouldn't, o
NEG (ii) wouldn't [can't, couldn't] [might n
* normally may in declarative, can in interrogative

( ) alternative forms
[ ] 'modality negative' forms

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
verbal (where verbal means 'functioning syntactically as

forms are the modal auxiliaries will would can could may m
ought to and need. Their properties are well known, but it
list them here:
(1) they have only finite forms: there is no to can

such as to be able are not modalities; see below)
(2) they form negative and interrogative withou
negative also being reducible: he cannot (can't),
doesn't can, does he?
(3) they are used as 'code verbs' in ellipsis: 'can you s
so can I
(4) they have three prosodic values, remiss (unstressed), ictus
(secondary stress) and tonic (primary stress): // A he can / go //,
// A he / can / go // , // A he / can / go // , the unstressed form
being normally reduced; these are systematic variants, differing
in meaning in a regular way.ll
They are further distinguished from the other verbal auxiliaries be get have
and do by the fact that
(5) they do not take -s on the third person singular: he can not he cans
Finally
(6) they do not occur in imperative
(7) they do not combine with each other12
The non-verbal forms are lexical items such as possible, certain, which may
occur in any of five different patterns, although most of them do not occur
in all five, e.g. certainly, it is certain that, I am certain that, he is certain to,
11 The absence of -s relates to the fact that these forms are not present but tenseless, like
the corresponding forms of the non-finite verb: can go is like to go, going (and not like
goes) in this respect. This is not true of the 'quasi-modals' (see below); can go in sense 'is
able to go' is present tense, so that the absence of -s is here unmotivated. (It is noticeable
that is to, which occurs as quasi-modal but not as 'true' modal, does have the form of a
present tense verb.) The absence of -s is one of the signs that modality rather than modul
ation is the underlying meaning of the modal auxiliaries.
12 Where there is double modality one must be expressed non-verbally, e.g. certainly ...
might (see below). Sometimes a speaker does produce a combination of verbal modals,
such as in he might ought to be here meaning 'perhaps he ought to be here'; but in such
cases the second of the pair appears always to be 'quasi-modal'. Professor William Labov
has drawn my attention to the occurrence of 'double modals' in certain Confederate
dialects, and I believe the same restriction applies there - at least the second member of
the pair must be a quasi-modal. Note also that verbal modality cannot be incorporated
into a non-finite clause; we have the proof that he can't have been here is ... but not his not
canning have been here is proved by ... (the form his not having been able to be here is of
course not equivalent, being again quasi-modal not modal).

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
there is a certainty that. Specifically:
(1) as adverb ('modal adjunct'): perhaps, possibly, presumably,

obviously &c.
(2) as adjective (predicative in impersonal matrix clause it is ...
that ...): possible, likely, obvious, conceivable &c. (The form
it ispossible (for ...) to ... is not a modality; see below.)
(3) as adjective (predicative in interpersonal, speaker-hearer matrix
clause lam ... that ..., are you ... that ...?): sure, certain &c.; also
doubtful (whether ...). (Compare I think that ..., &c.)
(4) as adjective (predicative in clauses such as he is sure to have known):
sure, certain, likely.
(5) as noun (complement in impersonal matrix clause there is a...
that...): possibility, chance, likelihood, presumption &c.
The verbal and the non-verbal forms do not correspond one to one. In
my own speech possible tends to go with may and perhaps with might, but
the two are interchangeable and other speakers probably have different
patterns. Nor do the different non-verbal forms of the same lexical item
necessarily correspond with each other: obviously is not the same as it is
obvious that, surely as I am sure that. But there are discernible groupings,
and a clear distinction can be drawn between pairs which are felt to be
equivalent, and thus reinforce each other (as 'concord') when both are
present, as in perhaps he might have built it, and those which are not equiva
lent and are thus cumulative in meaning, as in certainly he might have built it
('I insist that it is possible' or 'I grant that it is possible').
There is thus no one single place in the clause where modality is located.
It is a strand running prosodically through the clause;13 and this effect is
further enhanced by the fact that in addition to the forms above it may be
realized also by the intonation contour, or tone. For example the meaning
'possibility, but with an overtone of reservation' (i.e. 'maybe, BUT...') is
typically realized by tone 4 (fall-rise contour with rising onset and main
intensity on fall) in association with other modal forms, the modal elements
themselves carrying the tonic:
(5) //4 A he / might have / built it //4 A con/ceivably //
from which we know that the speaker is thoroughly sceptical about the
possibility which he is in process of conceding. This clause AS A WHOLE has
the features
(modality: ((committed: possibility) / (overtone of reservation)))
13 Cf. T. F. Mitchell, 'Syntagmatic Relations in Linguistic Analysis', TPhS 1958, 101-118.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
and these are realized, AS A WHOLE, by the tone, the m

verb, and the modal adjunct.
The system of modality may be set out in the form of a
r probable possible
possible /certain - virtually certain
- certain
> -neutral
r< < ---- undertone (tentative & c.)

-overtone (assertive &c.)
[positive
-negative
The basic distinction is that between 'probable' and the rest. This is a dis
tinction between the intermediate value in the speaker's assessment of prob
ability and the outer, or polar, values which are 'possible' and 'certain'. That
this is the basic opposition in the system can be seen from the negative. In
modal clauses, there are two possible domains for the negative: either the
modality may be negative (it is not possible that ...), or the thesis may be (it
is possible that ... not ...). In the case of the intermediate value 'probable'
these two are not in contrast: for the negative of
(6.1) this gazebo will probably have been built by Wren
it makes no difference whether we associate the negative with the thesis, as in
(6.2) it is probable that this gazebo was not built by Wren
or with the modality, as in
(6.3) it is not probable that this gazebo was built by Wren
- either of these can be expressed as
(6.4) this gazebo won't have been built by Wren

But this is not true of the outer values; there is a difference between
(7.1) it is possible that this gazebo was not built by Wren

and
(7.2) it is not possible that this gazebo was built by Wren

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
such that if the two are expressed verbally a different auxiliary is required:
(7.3) this gazebo may not have been built by Wren
(7.4) this gazebo can't have been built by Wren
At first sight, it would seem that we should express this by adding to the
intersection of 'possible/certain' and 'negative' a further system distinguishing
'modality negative' from 'thesis negative':
possible certain {
rnegative-thesis negative
negative ---- }
negat ive3modality negative
But this is surely wrong, since the total number of ne

greater than the total number of positive modalities.
with the complementarity of 'possible' and 'certain': 'it is
equals 'it is not certain that' and 'it is certain that ..
possible that'. Hence for each pair there is only on
(7.3) above can be analysed either (i) as 'possible' p
it is possible that this gazebo was not built by Wren
'certain' plus 'modality negative', it is not certain that
by Wren. Of the two, the former analysis is the pref
realization of modality is then the same whether ne
possible and/or may in each case.14
There is no such thing, therefore, as a negative mod
are positive. This is natural, since a modality is an assessm
and there is no such thing as a negative probability. A m
of course, with a thesis which is negative; but the modalit
to negation- it does not enter the system of polarity, wh
in the modality network. Negative polarity in the c
through the structural association of negative and m
don't think he built it), but systemically they are unrelate
14 This is always true for the non-verbal forms. For the verbal form
of must and can't: certainly this gazebo must/can't have been bu
'smeared' form this gazebo can't possibly have been built by Wre
certainly ... can't; cf. couldn't conceivably. There are also marginal
must in 'modality negative' form, e.g. John must be very worried
he may be; these are text negatives, or verbal crossings out, an
form of the negative shows that it is the modality that is being
might be pleased. - No he mightn't where the negation of 'unlike
interpreted as 'not ((merely) unlikely (but possible) but not (ev
can is occasionally used in positive declarative by syntactic 'bac
rogative: Can John be busy, I wonder? - Possibly he can be.
15 That is, if the clause is (independently) both negative and

