Full Text

kalbotyra
Variation of Conjunctive Discourse
Markers across Different Genres
Jungtukø vartojimo ypatumai
skirtingø þanrø tekstuose
Daiva VERIKAITË
Vilniaus pedagoginis universitetas,
Studentø g. 39, LT-08106, Vilnius
Santrauka
Straipsnio tyrimo objektas – jungtukø vartojimo
ypatumai vadovëlio ir mokslinio straipsnio þanrø
tekstuose. Tiriamas keturiø tipø jungtukø
(adityvø, adversatyvø, kauzatyvø ir temporatyvø)
vartojimas dviejuose ðimtuose vadovëlio ir
dviejuose ðimtuose mokslinio straipsnio þanro
tekstø puslapiuose. Reikðminio skirtumo tarp
tirtø jungtukø tipø vartojimo daþnumo dviejø
þanrø tekstuose nenustatyta – abiejø þanrø
tekstuose daþniausiai vartojami adityvai ir
adversatyvai. Þanro poveikis juntamas tik
konkreèiø jungtukø, priklausanèiø atitinkamam
jungtukø tipui, vartojimo daþnumui. Straipsnio
tikslas – remiantis þanro teorija nustatyti jungtukø
vartojimo daþnumà bei paaiðkinti jø vartojimo
ypatumus skirtingø þanrø tekstuose.
Reikðminiai þodþiai: a genre, conjunctive
discourse markers, the additive, the adversative, the
causal, the temporal, the frequency of occurrence
Summary
The article presents the results of analysis of
the conjunctive discourse markers in the texts of
the textbook and the research article genre. The
use of four types of conjunctive discourse markers
– the additive, the adversative, the causal and the
temporal – was analysed. No significant difference
between the frequency of occurrence of a particular
type of the conjunctive discourse markers was
determined. The additives and the adversatives
were the most frequently used conjunctions in the
texts of both genres. However, the frequency of
occurrence of a particular conjunction varried
across the genres and this could be explained by
the genre constraints.
Key words: a genre, conjunctive discourse
markers, the additive, the adversative, the causal,
the temporal, the frequency of occurrence.
The text presents a unit in which sentences are
related to each other logically. The logical
relationship of the text is generally referred to as
coherence. On a surface structure level, the
coherence of the text may be unmarked (implicit)
or marked (explicit). One of the most important
explicit markers of coherence are conjunctions. As
noted by Toolan (1998, 28), “cohesive conjunctions
have a ‘semantic signposting’ function. The semantic
or logical connection may be implicit between the
foregoing and following text, but the use of the
conjunction makes the connection more explicit”.
The explicit markers of conjunctive relations,
which in the literature are referred to as conjunctive
discourse markers, according to Halliday and Hasan
68
(1976, 238), are divided into additive, adversative,
causal, and temporal. As correctly pointed out by
Brown and Yule (1983, 192), “it is not the case that
any one of these formal markers stands in a simple
one-to-one relationship with a particular cohesive
relation: and, for example, can occur between
sentences which exhibit any of the four mentioned
relationships. Neither is it the case that the posited
relationships cannot be held to exist in the absence
of formal markers”. No matter that it is the
underlying semantic relation that actually has the
cohesive power rather than the particular cohesive
marker it is the presence of the cohesive markers
which constitutes “textness” (Haliday and Hasan,
1976, 229).
