t

i
t
Chapter 5
I

The texture of a text {

I
I

Introduction
that unity is a cnrcial attribute
I began the last chapter with the claim of textual unity: namely'
of texts,and went on to examine one source
Structure.Itriedtoshowthatthestructlrreofate*iscloselyrelated
tlre specific values of field'
to the context of situai"ion;;;;ffi1trat contextual configuration'
up a
tenor, and mode, *t'iciiiogtttter make
about the strrrcture of the text'
*ur. ."iiuiripi"ai"tio"r
can be used to be used as a pointer
t;;;;;;;fittt t"tt ittttt canThere
iust as the unfoldins is' thus' a two-
il1;?;;;il;;'tit'l "'"i"'tual
configuration'
*a ontextual configuration:
way relationrrrip u"t*iti tot'ti*tt"* onfirms th nature of
the on-going ,tru.t,,'""-oi i-r'" it*J a"n"o and
the contextuut *irii" ine latter acts as a point of refer-
"onng,irlii""l can appropriatdv appear when,
ence for deciding what kind oi elements
where, and how often'
In this chapter *t look at the second source of textual unity:
'ttutt
namely, texture.

What is texture?
Texture, like structure, can be shown
to be ultimately related to the
that lshall come back to in Chap-
context of situation' iiit lt " til"*"
let me Uegin UV u fti"t ait""*ti"" of
two examples @xam-
ter 6. Here,
ples 5.1 and 5.2).

ExamPle 5.1
girl
Once upon a time there was a little
and she went out for a walk
anO sne saw a lovely little teddybear
and so she took it home
*n"n she got home she washed it'
"nO

70
I
t Example 5.2
t He got up on the buffalo
have booked a seat
i have put it away in the cupboard
have not eaten it.
j
Faced with these two examples, any natural speaker of English is
i bound to say that Example 5.1 displays certain continuities that are
lacking in Example 5.2. One of these continuities is, of course, describ-
I able in terms of generic structure. Although the first passage is incom-
plete, it is a clear instance of a familiar genre; we have no difficulty
J in recognising it as an unfinished story. It is, however, doubtful if Ex-
ample 5.2 will be seen as representative of a genre quite so readily,
I though many of us who have taught a foreign language might not be
surprised to find that the four sentences of Example 5.2 have been lift-
ed from a foreign language teaching exercise. Now, even if we were
to accept that a foreign language teaching exercise represents a genre,
it appears undeniable that such a genre would not possess structure in
quite the same sense as that discussed in the preceding chapter. For
one thing, there is no discernible beginning, middle, and end in such
exercises. In fact, due to deplorable misconceptions about language,
the continuities in a language teaching exercise are normally strictly
meta-textual; there is a purely formal reason fcir grouping the sentences
of Example 5.2 together, which has very little to do with language as
used in everyday life.
But structural continuity is not the only kind of continuity. Ex-
amples 5.1 and 5.2 differ in another important respect; I would talk
of this difference in terms of texture. Thus I would claim that the first
of these examples possesses the attribute of texture, and that this at-
tribute is lacking in the second. What kind of continuities.do I have
in mind when making such a claim? To answer this question, let us
examine these examples a little more closely. Note that the first has cer-
tain kinds of meaning relations between its parts that are not to be found texture is a matter
in the second. It is these meaning relations that are constitutive of tex- of meaning relations
ture. For example, the third person, feminine, singular pronoun sfte
in each of its occurrences refers to the same little girl to whom the nomi-
nal group a little girl refers; il, on the other hand, refers to the same
lovely little teddybear to which a lovely little teddybear refers. A more
concise way of saylng the same thing is that she is co-referential with
a little girl, and ff is co-referential with a lovely little teddybear.If we
compare the two occurrences of ir in Example 5.2 with those in 5.1,
we note an important difference: it would make no sense to claim that
fl in either of its occurrences in 5.2 is co-referential with any other item
in the example. This is definitely not the whole story, but perhaps
enough has been said to draw certain conclusions:
It. fne texture of a text is manifested by certain kinds of semantic rela-
I tions between its individual messages. The nature of these relations
I and the lexico-grammatical patterns that realise them are discussed
in the following sections.

7t
 and text structure must be seen
2. At least in the first instance' texture
that from the point of view of
as separate ph.no'ntni'^w;il; but this does not affect
text structur., e*u"ipi?s'l-ii'i*otnplete' a passage possesses tex-
the claim that it f'uti"*t"t' So
to say that
is not to *ut" u'iv Juitn uuo't ttre specific structural status of
ture
to operate only in one direc-
that passage. fne 'eiationship appears
possess texture; it may or may
tion: whatever is (part of) a text must
not U. a complete (element of a) text' listener's perception of co-
3. The property it';UH to-the be described
".1"t
"r
herence. Thus in
Elalole 5'l would
";;;;;;;ilnce'
as possessing."ht;;;;;fiit"
i*u'ptt 5'i would be seen as lacking
is presented
nbtion of coherence
in coherence. n oi"uttio;-;llht
below.

Texture, cohesive ties, and cohesive

devices t
Examples 5'1 and 5'2 might lead t
The exaggerated difference between This is I

one to suppose,tt"t t"t'eit"*

is an all-o-r-none phenomenon'
of Texts 5'1 and 5'2 sill demonstrate' t
decidedlynot true,
';;;;;aG
Text 5.1
1. once upon a time there was
a little girl
2. and she went out for a walk
g. anO ;he saw a lovely little teddybear
4. and do she took it home
s.l"O *-n"n she got home she washed it
her she cuddled it
6. and when sle took it to bed with
7. and sle fell straight to sleep
a ritde wirebnEtr the teddybear
I ill ,i#!,11iil;i;; ."[u"o
"-"i
J!_*ith
opened his eYes
e. andPtarteo !o sngaf j1!g
.,i.liiiti?'iliirxi,"oovu"E to' manv manv -^:weeks and vears

I I. llil :f;;;'G i;AG""r oot

di.v;!e-.1gg +^ sah il
new :words rrom a
lr. illl;;'ffifo;iii'"n"0[it 'sed
to sav some
different countrY
how to spe'k Englistt' Scottish' and
13. and that's how she used to know
all the rest'
Text 5.2
1. the sailor goes on the ship
dog
2. and he's coming home with a
the girl.
S.;nO tn" dog wants the boy and
tn"V din't know the bea/s in the chair
+.
"nJ coming to go to sleep in it
S. rnO tn"-Oear's
6. and they find the bear in the
chair
7. they wake him uP
8. and chuck him out the room
9. and take it to the zoo
10. the sailor takes his hat off
the dog's chased the bear out the
room
ii. in"it chair what the bear was sleeping in'
"rO
12. and the boy *iff in
"iii-o*n

72
 It would be untrue to claim that
Text 5.2 is errtirery incoherent
rs possessesno texture, though it is or that
coherent than is a;ax;obvious that the text is less
Text 5.1. ftlr-*i.J, iwro questions:
l. How do Texts 5.1 and-5.2differ
in their texture, if they do? ll
2. If the two vary in the degree-;;'##;.e,
of language correlate *itt ttir-uu.iai;;;, what, if any, patterns
i-:
In the sections I. attempt to answer these questions.
However, before we can
-!elow,
examine und-.o36ur. the
Texts 5.r and 5.2, *. r""a specific
to u.;i;;;;ilT, the semanric texture of
grammatical parternsessentiar and lexico_
,o ,t.
shall discuss the linguisticr;i;;;;;; "i""uliln of rexture in general. I
tions I tar" ,air"a._ ffi;: I return to the two ques_
vr Ls^.urc oerore