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
Thus the primary option in the system is that betw

'possible/certain', which are differentiated by their r
other in the environment of a negative clause. Within t
contrasts with 'certain' as the lower versus the higher
within 'certain' there is a distinction between relative and
between what we have called 'virtually certain' and 'ce
Simultaneous with this variable is another one, whereb
be as it were either straight or modified, by undertone or
down', it has the sense of tentative, or else presumed by d
up', it takes the form of an emphatic assertion, or an a
with some kind of reservation; in the case of would the
what I expected', with an implication of perversity as in
(8) they would telephone just as I was going to sle
Note that this form with would, often treated as aberrant,

regular in the system.
The following paradigm illustrates the range of differ
probability that typically make up an English speaker's
Each one corresponds to one main cell in Table II,
columns from left to right. The negative form (i.e. the
modality when the thesis is negative) is shown on the righ
(i)
(9.1) probably this gazebo will be by Wren won't
(9.2) presumably this gazebo would be by Wren wouldn't
(9.3) predictably this gazebo would be by Wren (tone 1) wouldn't
(9.4) possibly this gazebo may be by Wren may not
(9.5) perhaps this gazebo might be by Wren might not
(9.6) conceivably this gazebo might be by Wren (tone 4) might not
(9.7) assuredly this gazebo should / ought to be by Wren shouldn't
(9.8) surely this gazebo should / ought to be by Wren
(tone 4) shouldn't
(9.9) certainly this gazebo must be by Wren can't
(9.10) obviously this gazebo must be by Wren couldn't
(9.11) surely this gazebo must be by Wren (tone 1) can't
One or two of these would be more natural in another form, e.g. (9.7)
where the likely form of expression would be I'm sure this gazebo ... rather
syntactically more delicate options arise, illustrated by forms such as he probably didn't
build it, I think he didn't build it, Idon't think he built it, I'm not sure he built it, &c.
16 With alternative non-reduced negatives should not be / ought not to be in (9.7) and (9.8),
and must not be in (9.9)-(9.11).

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
than assuredly ...; the aim has been to bring out the full regularity of the
paradigm. Some pairs are quite similar in meaning, but in no case, I think,
identical; for example, (9.8) and (9.11) both have surely in an intensified
form, but they differ in tone and in the modality in the verb. The former,
which has the 'virtually certain' form should together with the 'yes but'
falling-rising intonation, more readily admits the possibility of being wrong.
In fact, there is also a difference in meaning between what are treated here
as alternative (verbal and non-verbal) realizations of the same features, such
as possibly and may, but it is a difference in thematic structure and not in
modality: possibly it was Wren differs from it may have been Wren in that in
the former the speaker has selected the modality as his theme (see the final
section of this paper). In possibly it was Wren the meaning is 'as far as the
probabilities of the situation are concerned' or 'if you want to know what
I think,...'.
Modality is a form of participation by the speaker in the speech event.
Through modality, the speaker associates with the thesis an indication of its
status and validity in his own judgment; he intrudes, and takes up a position.
Modality thus derives from what we called above the 'interpersonal' function
of language, language as expression of role. There are many other ways
in which the speaker may take up a position, and modality is related to the
general category that is often known as 'speaker's comment', within which
a number of other types have been syntactically distinguished; like modality,
these are typically, though not uniquely, expressed by adverbs of different
classes, e.g. those represented by (taking just one example from each class)
frankly; generally; wisely; fortunately; officially; reasonably; personally;
incidentally; doubtfully.l7 Speaker's comment is then, in turn, one among the
syntactic complexes which together make up the interpersonal or 'social
role' component in language.
This, we are suggesting, is not a minor or marginal elementin language, but
one of its three primary functions, that concerned with the establishment of
social relations and with the participation of the individual in all kinds of
personal interaction. Language, in this function, mediates in all the various
role relationships contracted by the individual, and thus plays an essential
part in the development of his personality. To return for a moment to the
child, there is good evidence to suggest that control of language in its inter
personal function is as crucial to educational success as is control over the
expression of content, for it is through this function that the child learns to
participate, as an individual, and to express and develop his own personality
and his own uniqueness.18 Modality represents a very small but important
17 See Eirian Davies, 'Some Notes on English Clause Types', TPhS 1967,1-31.
18 See G. J. Turner and R. E. Pickvance, 'Social Class Differences in the Expression of

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
part of these resources - of the semantics of personal p

means whereby we express modalities are strung thro
woven into a structure, with other elements expressing di
This is why there is grammar in language; grammar is the
of strands from the various components of meaning into
we call linguistic structure. Into this go elements of th
experience, and of the expression of role and persona
those of discourse to provide the texture. If we take ap
order to see how it works, in this way we reveal its origin
functions.
We should not have advanced very far in the understanding of modality
if we had assumed that the modal auxiliaries in the verb were to be treated
on their own, and that the list of these could be taken as representing the
system. We have had to identify the CATEGORIES of modality and show how
these are realized; and the realization involves some fairly complex syntag
matic and paradigmatic patterns of relationship. Some general meanings have
been able to be recognized; it is clear that may has something to do with
possibility, for example. But we have not up to now accounted for all the
uses of may and the other auxiliaries. We have not, for example, accounted
for any of the following instances:
(10.1) You must build a gazebo.

(10.2) I can't build gazebos. If I could I would.
(10.3) Well you ought to be able to.
These have nothing to do with the speaker's assessment of probabilities. In

these examples the auxiliaries must, can &c. express various types of modula
tion of the process expressed in the clause; modulation in terms of permission,
obligation and the like. They are part of the thesis - part of the ideational
meaning of the clause.
Although by and large the same verbal auxiliaries are used for 'modula
tion' as for modality, there are some rather fundamental differences between
the two. One of these which appears most clearly relates to tense. The
modalities, being outside the ideational meaning of the clause, are also outside
the domain of tense; like other forms of speaker's comment, they relate only
to speaker-now. It is true that there exist different tense forms of the non
verbal expressions of modality, such as
(11.1) it was certain that this gazebo had been built by Wren until the
discovery of the title-deeds
Uncertainty in Five-Year-Old Children', Sociological Research Unit, University of

London Institute of Education, 1969 (mimeographed).

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
But this has become objectified, and is thus removed from the realm of
modality. Hence there are no corresponding verbal forms; we cannot say
(11.2) this gazebo must have been built by Wren until the discovery of
the title-deeds
Nor can it was certain that be replaced by certainly, which would give
(11.3) certainly this gazebo had been built by Wren until the discovery
of the title-deeds
At the same time, although modality itself is not subject to variation in tense,
it COMBINES freely with any tense. If it is expressed non-verbally, by surely &c.,
then all tenses of the non-finite tense system can occur: surely he built / builds
/ will build / had built / has built / will have built / was building / is building /
will be going to build / was going to have built / ....
If it is expressed verbally, by must &c., then the modality replaces the
primary tense and the tense system with which it combines is the non
finite one, in which the absence of primary tense leads to some neutralization.
One non-finite tense corresponds to up to three finite ones, e.g. must have built
corresponds to (i) surely ... built, (ii) surely ... has built, and (iii) surely ...
had built, as the following makes clear:
(12.1) surely he left yesterday he must have left yesterday

(12.2) surely he has left already he must have left already
(12.3) surely he had left before he must have left before you came
you came
Compare the non-finite form having left which is also the equivalent of all
three:
(13.1) having left yesterday, he ...