Daiva VERIKAITË
Introduction
þ m o g u s
i r
þ o d i s
2 0 0 5
I I I
However, “textness” of different texts with regard
to the use of cohesive markers might as well differ
considerably. Taxonomy of text variation is another
issue widely discussed by text analysts. The problem
that a text analyst faces is the problem of
terminology. The terms “register” and “genre” enjoy
the widest currency in linguistics. However, their
use has not been clearly defined to date. The
problem of the terminology lies in that genres are
often defined using extralinguistic criteria, following
the tradition of the classical Greek literary theory
which divided works into epic, poetic and dramatic
genres, and numerous sub-genres (such as tragedy,
comedy, satire). However, recently the situation
with regard to the use of the term “genre” has
changed considerably. Leckie-Tarry (1993, 27)
points out that the terms “register” and “genre”
appear to be of equal importance in the analysis of
written and spoken, literary and non-literary texts
and that “genre has assumed an important place
within functional linguistics, a place which might at
one time, seem to have been firmly, and exclusively,
reserved by ‘register’”. In his overview of theoretical
alternatives of register and genre, Matthiessen (1993,
232) supports the point of view by stating that “genre
is not a separate theoretical term <…>. One reason
the term was avoided early on was simply that its
traditional sense was far too narrow and associated
with literary varieties”. Moreover, genre theorists
claim that the concept of genre, with its dual
emphasis on all contextual levels and linguistic
structure, allows a dual focus, the focus of text as
product and text as process and eventually is more
universal then the concept of register. “Genres are
both ‘products’ and ‘processes’ – ‘systems’ and
‘performances’. Each time a text is produced so as
to realize and construct a situation-type it becomes
the model for another text and another situationtype. As a model, it functions like a static, finished
product or a system according to which new texts
can be constructed. Once the constructing begins it
becomes again a dynamic process, a ‘performance’
which will inevitably change the model with which
it begins” (Threadgold, 1989, 100). Therefore,
analysis of variation of different language means
across different genres helps in defining textual
characteristics of genres, the probabilistic, dynamic
aspects of their performance as well as their
schematic structures.
This study concentrates on analysis of variation
of conjunctive discourse markers in two genres –
the genre of the textbook and the genre of the
research article, which are found within a single
academic discipline. Inasmuch as “genres cut across
disciplinary boundaries and the subtle variations
across a range of disciplines appear to be more
significant in the way lexico-grammatical resources
and rhetorical strategies are exploited to give
expressions to discipline-specific concepts,
knowledge and its structure, modes of conducting
and reporting research, level of rhetorical intimacy,
and pedagogic approaches and concerns” (Bhatia,
2002, 25), the selection of a single academic
discipline for analysis does not reduce the validity
of the results.
As instances of a category of academic discourse,
the genre of the textbook and the genre of the
research article display significant overlap in terms
of what is known as the field of discourse, especially
patterns of specialist lexis and certain rhetoric
functions, however the two genres have their own
generic features. Firstly, textbooks present
established disciplinary knowledge, whereas
research articles present new and often contested
knowledge. Secondly, the textbook writer is regarded
as the knowledgeable disciplinary expert competent
to make an overview of the facts, ideas and
generalizations usually made by others and being
able to present them in a comprehensible way,
whereas the author of the research article is regarded
as a pioneer in a field presenting new ideas formed
on the basis of his/her own research. Therefore, the
relationship between the writer and the addressee
differs: the textbook writer is expected to be superior
to the addressee in many respects such as knowledge,
experience, competence, etc., whereas the
relationship between the research article writer and
the addressee is expected to be equal, balanced,
partner-oriented. Different communicative aims and
different relationships between the writer and the
addressee presuppose difference in use of linguistic
patterns, language means, discourse markers, etc.:
textbooks often display the use of rhetoric devices
that make the knowledge accessible to a wide range
of students which include description, definition,
classification and explanation, whereas in research
articles such devices are rare. Finally, textbooks
facilitate learning through the use of rhetoric
strategies, e.g. easification procedures, rhetorical
questions, non-linear devices such as charts,
diagrams, figures, pictures, or word glosses. Such
strategies are not often used in research articles
(Perepeèienë, 2004, 13).
Therefore, the aim of the article is to define the
variation of the use of the conjunctive discourse
markers in the texts of the two genres, namely the
ISNN 1392-8600
kalbotyra
69
Variation of Conjunctive Discourse Markers across
Different Genres
Jungtukø vartojimo ypatumai skirtingø þanrø tekstuose
genres of the textbook and the scientific research
article. The corpus was drawn from three textbooks
on linguistics and a number of scientific research
articles in linguistics written by different authors,
which involved 400 pages of the text1 . The mean
frequency of occurrence of a particular type of
conjunctions was calculated; the relative frequency
of each type of items under analysis was determined.
The Results
are expressed” (Swales, 1990, 62), they share certain
common objectives such as clear presentation of
information, encouragement of analytical thinking
and the need to raise reasonable argumentation. The
results of the relative frequency of different
conjunctive relations in the texts of the analysed
genres did not demonstrate significant differences
of occurrence; however, the comparative analysis
of each type of conjunctive relations in the texts of
the two genres indicated a significant difference in
the frequency of the use of different types of
conjunctions, namely the causal and temporal
conjunctions (for the results, see Fig. 2).