Cohesive tie
In talking about texture, the
concept that is most important
is that of
ti.' u ..iutio;;;;
ffiiffii::::Jl.iil1f cannot ri"* ii"
cannot appear i, a ii" ,irir,
*iitl,t
a retation berween ,"-1,"fllr-
reration herween th.n . L;;-;r;;# " ,iiri r,
;il:,ilJ:
"Tffi:jii
If you think of a text as
a continuous space in which
sages fo'ow each other, th;il;;;s"ff,at individuar mes-
of the tie-the A and thspj;q;fiil'iseparated runction as the two ends
A may be part of one message
nature of this link is semantic:
g from each other;
Iink between ttre two, aepi;ril ura pu'rr. or another. But there is a
il; ti'rn" two_headed arrow. The
I
the ,*. ,"i.rr of any tie
through some meaninedJ;;:s;.i are tied together
l'.,il'rr,i.
the messages o?ui"rr. "ir;;ilr"i;"#If;. ,u_ semantic
sis for cohesion between rerations
of meaning relation *,u, *"y There are certai;kinds
olii;;;il"". are the basis of
ti"., .f^irr"" liir_. in two members. For
rnsrance, take rhe first two the cohesion
. Example 5.3.
Example 5.3
I had a little nut tree
Nothing would it bear
But a silver nutmeo
And a gotden peai

Then thinkingof riute nfi treein line l

2 as member B you can as member
and i/in Iine
A
is the identity
see rhat ;;;;r;
rh. relation between the two
of reference.-The p."r.r"'ii're-fers
but the one thar has atread-l ro no other nut rree
il;;;;ili.,gq u, a tittle tree; the
situationat referenrs or uoiri
the discussion
;r.;il;;:"Jn,rr. rn rhe nur rirerarurl on
of textuar .orti*iiv,'ii.ir1'.'rutionrrrip
trty of reference is known as of situationar iden_
co_REFERENTTATITy.
Imagine now that we have
t*" (see Example 5.4).
co-referentialiry
"irr"l'r.n,"r.",
Example S.4
I play the cello. My husband
does, too.

73
 I
Then following the earlier practice, we could say that play the cello rets
is member A and does is a member B of the cohesive tie. But this time f two
the relationship is not of referential identity. The cello playing that I
do is a different situational event from the cello playing that my hus- ,- pau
band does. So the relation here is not of co-referentiality, but of the sorn
co-classification kind that could be described as co-cLAssIFICArIoN. In this type of mean- into
ing relation, the things, processes, or circumstances to which A and
B re-fer belong to an identical class, but each end of the cohesive tie
refers to a distinct member of this class. Thus there is a significant differ-
t'
*
FEr
Exe
ence between co-referentiality and co-classification- Exa
A third kind of semantic relation between the two members of a I
Exil
tie is exemplified by silver and golden in the last two lines of Example
5.3. Here the relationship is neither of co-reference nor of co-
I
classification; it is, rather, that both refer to something within the same :rs a
general field of meaning. Thus both silver and gold refer to metal, and terp
within metal to precious metal; their primary class affiliation is not that
identical-unlike two separate acts of playing the cello-but there is Iatte
is pa
a general resemblance. For want of a better term, I refer to this kind
co-extension of general meaning relation as co-ExrENsIoN. Thei
These three semantic relations of co-referentiality' o+hssification, Atrd
and co-extension are precisely what ties the two members of a tie, and that
the existence of such ties is essential to texture- The longer the text, DEIII
the truer this statement.
are o
thror
Cohesive devices-co-reference and co+lrssificelfrm sam€
These semantic relations are not independent of thlexio-grammatical the tr
patterns. It is not the case that they can be established randomly be- the e
tween any two types of language units; instead, thcre are very strong (Hall
tendencies for a specific relation to be realised by a clearly definable as G:[
set of items. For example, the relation of co-refcrentiality is typically I
iealised by the devices of reference, such as the pronominals tle', 'she', the s(
'it', eti. or by the use of the definite article'the or that of the demon- discu
stratives 'this' or 'that'. By contrast, co+I;assification is normally rea- this tl
lised either by substitution og by etlipsis- I should emphasise, perhaps, from
that this is a statement of what is typical; it does not fucribe all cases-
Either of the devices can realise either of the relations, but it is more ImpI
typical for reference type devices to signify c+'referentiality and for
'substitution and ellipsis to sigrrify the relation of co+lassification. I have In the
already given an example of substitution in Example 5.4; an example is esse
of ellipsis is given in the minidialogue Example 5.5. to son
tive sr
Example 5.5 force r

of the
ellipsis -'Can I borrow your Pen?' standir
-'Yes, but what hapPened to Yours? .o
ture ol
Here the nominal group yours is elliptical and its non-elliptical ver-
and te
sion would be'your pen'. Note that my pen and your pen are two dis- that u
tinct objects; they belong to the same class, but they are two distinct
members of the class, Thus the realisation of these two semantic

74
I relations_i.e.
;;;}r'jffii5ii"t"*,:1ffi;:ll';,x,ttion-tvpjcarrv
[i patte.n,ii,i ;;',}x]'i;?
::#,.JX,1f;.,1"",.*T[l1ll,: rexico_grammaticar
invorves

something that i
il:.T;1"il;:T,",'.'#."*;#**.,#*titt"it*tffi iH,}
Example
,ul,r.. B
*'[o'*
5.3 , rittre ,, tie type

ffil[;:
MemberB< -
;3iJ:' .yours
== ;:,'":".1;H:,
= co_classification
\_-___---___--

u. un i,p'r"i.i ;:l{,H"i.l*:i"H:
terpretation of / ii

mrmmffimffi
such devices become
cohesive_hrr", .^_-:.-.^':
^
:
co,EsrvE cohesive device

ii:"',"Ji:'1111'Ii.1ll:,*x":.*il,t-l*#f*;.,fl*liiT1;u;,

tt
ffix**fflffiil:,ffiii*?:l.T
jr*,r|,;;;,,",
:*,,*-&:,iT:J,L:#i;fi f :1;1".:_",*:,::,,;
[i':t{tit**:;,'l'i:,,,r','r'"ri,g,i,ii.dilr*,h.',';}hr,;
rrom e ;;;;s
il'j}: il; r"."ii
th

"
r
Hj: X,i:f :: I;:rrX ;[:.J li'j.,, *
Implicit devices and
their interpretation
In the above discussion
I
^^,-.^-l-_''."'.'""'

fifi*r*rutm*m"um
-f;;:-*l111+*:,..,iffi
ilT;;f
?trl,il,r#i.i[i#Tl#

75
 I
:

;
far as s1
environment-thecontext-relevantto,thetotaltext;and(2)thelin- lin- { ty is o'b
the lan-guase acompanying the
suistic environment - tt'l to-**'- of the in a par
r6t tlnt interpretation
Iuistic unit under ro.otl;;,';i;; source purelv contextual' \ sider, tt
ffi|,:",rffi;'ili""i*, .""ia either be co-textual or(Halliday & Hasan ceived a
The interpretu,'on-ii said to be
s-NpopHoruc
the sem
endophoric ties
interprt#tt#;;"f ini impticit ierm lies within the n
and nut tree' lt
1976) when the
co-text as, for tl' ""a little iirl or iltexture of a text: Cohei
"*u-rfii*i'tt
is really the endophoric ties that
are cruciaito the
.t

imolicit term can be sus-

unless an enaopnoric iir"tp*i"ii"' 9f tht- l One las
tained, cohesion *o'l;';;Il;
ptic"i'"d'. Noti that in Example 5'2' this is a
interpretation of anv of the
it is impossiure to sustli;";.";#;;;;ric * as anis
lation h
unfolds in time' the lin*t:::?:t::
impticit deivices.
tnc rauL that
31"1i1'fit;"ct
Lilven r@usseov
Lrr4L languale { of sema
permits . further factoring of en-
of a t"*t occur in succession' This .u. -- ,^..-,ror rmrrc mAv tended I
3il'Tilffi xirH;ffi :.\ild;1irr*'iJ:'3-1":::::'trh'll? Examfl
3iil1'Tl:'H":?ff ;:##ii;ilJt';-*15
either lollow or prsucuE Lrr4L ..oBrur
l:'::ff ::Ili:l,X'::
REF*1J^"Y- Len it follows its linguis-
t#;;*J-i.e. its LINGuI'TIC +:a ic (Halli- Exarpl
^N^Dr..rrrc
anaphoric rererence lli:'.'JffiHliP:L?ii:1lHX'I:':"li::l;"t?"*T,".flL'Jm:;
eili'#*""-",-':,:i::::'"'::*t"i;gf**i::il They d
HJt",l?JJffi and he 1

iffifi1ffi f::;;:;,iiJil;il;"t"-ti"e"i'ti"-'"r"'"ot'tn':lf,'X"l':
wnen tne urrprrurL ""^'^*"n Hasan 1976)' An Exarpl
N.v it tit*"o"" (Halliday &
is knor
cataphoric thus established
reference tnus estauusuEu
iii"ttt"tio.
rD
is given in Example 5'6'
I donl r
I wantil