(13.2) having already left, he ...
(13.3) having left before you came, he ...
It may be helpful to give some further examples:
(14.1) Smith can't be so busy (surely ... isn't)

(14.2) Smith can't have been so busy (surely ... wasn't / hasn't been /
hadn't been)
(14.3) Smith can't be working at this hour (surely ... isn't working)
(14.4) Smith can't be going to work all night (surely ... won't work /
isn't going to work)
(14.5) Smith can't be going to have finished all that work by tomorrow
(surely ... won't have finished / isn't going to have finished)

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
(14.6) Smith can't have been working all night (surely

/ hasn't been working / hadn't been working)
(14.7) Smith can't be going to be working at it all w
won't be working / isn't going to be working)
(14.8) Smith can't have been going to finish all that w
wasn't going to finish / hasn't been going to finish
going to finish)
and so on.19
Thus modality itself has no tense; but it may combine wi
of the verb. The categories of the second type (10.1-3),
which are not true modalities - they are a kind of quasi
to above as 'modulation' - have their own complete set of t
are not speaker's comments, but form part of the con
expressing conditions on the process referred to; they
modification by temporal categories. However, all their
present (and, very restrictedly, simple past) are realiz
auxiliaries, which have no tense variants, but by periph
be able to and be allowed to. Thus:
(15.1) Jones can / is allowed to go out today

(15.2) Jones (could/) was allowed to go out yesterday
(15.3) Jones will be allowed to go out tomorrow
(15.4) Jones had been allowed to go out before the rel
(15.5) Jones has been allowed to go out since this mor
(15.6) Jones will have been allowed to go out by the
(15.7) Jones was being allowed to go out at that time
(15.8) Jones is being allowed to go out just at the mom
(15.9) Jones will be being allowed to go out soon at th
(15.10) Jones was going to be allowed to go out until th
and so on. These 'modulations' catenate with a non-finite
which for its part is restricted to the unmarked 'tensele
infinitive. (This is a slight oversimplification as it ignores
terpreting forms like you should be going to finish it soon
finished it by tomorrow, as modulations. On the other
seen in such forms as you should havefinished it earlier is
these types are briefly discussed later.) So whereas in moda
19 And four more, to be exact (twelve non-finite tenses as against
M.A.K. Halliday, 'The English Verbal Group', 1965 (mimeograp
tense is based on a three-term system (past, present, future) but w
cursive. Note that going to and about to are regarded as (alter
secondary future. There may be a slight preference for going to in
non-finite forms.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
may have any tense but the MODALITY is outside it, in these 'quasi-modalities'
the MODULATION is subject to the full tense system, but the PROCESS that is
modulated is tenseless.
Modulation, when it is not expressed through the modal auxiliaries, is
realized not by non-verbal forms like possible, possibly, possibility &c. but
by verbal structures consisting of be + adjective + to, e.g. be able to, or be +
passive participle + to; e.g. be obliged to. These realizations are alternatives
and cannot be combined: 'ability' is expressed either as can or as be able to
but not both - there is no possibility of the two occurring in concord, as do
the two forms of modality. A combination of modulations is always cumu
lative, e.g. Jones must be allowed to go out, even where the same one appears
twice, e.g. Jones may be allowed to go out now, nurse 'you are allowed to
allow him to go out'. There would be no reason for modulation to extend
prosodically through the clause, since it is a part of the thesis rather than a
commentary on it.
The principal categories of modulation are given in Table III. There is
a fairly clearcut distinction between an 'active' and a 'passive' type, corre
sponding to the distinction in realization between adjective and passive verb
referred to above. In the active modulations, those of ability and inclination,
the modulation relates to and is intrinsic to the actor: Jones will / is willing to
drive, Jones can / is able to drive. Where the clause is active, the subject is
actor with respect to the modality as well as with respect to the process.20
In the passive modulations, on the other hand, the modulation likewise
relates to the actor, but is extrinsic: Jones may / is allowed togo out. Where the
clause is active, therefore, the subject is actor with respect to the process but
goal with respect to the modality. Here is a paradigm giving examples of the
principal types:
Active type: (i) (ii)

(16.1) Jones will / is w
(16.2) Jones can / is
Passive type:
(16.3) you may / can / are allowed to needn't can't
tell
20 If the clause is passive, however, there is a distinction between the 'able' and the
'willing': they were able to be helped means 'someone had the ability to help them' (cf. this
problem has been able to be solved); but they were willing to be helped does not mean
'someone had the willingness to help them' (hence we cannot say this problem has been
willing to be solved). The 'able' term is the odd one out in the modulation system as a
whole. Cf. Rodney Huddleston, review of 'Madeline Ehrman: The Meanings of the
Modals in Present-day American English', Lingua 23 (1969), 165-176.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
TAB
MODUL
ACTIVE PASSI
PERMISSION N
INCLINATION AND ABILITY OBLIGA
willing; insistent* allowed obliged, su
POS will can, may

NEUTRAL NEG (i) needn't can'
NEG (ii) won't mustn't,
OBLIQUE: POS would could, might should, ough

hypothetical, NEG (i) needn't shou
tentative NEG (ii) wouldn't mustn't
r I = __ ._--L^_--^ - _ .__-_----
able entitled desired, expected* d
POS can can shal

NEUTRAL NEG (i) needn't shan't, may
NEG (ii) can't mustn't, can't
OBLIQUE: POS could could shou

hypothetical, NEG (i) needn't shouldn't, migh
tentative NEG (ii) couldn't mustn't, could
* verbal forms phonologically weak in sense 'willing', salient in sense 'ins

** normally in first person interrogative only, e.g. shall we come with you

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
(16.4) you can / are entitled to tell needn't can't

(16.5) you should / ought to / are shouldn't, oughtn't to needn't
supposed to tell
(16.6) shall I / am I expected to tell mayn't, shan't needn't
(16.7) you must / are required to tell mustn't needn't
(16.8) you are to / are intended to tell aren't to needn't
In modulation too there is complementarityin the negative forms, although
with a difference. Corresponding to you can smoke are
(17.1) (process negative) you needn't smoke ('are allowed not to')
(17.2) (modulation negative) you can't smoke ('are not allowed to').
In the above list of examples, 'process negative' forms are given in column
(i) and 'modulation negative' forms in column (ii). The distinction looks
parallel to that between 'thesis negative' and 'modality negative' in the
discussion of modality above. But whereas in modality there are reasons for
recognizing only ONE system of positive/negative, that associated with the
thesis, in modulation there are two distinct systems, one associated with the
modulation and one with the process. We can have, for example, he is not
allowed not to tell.21 Furthermore, of the two it is the negative associated
with the modulation that is less restricted, being the only one that can be
expressed by modulation: there is no verbal modal form equivalent to Jones
is willing not to tell. Thus modulations are clearly subject to the system of
polarity, as they are to that of tense.
Some of the modulations are 'oblique' forms, used in environments
demanding sequence of tenses: past reported, tentative, and hypothetical.
In some instances these pair off with the simple forms: thus oblique could
corresponds to unmarked can, would to will, should to shall, was to to is to;
and, up to a point, should, ought to and need to to must. The correspondences
are not complete; but the set of oblique forms is, in fact, clearly specifiable,
since they alone occur in the particular pattern referred to earlier - that
exemplified by you should have known, they ought to have warned you. In this
pattern we may have
Active: could have but not can have
would have but not will have
Passive: could have but not can have
might have but not may have
should have but not shall have
21 We also find double negatives in modality, as in he can't not have done it. But in fact
both negatives are negations on the process, so that the non-verbal equivalent is certainly
he didn't not do it. Cf. he must not be here = apparently he isn't here.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
ought to have,
need have but not must have
was to have but not is to have
This is not to say that the forms on the right never occ
occur as modulations, only as true modalities, e.g. Smith
is certain that Smith knew'; the only one that never occ
but that is to be expected, since is to / was to neve
modality.
Forms like you should have known look like instances of the non-finite
tense (to) have known, which as we have already noted is in fact the one that
occurs in modality - in Smith must have known, for example. But they are not.
In you should have known it is still the modulation that is in the past; the
meaning is 'you were supposed to know, but didn't'. Thus there is an
explicit contradiction between the modulation of the process and the process
itself. Other examples:
(18.1) Jones would have driven you 'was willing to (but didn't)'
(18.2) Jones could have shown you 'was able to (but didn't)'
(18.3) they might have arrested you 'were entitled to (but didn't)'
(18.4) should I have apologized? 'was I expected to (I didn't)?'
(18.5) we needn't have given one 'weren't required to (but did)'
This form is thus the realization of the features 'modulation, past, unfulfilled'.
We are now getting near to the point of this lengthy discussion of modality.
It is clear that in modality and modulation we have to do with two different
systems which are at the same time in some sense semantically alike. Let us
recapitulate some of the differences between them. Modality is a system
derived from the 'interpersonal' function of language, expressing the speaker's
assessment of probabilities. It is therefore not subject to variations or
constraints of tense or polarity (or, we might add here, of voice): it has no
tense, voice or polarity of its own but combines freely with all values of these
variables in the clause. There is one exception to the latter: where modality is
expressed in the verb, it excludes the possibility of the selection of primary
tense. The verb may select either primary tense or modality but not both.
This is explained by the system of finiteness. The function of finiteness in the
verb is to relate what is being said to the 'speaker-now', both by allowing
options of mood and by giving a reference point either in time or in the
speaker's judgment. Hence the finite element always combines with one or
other of the two categories that serve to provide the reference point, namely
primary tense and modality; but they cannot both function in this way at the
same time.22 The two can co-occur if modality is expressed otherwise than
22 The fact that primary tense and modality are both realized by 'anomalous finites'