F ig . 2 T h e re lativ e fre q u e n cy o f e ach
typ e o f th e co n ju n ctiv e re latio n in th e
two g e n re s
80
70
60
50
40
30
20
10
0
69
55
45
additive
40%
adversative
36%
The frequent use of the additive and adversative
conjunctive discourse markers could be justified by
the genre constraints. Although “the genres of
textbook and scientific research articles vary in
terms of the mode or medium through which they
38
m
po
ra
l
al
te
us
ca
e
iv
rs
T h e te x tb o o k
ad
ve
T h e re se a rch
a rticle
As can be seen from Fig. 2, the frequency of
occurrence of the conjunctions representing additive
and adversative relations in the genres of the
research article and the textbook did not differ much
– 10 and 4 percent, respectively; however, the
difference in the use of conjunctions representing
causal and temporal relations – much more
considerable – 38 and 24 percent, respectively.
Therefore, the analysis of each type of conjunctive
relations found in the texts of the two genres can
point to the specific needs of a particular genre.
1
The author of the article is grateful to Olga Perepeèienë, a Master degree student, for contributing the data of her MA
paper to the present study.
2
The analysis of the conjunctive relations in this article was carried out following Haliday and Hasan’s (1976, 242)
classification.
70
Daiva VERIKAITË
causal
13%
at
tiv
di
temporal
11%
ad
Fig. 1 The relative frequency of
different types of conjunctions
62
48 52
31
e
Four major types of conjunctive discourse
markers – additives, adversatives, causals and
temporals were analysed2 . These connective forms
exemplify implicit relations between clauses. Their
presence in a text helps the addressees to construct
the text’s mental representation (Georgakopoulou
and Goutsos, 1999, 90). Additive is a generalised
semantic relation in the text-forming component of
the semantic system that is based on the logical
notion of “and”. Adversative is a relation that is
based on the notion “contrary to expectation”. The
expectation may be derived from the content of what
is being said, i.e. from the communication process,
the speaker-hearer situation. Causal is a relation
which includes general causal relations and specific
causal relations such as those of result, reason and
purpose. Temporal is a relation of sequence in time
(Halliday, Hasan, 1976, 256).
The results of the relative frequency of
occurrence of the four relations demonstrated that
the most frequently occurring conjunctive relations
in the texts of both genres were additive and
adversative, whereas causal and temporal were much
less frequently used (see Fig. 1).
þ m o g u s
i r
þ o d i s
2 0 0 5
I I I
kalbotyra
Additives
Following Haliday and Hasan’s classification
(1976, 249), four types of conjunctive relations of
additive type were analysed – simple (and), complexemphatic (furthermore, moreover, in addition,
additionally), comparative (likewise, similarly,
conversely), and appositive (that is, thus, for instance,
for example). As can be seen from the results in Figure
4, there is no significant difference between the use of
the additives in both genres. It is obvious that the
prevailing type of additives in the genres of the textbook
and the research article is that of the apposition, the
least frequent is that of the comparison (see Fig. 3).
Fig. 3 The relative frequency of
additives
Simple
19%
Complex
17%
Fig. 4 The relative frequency of
each type of additive in the tw o
genres
70
60
50
40
30
20
10
0
55
58
45
56
42
44
43
57
The research
article
on
n
ris
tio
pa
om
po
si
pl
om
C
Ap
C
e
ex
The textbook
pl
The most frequently used additive in the analysed
texts of both genres was for example – 30 and 23
percent, respectively. It occurred in similar contexts,
usually at the beginning of the sentence in the texts
of both genres.
The simple additive and, complex additives
furthermore and moreover were used more frequently
than their counterparts in both of the genres. The
m
Apposition
55%
Si
Comparison
9%
appositive thus and the complex-emphatic in addition
were more frequently used in the texts of research
articles. The difference in the frequency of occurrence
could be explained by the specific nature and
constraints of the analysed genres. As mentioned, the
genre of the research articles usually deals with new
knowledge. The aim of the research article is to
introduce novel ideas and to persuade the addressee
to accept the offered point of view. The use of the
specific additive conjunctions facilitated the
achievement of the set goals. For example illustrated
the propositions; thus drew conclusions, in addition
added information and substantiated the ideas. The
genre of the textbook having the aim to present
information and explain concepts and definitions in
the most comprehensible way demonstrated a higher
frequency of and, likewise, moreover and furthermore
which are information-adding and clarifying means.
