LACUS
FORUM
XXV
© 2009 The Linguistic Association of Canada and the United States (lacus).
The content of this article is from lacus Forum 25 (published 1999). This article and others
from this volume may be found on the Internet at http://www.lacus.org/volumes/25.
YOUR RIGHTS
This electronic copy is provided free of charge with no implied warranty. It is made available
to you under the terms of the Creative Commons Attribution-NonCommercial license
version 3.0
(http://creativecommons.org/licenses/by-nc/3.0/)
Under this license you are free:
•
•
to Share — to copy, distribute and transmit the work
to Remix — to adapt the work
Under the following conditions:
•
•
Attribution — You must attribute the work in the manner specified by the author
or licensor (but not in any way that suggests that they endorse you or your use of the
work).
Noncommercial — You may not use this work for commercial purposes.
With the understanding that:
•
•
Waiver — Any of the above conditions can be waived if you get permission from the
copyright holder.
Other Rights — In no way are any of the following rights affected by the license:
• Your fair dealing or fair use rights;
• The author's moral rights;
• Rights other persons may have either in the work itself or in how the work is
used, such as publicity or privacy rights.
Notice: For any reuse or distribution, you must make clear to others the license terms of
this work. The best way to do this is with a link to the web page cited above.
For inquiries concerning commercial use of this work, please visit
http://www.lacus.org/volumes/republication
Cover: The front cover of this document is licensed under the Creative Commons
Attribution-No Derivative Works 3.0 license (http://creativecommons.org/licenses/bynd/3.0/) and may not be altered in any fashion. The lacus “lakes” logo and Claremont
Graduate University logo on the cover are trademarks of lacus and Claremont Graduate
University respectively. The Claremont Graduate University logo is used here with
permission from the trademark holder. No license for use of these trademarks outside of
redistribution of this exact file is granted. These trademarks may not be included in any
adaptation of this work.
FUNCTIONAL SENTENCE PERSPECTIVE, FOCUS OF
INFORMATION AND SEMANTIC RELATIONS
Michael Cummings
York University, Toronto
1. The problem. Katherine Mansfield’s short story ‘At the Bay’ (1945:205–45) is a
wistful account of a summer’s day in the lives of New Zealanders on holiday.
Various members of two neighboring families interact while reflecting on freedom
and constraint, affection and communication, the beauty and seeming indifference
of nature. Jan Firbas takes up a segment of this text (Figure 1) in his book on
Functional Sentence Perspective (1992), finding that it offers a useful variety of
semantic roles in clause structures for illustrating the dynamics of information. The
same segment is considered in this paper because the Firbas treatment offers a
ready-made and authoritative example of Functional Sentence Perspective (FSP)
analysis which can be usefully compared with other procedures for analyzing the
contrast in clause structures between ‘given’ and ‘new’ information.
In FSP terms, this contrast is a distinction between Theme and Rheme in sentence and clause structures, with Theme as a label for the elements which advance
the flow of information least, and Rheme as a label for the elements which advance
its progress most. Within the terminology of the Systemic-Functional theory of
Focus of Information (fi), a similar distinction is made between given and new
information, with given as the label for the elements of clause which show the
greatest continuity with the preceding discourse, and new as the label for the elements which the speaker of utterances focuses on as providing most novelty or contrast with preceding discourse. Neither of these approaches is primarily oriented to
the structure of discourse, however. FSP accepts context dependency as a constraint,
but relies primarily on clause semantics. fi explicitly identifies the stretch of new
information in clause structure, in whole or in part, without developing a systematic description of the given information as such.
One approach more specific to discourse would be to examine types of semantic relations among elements of clause structure, and to measure relative givenness
or newness of such elements from the nature of their relationships with preceding
elements. For example, iteration could be taken to represent unqualified continuity, and absolute variation could be taken to represent unqualified novelty. Such a
procedure could avoid some of the indeterminates detectable in FSP, such as the
identification of semantic roles, and could add to the descriptive power of the fi
treatment of given information. The main purpose of this paper is to hypothesize
192
Michael Cummings
(1) In a steamer chair, under a manuka tree that grew in the middle of the front grass patch, Linda
Burnell dreamed the morning away. (2) She did nothing. (3) She looked up at the dark, close, dry
leaves of the manuka, at the chinks of blue between, (4) and now and again a tiny yellowish flower
dropped on her. (5) Pretty—(6) yes, if you held one of those flowers on the palm of your hand and
looked at it closely, it was an exquisite small thing. (7) Each pale yellow petal shone as if each was the
careful work of a loving hand. (8) The tiny tongue in the centre gave it the shape of a bell. (9) And
when you turned it over the outside was a deep bronze colour. (10) But as soon as they flowered, they
fell and were scattered. (11) You brushed them off your frock as you talked; (12) the horrid little things
got caught in one's hair. (13) Why, then, flower at all? (14) Who takes the trouble—or the joy—to
make all these things that are wasted, wasted… (15) It was uncanny.