ExamPle 5'6 Mo
I shall be telting this with a sigh second r
Somewhere ages and ages hence:' ' no idea
f*o to"as diverged in a wood' and l- whateryr
I took the one less travelled bY' between
and that has made all the difference' they an
Frost's'The road not taken''
This is the last stanza from Robert be interpreted bv refer- Exampl
;iil; nrri ri". *iu
Here the demonstrati;ffi
io in"t" a cataphoric co-referential They aC
ence to lines 3-5 of rc stanzi' "tist and he I
---- between this and lines 3-5-'
".il"ti""-,i"
it. interpretatio" * impficit device is sai'l to be rxopnonrc
reference "f lig gutside tae co-text and can
exophoric
when the source for its i;;;dti"" Thr
of the context. Imagine a situa- referent
onlv be found trrro,rgr, #el'i*i""tion at some toy' *1Ittt ing of E
tion in which a small child is hammering lway concentrate on writ-
a good deal of ooi,"
'"iillri -"tn* i";li"gio
In"t tn" might say to the
in the sr
treatme
ing a confer.."" ,up"r'.^tl i"'iighlipi"bd;t" as Exan
child:
Exampl
ExamPle 5.7
to work' I don't u
Stop doing that here' I'm trying so you (
is highlv implicit; pd ryl1 of
The first message of Example 5.7 except by ret.erence Ihr
the items doing, *oti iia iiretan ue interoreted interpreted- im-
to the immediate .o"'i*Ioi tituution' Exophorically so
l. itthn
the text and its context
plicit devices ut' opaque link between sectir
"r"u," interl

76
 far as speakers oufsidp rh-
^^-.^---

g*+ffi$+lfi#r*lr*fnrml-
tt ,",,iu,Ii"" il11i3,1'Ji'Ji:.'j,Y,",TlX:' u."uu,. oi
" "oifli,l'ties witr,
see p' 7e-82
Cohesive interpretation
and cohesive tie
:!!',lT:
u, u, irru."tr,I;',;
i;,ff-:::*
tn. i:.?' "'. i.",.
tnterpretarion of r
*Ti,,q co-exrension, and
1o
in_p.ir.rpl.,;;;;;;",1" rmpricit rerm musr be.seen
ration berween the
o r . em a n,il il i::*t*y, if i?*:,,* fjfi
i c ;r;t t*; ffi
ffi ;:";:i",ff TT:;iil;H#ffi nl',lJ,ft il:lil[i,J.,1",#;
Example 5.8
They asked the
sailor for some food
and he gave them
rr"r oioi""j.]"
"
Example 5.9
I don,t want this one
I want that one.

Most of us wh,
:.^.?y.il;s;J;:r,ii"'.'.1,Iili,,T?[rJ:#,r:1,il..1,ieem.inthe
no rdea whether the

|;:r;ii#;{dfl fi fi lljflfrH'#}ffi
Example 5.10
They asked the sailor
and he tound them
for some food
ffi
i, ,n" o"iio#Jirn" orn.
The reason whv m^or
- oh^^r,^_-
T,ix,ii:
,'lij:,,jt#l:^j{fH#"Tli-[,:ii,,ti*;,,i.{:d':,""fi we would treat ne o
rrearment wourd nol
. ," po"i6r.;IH;"ffiI#:!
51t--!o-91*rii"rrirrri',".',''
as Exampre 5.rr. ffiJlJi:i3,l;fltli
Example 5.11
I don't want this one
so you can have it.

have Iaboured this point

t- ..I because

i::i,i"Ji' ld;fjr'i# oevtces' and 'Implicit made in -'"r'j

on some.or the statemenrs
the previous
rnterpretation,; devices "
rir".i.

77
 rf;
the semantic rela- ct
of the basis 'of q91c;ivine
2. it raises the question and u
tions of .o-"ft'"n"t Inj'tt-tr*tification: devices in creating
texture'
3. it is relevant'o tt't"Jit-"i""*opt'o'i" is estab-
G€
said earlier that cohesion
To take the first o"t"i i"t'
j''int"tpttt:1-h" reference to some rtem
' TI
lished when un i*pritifiiuitt it-go"t' uut r*utnnio 5'8 and 5'9 clearly na
of the text. This is t"""Ji'
"' in
;:#;;;;;,;riq:::ii*1.'"'$l:*ili'.:*xxl#:*:"##Hm;
remarns.ulrrt':I:';'.:,
tcr
ific measing ;:;/;'
the-seman-
the similarity of dd
i,,poiiu",6t"*tu,,.ilili'ff X,,:fl i;:{f,"'+ff ,'it'Jfrll'"i,:it",TT.'.
than.1t,:-":'::l: the
pr
ii.-"Lnt.',t rather #;;;i;rt,,." oot because be
of
*ri:h:rl""lt".'ln:"#Jilii::"TH"""ir'"i"'"'p':t'1t"::
ract ,n"i?'0",,i:ii*
tina t'r:*X';
ck
crincires
'""-^do
the "r-i:lif,'1fi"'$*ffi
exophol is
t o* about exophora? Are rel
..dil.;;,"#'#;i;'*,;'g,:Tyil;l"Y3::*T'f.'i:f':l:l:
to prove 1
created what
na
attempted to
in order to dt'non't"ll: 'httiil;;'n31,1"* '"ttallv UT
th

**l***x*dJnflHtffiffi Er
Ar
th
;klllt;il"'*:"".'.'"Xl;Fi!!F:iTJt'"'n"'spokinorw-ritten
ii muctr ta'ge' - non-#"i*'r-texts-
which dis-
sh
is typically p'oouttt-'l
praytheruurange.ffi ."iii**:;t":::f Uf*gf;?*iJl,lXi,l; Th

;;'"d"'Forexampre'we ts
:aJi';'E'h'#f x[l,:'t"ilxl?h-ffi
not'tt"'9 glo*" strudure; and to con- sh
might hu" to tu"iluiitii'Jo o'a"' to *"Ippott our starcmenT
u!ou' Ro
cede thisis quite Jft;-' illn account' slnce
into wi
ii"t tute non-miniiai texts
texts in general' *" mifriJii"ttt well' while the
this witl permit g"*''"il'otio* "uoot "t IU€
*""'i'J:J"'lt-""*:m:,mffi -,f :""-:$.*if ii;;in- afi
wa
which have "'-,J#$htffi
ilG'iur" to arrive at the ali
"'.'p::'^T:'lsy;T;H# tior
[*,n*:'l;3J;TJ"":i;hlLfi?ffi 'a;tothecontextorsitua- pol
;i;[:'ffilt,'Xiffi#ft;1fi'-*#;"#itt"on"'inereralionwith ext
il'*:''"JI: AIT
ffi .:Irffi'rqjl*fl'$::iffi"1ffi tat
'ltnl
,i"*d{fl4i=ffi,8+*s*+,m3}*
lror Consider an extract
,
Co
i. il l*""s'
il ffi;no1m; (see ExamPle 5'12)' Let
Fh
ExamPle 5'12 Exi
on broken !!t"b:'-
Upended, it crouches
About to fot'J"tO' No longer threatened
"n Co
eii into this vigilance
""Pti""O core' Yo
;;rd enmitY from its hollowed