Adversatives
The results of the analysis demonstrated that
the adversative proper and the contrastive
adversatives were more characteristic of the
textbook genre, whereas corrective adversatives
enjoyed higher frequency in the texts of the research
article genre (see Fig. 6).
Fig. 6 The re lativ e fre que ncy of
e ach type of adv e rsativ e in the
two ge nre s
Fig. 5 The relative frequency of adversatives
100
80
Dismissive
5%
60
Contrastive
3%
87
89
56
44
40
5050
13
20
Dismissive
Corrective
Contrastive
Adversative
proper
82%
The res earch
article
The textbook
11
0
Adversative
proper
Corrective
10%
ISNN 1392-8600
Four types of adversative relations were analysed
– the adversative proper (but, however, though,
although), the contrastive (on the other hand, in
fact), the corrective (on the contrary), the dismissive
(in any case).
As can be seen from Figure 5, the most frequent
of the four types of adversative relations is the
adversative proper.
71
Variation of Conjunctive Discourse Markers across
Different Genres
Jungtukø vartojimo ypatumai skirtingø þanrø tekstuose
The distribution of the most frequently used
adversatives however and but was different in the
two genres. Contrastive however, adversative proper
though, although, corrective on the contrary were
preferred by the authors of research articles, whereas
but was much more popular among the authors of
the textbooks. The difference in the frequency of
occurrence could be explained by the genre needs –
active persuasion in the case of the research article
and reserved clarification in the case of the textbook.
The genre of the textbook often displayed the use of
rhetoric devices employed to make the knowledge
accessible. Such rhetoric devices as classification,
definition, description were realized in the genre of
textbooks through the use of the adversatives proper
in order to present and explain information. The
authors of the research articles, on the other hand,
while trying to influence the audience and present
arguable information, employed the rhetorical
strategies of comparison and contrast. The
adversative proper however and the corrective on
the contrary were used to introduce new arguments
and eventually convince the audience.
Causals
Conditional
35%
Respective
3%
Specific
8%
General
54%
The most frequent causals were for this reason/
for this purpose (27 percent) used in the textbooks
and so (19 percent) used in the research articles.
Therefore was frequently used in the research articles
(17 percent) and then in the textbooks (14 percent).
The analysis of the results of the relative
frequency of the occurrence demonstrated the
following tendency: the most frequently used causals
72
80
60
40
20
0
74
54
46
4753
56
44
26
The research
article
e
l
ct
tio
pe
es
R
di
on
C
iv
na
c
ifi
ec
Sp
en
er
al
The textbook
in the textbooks for this reason/ for this purpose, so,
and then were much less frequent in the research
articles, whereas the most frequently used causals
in the research articles therefore, since, and hence
were the least frequent in the textbooks. The
difference in the frequency of occurrence could be
explained by the needs of a genre – textbooks
presented generally accepted theories and facts
attempting to establish a single voice of authority
and encourage analytical thinking. Therefore, the
general causal so, the conditional then, the specific
for this reason/ for this purpose helped to facilitate
the understanding of information making hints in
advance, labelling and recapitulation. Research
articles are valued as sources of new knowledge,
which aim at presenting the wide scope of issues
and multivoicity of views. The general hence and
therefore, the general emphatic because, the
conditional since were frequently used in the
research articles to foster interest, develop
argumentation and active collaboration with the
addressee.
Daiva VERIKAITË
Fig. 7 The relative frequency of
causals
Fig. 8 The relative frequency of
each type of causal in the two
genres
G
The general causal relations which are expressed
by so, hence, therefore, consequently along with the
specific causal relations – that of reason (for this
reason, on this account), that of result (as a result, in
consequence) and that of purpose (for this purpose,
with this in mind), conditional (under the
circumstances) and respective (in this respect, with
regard to this) were analysed. As can be seen from
Fig. 7 and 8, the general and the conditional causal
conjunctions were much more frequently used than
the specific and the respective causal conjunctions.
However, the comparative analysis of the frequency
of occurrence of each type of causal conjunctions in
the two genres demonstrated that the difference in
the use of conditional conjunctions was the most
obvious – they were twice as frequent in the research
articles as in the textbooks.