(16) On the grass beside her, lying between two pillows, was the boy. (17) Sound asleep he lay, his
head turned away from his mother. (18) His fine dark hair looked more like a shadow than like real
hair, (19) but his ear was a bright, deep coral. (20) Linda clasped her hands above her head (21) and
crossed her feet. (22) It was very pleasant to know that all these bungalows were empty, that everybody
was down on the beach, out of sight, out of hearing. (23) She had the garden to herself; (24) she was
alone.
(25) Dazzling white the picotees shone; (26) the golden-eyed marigold glittered; (27) the nasturtiums wreathed the veranda poles in green and gold flame. (28) If only one had time to look at these
flowers long enough, time to get over the sense of novelty and strangeness, time to know them! (29)
But as soon as one paused to part the petals, to discover the underside of the leaf, along came Life (30)
and one was swept away. (31) And, lying in her cane chair, Linda felt so light; (32) she felt like a leaf.
(33) Along came Life like a wind (34) and she was seized and shaken; (35) she had to go. (36) Oh dear,
would it always be so? (37) was there no escape?
(38) Now she sat on the veranda of their Tasmanian home, leaning against her father's knee. (39)
And he promised, (40) 'As soon as you and I are old enough, Linny, we'll cut off somewhere, we'll
escape. (41) Two boys together. (42) I have a fancy I'd like to sail up a river in China.' (43) Linda saw
that river, very wide, covered with little rafts and boats. (44) She saw the yellow hats of the boatmen
(45) and she heard their high, thin voices as they called… (Firbas 1992:26–27)¹
Figure 1. Annotated extract from Katherine Mansfield, ‘At the Bay’.
how the results of such a procedure would compare with the results of FSP and fi
analyses. Another purpose will be served by comparing the results of this procedure
with that obtained by a computational analysis of semantic relations in text
(Cummings 1995, 1996, forthcoming).
2. Functional Sentence Perspective Analysis. FSP recognizes three factors in
the assignment of information status—context dependency, semantics, and linearity—prioritized in just that order (Firbas 1986:43–47). Context dependency means
semantic continuity with elements from the preceding discourse, and is frequently
realized by reference items like anaphoric pronouns, or by lexical iteration. Where
context dependency is not observed, semantic roles ordered on a scale of increasing
functional sentence perspective, focus of information…
193
informational novelty are assigned, such that a Setting (Set) is least informationally
dynamic, and a Further Specification (FSp) is most:
Set<Pr<Ph<B<AofQ<Q<Sp<FSp
Setting
Presentation
Phenomenon
Quality Bearer
Ascribing of Quality
Quality
Specification
Further Specification
Set
Pr
Ph
B
AofQ
Q
Sp
FSp
Where neither context dependency nor semantics is at issue, then simple left to
right linearity is taken to realize increase in informational dynamism (Firbas
1992:21–40, 66–74). Context dependency cannot simply be equated with reference
or lexical cohesion, since anaphoric pronouns or lexical iterations sometimes realize
items which are seen to be entirely new within the flow of information (ibid:31–37).
Information status in clause has its own notation. Least dynamic elements
belong to the Theme stretch, and most dynamic to the Rheme. Between these is the
Transition. Degrees of dynamism within these categories are distinguished such
that dynamism increases across the categories Transition Proper, Transition Proper
oriented, and Transition; and across the categories Rheme and Rheme Proper:
Theme
Transition proper
Transition proper oriented
Transition
Rheme
Rheme proper
Th
TrPr
TrPro
Tr
Rh
RhPr
Additional notation in the formal analysis includes:
context dependence
linking and boundary function
d
+
To demonstrate fsp analysis, sentence (6) from Figure 1 is selected because its
clause elements show a useful variety of semantic types:
194
Michael Cummings
(6)
yes, if you held one of those flowers on the palm of your hand and
looked at it closely (Set, Th), it (B,d, Th) was (+, TrPr; AofQ, Tr) an
exquisite small thing (Q, RhPr).