78
III *,*,,*'J**,:trffi
#{l!r:*l},,3,:**Ji*

II ,;li,:m
it*iui'xfilo-,,ir#,.#H",r,:*xfrf,i,,,:!jtl1ffi
##tri+i!{{}iftr{if*l,,*:tr",w

lfirufftm**rum*ffi
n:,][,ffirffi
I
-- "^tenslonal ti
*l'::t^-*'exrensiou
o*o u-"*r^Tll
Ijf", '.
5.3
A
silver

*"n'*,,-
tie-fype
\,------2'o"n
wiri.n.ne-oi:j;ffi
[ j;i.;f j.,,:;r****,r*r,*.*r*:i""i*
,,,*r*ffiurejr see p. 7j

79
 else in order
in this tie is implicit; we do not need to refer to anything
silver and golden-w.e only need to know the lan-
to interpret the terms
linguistic units
guage. The two terms of a co-exGnsional tie are typically
items'' The relation of
it ui*. refer to as 'content words' or 'lexical any
,
;;:;*i.*i";, described earlier, naturally does not exist between
to state under what
two randomly co-occurring lexical items' So we need
;;;iiti;;r r";h a relation lo*.t about.leave
'The same general field of mean-
unelaborat-
ing' is a vague expression. And if we the expression
could be taken as constituting
il,il ,n'y t ina^of meaning a chain which the members
association
ul"iuiio". we could end up-with i! twig' branch' trunk'oftree'the
iii.,i.i. iJ roio*st flo*ir,'petal, stem, stalk, list we have ended up
*L"i,l"g, faggot, tina.i, fire, flame' In this which it is not very
g.oupi"gli.miiucf, as'flower'and'flame', between
if we
;hat kind of general me.anTg relation obtains' But point
;il;;;t is no
examine the list, we woulifind that in thii collection
there I

ui *t i.t we could stop on the ground that the members

of the pair
r"n. of each consecutive pair
ir" ,"i-i"r"tJ.r"andg-*ir.. members
and'stalk'
I
,uat ut, say,'fl ower' uari'patul','petal' and'stem" and'stem' i
meaning retation, but the further apart the items are the
,t
"* "'.r""ri
moredifficultitiStorelatethemtoeachotherSemantically;forinstance,
to do is to
.ontiaa, 'flower' and 'faggot'. So obviously what we have t
delimit
--._ the notion 'general field of meaning" s
io achieve this end, I have used the traditional concept of sense
generally recog-
c
three sense relations relation with certain addiiions. The three sense relations
on sernantics are those of svNoN-rrrv' ANToNYMY'
nised in.the literature E
Whenever two lexical expressions stand in any of these
u"J T
"""o"ruy.
relations, a cohesive tie is established'
In svNoNvuY, the experiential meaning of the two,lexical mean-
items L

synonymy is a total oYerlap oI

is identical; this does not mean that there
-r;;'-ti*pfv that so far as oneinkind
i"gt, of meaning goes' they 'mean the
a
the standard tiie.atute semantics, for example' mentions such o
p"Itt v".run' utta 'lady', 'buy' and 'purchase" and 'smile' and 'grin" I
etc. "t
antonymy ANToNYMYcanbedescribedastheoppositenessofexperiential
golden are
,,.urri.rg; the members of our co-extensionaltie silver and
an example of this kind of meaning relation'
hyponymy I{YPoNYMYisarelationthatholdsbetweenageneralclassandits
sub-classes. The item referring to the general class
is called suprn-
L""^or.; those referring to its sub-classes are known as its Hvpot{v\'Is'
if *" t"t. onimal .n i*"mple of super-ordinate then its hyponyms
are
"s Note ih^t
cdt, dog, bear,etc. dog, and beor ate also semanti-
i"f"t.I as the co-hyponyms "ot,
of the superordinate onim.o-li-,
""frvThe lexicon of a language is organised into a hyponymtc hlerar-
in
chy, so that we trave Ofiring degrees of-generality' For example'
Enelish. the most general anl thJrefore the super-ordinate par excel-
";;u;;i;;.rt.*.tn]n!;,*hich can be used to refer to almost anythins.
blue'
Consider also the gridation of generality in food, fruit, berry,
ittention to the fact that when we have
;;;;;.-Arif,irpoi"t1"t me draw
for example, between cat and dog' we
a reiation of co-hyponymy, as
The distinc-
ufto think of iti" t"tution as that of weak antonymy'
"u" between a certain kind of antonymy and co-hyponymy is not easy
tion

80
 to draw. On t
poses, ,ir.. ,^l,t"ro,jl-.llula, this ma
ffi jt.;l,xffi
+,:li,',n,J;ili,: #13i+r::iir,:,,ilfr
ro cohesion :i:,;,,.T:
in either
MERoNyMy: ,n.t11T"I^:..ognir.a ,!ltE case.

i:,J,"*!:i;;;;i;;21;;-,xTi*di:,1!i",,-,",,,y#.tr;l meronymy
ff tff
rerarion. "T3lf.i;+,Jtri,::il,trj;yJ,jtg,:,Jili["J.'.],#;ullll,i
I ve ; :: :: v.rpear(ing, i, r..oi,iir.i,;:,;
repe
ha
on o r
;
"ot
:?:Hrr. ;ffj
rir

tar sety sim

itar
i
..,,:i,[,TIi,r:l::I"
t

::ljcat
rr-e
unir
s
:
creates
am
I,l {, *r. ;ril,,
u ."J"ll^i' ^llltcat .u ir. The
n
.",".;i;il1;fj,.lil.,j"l"r."riffi
";:;^: repitition of
lexical
'war urrl[ as tn Exampl"li:l;T,#:,HHf g,!;T:ff items
.1
Exampre s.r4
There were childr,

,H"fi#i:iiTlT,,i:;',1H:[",1i,, on the srides,

"n,,o,en and chi,dren on

iffi :,,,;;fr4;fr#f,jf#1i,,ffi .*ffi{.{",.hi,ffi

Example 5.15 fr

r::11,:',:1rffi :?,;"Tl#"*:ir?T;rri?r
1**:::,::r,ff th e coh esi ve
:J device
#1,0ff1lr;i,,1T
;;;,;j::,,:n:n,';;lp:"lr;lii#ffi i{"tit#*:tlgli,?;Ti
il'#;11,,'ii';:
irl : S:,i."'or,vr"r"I-1uove
the^relatio,
oi,",,..Jli ate general
in nar
u': ritr te u
i}llll llct o r. Engris h. il:;#fy^ a,t * ""," ri iy*'
-
ii!, f""J#i.T'i'
ff S#::,?:ii,,,..1sri;ii;,.r;,."_iff :r,;;1;^g;"J,;;;:i{:,:
,,,i;;:"::u"!,:,H,:,W:;r,!,::;i::::!i;::llHlltf
cxample ,
,
r:ir;;##*iitr*## {fl+:ili{tffi r, r;*
"."".11^'I'-

nettro 'r"^ ,l' l?" ut

IJ1"'d:1;i.:'{,Hl#.fl.}5$!?'r:iT#tff::ttn*S
u,e. rrres J ;,#X "::l I ;i;;,;,"J,:;,,;. ifi1 ; ;;; :::::l I
r nj co unc_
ti "
) il; ; . *r,i, jfi
m essage(s rt 'r.-r".' GA NIC
;;'" :i:T::,;;;[
towin s wh--^ _.' "' :'l; -oR ^ componenrs,
,^f #,::3^C^.,

r'- :.:it;;'r;;U1,"_,,#!,;:H.Jx*:XITii:;X*{*fu
quEsuofl-answpr '^'
."'
uer! sleepy.
Adjz
t97st. aro. .:_:.""" 'r'1lll1-:.rptiance
varrerr or (5c :1
organic .of,"rir" iil