þ m o g u s
i r
þ o d i s
2 0 0 5
I I I
kalbotyra
Temporals
Various temporals representing different types
of temporal relations were analysed to identify
tendencies of the genre constraint: simple (then,
before that, previously, earlier, afterwards), conclusive
(finally, at last, in the end), sequential (first…then,
first…second, first…next, secondly), and summary
(to sum up, in short, briefly) were subjected to
analysis.
Fig. 9 The relative frequency of
temporals
Summary
20%
Simple
38%
Sequential
31%
Conclusive
11%
As can be seen from Figure 9, the most frequently
used temporal conjunctions were the simple and
the sequential. The conclusive and the summary
temporal conjunctions prevailed in the research
article genre, whereas the sequential conjunctions
were much more characteristic of the textbook
genre (see Fig.10).
Fig. 10 The relative frequency of
each type of temporal in the two
genres
86
80
100
80
60
40
20
0
69
4753
20
31
14
The research
article
y
l
m
m
ar
tia
Su
en
qu
Se
us
cl
on
C
Si
m
pl
iv
e
e
The textbook
The analysis of a particular temporal conjunction
use within the class of a particular type
demonstrated that the frequency of occurrence was
indirectly proportional, i.e. those that were frequent
in the textbooks were much less frequent in the
research articles and those frequent in the research
articles were less frequent in the textbooks. For
instance, the simple then was the second most
frequent in the textbooks; however, in the research
articles it was one of the least frequent temporals or
the third most frequent discourse marker
representing summary temporal relations – to sum
up – was the least frequently used in the textbooks.
The results of the analysis demonstrated one
exception – the temporal next was the most
frequently used in both genres. Differences in the
use of the conjunctions could be explained by the
peculiarities of the genres. The textbooks tried to
facilitate learning through the use of access
structures and easification procedures such as
metadiscourse, lexical familiarisations and word
glosses to lead students to a discipline-based
understanding. Such conjunctions as the simple
temporal before that, then and the sequential
first…then, next, frequently used in the textbooks,
facilitated comprehension in particular and the
learning process in general by organizing the linear
structure of the text and signalling the relationships
between passages. The research articles, on the other
hand, had the communicative purpose of presenting
new ideas, persuading the audience and, in a way,
convincing the addressee to take up the presented
point of view. The macrostructure of the articles,
usually fitting the frame Introduction-MethodResults-Conclusion format, demanded suitable
transition from one part to another. Therefore, the
sequential firstly, secondly, thirdly, next served as the
most suitable discourse markers of transition
between parts of the frame, the summary to sum up
and the conclusive finally were used to logically
conclude a part of the frame before transition to
another part of the frame.
To sum up, the use of the four types of conjunctive
discourse markers in the textbooks and the research
articles in linguistics was found to be similar.
However, the use of a particular discourse marker
representing a particular conjunctive relation varied
in the analysed genres.
The analysis of the distribution of the conjunctive
discourse markers in the genres of the textbook and
the research article demonstrated the following:
1. Of the four types of conjunctive relations, two
types – the additive and the adversative – were
significantly more frequently used in the textbooks
and the research articles than the other two types –
the temporal and the causal.
2. No significant difference between the frequency
of occurrence of the additive type conjunctions in
general as well as of a particular type of additive –
the simple, the complex, the appositive or the
ISNN 1392-8600
Conclusions
73
Variation of Conjunctive Discourse Markers across
Different Genres
Jungtukø vartojimo ypatumai skirtingø þanrø tekstuose
comparative – was found in the texts of both genres.
Therefore, the use of additive conjunctions in the
textbooks and the research articles did not differ.
3. Adversative conjunctions proper were the most
frequently occurring adversatives in both genres.
However, corrective conjunctions prevailed in the
genre of research articles, whereas contrastive
conjunctions were more characteristic of the
textbook. The adversative however and the
corrective on the contrary were much more
characteristic of the research articles, whereas the
contrastive but was significantly more frequently
used in the textbooks.
4. Although the general frequency of the causals
and the temporals was equally low in both genres, the
relative frequency of the occurrence of the causals
and the temporals in the texts of the research articles
was significantly higher than in the textbooks.
5. Of all types of causals, the general and the
conditional prevailed over the other types of the
causals in the texts of both genres. Although there
was no significant difference in the frequency of
occurrence within the types of general, specific and
respective causals in both genres, the conditional
causals were significantly more frequent in the
research articles.
6. Of all types of temporals, the simple and the
sequential temporals prevailed over the other types
of temporals in both genres. Although the simple
temporals were distributed with equal frequency in
both genres, the conclusive and the summary
temporals were significantly more frequently used
in the research articles, whereas the sequential
temporals were significantly more frequent in the
textbooks.
7. Therefore, it could be presumed that the
frequent use of the additives and the adversatives in
both genres point to the necessity of both genres to
present and clarify information, encourage
analytical thinking, develop reasonable
argumentation.
8. The relatively more frequent use of the
temporals and the causals in the research articles
point to the specific needs of the research article as
a genre: the function of the temporal conjunctions
to express sequence and progression and the
function of the causal conjunctions to signal
subsequence and results make them necessary
attributes for maintaining the frame IntroductionMethod-Results-Conclusion of the macrostructure
of the research articles.
References
Bhatia, V. A Generic View of Academic Discourse.
Academic Discourse ed. by J. Flowerdew. London: Longman, 2002.
Brown, G., Yule G. Discourse Analysis. - Cambridge:
CUP. 1983.
Georgakopolou, A., Goutsos, D. Discourse Analysis. Edinburgh: EUP. 1999.
Halliday, M.A.K., Hasan, R., Cohesion in English.London: Longman. 1976.
Leckie-Tarry, H. The Specification of a Text: Register,
Genre and Language Teaching. Register Analysis.
Theory and Practice ed. by M. Ghadessy. London&
New York: Pinter Publishers. 1993.
Matthiessen, C. Register in the Round: Diversity in a
Unified Theory of Register Analysis. Register
Analysis. Theory and Practice ed. by M. Ghadessy.
- London& New York: Pinter Publishers, 1993.
p. 221-292.
Perepeèienë, O. Conjunctive Discourse Markers in the
Genres of Textbooks and Scientific Research
Articles. MA paper. -Vilnius: VPU, 2004.
Threadgold, T. Talking about Genre: Ideologies and
Incompatible Discourses. Cultural Studies 3, 1989.
p. 92-118.
Swales, J.M. Genre Analysis – English in Academic and
Research Settings. Cambridge: CUP. 1990.
Toolan, M. Language in Literature. An Introduction to
Stylistics. London: Arnold. 1988.
Blackmore, D. Understanding Utterances. An
Introduction to Pragmatics. - Oxford: Blackwell
Publishes. 1997. p.1-50.
Collins, P.C. Cleft and Pseudo-cleft Constructions in
English Spoken and Written Discourse. ICAME
Journal 11: 5-17. 1987.
74
Douglas, L. N., McEvoy C. L., Dennis S. What is Free
Association and What Does it Measure. Memory
and Cognition, 28/6: 887-899. 2000.
Evan, G. Pronouns. Linguistic Inquiry, vol. 11/2: 337362. 1980.
Hofmann, T. R. Paragraphs and Anaphora. Journal of
Pragmatics, 13: 239-250. 1989.
Daiva VERIKAITË
Sources
þ m o g u s
i r
þ o d i s
2 0 0 5
I I I
kalbotyra
Roelofs, A. Phonological Segments and Features as
Planning Units in Speech Production. Language
and Cognitive Processes, 14/2:173-200. 1999.
Threadgold, T. Talking about Genre: Ideologies and
Incompatible Discourses. Cultural Studies 3: 92118. 1989.
Stainton, C. Language Awareness: Genre Awareness.
A Focused Review of the Literature. Language
Awareness 1 (20): 109-122. 1989.
Ventola, E. Problems of Modelling and Applied Issues
within Framework of Genre. Word 40 (1/2): 129161. 1989.
ISNN 1392-8600
Kempson, M. R. Semantic Theory. Cambridge:
Cambridge University Press.1-50. 1995.
Nielsen, S. Bilingual Legal Lexicography – A New
Theoretical Basis. Kalbotyra. 2001. 50 (3): 85-91.
2001.
Nunan, D. Research Methods in Language Learning. Cambridge: Cambridge University Press. 1-100.
1992.
Pierrenhumbert, J., Nair, R. Word Games and Syllabus
Structure. Language and Speech, 38/9: 77-114.
1995.
75