[if (TrPro) you (B,d, Th) held (+, TrPr; Q, Tr) one of those flowers
(Sp, Rh) on the palm of your hand (FSp, RhPr)]
[and (TrPro) looked (+, TrPr; Q, Tr) at it (Set, d, Th) closely (Sp,
RhPr)]
(ibid:71–79)
The Theme elements in the sentence are semantically Setting and Bearer—one of
which, ‘it’ , has context dependency (d). The item ‘was’ actually realizes two informational categories, TrPr and Tr because it implies a ‘temporal and modal’ feature
(ibid:70–73) as well as an Ascribing of Quality. The Rheme (italicized), which consists solely of a RhPr, plays the role of Quality. The first Theme element includes
two embedded clauses, and the analysis of each of these is appended in brackets. In
the first embedded clause, the core semantic roles are Bearer of Quality realized by
Subject, Quality realized by Predicator, Specification realized by Complement, and
Further Specification realized by Adjunct, in an ascending order of dynamism. In
the second embedded clause, however, the Thematic Setting comes between the
Transition and the Rheme Proper.
3. Focus of Information analysis. In the theory of Focus of Information, priority is given to spoken text, so the analysis of written text is as if recited. Prosodically,
a text is constituted of roughly isochronous feet, each foot consisting of a stressed
syllable and possibly one or more unstressed syllables. Sometimes the stressed syllable is replaced by a silent ictus. Intonationally, a text is constituted of tone groups,
each tone group distinguished by an intonation contour and climaxed by a foot
whose stressed syllable is ‘tonic’, that is, bearing the most radical change in pitch
within the contour (Halliday 1967b:200–11, 1994:295–98).
Informationally, the tone group maps together with the ‘information unit’,
whose ‘focus’ is on the grammatical element distinguished by having the tonic syllable. The typical or unmarked place for the focus is in the last lexically realized element within the information unit, as in ‘I’d like that!’ (tonic syllable bolded). When
the tonic syllable falls elsewhere, the focus is marked, as in ‘I’d like that!’, which
shows the focus on a non-lexical element. Another instance of the marked focus is
when it falls upon a lexically-realized element which is other than final.
Grammatically, the clause or clause segment mapped together with the tone
group/information unit is divided linearly between an optional given segment and
a default new segment. An unmarked focus of information marks the end of the
new information segment, while the beginning is left indeterminate. A marked
functional sentence perspective, focus of information…
195
focus designates its grammatical element as the sole new information and projects
given status onto the rest (ibid).
An fi analysis of (6) recited might look like the following:
(6)
a.
b.
c.
d.
// yes, //
//. if you / held one of those / flowers on the / palm of your / hand //
// _ and / looked at it / closely, //
// _ it was an ex / quisite small thing. //
Notationally, each of four separate tone groups/information units is bracketed
with double slants. Feet are separated by single slants. Tonic syllables are bolded and
silent ictus is represented with a carot. Stress always occurs only on the first syllable
or as a silent ictus in each foot. The focus in each of the first three information units
is unmarked. The implication for the distribution of given/new information is that
the clause element terminating each of these information units also terminates the
stretch of new information in the clause or clause segment which is mapped
together with that information unit. However, the beginning point of the new
information in each information unit is left indeterminate. On textual grounds, systemicists would have no trouble finding that ‘if you’ in (6)b is a given information
segment. The fourth tone group/information unit shows a marked focus of information, distinguishing the modifier ‘exquisite’ alone as new information. Important
differences from the fsp analysis in the fi analysis include the binariness of the
given/new distinction, its confinement to clause or less inclusive grammatical unit,
the potential indeterminacy of boundaries, and the potential for distinguishing
given from new within grammatical units below the rank of clause.²
4. Semantic relations analysis. An analysis of semantic relations implicit among
various formal items of the text, regardless of sentence or clause structure, is the
basis for a third approach to the given/new distinction. The semantic categories
employed here are conventional in the literature on semantics (e.g., Lyons 1977:
291–95, 311–17), but the taxonomy to be presented is particularly indebted to Firbas
1992. Firbas (31–32) offers an ordered series of semantic categories as a basis for
explaining the perception of context dependence or independence in fsp. For the
semantic relations (sr) analysis of text, this series is here expanded to include
the following categories; the order from the most connected information to the
least is offered strictly as hypothesis:
Semantic category
iteration
substitution
close synonym
superordinate
Means of realization
lexical repetition
pronoun or deictic
lexical item
"
196
Michael Cummings
(Semantic category)
hyponym/meronymy
co-relation
loose synonym
mere association
variation
(Means of realization)
lexical item
"
"
"
"
This scale is implicitly taxonomic insofar as the first eight categories represent
association, and the last the opposite; moreover the first four represent reference,
and the other four associations do not. The first four categories have a high potential for suggesting given information status, and the remaining categories have a
progressively higher potential for suggesting new information status.