8I
 l
+
Table 5.1 summarises the devices discussed' E
5.1 Summary of cohesive devices l Ji
Table n
NON-STRUCTURAL COHESION l
COMPONENTIAL RELATIONS ORGANIC RELATIONS E
1

tie relation *
Device A: Conjunctives A
IJ] A: Reference e.g. causal tie Ii
a l.Pronominals concession tie I
E] 2. Dqmonstratives TI
o 3. Definile aJticle
'4.
h
C)a Coinnalatll-b.9 in
-tl.l B: AdiacencY Pairs
<= e.g. Question (followed bY) y(
OZ answer; y(
FI! B: Substitution
offer (fotlowed bY) g
& Ellipsis
z l. Nominal acceptance; T€
z 2. Verbal order (followed bY) ol
comPliance . . . th
& 3. Clausal

A: Genetal -l Continuetives t--gAT

t! (e.g. still, alreadY. . .)
l. Repetition I w
a
r! 2. SynonYmY I co-classification te
icn 3. Antonymy I ot
b(
4. MeronYmY co-extenston
_J aE

?H B: Instantial )
ht
l. Equivalence I co-reference
X 2. Naming [ 9t .- ti(
3. Semblance J ti(
IJ] co-classlltcatton
J
th
STRUCTURAL COHESION fo
A: Parallelism er
B: Theme-Rheme DeveloPment su
C: Given-New Organisation l.
2.
3.
The interdependence of grammaricrl end lexical cohesion
4.
I suggested before that even if two implicit tcrms remain un-interpreted,
j-g, of ry
,.ii"ii"*ples 5.8 *a it is stili possible to perceive relations
Example 5'12 cl.
;;-;;f;;;;"; and co-classification tet*een them' With of both a specific
i ai"* to the fact that even in the absence t-.
"tt""tion and any situational
il;;i;,i;;"f;r""t clues^, there are occasions when
ffi;*tbl" to provide an interpretation oflargely the implicit device. I went tr
.,n io ,.rssest th;t both these things happen because of the s.eTan-
i* iir""iffi*'-ri"tuio"A-ir,rough t"*iiil ties. ln a text of non-minimal
,ir", th"r" normally occur many such threads of semantic relation' and
the
their simultaneous operation ijimportant in the resolution of
both
d
uUor" ptoUf"-s- The moral from this is easy to draw: to-be.effective' Ir
;;"-;;;i."1 cohesion lequirgslhe5gpgort of lexical to[:t'.o.1 , of
However,therelationshipisnotsoone-sided:tobeetlecttve'Iexl- th(
cA coilision, i" it. t"*, i"q"ires the support of
grammatical cohesion'
is esiential, as can be
rq
i;;;".iprr;ity of thes; tw; kinds of iohesion as
.e"., fto* ExamPles 5.16 and 5.17'

82
 Example 5.16
John gets up
next t; ni". - -"lY' we bought him a tie. He
ei
loves peaches.
My house is
Example S.l7
,;HT'i,yJ":ffi
f ,,"f
ff :Al?:i:,ffi:ifi enrerswork
In Example 5.Ir
#:?,i#ff l;8ff
ft:r srammar
1e, _him, ana ni.lllj: that wou rrr prevenr
11 1,.,*",; #,:;##fltlditicar.reason
#; jlHi;::,,,*l*;;T;l;flffi iJ: j*g*:ifi:!,i;Zt'!:i,!;r,

*r,ffiffiffi

*ffim'**rup1gpgf eep7o

gg*firy*ffii***r#
cl. 1 cl. 2
cl. 3
c[. 4
she went walk cI .5

*. go, tom. rt.f,

**x+:ru,nfi*-risffir;+ffi
jfrdltii.;q{l**;liil,[h]1',:ffi
f
,.norrr.i"', $i,Tl#$
##l:lfl

83
 te5
atenessofthemetaphor.threadsofcontinuityrunningthroughout.the
;;;;'; desc.iUe tt e simuiianeous operation of many cohesive chains, b€

each of which suppo.tt ;A refinei the domain of meaning for the ra'
of language' we hardly
others. This is on. ..u*o*hy, in natural uses an
ever notice ambiguities. b€
is conEsrvr
cohesive chains A technical term that has appeared in this discussion in Figure ite
cHArN^. wt ut i, u .ot.ri* .iiui"i er the analysis nrovi_dg{
which is related
i.: ,t o*r, a chain is fot..,ea by a set of items each ofco-classification' rh,
pl
to the others by the semantic reiation of co-reference' tir
into account' we can
u"a7ot co-extension. Taking the type of relation
two types: CHAINS and srullenIrv
il;A;ise ctrains into IDENTITY
Figure.5'3' Thus chain idr
cHAINS. Again, Uott oiJt'ete ate ixemplified.in
ifiuit' The relation between the ra
an identitY chain I with girl, she,.t.. it'u"'iit'iitv every member of of
members of an identitv .tt"i" it tt'at of co-reference:
or.whatever' as in this chain' is
the chain refers to tha ,u-a thing, event,
it.* r.t.rr"io tt" t#. girl. This particular identity chain ge
where each
i;;;;-;;h;ritiu., l...lii""t fto* tiecharacteristic
beginning to theend
of short "11-9::.t'
narratrves:
an
m(
ifris, t would suggest tentatively, is a
text-exhaustive fo
texts of this categorr-,"tt'"ffV contain at
least one
an
identity chain.
which is provid-
Now, turning to similarity chains, an example of
sel
a similarity chain members of a if
ea Uv ctain z iniig"i. i.i*ittt went, walk' etc': the
to each other either by co-classification or to
similarity chain are retated
that refer to non- thr
.o-.*i.niio.r. Each such chain is made up of items etc'' or to mem- ch
il;;;:J;;bers of the same class of things' of
events'
things' events'. etc' gr(
t.tt
- --- oJ
"on-identical
but related classes
and similarity chains is important' in!
The distinction between identity
Let take the iden- ho
relating both to the n;iio; of text and of context'
us
same'thing' ch
;t,;.#; fl;sr. Each ii.* l" an identity chain refers to the
of
word'thing; should be interpreted as covering airy cldssthe
i;i;;;ih. p"iaao*icaitv,-t
)J.r..ii. o*.r.., the extra-linguistic identity of I
to
develop this point a little. while
ifi;i;-i;aterial to l.rtu... Let me items wt
I
as I' me' my' These make of
;;ff"; ittis cttapte, r iru,. used such referring to the same extra-
I
cel
an idJntity chain, each item in the chain
this text' Ru-
iinguitti. iting: Ruqalya Hasan'-No*-,.independent of
rh,
at'Mac- ch
qul"vu ffuru" iittre same person who rvill be talking to.students
time. r find that it is not possible to ral
;;;;u;il;iiitv ir, "-i&-*..r,, as'l find ' 1l-t1-T:tl?.1"'"" l9
give talks without,r.h.tptt"ions
and.inmyopinion...,.Theseexpressionswerepresentinmyearlier of
probably be present
talks, they are present today, and they will most
what I am int
i1 in future talks as well. I am-sure that you can anticipate
extra-
of 'referring to-the same 5.1
* about to say: if we take the criterion
first person
1i
1l
il;;irii. th1", Iteraily, then all of these variants of the Tc
* an.identity be
,in"grru. pro"Iun *iliiorm but one identity chain.-Such and
.#;;"y definitely have uses in the constiuction of biographies co
qrit. far notions of textual unity Ta
case histories, but it i. useless so as
and textual identity ur. .on.""'"d' So we come
up with the rather in- an

84
'ri

\.
 teresilng con

oen ti y .rr di
i t
ratron are comt
ain r: i? ffiif:*i,flff:jj rreren ce bet
ween si m

fi rr"r,#-'
,'.r:;ir*+*;ri*,t#i*i,;dtlftir,+ffi
,,a"il.",,"",'aTf
fi

3,T,lf;f i;:,ei'iffi:1,:'."iti*,::i:if*tr,.,,m.
il:ffi :i'Jfl ,l#fl :,#:l;iff i:r-1i"*'i''l;H:[1;J,T:'i
ua,o*,"".*ilTfiy;gi:',Jt,"::Ifi?,T;::,r*:i:l..X,i,,ro.},-" i
t