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
in the verb; but it remains true that, because of the deictic function of fi
niteness, neither can be present except in the environment of a finite clause.
The system which we have called 'modulation' is very different: it is
ideational in function, and expresses factual conditions on the process
expressed in the clause. These are as it were 'quasi-modalities', in that they
may be realized through the medium of what is essentially a modal structure
- by finite verbal auxiliaries; but they are not themselves necessarily finite,
and they carry the full range of options in tense and polarity, while the main
verb with which they catenate must be non-finite and tenseless. On the other
hand, they do not display a voice option; each one is either inherently active
or inherently passive. Consequently, while they may combine with either active
or passive in the main verb, the nature of the active/passive opposition in
these cases varies according to the particular type of modulation involved
(see n. 20 above).
The complex nature of the relationship between modality and modulation
is brought out by a consideration of the ambiguities that arise - which
appear sometimes as ambiguities and sometimes as blends. In the first place,
naturally no ambiguity arises at all where the realization is other than by a
modal auxiliary, since in such cases there is no overlap.23 Where a modal
auxiliary is used, ambiguity arises under the following conditions:
(i) non-oblique modal + simple infinitive (e.g. must do)

(ii) oblique modal + simple infinitive (e.g. should do)
(iii) oblique modal + past infinitive (e.g. should have done)
These may be either modalities or modulations. (But not non-oblique modal

plus past infinitive, e.g. must have done, which as we have seen can only be a
modality.) We shall look briefly at each of them in turn.
Under condition (i), if the form in question is a modality then the tense
must be simple present: John may go = possibly John goes (and not possibly
John is going). Any given instance will therefore be ambiguous only where this
tense makes sense, for example John must go round the world about once a
month these days, i.e. 'he certainly goes ...'.24 It is difficult, for example, to
interpret you must tie a knot here as a modality: 'I'm sure you tie a knot
here regularly'. If the form is a modulation this restriction does not apply,
reflects the similarity between them; they are both 'deictic' in the extended sense, and
differ merely in the type of deixis involved. Cf. Charles J. Fillmore, 'Deictic Categories
in the Semantics of come', FL 2 (1966), 219-227; and Rodney Huddleston, 'Some Observa
tions on Tense and Deixis in English', Lg 45 (1969), 777-806.
23 Except that the form it is possible for ... to ... can occur in both systems (and perhaps
other locutions withpossible).
24 The difference is realized through the stress pattern. In modality, the must is salient:
I/ A he / must go ...; in modulation, it is weak: // A he must / go ....

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
since the tense is in the modulation and here (mod

simple present is the norm: John must go = John is re
number of ambiguous INSTANCES is not as great as mig
In action clauses, which typically have the be ...ing for
of 'present', the simple present tense is fairly restricted
normal in clauses of other types - mental process
clauses - the former are less subject to modulation a
given above, Mary can't think!). So the most usual
INSTANCES is the relational clause with be and similar
pretation, as modality or as modulation, often depen
complement; he must be crazy is unlikely to mean 'he i
- thought it might, in referring to an actor - whereas
interpretation 'is required to behave ...ly' is the m
examples with careful and careless at the beginning o
however is common to all instances of the non-obliq
namely that it is necessary to distinguish the two in
truly ambiguous, and the hearer has to select one or th
This is much less true of the oblique forms, co
especially those with past infinitive: could have, sh
find instances which are more like blends, where t
requirement of selecting just one OR the other inter
he could have escaped if he'd tried, 'that he would ha
is possible' or 'if he'd tried he would have been able
home team ought to have won, if they'd had reasonable
have won is predictable' or 'they would have been un
The distinction between modality and modulation ten
a hypothetical environment. This is true even with
there is a hypothetical element somewhere around (for
could or might = possibly would, e.g. he could finish it
modulation, e.g. they ought to be able to get it right); o
are ambiguous, or even seem to permit both dis
interpretations (e.g. they couldn't trust Smith 'sur
wouldn't be able to', 'surely they wouldn't be able to')
Some interesting further light is shed on this by a
tives. There is a general constraint whereby encodin
the leader is John) are not subject to active modul
have the leader is willing / able to be John. To any clau
of thematic variants which are in fact encoding equa
them corresponds (among many others) the one who
where the matrix clause is of the type of the leade
clauses cannot be modulated by ability or inclination

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
told them could have been Smith is not ambiguous in spite of the could have
been: it can only be a modality 'possibly the one who told them was Smith'.
If, on the other hand, the could have had been in the constituent clause,
which is not an equative, the clause would have remained ambiguous: the
one who could have told them is Smith. Passive modulation, which in this and
other respects is intermediate between active modulation and modality, can
be combined with encoding equative, but the result is incongruent: the
leader must be John (in sense 'we demand that ...'; cf. the one who told them
should have been Smith 'it was supposed to be Smith who told them, but it
wasn't - i.e. someone else did').25
Summarizing: not all forms with a modal auxiliary can belong to both
systems. Non-oblique past tense forms, such as can have ..., can only be
modalities; likewise some equatives; and some negative forms are specialized
to one or the other (these have not been discussed, but are shown in the
Tables). Among those that are ambivalent, we find a range of semantic
differentiation: clearly ambiguous pairs at one end, e.g. may do = either
possibly does or is allowed to do, the two being quite distinct, and blends at
the other, e.g. might have done = either possibly would have done or were /
would have been allowed to ('but didn't'), without any very clear distinction
between them. Blending is associated with the remote or hypothetical end
of the scale; the more immediate the environment, the more discrete the
meanings of the two systems.
How do we account for this? Let us represent the systems of modulation
in a simple network as we did with those of modality earlier:
inclination
active- l
Ability
i-permission
Z passive -ncssity-obligation
o0 -compulsion
- -neutral
0 - oblique (hypothetical & c.)
Positive
-- negative
A noticeable feature is that, although as set out independently, the two

systems look very different, in fact they match very closely term for term, and
25 For a discussion of 'encoding' and 'decoding' equatives see M. A. K. Halliday, 'Notes

on Transitivity and Theme in English, II', JL 3 (1967), 199-244, esp. pp. 235-237.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
this is obviously not to be dismissed as coincidence.26 T