A semantic relations analysis of (6) would simply label formal items with the
semantic categories realized:
(6)
yes, if you (sub.) held (v.) one (mer.) of those flowers (it.) on the
palm (v.) of your (sub.) hand (mer.) and looked (it.) at it (sub.)
closely (v.), it (sub.) was an exquisite (v.) small (c.s.) thing (sup.).
key:
iteration
substitution
close synonym
superordinate
hyponym/meronymy
co-hyponym/co-meronym
loose synonym
mere association
variation
it.
sub.
c.s.
sup.
hyp./mer.
v.
Each of the semantic categories applied to lexical items or anaphoric substitutions represents a relation to at least one item previous in the text within a span of
30 items.³ Secondary relations (i.e., to another previous item and representing a
relation lower on the scale) are omitted. Items which can conveniently be thought
of as given information on this basis include ‘you’ (sub.), ‘flowers’ (it.), ‘your’ (sub.),
‘looked’ (it.), ‘it’ (sub.), ‘small’ (c.s.), and ‘thing’ (sup). A comparison with the fsp
and fi analyses shows significant variation and reinforcement among the three. All
agree that ‘you’ is given. However, the fsp and (by implication) fi analyses include
‘flowers’ and ‘your’ within stretches of new information. Only the fsp analysis
specifically agrees that ‘it’ in the second clause is given. The fi analysis of the last
clause would agree perfectly with sr, but the fsp analysis would agree only that ‘it’
is again given, including ‘small thing’ within new information.
functional sentence perspective, focus of information…
197
5. Computational analysis. A fourth method is a computational analysis,
Analyzing Semantic Connectivity in Texts (ascit). Like sr analysis, it equates
givenness with connectivity. However it tries to establish the relative connectivity of
each lexical item or lexical substitution in a text to each preceding item in a span,
thus treating givenness as a field effect dependent on the aggregate semantic relations of the nodal item. The semantic distance of that item from each of the preceding items in the span is computed from a semantic graph which links each of the
lexical referents of the text in a network of transitive relations. Givenness for any
one item in the text is perceived as a numerical output of the equation
Gnet = 1/c1 + 1/c2 + . . . 1/cn
in which the givenness (Gnet) of a nodal item is the sum of the inverse of the cost
(c) of traversing the graph from the nodal referent to each of the other referents in
the span (raised to some power for purposes of scale). Computer methodology has
been fully described elsewhere (Cummings 1987, 1995, 1996, forthcoming).
An excerpt from the semantic graph database for the Mansfield text is represented in Figure 2. Each node in the graph is a lemmatized lexical item from the
text, linked together in Systemic-Functional system network notation (cf. Halliday
1967a:38). Links between semantically associated nodes are costed such that the
identity relation is given a cost of 1, mother-daughter a cost of 2, sister-sister 3 and
so on additively. Some nodes have been postulated to explain evident co-relations,
and these have been bolded in Figure 2.
Results for a section of the text containing sentence (6) are represented in Figure
3. Interlineated with the text is a graph line whose height at each lexical item or
lexical substitute indicates degree of givenness on a scale ranging from (0.0)1
to (0.)15—values of Gnet divided by the size of the span (30) to give an average.⁴
The horizontal between values 0.03 and 0.04 represents the most plausible demarcation between given and new for the whole output. The wave shape showing
oscillation between given and new is typical for the results of all such analyses. The
analysis of sentence (6) indicates that the results of the sr method are matched by
the results of the ascit procedure. Items which in the former represented semantic
relations suggesting given information (it., sub., c.s., and sup.) here have average
values from 0.18 to 0.04, and those items which represented semantic relations suggesting new information (mer., v.) here have average values from 0.02 to 0.01.
However there is no consistent range of values corresponding to each of the relations in the sr scale; e.g., items in (6) ascribed identity relations here show variously
values of 0.14, 0.06, 0.18, 0.04, 0.07, and 0.10, due to the field effect.