,,
,Til:?x;$.,.rtt,s,ir,"y-ffi jl#i::f"*qrr,'i*is.,'le-specinc
j:
:i
I
tI
:"':i; iffi, s#i.
,
ft iit',T:?T:H li lffi r::;il
c gro pin g *,,,
*J'}# .t
J
I
I
:!:t i, ir, i"*1''
t
u
",I"lli"i'.| I : :'ffi j "j"r. n rro m
ti o .l
fl :. I
"" J,

#t.{
t, I t
ii;, f

ffi *i*it*i;I.,

85
 ,
;
I

:
The texture of Texts 5.1 and 5.2 wl
I In
Look again at Texts 5.1 and 5.2. thi
see p.70, 7l Whenever I have presented these two texts to informants, they have rq
unanimously agreed that Text 5.2 is less coherent than 5.1 (which is atr
not to say that 5.2 is a non-text). An explanation of what this judg- ir
ment correlates with in patterns of texture is difficult to find, so long im
as grammatical and lexical cohesion are examined separately. You will isr
probably be surprised to learn that the number of grammatical cohe- he
sive devices in the two texts is identical as shown in Tables 5.2 and 5.3. inl
trt5
Table 5.2 Grammatical cohesive devices in Text 5.1 thc
2. she 3. she thc
4. she it 5. she she it tha
6. she it her she it 7. she
5.2
8. she SE it the his 9. SE her
10. she the ll. the she it
tho
12. she it it 13. she +the***
8ra
Table 5.3 Grammatical cohesive devices in Text 5.2 etul
l. *the *the 2. *he
sagt
3. the +the +the 4. they? +the +the itetr
5. the it 6. they? the the diEr
7. they? him 8. SE? him +the the
9. SE? it +the 10. the his
l9&
I l. the the the 12. the their? the
The
As is obvious from Tables 5.2 and 5.3, the texts do not differ cru-
cially in the frequency of grammatical cohesive devices; nor do they Trb
differ greatly in the patterns of lexical selection, or even in the propor- t-
tion of devices that are subs'med 11 sfiains. Table 5.4 presents some 2-
facts regarding grammatical and lexical cohesion in the two texts. 3-
4-
Table 5.4 Grammatical end lexical c.ohesive devices in Texts 5.1 5.
and 5.2 6-
Text 5.1 Text 5.2 7-
8.
l. grammatical cohesive devices ,30 30
2. frequency of I per clause 2.3 2.5
9-
3. percentage of I entering in chains 97t 93
10.
4. explicit lexical tokens 47 37
ll.
5. cohesively interpreted lexical tokens 27 30
12.
6- total lexical tokens 74 67
r3.
7. 5 as percent4fe df 6 36 41
8. percentage of I interpreted
Tabl
anaphorically 97 60
9. percentage of I interpreted t.
exophorically -1 27 2.
10. percentage of I interpreted 3.
4mbiguous l3 4.
5.
SE = subject ellipsis Let me first gloss the unfamiliar terms. SE in Tables 5.2 and 5.3 6.
stands for subject ellipsis; the first example occurs in message 8 of Text 7-
5.1: 8.
when she got up 9.
and [SE] combed it 10.
11.
12.

86
*

where SE will be interpreted as she, i.e. (aforementioned)

rittte girr.
In message 13 of Text S.l, the has several asterisks
1*; attacneJto it;
this.is to sensitise you to the fact that it occurs in a semi-fixed
exp-
ression all the rest. In Table 5-3, severar grammatical
devices have
an.interrogative (?) or a cross (+) attached to them; tt . ior*"i
indicate that the interpretation of'these is problemati.-tt.v.o"ra
i, to
u"
interpreted in more than one way; the .ror. i, to indicate
that the device
is exophoric. Table 5.4 (line 3) presents percentages
of grammaticar co-
hesive-d:vices entering in chains; theseihains are formal
as described
iiHalti$av * t{3rq, 09i6) and do not necessarity correspon;i;i;.r_
tlty and,/or srmirarity chains. The term 'explicit lexical token, refers
to
the content words in the texts, which appear as content
words from
the start; by contrast 'cohesively interpreted lexical tot.nr'-ui"
tt or.
that are arrived at when the grammaticar cohesive devices
toi ruLr.,
5.2 and 5.3) are interpreted.
Tables 5.5 and 5.6 present the-total picture. In these two
tables,
those lexical items are underlined that ariihe inte.p.eiatio;;i;;;.
grammatical cohesive,device. For example, in
Tabie S.S, m.ssage Z,
glrl is underlined; this lexical item is the inteipret ation
of she from"mes_
sage 2 of rext 5.1: she went out
for a wali. you will note that some
items are underlined with broken lines; each of these is
a noun mo-
dified by the.For example, Text 5.2, line l, reads: the soilor
the ship. Given the meaning of the irrariaay & Hasan rgioiHasan iirio,
1984c), the modified noun refers to a uniquely identified
The cross (+) marks exophorically intirpieted the.
Table 5.5 Lexical rendering-Text 5.1
tt"i.o inr.g(0.
t
l Iittle girl was
2. girl went walk
J. girl saw lovely little teddybear
4. irl took teddybear home
gril
5. glrl eot home girl-washed teddybear
6. girl to.ok-to-bed teddybear girl girliuddled teddybear
-

7. glrl letl-to-sleep straight

8. girl gotup girl combed teddybear little wirebrush
teddybear opened-eyes
!eggyqcer
9. teddybear started speak girl
10. gir.l had LeqrylBI many-rnany years weeks
ll. LedQy_bgef got dirty girl washed teddybear
12. girl brushed E44ybegl-tg44ylggl;ay:ome new words
different country
13. girl know speat fnefish S6ttish u11-*11r"-r.r1***

Table 5.6 Lexical rendering-Text 5.2

l. * tetlpt go + ship
2. sailor come h-olne dog
3. dog want +boy +girl
1. railor lqy g[-dgg-tno* * beal was * chair
). bear come go-to-sleep chair
9. sailor glsg Ig! find-bear chair
7 sailor glqg !9y
lgy g!1! wate--up-U-e-ar
sailor glsg boy gtt! chuck_out bear * room
!. sailor
?. sailor qsg lgy gI4 take bear *-Ioo
10. take-off sailor hat-
I l. aog chased beaTloom
12. Eof sir sailoi !g61-!o:f girl chair bear sleep