able; or if not, then either possible or (virtually or abs
of 'willing; or if not, then either permitted or (virtu
compelled' are in some sense to be equated. It is not ve
their identity through labels, but this is not an unfamiliar
linguistic systems is a use of language rather than an essen
a description of language, and the categories do not cor
any everyday concepts, or to any logical structures eit
sionally, use the modal auxiliaries themselves as labels; t
in fact reduce taxonomically to classes of modals, but if
non-oblique forms, and ignore neutralizations and diver
represent it thus:
E'will'
MODALITY/ j . _____should'
MODULATION r[ non - oblique must'
oblique
We also ignore here sub-categories that are not readily identifiable across
the two systems. If we now attempt to label the conflated categories in
some metaphorical way, based on 'content-substance' as grammatical labels
always are, we might arrive at something like the following:
uncommitted
--^ r- weak
committed---
L- strong -, >
absolute*
MODALITY I neutral*
MODULATION - oblique
r-positive
Negative
26 There is also movement from one to the other, and dialectal variation. For example in
American but not in British English has to is used as a modality: that has to be the best
dinner ever! There is also some transfer of non-verbal expressions, e.g. possible (cf. n.
23 above): it is possible (for John) to ... cannot be used in sense 'John has learnt to' (e.g.
swim), but can in sense 'John has the capability, strength &c. to' (e.g. climb that mountain).
Cf. is supposed to, is expected to where the interpretation is related to person: these forms
are more likely to be modalities with subject he ('probably is/will'), modulations with
subject you ('ought to').

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
where 'uncommitted' means 'probable (modality) or willing (modulation)',

'committed' means 'possible/certain (modality) or permitted/compelled
(modulation)', and so on. The network is corrected to show that 'relative'
(i.e. virtually certain/compelled) is found only in the oblique form: it is
only in hypothetical, tentative &c. environments that there is a distinct
category of 'nearly certain/compelled but not quite'.27 The forms themselves
are set out in Table IV; again, only categories common to both systems are
shown, but with their specific as well as their general labels and with negative
forms included to indicate where these differ, as between the two systems -
for example, the negative of must is can't in modality but mustn't in mod
ulation.
The partial reduction of modality and modulation to a single network
expresses the closeness of fit of the two systems; but although they are
systematically related they are not identical. This 'same but different'
phenomenon is nothing new; much of linguistic description consists in
accounting for the fact that things are different and yet identical at the same
time - the notion of realization expresses just that. But in this instance the
'same but different' phenomenon rests on the functional diversity of language.
Modality and modulation are the same system in different functions, in the
sense in which 'function' was used at the beginning of the paper: the one is
interpersonal, the other ideational. In both cases we have to do with some
kind of qualification of the process expressed in the clause, or rather of the
complex of 'process + participant'; either this qualification resides in the
speaker's own mind, or it resides in the circumstances. If the former, then
it is interpersonal in function: it relates to the speaker's own communication
role. The declarative mood is tempered with a modality showing the value
attached to the declaration. If the latter, then it is ideational in function: it
relates to a particular part of the content of the clause. The transitivity
structure is accompanied by a modulation showing the conditions that
circumscribe the process.
But the likeness is not merely one of parallelism; there is also some actual
overlap between the two systems, and this accounts for the blending that
can occur. In the passive type of modulation - permission or necessity - the
qualification comes from outside the participant; he is committed by some
one or something other than himself, e.g. Jones must resign. The agent in
the process (Jones) is the goal of the modulation. Then what is the agent of
27 It looks as though the relative/absolute distinction is of fairly recent origin and is still
only in process of emergence: the simple opposition 'strong: neutral (= absolute)' must I
'strong: oblique (= relative)' should is being replaced by one in which oblique/neutral
and relative/absolute are independently variable, but the combination 'neutral: relative'
is not (yet) found.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
TABLE IV
MODALITY/MODULATION (unified system)
UNCOMMITTED COMMITT
PROBABLE/ACTIVE POSSIBLE-CERTAIN/
WEAK |CERTAIN/N
PROBABLE/WILLING POSS
RE
VIRTUA
POS will can, may

NEUTRAL NEG won't ay not (modality)/ c
NEG won' t may not (modality)/ c
needn't (modulation) m
OBLIQUE: POS would could, might should, oug

tentative,
hypothetical NEG wouldn't might not (modality)/ shouldn't, oughtn't to c
etc. needn't (modulation) mustn't

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
the modulation? Very often it is the speaker. It is the speaker's opinion

that there is some qualification on the process; his own judgment is involved
- in which case it is very close to a modality. The difference between I think
it possible and I permit it is small, especially in a hypothetical or tentative
environment; the I is prominent in both cases. Note that this lies behind the
difference between the two forms of modulation: between Jones must resign
and Jones is required to resign. The first is speaker-based, meaning 'I insist
on it'; hence it is realized by a modal auxiliary. The second is external to the
speaker, and means 'someone else insists on it'. Compare you can go now 'I
permit it', you are allowed to go now 'someone else permits it'.28 Modulation,
especially of the passive type, is a condition imposed by someone; and if that
someone is the speaker himself then it becomes a kind of modality - the
speaker in his normal, modal function interfering as it were in the event,
in the ideational content of the clause. Hence the term 'quasi-modality'
which I used above.
Modality, then, is the speaker's assessment of probability and predictability.
It is external to the content, being a part of the attitude taken up by the
speaker: his attitude, in this case, towards his own speech role as 'declarer'.
It is thus clearly within the interpersonal component; but at the same time
it is oriented towards the ideational, because it is an attitude towards the
content that is being expressed. Modulation, on the other hand, is part of the
ideational content of the clause; it is a characterization of the relation of the
participant to the process - his ability, &c., to carry it out. But while reference
to ability does in fact characterize the participant in question - Smith can
swim is a fact about Smith - reference to permission or compulsion does not.
Jones must swing is not a characterization of Jones' participation in the
process but of someone else's judgment about Jones' participation; and that
'someone else' is, typically, the speaker. Thus the same forms can be used to
express both. When we say that the opinions a person expresses often tell us
more about the speaker than about the subject he is pronouncing on it is
likely to be his use of these 'quasi-modalities' that we have in mind: his musts
and mays and shouldn'ts. So while modulations are incorporated into the
thesis as ideational material, they represent that part of it that is oriented
towards the interpersonal - it is the content as interpreted by or filtered
through the speaker that is being expressed.
It appears, therefore, that the similarity between the two systems has two
aspects to it. On the one hand, there is a semantic region where the two
functions, the ideational and the interpersonal, overlap, that of speaker's
commentary on the content. The interpersonal function includes all kinds of
28 The position is more complex if the main verb is passive: cf. the difference between
John must be warned and John must be sacked.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
commentary by the speaker; where his comment sp

probability of the content it is expressed through t
modality. The ideational function includes all aspects
that specifically involve the speaker are expressed
system of modulation. Hence the two overlap in me
possibility of semantic blends. On the other hand, the t
a certain point (in delicacy), formally identical, so th
up a single syntactic system, on the lines suggested
operates in different functional environments. The
either from the interpersonal function, in which case i
or it may be entered from the ideational function, in w
modulation. Hence the two are parallel but distinct,
bility of ambiguities. But the explanation is a funct
and both the ambiguity and the blending are accounted f
It was suggested at the beginning that the interna
grammar of languages provides strong reasons for
approach, since grammatical systems group themsel
systems and these groupings turn out to reflect a ba
in language. But the present illustration has gone f
demonstrating the assignment of systems to functi
ality and modulation do derive from different function
they are related to different neighbour systems in the
modality to mood and other interpersonal systems of
to transitivity and the grammar of processes and par
also remarkably similar, so that having once taken t
put them back together again; and this reveals anoth
tional diversity, namely that it provides the conditions
in the grammar what are essentially distinct sets of
interrelation between them, in realization and in m
schematically in Table V; the actual picture is, as w
somewhat more indeterminate than this implies. For
paper we shall consider a second problem, this time
of mood, which likewise involves the interrelation o
but in a rather different way.
The second topic to be considered involves an expla

different kind. In the previous example we sought to ex
'same yet different' type of grammatical and semant
two systems, and this was seen in terms of functional
modality and modulation were suggested to be, fun
options in different functional environments. Her

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
TABLE V
Interpersonal component ( Ideational compon
speaker's process and

comment participant
/ ...on conditions \
probabil
/n
not part of content pa
finite only \ finite
no tense, polarity or voice selects for tens
(but combinable with all modalities modul
variants of tense, form of pr
polarity and voice) \ each one
probable f uncommitted w
possible / p weak \ w
\^possible . /virtually = committed
-cerain certain certain strong per
-certain cern abso
absolutely (passiv
certain / \ / \
Cmo/and/or
modal andor verbal
verbal('modal)
'moda') or
or cate
cat
adjunct' auxiliary
// | \\ 4 neutral
impersonal ... personal ( non-oblique) (passive type
(adverbial; nominal- be + passive + to be +
adjectival; verbal locutions) oblique

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
particular set of related structures, and attempt to explain

variation among them; this also raises questions of lingu
the variation seems to be relatable to an associated struc
functional origin.
As examples we may cite the following mood paradigm
(19.1) Sir Christopher Wren built this gazebo

(19.2) did Sir Christopher Wren build this gazebo?
(19.3) who built this gazebo?
(19.4) what did Sir Christopher Wren build?
The question is: why do these differ in the way they do -

in which the elements occur? Here it will be necessary to t
third of the generalized functions of language, that referr
'textual' or discourse function.
Let us establish first what the difference is that is under
If there is a WH-element, this comes first. This is, of cour
the 'good reason' principle: it comes first unless there
putting it somewhere else.29 But what constitutes 'good rea
is itself one of the things we are interested in. (ii) If there
the 'modal block' comes first. The modal block is the s
tical subject') plus the finite element of the verb. With
either (a) the finite element of the verb comes first, in wh
dissociated from the lexical element (did Sir Christopher
Sir Christopher ...?); or (b) the subject comes first. We ignor
clause without a modal block.
The contrast has been expressed entirely in terms of wha
this reflects the fact that there is only one semantically s
sequence in the English clause, namely first place. Nea
sequence in the clause is statable in terms of first posit
overall variation in the permitted ordering of elements; bu
simply from one general principle, which is that it mus
elements, and all combinations of elements, to come in
requirement largely determines the range of options availa
The question can thus be formulated in these terms: (i
element come in first position, if there is one? (ii) Why
element, does the modal block (subject plus finite verb) com
What constitutes 'good reason' for a departure from this
within the modal block, does (a) the finite verb come f
interrogative and (b) the subject if it is declarative?
29 I.e. this represents the unmarked option.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
Since the discussion largely centres around structural functions, it is

important to make clear in what sense the ambiguous term 'subject' is - and
is not - being used. It is customary to recognize three 'types' of subject, the
grammatical subject, the logical subject and the psychological subject; and
although these terms are preposterous ('grammatical subject' suggests a
structural element whose only function is to be a function in grammatical
structure, and the other two are no better), the categories are on the whole
clear enough: in
(20.1) Sir Christopher Wren built this gazebo
Sir Christopher Wren is at once grammatical, logical and psychological

subject, whereas in
(20.2) this gazebo I was left by my father

the three are dissociated - this gazebo is psychological subject, I is gram
matical subject and my father is logical subject. In the present discussion,
'subject' will be used to mean grammatical subject and nothing else; log
ical subject will be referred to as 'actor'. The category of psychological
subject is a mixture, involving both the concept of 'given' and that of
'theme'; these are discussed at some length, and distinguished, below.
The first two questions therefore amount to asking why the three types of
indicative clause differ in just the way they do, such that the first position is
occupied in each instance by one particular element: by the subject if the
clause is declarative, by the finite element of the verb if it is a polar, or 'yes/no',
interrogative, and by the WH-element if it is a WH-interrogative. There ap
pears to bejust one answer, the same answer in each case: because, in English,
first position is thematic.
This is an 'explanation' only in the limited - but important - sense of an
interrelation with other patterns within the language. There is no automatic
reason why declaratives and the various types of interrogative should differ
in just this respect; but the difference can be related to other factors in the
grammar of English. In order to establish this particular relationship we must
first draw the distinction referred to above: between the theme, in a theme
rheme structure, and the given, in a given-new structure. These two func
tions, although sometimes treated as if they were one, are in reality quite
distinct.30
30 In Mathesius' original formulation the two are conflated: the 'theme' is 'that which is
known or at least obvious in the given situation and from which the speaker proceeds in
his discourse' (quoted in Jan Firbas, 'On Defining the Theme in Functional Sentence
Analysis', TLP 1, 267-280, p. 268; the translation from the original Czech is by Firbas,
who also added the last three words). Subsequent studies in 'functional sentence perspective'
have separated out these two components, or simultaneous structural configurations, that

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
By the 'given' we understand that part of the message

English by the intonation, to constitute a link in the ch
every unit of the message - every 'information unit',
intonation contour - there is one part, not necessarily
constituent of any other kind, that the speaker signals
is, that he explicitly offers as non-recoverable inform
additional to or contrastive with what the hearer is bein
available to him. Optionally, the speaker may associ
necessarily preceding it - another part of the message,
offers as recoverable information: it is the known part,
the speaker is putting over as known - with consider
rhetorical effect. This part of the message, which we s
has a specific function in the textual organization: it lin
unit to the rest of the discourse. It is often explicitly
with the preceding text through one or more of the set
lexical relations that are available for the purpose: lex
synonymic allusion; grammatical reference, substitution
and conjunction.
All text in spoken English is organized in informatio
information unit is structured as a configuration of 'given
The realization of this structure involves intonation and
reaches of the phonological hierarchy; the details have
number of studies.31 One information unit is realized a
contour, or 'tone group'; broadly, the 'new' element is
tonic element in the tone group, that which contains t
Anything which follows this is necessarily 'given'. So, in
(21) all the examination papers are to be marked out
where the tonic element consists of the single word all

the tone group, it is possible to reconstruct quite a nu
the preceding conversation, because everything following t
fall within the 'given' element in the information unit.
On the other hand, what precedes the tonic element m
is not necessarily so. Here is a typical passage, shown with
bounded by // and 'new' elements italicized:
of 'recoverable and non-recoverable' (my 'given-new') and that of

tion' (my 'theme-rheme'), though in various different ways; my
closest to that of TravniCek quoted in Firbas, op. cit. For a fuller di
suggested here, see ref. in n. 25 above, esp. pp. 200-223.
31 For a detailed textual study see Afaf Elmenoufy, A Study of th
the Grammar of English, 2 vols., Ph.D. thesis, University of London

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
(22) // my side of the road // was completely blocked // and unfortu

nately // I went into a four wheel skid // before coming up to this
Mini // and at that time // I didn't know how to deal with a four
wheel skid //32
In before coming up to this Mini, the tonic is at Mini and what precedes is
given; whereas in I went into a four wheel skid, the tonic is at skid but all
except Iis informationally new.
For the present purpose it will be useful to note just one further detail of
the system, namely the distinction between unmarked and marked informa
tion structure. In the unmarked instance, the new element comes last in the
information unit; this has certain consequences which again may be brought
out by an example:
(23.1) George takes his wife to the movies