6. The complete text. A comparison of results from each of the four methods
applied to the whole of the Mansfield text confirms the comparison of results based
on sentence (6) alone. Throughout the text, the given/new distinction based on the
198
Michael Cummings
species
1
picotee
| flower
2
marigold
nasturtium
part
garden
petal
3
4
| leaf
2
part
4
tree
topography
manuka
grass
| cane
beach
river
nature
gas
flame
window
parents
father
mother
family
boy
individual
Linda Burnell
non-family
person
5
boat
oral
head
body
leg
ear
eye
hair
foot
| boatman
tongue
6
7
| voice
knee
hand | palm
Figure 2. Extract from semantic graph of Mansfield text.
ascit average values is almost completely consistent with the same distinction
based on semantic relations. Marked and unmarked focuses of information, based
on the given/new distinction derived from either of these two methods, yield not a
necessary intonational reading of the text, but at least a reading which is nearly
everywhere plausible.
However, comparison between these three sets of results and the result of fsp
analysis shows both consistencies and significant inconsistencies. There is a perfect
correlation between sr category ‘substitution’ and fsp ‘context dependency’. But
iterations in the fsp analysis are only perceived as context dependent about half the
time (even omitting from consideration iterations which are modifier elements in
functional sentence perspective, focus of information…
199
1
5
10
15
(2) She did nothing. (3) She looked up at the dark, close, dry
1
5
10
15
leaves of the manuka, at the chinks of blue between,
1
5
10
15
(4) and now and again a tiny yellowish flower dropped on her.
1
5
10
15
(5) Pretty—(6) yes, if you held one of those flowers on the
1
5
10
15
palm of your hand and looked at it closely, it was an
1
5
10
15
exquisite small thing. (7) Each pale yellow petal shone as if…
Figure 3. Extract from ascit output for Mansfield text.
rhematic nominal groups and therefore not affecting perceived givenness of the
group). The only instances of close synonyms are also as modifier elements in rhematic nominal groups. Only one out of four superordinates is perceived in the fsp
analysis to be context dependent. Hyponyms, meronyms, co-relations and variations show consistency with fsp analysis insofar as it never designates them as context dependent. However, such elements are often declared Thematic on the basis
of the fsp semantic scale. These results fully exemplify the difference between
methods based on discourse semantics and a method based on the semantics of
sentence or clause in isolation.
200
Michael Cummings
¹ I am indebted for the transcription of this text to Mr. Zhang Ren, Graduate
Programme in English, York University.
² fsp does analyze the structure of noun phrases in terms of theme/rheme, but at a separate rank of information structure. The method was not carried through in the specimen analysis quoted (cf. Firbas 1992:83-86).
³ Previous experiments with the ascit method has suggested optimal results for modern English text with a span set for 30 items (Cummings 1995:451).
⁴ Values of Gav greater than 0.15 have been reduced to that value in the diagram.
REFERENCES
Cummings, Michael. 1987. syspro: a computerized method for writing system
networks and deriving selection expressions. Language topics: Essays in honour
of Michael Halliday, ed. by Ross Steele & Terry Threadgold, vol 2:45–64.
Amsterdam: John Benjamins.
——. 1995. Structural semantics as the basis for theme/rheme. lacus forum
21:443–59.
——. 1996. Computational analysis of Old English lexical cohesion. lacus forum
22:293–303.
——. forthcoming. Intuitive and quantitative analyses of given/new in texts.
Communication in linguistics, ed. by J. DeVilliers & Robert J. Stainton. Toronto:
gref.
Firbas, Jan. 1986. On the dynamics of written communication in the light of the
theory of functional sentence perspective. Studying writing, ed. by C. R.
Cooper & S. Greenbaum, 40–71. London: Sage.
——. 1992. Functional sentence perspective in written and spoken communication.
Cambridge: Cambridge University Press.
Halliday, M. A. K. 1967a. Notes on transitivity and theme in English, Part 1.
Journal of linguistics 3.1:37–81.
——. 1967b. Notes on transitivity and theme in English, Part 2. Journal of linguistics 3.2:199–244.
——.1994. An introduction to functional grammar. 2nd ed. London: Edward
Arnold.
Lyons, John. 1977. Semantics. New York: Cambridge University.
Mansfield, Katherine. 1945. At the Bay. Collected stories of Katherine Mansfield,
205–45. London: Constable.