87
 Returning now to Table 5.4, line 7 shows what percentage of the
total lexical tokens is arrived at through the interpretation of the gram-
matical cohesive devices. So far the differences between Text 5.1 and
5.2have been statistically insignificant, but the last three entries ap-
pear different. Of the grammatical cohesive devices of Text 5' 1, 97 per
Text 5.1 is highly ur" anaphorically interpretable. This means that the text is highly
self-sufficient ".nt
self-sufficient; to understand the speaker's meanings, one needs simply
to know the English language. Not so with Text 5.2, where 40 per cent
of the devices cannot be interpreted by reference to the text;27 per
Question 2, p.73 cent are exophorically interpretable while l3 per cent are ambiguous.
We are now in a position to revive question 2 raised at the begin-
'Ifthe two varY in ning of this chapter, rephrasing it, in the light of our findings, as fol-
the degree of lows: can the listener's perception of varying degrees of coherence
coherence, what if between Texts 5 . I and 5 .2 be correlated with the differences in texture
any patterns of indicated in the last three entries of Table 5.4?
language correlate
with this variation?'
Texture and textual coherence
Exophora I
(
see 'Cohesive There can be no unequivocal answer to the question raised above. I
have argued above thit although exophora reduces the possibility of
I
interpretation and t
cohesive tie', P.77 interpretation, it does not necessarily prevent the formation of cohe- (
sive ties; and to this extent it does not militate against texture, p^articu- I
larly if we find that relations of co-reference and,/or co-classification t
are not being adversely affected by the presence of exophora. What is C
the position with regard to Text 5.2? t
the origins of Texts Here the history of the data is relevant. These stories were collect-
t
5.1 and 5.2 ed in Bernstein's Sociological Research Unit (University of London) 4
in the mid-1960s from children who were asked to tell a bedtime story
to a teddybear about a sailor, a dog, a boy, and a girl' All five charac- t
ters werspresented in toy form to the children. Thus the meaning of t
the in the sailo,, type of phrase was clear to both participants. Moreover, C
in all cases the exophoric device is the.ln a group such as the soilor, n
someone who does not know the history of the data is likely to ask: d
which sailor? However, it is doubtful that the absence of an answer li
to this question will make the reader perceive Text 5.2 as less coherent, ir
especially since the co-referential link between the suilor of line I and c
th-e sailor of line 10 does not appear to be in question. There are al-
tl
together eight occurrences of exophotic the: in Tables 5.3 and 5.6, items n
with cross mark ( + ) where a new referent is introduced ex- c
ophorically, for example the soilor, the ship (line l), the boy, the girl
(line 3). Of these only the sftrp (line l), the room (line 8), and the zoo
(line 9) did not appear in the instruction given to the child. The ex-
ophora of the zoo is a formal exophora (Hasan 1984c) which is the
(
leist opaque of the exophorics; the ship and the roorn become less I
problematic because of the semantic relation between sailor and ship tl
and home and room.If Text 5.2 is perceived as less coherent than Text o'
5. 1 , the reason cannot lie in the variation of exophorically interpreted si
grammatical devices.

88
 tl

#f4gg,ir{figgfr,1ryfrffi;,1*r,l
t;d
#;ttf *,"*. *,;
89
see p.82 of ambiguity under focus. The discussion regarding 'The interdepen-
dence oigrammatical and lexical cohesion'also carries the same impli-
cation. Iimight therefore be illuminating to look into the identity and
simitarity chiins formed in the two texts. Perhaps this examination
see 'The inter- would ui on"" provide an explanation for the occurrence of ambiguity
dependence of and reduced coherence. Table 5.7 presents the chains from Text 5.1.
grammatical and
lexical cohesion', Tabte 5.7 Cohesive chains in Text 5.1
p.82 Identity chains: (a) girl (17)
(b) teddYbear (14)
(c) home (2)
Similarity chains: (d) was got (= became)
(e) went walk got (: reached)
(0 lovelY dirtY
(g) wash (2) comb brush
(h) took had (: owned)
(i) weeks Years
0) manY (2) some
(k) new diflerent
0) sPeak (2) saY
(m) took-to-bed fell-to-sleep got-up
opened-eyes
(n) words English Scottish all-the-resl
(o) little (3)

The numbers in brackets show how many tokens of the lexical unit
occurred in the text. of the total tokens, 90.5 per cent.(67 out of 74)
are subsumed in cohesive chains. How does this compare with Text 5.2?
Here we face a problem. It is difrcult to decide what goes into the
identity chains and what the total set of lexical totens for Text 5.2 is,
without resolving the ambiguity. One solution is to ignore those iden-
tity chains relating to sailor, boy, girl, or dog, and examine the rest
of the text. In thaicase, our findings will be as displayed in Table 5.8.
Table 5.8 Cohesive chairs in Text 5-2
Identity chains: (a) bear (8)
(b) chair (4)
Similarity chains: (c) come (2) go take
(d) go-te'sleep wake-uP sleeP
(e) fnd chase-out chuck-out
(f) home room (2)
When sailor, dog, boy, and girl are ignored, the total number of
lexical tokens in Text 5-2 is reduced to 33; of these 25 are subsumed
into chains. By comparison with Text 5. I , only 76 per cent of the tokens
are in cohesive chains. In respect of cohesive chains then there seems
to be a signfficant difference between Texts 5.1 and 5.2. But what is
the interpretation of this difference?
One obvious interpretation is that lexical selections in Text 5.2 do
not divide themselves into a homogeneous set of semantic groupings.
The fairly large percentage oftokens that fall outside chains-i.e. are
pERFmRAl-prevent a consistent reconstitution of the field of the text.
This can then be seen as part of the reason why Text 5.2 coheres less
well than Text 5.1.

90
 Chain inleracfion
Convincins

gruff*ff*mmffi
Exarnpte
---'r q usr such
as follows.
S.lg

itr*iinn#r:[g;;l;:r""*'

i,itl*;l;i]Hi:1;mt:r,fl :}#.,#il,"Tij#
.*;::":.ff H,;;.Tff llT.;:il#.,r*l*":,***,r*,..
lilfifi i#*#ili,*':itTxruu::i,:r;:,",:::!;:Iiiirii
;t:#inT.,m*l*i:ffi r #;fu rr,:, ;;i!T,
*'l**:d{}t**[ffiffiffiffi
*ffmgg*ffigffi$1
i ;il :ffJ;i ; :T*i::.^T-:T
:romtfu,'ffi
ff ',"99dq";, f i..,11?,,,i.;1":,: ,*?*t
l:.,",.

ff*#fifi**rfu'rfffffi
Ji,fi;
L' a'I n
y:*"u..,,r,li
torma'i""
i"i ".#,e,i g
lil,{f *,. #l*fi,i
ffi
91
 would be a measure of chain interaction. But this is surely wrong rrte
since a random list of clauses or groups would not necessarily be
coherent; nor does chaining entail coherence (see discussion of Ex-
ample 5.18 above).
2.The second reason is deeper still. The recurrence of a relation be-
tween two chains is indicative of two vectors of unity. The first vec-
tor of unity is indicated by the semantic similarity that permits
members to be part of the same chain; the second vector of unity
indicates the semantic similarity that unites at least pairs of mem-
bers from two chains. The rationale for this is simple to find: in a
coherent text one says similar kinds of things about similar phenome- q
na. For example, the girl in Text 5.1 does not simply go home, she
also gets home; she does not simply fall asleep, she also wakes up, I;
and so on. L:
When the text is not too long, the chain interaction within it can is di
be visually displayed. This visual display highlights the continuities and ofd
the discontinuities in the text. Figures 5.4 and 5.5 display the chain inter- mort
aqtion in Texts 5.1 and 5.2 respectively. ier ti
Figure 5.4 Chain interaction in Text 5.1 furet
(a) (e) wift
hd,1
I
ence;
Any I
ia
la
lll at
took-to-bed h
fell-to-sleep (m) iv al
got-uP
var
Those
ii
app€n
AS CEI
CEITrI
tokens
l. Rd
_cta
I (a).
o)
2. Pq
ofr
Each rectangle in these figures represents a (part of a) chain; the Having
chain labels used here are the same as in Tables 5.7 and 5.8. If Figure state fa
5.4 is compared with Table 5.7, you will see that (a) gil contains 17 herenq
members, though the rectangle (a) in.Figure 5.4 contains only ll of l. The
these: this is because only I I of the 17 members of chain (a) qualify on6
as interacting with some other chain(s). Thus although the rectangles 5. I,
bear chain labels, they need not represent complete chains. When there 5.2,,

92
 Figure 5.5 Chain interacfion
in Text S.2
(a)

If";
I u.r,
r+_l I

1/ffi: /r

@/
js chain
interaction rrrra i*^_

ffid,1;fri*d#$#l*niffi:i-jtffi
I;tr[ji"T,:i:::1f iiff i:?;.,,,i.i,nr.i#Hfi .i;i:.,:ff
:
ence. rrrei'"# ;.: HXX:i,#'r:?i"i::man number to arow
easy rerer_
1ry t*o chains Ir

j pfl+ffi d*#.trud:fft,Ti,,t H !ii;r*:;;t,

are.in 'saying ' -' "'rcrrrprc' f
]." u.. text, rcrqti^^ /c_
: i, .attiib;:":l,l;:?1.%::,;11lrl. , said words)