These presuppose, respectively, contexts such as the following:
(24.1) I don't know any man who takes his wife to the movies
(24.2) Does George approve of his wife going to the movies?
(24.2') Does any man approve of his wife going to the movies?
(24.3) Who does George take to the movies then?
(24.3') Does George go all by himself to the movies?
(24.3") Most people go to the movies by themselves
That is to say, for (23.1) we must have an environment such as (24.1), in

which the taking, the wife and the movies are all recoverable. (23.2) presup
poses just the wife and the movies; and we can actually specify the nature of
a clause (23.2') corresponding exactly to the environment (24.2') and differing
from (23.2) (i.e. having George within the 'new'), but this requires reference
to rhythm.33 Similarly, (23.3) presupposes the movies only (with, again,
variants (23.3') and (23.3")). However - and this is the point - (23.4) pre
supposes nothing. It is not necessarily the 'answer' to anything at all; it may
just be the beginning of a discourse. This is the sense in which it is unmarked.
32 Ref. as in n. 31; vol. 2, p. 119. For the tone group, see ibid. vol. 1, parts I and II. Cf.
F. Danes, 'Sentence Intonation from a Functional Point of View', Word 16 (1960), 34-54.
33 (23.2) // A George / takes ... // with George weak;
(23.2') // George / takes ...// with George salient.
Likewise (23.3) // A George takes his / wife ... //;
(23.3) // A George / takes his / wife ... //;
(23.3) // George / takes his / wife ... I/

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
The unmarked form is unique in that it does not REQ

should be recoverable. Thus the unmarked form is ver
marked forms in its textual function, its significance fo
discourse.
While therefore there is some tendency for the 'given-new' structure of
information to be reflected in the sequence of elements in the clause, in that
in the unmarked form of the information unit the 'new' element occurs in
final position, this is true only to a very limited extent. In the first place, even
in the unmarked form there may be no 'given' element preceding the new;
and in the marked forms the order is actually partly reversed, with at least
some of the given component occurring after the new. In the second place,
the domain of 'given-new' structure is not the clause but the information
unit, which is coextensive with the clause only under 'good reason' conditions;
it is often shorter and sometimes longer than a clause, so that even where an
information unit is structured in the unmarked form, as given followed by
new, this does not by itself specify any ordering of clause constituents. The
mapping of information structure onto clause structure is a distinct relation
with its own significance as a semantic variable.
This is important to the present discussion, since it establishes that first
position in the English clause does not express the function 'given'. Yet at
the same time first position in the clause is structurally significant. In fact,
it expresses something else, something which is also concerned with the
organization of a text but is nevertheless a distinct structural function. First
position (and note that this does mean first position in the clause, and not in
the information unit) expresses the function of 'theme'. What the speaker
puts first is the theme of the clause, the remainder being the 'rheme'. While
given-new is a structure not of the clause but of the information unit, and
is realized not by sequence but by intonation, theme-rheme on the other hand
is a structure of the clause, and is realized by the sequence of elements: the
theme comes first.
The theme is thus a function in clause structure, like the subject or the
actor. It is in fact what is meant by the psychological subject once the
notion of 'given' is abstracted therefrom. The meaning of theme is not the
same as that of given, although the two functions are often realized by the
same element, or overlapping elements - not as often in real life, however,
as in the isolated examples that are usually brought up for discussion, which
have unmarked information (given-new) structure precisely because they
have no context. The present discussion of the order of elements in the
different classes of indicative clause will in fact serve to shed some more
light on what the meaning of theme is; this has always been the more
difficult of the two. In principle, the theme is the point of departure - the

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
takeoff point of the clause; and the significant fact about it is that the speaker
is free to select whatever theme he likes. His choice is not determined by
the context. He may select an item that is anaphoric, and thus links with
what has gone before; but he is not obliged to do so, and he very often does
not. In any case, there may be nothing of the kind that he can choose as the
theme: not every clause has an anaphoric or other kind of 'given' element
in it, whereas every clause must have a theme.34
The thematic function as such is peculiar to the clause. It may nevertheless
be relatable to a more general meaning of first position in English, since in
both the verbal group and the nominal group first position realizes a struc
tural function which relates to 'speaker-now' - which is 'deictic' in the
extended sense:35 namely primary tense or modality in the verbal group
(see p. 342 above), and deixis in the narrower, more usual sense in the nom
inal group. The function of theme can be regarded, by a further extension,
as the deictic element in the structure of the clause, in that it defines the
speaker's angle on the content: so in
(25) this gazebo was built by my father
this, the past finite in was, and my relate the ideational elements (goal,
process and actor) to the speaker-now, while this gazebo relates the clause as
a whole to the speaker's current perspective. Note that it has to be theme, not
'given', which fulfils this function: the given is hearer-oriented and context
bound, whereas the theme is speaker-oriented and context-free.
The thematic function thus gives the clause its significance as a component
of a text. While the information unit structure, in terms of given and new,
gives the message coherence with what has gone before, the organization of
the clause into theme and rheme gives it coherence within itself. The theme
is what turns an ideational structure into a message; we may have a pattern
of content such as actor - process - goal, but only when the function of theme
has been mapped on to one or more of these elements does a message result.
In (20.1) above Sir Christopher Wren is actor and also theme; in (20.2) this
gazebo is goal and also theme. Thematic status can be made more explicit in
various ways, e.g. by as for, as far as ... is concerned; by 'reprise' as in this
gazebo ... it; and by a (marked) association with the function 'new' in the
form it was this gazebo that....
First position in the clause, therefore, is structurally significant: it is the
34 That is, of the types under discussion (independent indicative clauses); and with the
possible exception of those beginning with dummy it and there, which may be best regarded
as having no thematic element in their structure.
35 The significance of 'first position' as a general realizational category in English, having
a similar meaning in the clause, the verbal group and the nominal group, was pointed out
in discussion by Gunther Kress and William Downes.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
realization of the category of 'theme', which is an

the textual function of language. If there is a WH-elem
characteristically occupies first position. The WH-ele
in clause structure, one that derives from the interp
component. It expresses the communication role taken
also that assigned by him to the hearer: it is a requ
action, the provision of a specific piece of information
in first position: that is, in the thematic position in th
theme of a WH-clause is the request for informati
particular missing piece should be supplied. This, in
speaker's reason for communicating anything at all.
There is no intrinsic reason why the WH-element sho
In many languages, it occupies whatever position in t
to its status in transitivity. Where the WH-element is
this must be because first position has some indepen
icance as the expression of another function with whic
typically associated, a function that will automatic
WH-element if the clause is of the WH-type.
We can test this by looking at the other kind of in
type. This is also characterized by its having a parti
position, here the finite element of the verb. Is there a
for associating this with the function of theme? Ve
polar question, like a WH-question, is a request for
case, however, the information is specifically and only
answer yes or no? The request for information about p
a yes/no question, in just the same way that the requ
the WH-type is the theme of a WH-question. Polar
finite element of the verbal group; hence in polar inter
comes first. Moreover it is made prominent by bein
lexical verb, even where in the declarative the two w
(26) did (didn't) Sir Christopher Wren build this g
Note that, in languages where the 'yes' and 'no' of d

in English, 'the answer is in the positive' or 'the ans
but 'you are right' or 'you are wrong', it is the reque
'am I right or wrong?' that would be thematic; but this
of polarity in the verb, so that in those languages th
to associate the verbal element with thematic function.
If theme is the function typically assumed by th
element in an interrogative clause (i.e. 'this clause i
request I have for information'), is there any comp

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
clarative clause? It is clear that first position is significant in declarative

clauses, as we can see by frontshifting each element in turn and holding
everything else constant; and the function most typically associated with
that of theme is the subject function. Other things being equal, in a declara
tive clause the subject is the theme. However, this is a much less strong
tendency than that which operates with interrogatives: the conditions of
'good reason' which lead to the selection of a theme other than the subject
in a declarative clause are much less stringent than those required to override
the selection of the information-seeking element as theme in an interrogative
clause. There are two reasons for this. On the one hand, there is no function
in declarative with such strong claims for thematic status, because there is no
sharp distinction corresponding to that in interrogative between the infor
mation-seeking element and the rest.36 On the other hand, there are various
non-ideational elements in declarative clauses which are also likely themes,
for example modalities. The reason why probably, perhaps &c. normally
appear in first position is that, if the speaker is expressing such an opinion
at all, he is very likely to give it the status of a theme: 'as far as my own
assessment goes, ...'. But with these provisos, the subject is the 'natural' theme
of a declarative clause, because it is structurally bonded to the finite verb,
and it is finiteness that gives the clause its status as a potentially independent
message.
This in turn is one of the reasons for choosing a particular element as the
subject - because it is the unmarked choice for theme. Of course, something
must come first; this is a property of any expression. But it is precisely this
property that is being exploited in order to endow a clause with the value of
a message: the linearity is used as the basis for the essential message structure,
that of theme-rheme. Moreover, the thematic status is not limited to just one
element in the clause; any combination of elements can function as theme,
through the use of the structural device of nominalization. For example, in
(27) the one who built this gazebo was Sir Christopher Wren
36 Note that other elements in an interrogative clause resemble those of a declarative

clause in that they are not information-seeking. A distinction is to be drawn here between
two different but simultaneous functions of the information-seeking element. On the one
hand, it specifies what information is not known to the speaker. (For a most penetrating
discussion of this aspect see Jan Firbas, 'Some Thoughts on the Function of Word-Order
in Old English and Modem English', Sbornik Praci Filosofick6 Fakulty AS (Brno 1957),
72-98, which was not known to me when the present paper was being written. Firbas also
groups together the WH-element and the finite element of the verb, and refers to this
category as 'indicator of the want of knowledge on the part of the questioner'.) On the
other hand, it expresses that the fact that the speaker lacks this information is significant,
thereby explaining his choice of role as a questioner; it is this that makes it the speaker's
natural point of departure, and hence the unmarked theme of the clause. (See discussion
in M. A. K. Halliday, ref. as in n. 25, pp. 212-3.)

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
the process (build) and the goal, or more strictly speak

(this gazebo), together constitute the theme of the clause.
Thus the variation in the sequence of elements in clauses
in English may be explained in terms of theme. The in
WH-interrogative is the WH-element, that of a yes/no
polarity-carrying element, so that these come first; and b
over the subject, which, likewise for thematic reasons
complex factors involved, typically occupies first position
The organization of the clause into a theme and a rhe
feature which, like other structural features, derives from
it is a configuration of functions expressing a particul
total meaning of the clause, namely the 'textual' compo
a message. The variation in sequence in declarative and
exemplifies, in fact, the interaction of the textual and
functions. In this part of the grammar, the (interpersonal
which is concerned with the speaker's choice of speech r
with the (textual) system of 'theme', which is concerne
tion of the clause as a message; hence the elements of stru
the mood system, the WH-element and the constituents of
- subject and finite verb - become associated with the e
deriving from the theme system, the theme and the r
of elements in the structure of the English clause is the p
action between these two functions of language.
The study of language within a social and cultural fram

some account of linguistic functions, and the 'multip
language has for a long time been familiar in such con
understand the use of language in a particular culture o
the specific situation types of that culture, we have to
fact that language serves a variety of different ends, and it
can be understood only as relating to those ends. This i
more recent field of language acquisition studies; learning t
is, in effect, learning the functions of language, which
context for and give significance to its structures and
code and register, of speech acts, of context of situat
relate linguistic features to the functions which language
37 Malinowski distinguished the 'active', 'narrative' and 'magical'

(ref. as in n. 3 above). He also considered (ibid.) that 'Language in its
real categories derived from practical attitudes of the child and of p
to the surrounding world', later discarding the notions of 'primiti
'primitive languages', and generalizing his functional approac
headings 'pragmatic' and 'magical') to apply to all languages.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC
The present discussion has attempted to go rather further, making a

somewhat different and perhaps less self-evident point. This is that, aside
from questions of language and society, or language and culture, the internal
organization of the linguistic system has itself a functional basis, so that in
order to understand the nature of language it is necessary to start from
considerations of its use. The point is not a new one - it was made by
Malinowski in 1923 - but it is often taken to refer only to some marginal
features or specialized uses of language. Hence the examples treated here
have not been drawn from situationally dependent areas of language, such
as greetings or other restricted language systems, but from the most general
and central area of the grammar, that which in pedagogical linguistics is
sometimes called the 'common core'.38 It is impossible to conceive of any
variety or use of English, except the most restricted, without declaratives and
interrogatives, without thematic structures, or without modalities and
modulations. Yet all of these raise questions of language function; not
merely in the sense of that they can be shown to derive from different
functional origins - to be there for different reasons, as it were - but in the
sense that certain of their internal features, both as systems and as realiza
tions, seem to be most readily explainable in terms of their functional
diversity. Not only does any system of linguistic features derive from some
function of language, perhaps more than one, through which it can be related
to situations of use, but in addition, this fact may also determine, at least in
part, the form taken by that system and its exponents in the grammar.
University College, London
38 The study of'system networks' in the grammar leads empirically to generalizations based
on function: the systems of options which show a high degree of mutual dependence turn
out to be those which are functionally related. Cf. reference in n. 8 above.

131.104.16.121 on Sat, 19 Oct 2024 13:39:37 UTC

Halliday FunctionalDiversityLanguage

Uploaded by

Copyright:

Available Formats

Halliday FunctionalDiversityLanguage

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Halliday FunctionalDiversityLanguage

Uploaded by

Copyright:

Available Formats

Functional Diversity in Language as Seen from a Consideration of Modality and Mood in

Stable URL: https://www.jstor.org/stable/25000463

This content downloaded from

FUNCTIONAL DIVERSITY IN LANGUAGE

Recent work in a number of different areas within linguistics has tended to

Foundations of Language 6 (1970) 322-361. All rights reserved.

This content downloaded from

however, it is no different in kind from the elements of structure of the adult

This content downloaded from

typically, exposed to a large quantity of highly structu

6 Cf. also the discussion in John Morton, 'What Could Possibly

This content downloaded from

This content downloaded from

the one who died was Smith this would be a differen

The first topic that we shall examine from a functi

(1.1) you must be very careful

This content downloaded from

Adverbial 'MINOR PROCESSES' narrowing COMMENT

This content downloaded from

are'; an example such as you must be very sympathet

(2.1) possibly this gazebo was built by Sir Christop

We can express more or less the same content by saying

(2.2) this gazebo may have been built by Sir Chris

with may have been instead of possibly ... was; and

(3.1) surely he'll stop talking soon

(4) could they perhaps have left a note somewhere?

i.e. 'do you consider that there is a chance that ...?'

This content downloaded from

UNDERTONE: presumably perhaps assured

OVERTONE: predictably (tone 1) conceivably (tone 4) su

* normally may in declarative, can in interrogative

This content downloaded from

verbal (where verbal means 'functioning syntactically as

(1) they have only finite forms: there is no to can

This content downloaded from

there is a certainty that. Specifically:

(1) as adverb ('modal adjunct'): perhaps, possibly, presumably,

(5) //4 A he / might have / built it //4 A con/ceivably //

(modality: ((committed: possibility) / (overtone of reservation)))

13 Cf. T. F. Mitchell, 'Syntagmatic Relations in Linguistic Analysis', TPhS 1958, 101-118.

This content downloaded from

and these are realized, AS A WHOLE, by the tone, the m

r< < ---- undertone (tentative & c.)

(6.1) this gazebo will probably have been built by Wren

it makes no difference whether we associate the negative with the thesis, as in

(6.2) it is probable that this gazebo was not built by Wren

or with the modality, as in

(6.3) it is not probable that this gazebo was built by Wren

- either of these can be expressed as

(6.4) this gazebo won't have been built by Wren

(7.1) it is possible that this gazebo was not built by Wren

This content downloaded from

(7.3) this gazebo may not have been built by Wren

(7.4) this gazebo can't have been built by Wren

But this is surely wrong, since the total number of ne

This content downloaded from

Thus the primary option in the system is that betw

(8) they would telephone just as I was going to sle

Note that this form with would, often treated as aberrant,