#,xT,,'nH**f,liIffi
:ttr#,:iffi
AS CENTRAI, ,^--,,
cENrRAr.
r.oKens
w;fir;
of a text:

mmffimffiffifil
fiif#l*"'1,1*rffTi'J#Tfl
il - "
"i#:ri:'
,{{##i'ui ,u*i

93
 the central tokens to the non-central
1
2. The higher the proportion of tokens
ones, the *o.. .ontl"'ii^ttt
it.irt likelyto be' Theofcentral
th9 relevant h
per
of rext 5.1 (see Fi;;;;'i';i ton-ttitutt.65 ^cent
figur. is onlv 36 per cent' e
tokens while for rliiJ.z, ihir
pi"tuit of interaction' the more coher- is
3- The fewer the bt";; il t-ht itrteracting chains is relat- s
In eigull;.;,d;;r* set of
ent the text. of s
as rocer CHATNS' each
. ed, with chains ("i';;;'iui'i'nttioning s
focal chains ;;;;"t of-other chains' [n Figure 5'5'
which interacts wrth'a ia;;; a
there is a clear break' amounts d
above are ordered' The first
The three features mentioned p
tJ;;i;;;;;re itt^ttt" text should be such as to
to saying that the i!
otnlli'i referential domains' lf and
when
establish un.qui,ocaltv;;'-"fi of a text will fall within
s
this happens, ttre majlr"iivt;il;i;;;iiort""t is a necessarv condi-
tl
chains, leaving o't b;;;i ilJ;;tfr;;
few' This
to textual unitv'
essential
si
tion for the second;;iil;lf;tt*g1'thtrs
*t'i".rt tr'" tan"" of coherence is
built'
and cohesion i, tht fJ;#;; "' t
but not sufficient by itself'
Like all foundationsliiit'""tttary tl
Thesecondstatem;n;;;;;;iti"theclaimthatsimplvtheestab-
is not enough. Identitv and
t
lishment of the definitili.;;;aiaomains components alone- such U

similarity should .rot L"Ji.it"dto.message

formation; ihe notions of identi-
fi
identity and similarity '-'J"tii" ttt"in to the content of the message
tl
ty and similarity tn"'iri "i*'u" "*iioato
when speakers'are enea8e'd^ll'h' t
as message. r, similar n
Drocess
"o*i'J'ipu'rt*t'
of creating , i.'t.'r"r,text, they stav with the same and
in which I
t@"t1n" of affairs are
ihines long enough t' tlf-*-i'"* states
n
ileslsim; unaii-itut things are implicated' I
ih9 process of creating coherent
The third *u,"*lni'A'il-s thal between the things one ts
a

texts involve, u., ira'i'J#o-"1?i.rutionships break, i" (

'on about'. rn. o"t"o?"; tt'"i
; ;*pleie :-hT, ]1t:i::9i i
place:1r*titi"" from one topic to the next is a mergtng
does not take
rather than a clear boundarY' a
three phenomena as coHESIvE
cohesive harmonY
I have referred;il;"- of these could be formulated
t
HARMoNY; una u u'Ll#-"iJt uuout coherence t
thus: (
of variation in the cohesive
variation in coherence is the function I
harmonY of a text' I
it brings together lexical t
It is harmony in more than one respect: them to semantic con- i
and grammati"'f toi'L'in" atui"tt'
subjectine
io.oiili*iii"iir-"ri,v.this iJ as it should be; a text' J
I
siderarions or
after all, is not a
but of meaning' Secondly' it is'harmony I
"Jililtt- t"tp* 9f-two microfunctions: the
because it t armoJset in" the
textual
chains
(

.and the experientii"]it ."ip'", ofthe textual function are I

" thecxperiential function at the rank i
and the infeiactio#; ,irr"-..,,piri.r is built upon' Thus cohe-
of clause and grooi it *t'"i the interactionfunctions find their expres- I
, sive harmonv it
"ti;;;;;;fil"*1#two doubt' the concept of cohesive
sion in one signifiiiiJt;;1"' No
(

uv uringine in the loeical and interper-

harmony can be dffi;; *n""a that
sonal functio", i#;;; ;i"i"tt'
ir tnit tu-n u" done' it will show concept.
murtirunctionality is a fruitfur
even where text is Joncer-ned,

94
 Texfure, cohe,
and the teacher
In recent
r"u.r. *^-*nce'
,"J3il,il;i1;i;fffi dr,",?ln:iB'xJ1T:1l"^lri.approachFor
ir,ir*.'i'n,?1i1fl:: tt " u'rr,,pli-o; ;':?'3i:: 11901 t[ri .ol-...r."
'rv
;xff ,',:x;f"yti{fi ;J'lX;:*,'Xl.J:'i1'{qTff lil:,i,ti;I".,
jflx'j:riljfl :ffilff :ilH#xI.:,:ifl ffill'#iiilt,l;:.
;,,i]:I " of some wil; ;:,f:.rr.h
un urrrmptio, rrsrener?
;r":,1;:l ol
lyglText ttrc cose ws to
i"::^:"1:
tsSUeS Of terfrr,o
s.zz o,,^",ilu.
\<qsstlOnS Of
fh;c Li,
,r"riJJ _rritve
^-r-'
,Tsi:,.Tt"T:H,,:i;iL:T",il.,:R#i{Tslunt";;#;:tiil
i[:H{{lirLHt*:.il'Jil';iT.'"'"1t##fi
,*?*I}:qlf

II **[+.'i:l[g*;',r,,,*tlli:l*ii#d;]#,xn;:n,,*.
i:{Jf#lqf;x.:,:*#}[:tl**i-liit'+;rr,t*
t i:ififril";F*'::,itffiil*,1#$#*isru
i, merit. And i,
""r-I
j:1" rl..discourse urirT^ill"qent. He or she_
worku.,r,.r,,jll;i1r,irqi;r;";i:li:,T.f.,,:oT"r,
ar rhe meanir"::,::,:^ooubt.wished;L:,1
relations-in"ruairr*,,t^t"o,y:l[,
.;[i; its own
:hl ,n. lirlriron
,il; ffi#rle does not
,ll
gnry do this bJ
ft,l;:Tf,3ff :[,H,,Xffi,,j:flXH'+ff iil:f fi
llft ff [TX,
,*:'"'**fr i,:j.^o-:*il:xi.x#:x,ff :;I::jTtiymuu,ismuton-
l:i'ilfi il;fl :;11T9ifl,":i:F:,,ffi:?1T:S:'r"J,,.v""on-
.u,,..or,r,.i..i*a"wT,lffi
,;:ffj",:lfl ll:i,"",.r,i.Iii,Tf :m

fiu,,smffmffim*fffififfi
rffiilffi,:e1ffi,mffi,*l'::ffi
:3n#iihl",i"xri;*,#iffir"#,tr;;i.HTu,U*:ff
jjffi

95
 to put in 60 per cent of pronominals , 20 per cent of definite articles,
3 causal relations, and by making the lexis hang together in chains. The
cohesive devices create texture because they establish relations of mean-
ing. The incoherence of discourse is often a pointer to an inuuiiiiv to
organise the relevant meanings in relation to each other. A teacher can
assist by pointing out what semantic consequence the choice of a par-
ticular pattern of wording has; for example, what difference of *iun-
ing there is between the following: select o tube and put it in the bottte
and select a tube and put one in the bottle.It is thise kinds of deep
semantic differences that the mere assumption of coherence witi not
and does not handle. The infra-structure of all assumptions about co-
operative acts of doing and saying is, in the last resori, social. The as-
sumption of coherence can be sustained so well because human lan-
guage has the resource for indicating coherence, while the nature of
language as a resource has developed in a particurar way because it has
had to serve the needs of the community. our task ii to understand
the specific nature of these resources-not simply to hide behind the
mind and the intention of particular speakers and listeners.

I
I
I
t
1
F
I
h

I
7
Il
lo
alr
qu
w(

96
:---
.l.l- i l:::=: