The Effect ofDiscourse Markers on the

Comprehension ofLectures*
CRAIG CHAUDRON AND JACK C. RICHARDS
University of Hawaii at Manoa

INTRODUCTION
With greater numbers of non-native speakers of English entering English-
medium institutions of higher education around the world, the provision of
appropriate English-language training as a preparation for academic study
becomes increasingly important. In the United States, all major universities now

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

offer specialized English language courses for foreign students, dealing with the
use of English in such areas as advanced reading, writing term papers, dis-
cussion skills, lecture comprehension, and note taking. The growing demand for
such courses has led to the need for research into the processes underlying
academic performance in a second or foreign language, which can provide input
to teacher training as well as to the development of appropriate curriculum and
instructional materials. This paper deals with the university lecture and
describes a study which was conducted to investigate how different categories of
discourse markers affect the degree to which foreign students understand
university lectures.

1. LISTENING COMPREHENSION PROCESSES

Research on the processes of listening comprehension in both first-language
and second-language listening is in its infancy. Still, a considerable body of
related research into the cognitive processes underlying comprehension has led
to a greater understanding of the nature of listening comprehension (e.g. Clark
and Clark 1977; Richards 1983). Two basic processes that have been identified
are referred to as bottom-up processing and top-down processing. Bottom-up
processing refers to the analysis of incoming data, and categorizing and inter-
preting them on the basis of information in the data. With respect to the compre-
hension of language, bottom-up processes would be, for example, those which
assign grammatical status to words on the basis of syntactic and morphological
cues and which assign topics and meanings on the basis of syntax and word
order and the meanings of lexical items used in the message. Top-down
processing makes use of prior knowledge as part of the process of comprehen-
sion. This may take many forms, including expectations about the topic and
structure of a piece of discourse based on real-world knowledge and reference
to various types of frames, schemas, and macro-structures (see van Dijk 1977,
Rumelhart 1977, Meyer 1984, and especially L2 research on schema theory by
Carrell and Eisterhold 1983, and Carrell 1984). Top-down processing involves
prediction and inferencing on the basis of hierarchies of facts, propositions, and
expectations, and it enables the listener or reader to by-pass some aspects of
bottom-up processing. On encountering the topic of 'going to the dentist', for
example, we refer to knowledge about the participants in the situation, their
roles and purposes, and the typical procedures adopted by dentists and their
Applied Linguistics, VoL 7, No. 2,0 Oxford University Press, 1986
114 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

consequences. Comprehension is viewed as a combination of both bottom-up

and top-down processing. As Adams and Collins put it (1979):
An important aspect of a schema-theoretic account of reading comprehension is that
top-down and bottom-up processing should be occurring at all levels of analysis
simultaneously. The data that are needed to instantiate orfillout the schemata become
available through bottom-up processing; top-down processing facilitates their assimi-
lation if they are anticipated or are consistent with the reader's conceptual set Bottom-
up processing ensures that the reader is sensitive to information that is novel or that
does not fit his or her ongoing hypothesis about the context of the text; top-down
processes help the reader to resolve ambiguities or to select between possible inter-
pretations of the incoming data. Through the interactions between top-down and
bottom-up processing, the flow of information through the system is considerably
constrained, (p. 5)

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

The interactions of these two levels of processing are well illustrated in the case
of understanding lectures.

2. LISTENING TO LECTURES
The function of lectures is to instruct, by presenting information in such a way
that a coherent body of information is presented, readily understood, and
remembered. Several distinct modes of lectures have been distinguished in
research on the nature of academic discourse. Dudley-Evans and Johns (1981)
establish three categories of lectures: a reading style, a more informal con-
versational style, and a rhetorical style:
Style A—'Reading Style'. The lecturer reads from notes, or speaks as if he was reading
from notes. Characterised by short tone groups, and narrowness of intonational range.
Falling tone predominates: level tone (in Brazil's terms, signal of withdrawal from inter-
action) may also occur.
Style B—'Conversational Style'. The lecturer speaks informally, with or without notes.
Characterised by longer tone groups and key-sequences from high to low. When the
lecturer is in 'low key1 at the end of a key sequence, the speaker may markedly increase
tempo and vowel reduction, and reduce intensity.
Style C—"Rhetorical Style'. The lecturer as performer. Characterised by wide into-
national range. The lecturer often exploiting high key, and a 'boosted high key'.
Frequent asides and digressions marked by key and tempo shift—sometimes also by
voice-quantity shift (1981:34)
The present study deals primarily with lectures in the reading style.
In view of the didactic focus of lectures, the structuring and organization of
information within a lecture has been assumed to be an essential aspect of its
comprehensibility. Within applied linguistics, researchers interested in identi-
fying those aspects of lecture structure which might be relevant focuses in train-
ing non-native listeners have examined how the organization of information
within a lecture is signalled. Cook (1975) describes the macro-structure of a
lecture as being composed of a number of 'expositions'. These consist of an
optional episode of expectation, an obligatory focal episode, an obligatory
developmental episode together with optional developmental episodes, and an
obligatory closing episode. At the level of micro-structure, episodes are
described in terms of moves. For example, a concluding move is a justificatory
statement, a focal episode with a concluding function, or a summary statement;
 CRAIG CHAUDRON AND JACK RICHARDS 115

a summarizing move gives a resume of the immediately preceding discourse.

Various related approaches to the analysis of written rhetorical structure have
been pursued by Meyer and her associates in first language (LI) research
(Meyer 1975, Meyer et al. 1980, Meyer and Rice 1982) and by Carrell (1984),
who has investigated its effects on second language (L2) reading comprehen-
sion. Both Cook, and Murphy and Candlin (1979) focus on how the rhetorical
organization of a lecture is signalled. Cook examines the functions of connec-
tives and other devices which serve as indicators of topic continuation.
Discourse markers which signal the information structure of discourse by
emphasizing directions and relations within discourse include phrases such as
'Now, getting back to our main point ...'. Murphy and Candlin identify a
number of markers of the rhetorical organization of lecture discourse, including

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

what they refer to as markers (e.g. 'Well, right, now"), starters (e.g. *Well now,
let's get on with ..."), and metastatements (e.g. 'I want to mention two types of
generator*). These discourse signals also reflect the interactional nature of
lectures and are probably more frequent in a conversational style than a reading
style.
Studies of listening problems for non-native listeners have confirmed that
they do indeed have difficulties recognizing signals and markers of organization
of information within lectures. In a study of lecture comprehension problems
encountered by Chinese students at the University of California at Los Angeles,
Yuan (1982) observes:
In general, the subjects were rather weak at paying attention to the sequence of the
lecture because of their neglect of the logical connectors of sequence and their lack of
recognition of transition from one main idea to another. Besides, they paid more atten-
tion to decoding the speech sentence by sentence than to extracting the science infor-
mation from the lecture through understanding the rhetorical nature and functions of
both textual and lecture discourse, (p. 48)
There have, furthermore, been a few experimental attempts to investigate
what features of lectures might aid L2 learners' comprehension. Several studies
(Long 1985, Cervantes 1983, Kelch 1985) explored the effects of repetitions,
paraphrases, rate of speech, or syntactic complexity. Two studies examined
features of discourse organization. Chaudron (1983) studied the effects of topic
signalling in experimental lectures on ESL learners' immediate recall of the
topic information. He found that recall was significantly better for a repeated
topic than for more complex signalling of topic change. Sawa (1985)
manipulated two factors in recorded lectures: the repetition and paraphrasing of
information, and signalling of major segments and emphasis (especially by
rhetorical questions). He tested the effects of each of these separately and in
combination, on intermediate-level ESL listeners' post-lecture free recall.
Sawa's results showed no significant overall difference in recall between lecture
versions, but this lack of differences may be attributable to the generally low
recall rate of the subjects on the free recall. A more immediate measure such as a
cloze procedure may reveal differences. Sawa's subjects also had very little time
available for recall (about four minutes for a six-minute lecture).
Some ESL instructional materials for the teaching of comprehension of
lectures acknowledge the role of discourse markers in aiding comprehension,
and give practice in recognizing different kinds of discourse markers and their
116 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

functions within lectures. In Listening and Learning (Young and Fitzgerald

1982), for example, exercises are provided which train learners to identify
discourse markers with the functions of addition, comparison, contrast,
exemplification, explanation, restatement, result, sequence, summation, and
transition.
The research cited above suggests that, in addition to possible prior
knowledge of the topic of a lecture, the L2 listener may benefit from knowledge
of the macro-structure and discourse organization of lectures. Prior knowledge
of this sort helps top-down processing by initiating expectations and predic-
tions about the lecture. These expectations are then confirmed and supported
by the speaker's use of discourse signals of the relationship between successive
episodes and moves within the lecture. Such discourse signals, like Murphy and
Candlin's starters and metastatements, and the signals of high-level information

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

described by Meyer etal. (1980), could be termed 'macro-markers'.
At the same time, some research suggests that the speaker's signals such as
Veil', 'so', 'now5, often used as signals of lower-level information in the text (cf.
Meyer et al. 1980), and which could be called 'micro-markers', serve as filled
pauses giving listeners more time to process individual segments of a piece of
discourse; they hence provide more opportunities for bottom-up processing.
Research on pausing phenomena in teacher talk (Chaudron 1983; Chaudron
forthcoming) has found that both filled and unfilled pauses increase in
frequency or in length when native speakers address learners of low proficiency,
presumably reflecting the speaker's attempt to provide greater opportunity for
bottom-up processing. Such pauses provide further time for processing, and
assist in segmenting discourse into meaningful units for higher-level processing.
However, the precise functional effects of different kinds of markers (e.g. those
functioning as signals of the macro-structure as opposed to those functioning as
fillers) require further clarification. The LI reading research (e.g. Meyer 1975,
Meyer et al. 1980, Clements 1979) has not manipulated the use of different
types of markers, sufficing mainly to investigate the presence or absence of
markers in the texts. It is this question which the present study seeks to address
in regard to L2 listening comprehension.
3. RESEARCH QUESTIONS
The present study is intended to explore the effects of discourse signals and
markers in lectures on second-language learners' comprehension. The relation-
ship between top-down and bottom-up processing was targeted by posing the
following two questions:
1 What is the effect on L2 learners' comprehension of lectures of the use of
discourse markers which indicate the overall organization of lectures—that is,
macro-markers which signal the macro-structure of a lecture through high-
lighting major information in the lecture and the sequencing or importance of
that information?
2 What is the effect on L2 learners' comprehension of lectures of the use of
micro-markers, which indicate links between sentences within the lecture, or
which function as fillers?
In order to answer these questions, a lecture on American history was
prepared, based on a natural, live performance on the topic by an ESL teacher.
 CRAIG CHAUDRON AND JACK RICHARDS 117

Four different versions of the lecture were recorded, each version including a
different combination of macro and micro discourse markers. These lectures
were then played to second-language learners of different ability levels, and
measures were taken of their comprehension.
4. HYPOTHESES
It was hypothesized that learners would comprehend the information in the
lecture better when more discourse markers were included, in particular
H I : L2 learners would comprehend the lecture better when micro-markers
were added than when no markers were added.
H2: L2 learners would comprehend the lecture with macro-markers better
than the lecture with only micro-markers.

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

H3: L2 learners would comprehend best the lecture with both micro- and
macro-markers.
In order to test these hypotheses, the following procedures were employed.
5. METHOD

5.1 Materials
A video tape and transcript of a natural lecture presented to ESL university
students was used as source material. This lecture, which dealt with the
expansion of the United States from thirteen colonies to an imperial nation, was
condensed to a seven-page written passage. This passage, the 'Baseline' version,
did not include any special signals of discourse organization or linking between
sentences, other than what was absolutely necessary to convey the meaning of
the lecture.
A second version of the lecture, the 'Micro' version, was then constructed, in
which various markers of intersentential relations, framing of segments, and
pause fillers were inserted. No other changes in the content of the lecture were
made. Examples of these markers are (see Appendix A for a complete listing of
the markers used):
temporal links—then, and, now, after this, at that time
causal links—because, so
contrastive relationships—but, actually
relative emphasis—you see, unbelievably, of course
framing/segmentation— well, OK, all right?
Every attempt was made to keep any of these markers from adding semantic
information to the lecture. The relationships encoded by the markers were,
typically, already evident in the content of the text
A third version was also constructed, called the 'Macro' version, containing
signals or metastatements about the major propositions within the lecture, or
the important transition points in the lecture. For example (see Appendix A for
a complete listing):
what I'm going to talk about today...
let's go back to the beginning...
this brought about new problems...
118 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

Again, these added no new prepositional content to the lecture.

Thefinalversion was simply a combination of versions (2) and (3), called the
'Micro-Macro' version. The following are examples of sections showing all four
versions:
Baseline versions:
The United States came into existence officially in 1783 after eight years of
war...
By 1803, the original thirteen colonies had doubled in size...
Micro versions:
Well, the United States came into existence officially in 1738 after eight years
of war...

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

And so, by 1803, the original thirteen colonies had doubled in size...
Macro versions:
To begin with, the United States came into existence officially in 1783 after
eight years of war...
What we've come to by now was that by 1803, the original thirteen colonies
had doubled in size...
Micro-Macro versions:
Well, to begin with, the United States came into existence officially in 1783
after eight years of war...
And so, what we've come to by now was that by 1803, the original thirteen
colonies had doubled in size...
These four versions were then separately recorded by a male speaker in a
professional studio, at a moderately slow rate of speech. Table 1 shows the total
durations and rate of speech for each lecture version.1

5.2 Comprehension measures

Three measures of the learners' comprehension of the lectures were pilot-tested
and revised before inclusion in the study. The first was a recall cloze measure,
which involved a sample of sections from the lecture. This sort of test has been

Table 1: Duration and rate of

lecture versions

Lecture version Duration Words Rate (wpm)

(includes
8min.
of pauses)

Baseline 23:02 1760 1173

Micro 24:46 1910 114.0
Macro 25:46 1914 107.8
Micro-Macro 27:17 2064 107.1
 CRAIG CHAUDRON AND JACK RICHARDS 119

shown to be a relatively reliable measure of comprehension (Henning et al.

1981,Chaudronl985).2
The second comprehension measure was a ten-item multiple-choice test with
four alternative responses, covering the material in the entire lecture. The third
was a ten-item true-false test covering the entire content. These were to be filled
in by the listeners following the completion of the lecture.

5.3 Subjects
Two groups of subjects were tested using these materials. These will be referred
to as the 'pre-university' and the 'university' groups. The pre-university group
were 71 ESL students enrolled in sections of two listening courses in an inten-
sive English language program at a private college in Hawaii. The university

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

group were of a higher level of proficiency: 81 ESL students enrolled at the
University of Hawaii, who were taking an ESL listening comprehension course
in addition to their regular courses of study. These students were of mixed but
predominantly Asian and Pacific ethnic and linguistic backgrounds.
Some placement test scores and standardized test scores were available on
these students which were used later in the analysis of the results.

5.4 Design and procedures

The four versions of the lecture were assigned at random to different classes in
the respective subject groups.3 The researchers met the subjects at their regular
class times with the teachers present, and explained the purpose and procedure
of the study. Whereas the pre-university group were met in their classrooms, the
university group met the researchers in a language laboratory, where each
subject could do the exercise in an individual laboratory booth. The subjects
were given a two-page set of instructions with examples of how to do the cloze
procedure, and of true-false and multiple-choice responses. They were then
given two short practice listening passages employing the cloze procedure,
followed again by comprehension questions. No note-taking was allowed.4
Following the completion of the lecture, the cloze response booklets were
collected, and the subjects were given first the multiple-choice questions to
answer, then the true-false quizzes.

5.5 Analysis
All the subjects' responses were scored either right or wrong. An exact-word
scoring method was used for the cloze items, with only minor errors in spelling
or grammatical form being considered acceptable.5
Pearson product-moment correlations were calculated between the depen-
dent cloze and comprehension measures and the various standardized and
placement test scores for the subjects.6 The dependent measures were analysed
separately in an analysis of covariance (SPSSX MANOVA procedure, SPSS
Inc., release 1.1), with lecture version the independent factor (four levels), and
different standardized and placement tests as covariates.7
Unexpectedly, the placement scores for the pre-university group consisted of
three test administrations, separated by six-month intervals. As a result, the
analysis was conducted on two separate sub-groups, each representing a test
administration. This left an N of 33 in the one group and 32 in the second. The
third group was too small for further analysis.
120 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

In addition, a priori comparisons were made according to the hypothesized

direction of effect between the lecture versions. Thus, the Micro-Macro version
was tested against the Macro version, the Macro against the Micro, and the
Micro against the Baseline. Where a significant ANCOVA was obtained, post
hoc comparisons were also made, to determine which versions were different
from others.

6. RESULTS
The results will be reported separately for the two groups.
Pre-university group
The cloze measure for this group as a whole was very low (X •** 19.6, out of 79
possible). Table 2 reports the means for the group as a whole and separately by

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

both version condition and by test group (test date 1—TD1 —was the winter of

Table 2: Means on cloze recall for

pre-university group

Lecture version TDI(N -33) TD2(N -32)

mean s.d mean s.d

Baseline 18.5 ±7.7 25.0 ±8.3

Micro 9.83 ±5.4 20.42 ±10.5
Macro 24.1 ±10.1 22.17 ±8.3
Micro-Macro 13.17 ±8.4 23.09 ±12.9
Overall means 17.48 ±9.4 22.18 ±10.5
Grand mean 19.8

1984, test date 2—TD2—was late spring of 1984). Whereas the cloze had a KR-
21 of .87, considered adequately high, the multiple-choice and true-false tests
were so low in reliability (.2 and 0.16, respectively), that they will not be con-
sidered in the analysis of the lecture versions here.
Table 3 shows the results of the Analysis of Covariance on this cloze measure,
as differentially affected by the lecture versions, for TD1 and TD2. The version
means as adjusted by the covariates are also displayed here for TD1.
It is evident in Table 3 that there was a significant effect for version on cloze
responses with TD1 (F(3/27) - 4.99, p < 0.01), but not TD2. The a priori
comparison of differences between means snowed that there were, no
differences between the adjusted means for the Micro version and the Baseline.
Supporting the hypothesis, the Macro version was significantly superior to the
Micro version (p < .01), but in conflict with the hypothesized effect, it was also
superior to the combination Micro-Macro version. Post hoc comparisons
revealed no differences between the Micro-Macro and the Micro or Baseline
versions.
It was therefore of some interest to determine what difference there might be
between the two test groups. Based on the cloze recall outcome and all other
 CRAIG CHAUDRON AND JACK RICHARDS 121

Table 3: Analysis ofcovariance of cloze scores

by version

Source SS df MS F

a. Test date 1
Constant .78 1 78 .01
Regression (co-variates) 473.41 2 236.71 4.3 •
Version 822.71 3 274.24 4.99**
Within 1484.14 27 54.97
b. Test date 2
Constant 173.10 1 173.10 2.49

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

Regression (co-variates) 1592.86 2 796.43 11.47*"
Version 244.82 3 81.61 1.18
Within 1804.88 26 69.42

•p<.05
«p<.01
*~p<.00l

Version Adjusted mean (TD 1)

Baseline 16.08
Micro 12.01
Macro 25.61
Micro-Macro 13.58

placement test scores, the later test group (TD2, summer 1984) performed
significantly better than the earlier one (t-tests, p < .005). The relevance of this
difference will be discussed below.
University group
The mean cloze score for the university group was 35.8 (n = 81), considerably
higher than the pre-university group. Table 4 reports the means for the group as
a whole and separately by version condition. The cloze in this case also had a
high reliability (KR-21 r =• .89) and the multiple-choice and true-false tests

Table 4: Means on cloze recallfor

university group

Lecture version N Mean s.d.

Baseline 18 31.89 ±9.4

Micro 26 32.12 ±13.59
Macro 18 42.94 ±10.18
Micro-Macro 19 37.95 ±14.3

Overall mean 81 35.8 ±12.84

122 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

improved in reliability (.32 and .39, respectively), but not enough to warrant
their being included in further analyses.
Table 5 shows the results of the Analysis of Covariance on this cloze measure,
as differentially affected by the lecture versions. The version means as adjusted
by the covariates are also displayed here.
Just as with the TD1 pre-university group, Table 5 shows that there was a
significant effect for version on cloze responses (F(3/76) — 3.10, p < .05),
adjusting for the listening placement scores as covariate. The a priori com-
parison of means showed in this case that there were no differences between the
adjusted means for the Micro version and the Baseline, nor between the Micro-
Macro and Macro versions, but that the Macro version was again significantly
superior to the Micro version (p < .01). Post hoc comparisons revealed no
differences between the Micro-Macro and the Micro or Baseline versions,
although in this case the Micro-Macro version appeared to produce better

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

results than the Micro or Baseline.
Since the cloze measure was also significantly correlated with the reading
placement test scores, and it could be argued that the cloze is a reading measure
as much as a listening measure, these scores (which were available on nearly all
subjects, N — 79), were used as a second covariate. Very similar results were
obtained.
To summarize, what we have found is a consistent result across groups that
macro-markers, that is, the higher-order discourse markers signalling major
transitions and emphasis in the lectures, are more conducive to successful recall
of the lecture than micro-markers, that is, lower-order markers of segmentation
and intersentential connections. This was a significant difference in the case of
the university group and the less proficient pre-university group, and a similar
trend was shown for the TD2 pre-university group.

Table 5: Analysis of covariance of cloze scores

by version

Source df MS F

University group
Constant 64.14 1 64.14 .49
Regression (co-variates) 1592.84 1 1592^4 12.14—
Version 1218.40 3 406.13 3.10*
Within 9967.43 76 131.15

• p<.05
•** p<.001

Version Adjusted mean

Baseline 32.98
Micro 32.33
Macro 42.21
Micro-Macro 3732
 CRAIG CHAUDRON AND JACK RICHARDS 123

7. DISCUSSION
In the light of the theory of information processing and top-down compre-
hension of discourse, the finding that macro-markers led to better recall of the
text material than micro-markers should not be surprising. The learners are
evidently aided in organizing the major ideas in the lecture from the guidance of
the lecturer's signals of major segments and emphasis. These help them
construct appropriate schematic models of the major portions of the lecture,
even if they lack sophisticated understanding of the content or the rhetorical
structure of expository speech.
Why do the micro-markers not aid the learners' retention of the lecture
content? For one thing, these markers probably do not add enough content to
make the subsequent information more salient or meaningful. For another, the

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

quantity of the markers scattered throughout the lecture probably results in
making the entire lecture merely appear less well organized, a notion that is
implied in research by Hiller et al. (1969), who found that hesitation markers
and vagueness in teachers' classroom speech detracted from learners' retention
of instructional content. Third, it may be that the baseline lecture version was
already slow enough for these learners to derive the most benefit possible from
the pace of the lecture, and no further segmenting orslowingdownof the pace of
delivery of important information would aid further retention of information.
Fourth, however, there is the possibility that the dependent measures were
simply not sensitive enough to the differences in comprehension between
listener groups. However, the virtual identity of the means for the Baseline and
Micro versions suggests that this is not the case, or if it is, then it would require
only the most sophisticated experimental techniques to reveal a difference, in
which case there would be little practical application to teaching situations.
A second issue concerns why the Macro version was consistently superior to
the combined Micro-Macro version. Procedural explanations come first to
mind: Perhaps the particular groups hearing these passages were less proficient,
or the actual recording of the version was inferior in some respect However,
there is no reason to suspect either of these explanations. For one thing, the pre-
tests showed that the different groups were quite homogeneous in proficiency.
Second, subjective listening to the lectures produces no impression of differ-
ences in quality among any of the versions. The only obvious differences are
those of length, yet the total length should not have been an inhibiting factor for
the Micro-Macro version, since it was interrupted at the same regular cloze
intervals as the others. The only evident explanation is that the further addition
of micro-markers to the macro-markers achieved the same result as the Micro
version alone: it increased the listeners' attention requirements without adding
valuable information in any way, thereby detracting from the effect of the
macro-markers alone.
A rather important point arises from this apparent differential effect of
macro- and micro-markers. Whereas the micro-markers included here are of
relatively less semantic value in the lecture information, and often do little else
than allow the speaker time to plan the next utterance, the macro-markers are
explicit expressions of the planning of the lecture information. In using micro
discourse markers, the speaker is merely succumbing to the necessities of on-
line discourse production, while in the case of macro-markers, the speaker must
124 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

devote some attention to the particular phrasing and placement of the expres-
sion. The anticipation and processing by the listener follows accordingly: one
leams to disregard all the minor pause fillers and redundant intersentential
connectors, perhaps making use of the time to process the significant parts of
the text. But on the other hand, the listener knows that paying attention to
markers of the overall organization of the text is a critical skill for the compre-
hension of the information conveyed by the lecture.
A final comment is warranted concerning why the TD2 group did not show a
significant effect like the TD 1 pre-university students and the university group.
One explanation is that these subjects may have been a more homogeneous
group, since most of them were recent arrivals in the USA, just enrolling in the
summer session of the intensive program, and relatively homogeneous in ethnic
and language backgrounds (mainly East Asians). In these respects they were

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

different from the TD 1 group, who had been living in the USA for a longer time
and were ethnically more diverse. Unfortunately, comparisons between these
two groups' proficiency are not possible due to the different test dates and non-
standardized forms. Yet, it is plausible that the TD1 group had greater familiar-
ity with the spoken language (and therefore, facility with bottom-up processes
like word recognition and sentence parsing), owing to their longer residence in
the USA. This would allow higher level, top-down processing more time and
cognitive effort (cf. Hasher and Zacks 1979) to play a greater role in listening
comprehension. On the other hand, the recently arrived TD2 group, although
capable of placing at equivalent levels in the pre-universiry program on the basis
of solid background in academic, literary English (i.e. that used on the place-
ment test), would require a greater degree of attention to bottom-up processing
and therefore be unable to benefit from the signalling of the macro-markers.
8. CONCLUSION
We have attempted here to explore the relationship between modifications in
oral discourse and their effects on second-language learners' comprehension of
the information conveyed. This issue is of ultimate importance not only for the
language teacher who trains L2 learners in listening skills, and the curriculum
developer who devises a program or materials that achieve that training, but
also for teachers and lecturers who teach content subjects to non-native
learners.
The most immediate implications of this study are for these teachers and
lecturers. It is important to realize that a lecture read from a written text will
usually lack the kinds of macro-markers found in the more conversational style
of teaching. A lecture which uses more macro-markers is likely to be easier to
follow. On the other hand, an over-use of micro-markers possibly detracts from
the overall coherence of the lecture. For the curriculum and materials
developer, and for L2 teachers, the macro-markers probably constitute a
relevant focus for second-language classroom activities and instructional
materials. Such a focus is often lacking in current published materials.
Further research, however, is necessary to replicate the present findings, and
to determine whether materials and instruction which exercise learners'
recognition of such markers bring about a higher level of comprehension.
{Received October 1985)
 CRAIG CHAUDRON AND JACK RICHARDS 125

NOTES
• An earlier version of this article was presented at the 19th Annual Convention of Teachers of
English to Speakers of Other Languages, 10 April 1985, in New York City. The authors would like
to thank all those who kindly gave their time, and made their classes available for this research. They
also are grateful to the students who co-operated in doing the lecture comprehension task. In
particular, we would like to thank Janet Leister, Ann Chun, Steve Taylor, and Bruce Horton, of
Hawaii Pacific College, and Rodger Snow, Tom Grigg, Kathy Rulon. Patricia Card, Philip Pinsent,
and Chas Mason, of the English Language Institute, University of Hawaii at Manoa.
1
The total duration of the lectures varied between 23 and 27 minutes because of the added
material in each of the versions from Micro to Micro-Macro. These times include eight minutes of
pauses, because there were 13 pauses of 40 seconds inserted at systematic intervals, as described in
the next section.
2

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

At about one to one-and-a-half minute intervals, a 40-second pause was inserted into all the
versions of the lecture. Corresponding to each pause, a special response answer sheet in a bound
booklet was prepared which summarized the lecture content as a short text, virtually verbatim from
the lecture, but with six or seven selected vocabulary items deleted as cloze items. On hearing a tone,
the subjects turned a page in their answer booklets, and filled in the blanks in the short passage in the
booklet. A second tone indicated the end of the pause, at which point they were to be prepared to
listen again. There were 13 such pauses in the lectures, and a total of 79 cloze items. The duration of
the response pause was intentionally short, and the subjects were prohibited from reading the
passages ahead of the interruption by a shielding mechanism, in order to enhance the effectiveness
of the test as a measure of short-term recall, rather than as a measure of reading ability.
3
Due to obviously unequal numbers of subjects in some classes, and to the fact that the pre-
university subjects were divided into two ability levels, several classes were split up at random into
different lecture version treatments, in order to obtain as balanced a distribution as possible across
lecture versions of subjects of different abilities.
4
During the practices, the researchers observed to make sure that the subjects were performing
correctly, and not looking ahead at the cloze passages until the proper interruption in the lecture. As
an aid for the subjects to follow references to geographical areas in the lecture, a map of the United
States was provided. It was believed that the majority of the subjects learned the procedure quite
readily, and only a few needed to be reminded during the first few interruptions not to look ahead at
the cloze passages or to take notes.
5
Prior research with cloze procedures has demonstrated that an acceptable-word scoring only
slightly increases reliability (and of course eases difficulty), but in this procedure, it was felt that only
an exact-word scoring would technically qualify as a correct measure of listening comprehension,
rather than reading comprehension.
* Due to missing data in some cases, not all analyses could include the total number of subjects.
Kuder-Richardson 21 reliability estimates were also made of the dependent measures. Although
this is ordinarily not considered appropriate for a cloze test, the fact that only one right answer was
considered warranted using this generally conservative estimate.
7
The reason for using co-variate$ is that variance in proficiency should be removed from the
dependent measures in order to determine the difference between the four versions independently
of learners' ability. The cloze measure was highly correlated with the placement test scores (/• ranged
from 0 J 9 to .56, p < .001, between the cloze, and listening, reading, and TOEFL scores for the
groups), so especially the listening placement tests, for which all subjects had scores, were used as
co-variates (Michigan and CELT tests for the pre-university group, and the Plaister Auding Test for
the university group).

REFERENCES
Adams, M. J. and A. Collins. 1979. 'A schema-theoretic view of reading' in R. Freedle
(ed.): Advances in Discourse Processes, VoL II: New Directions in Discourse Process-
ing. Norwood, N. Jj Ablex.
126 DISCOURSE MARKERS AND THE COMPREHENSION OF LECTURES

Carrell, P. L. 1984. 'Evidence of a formal schema in second language comprehension.'

Language Learning 34/1:87-112.
Carrell, P. L., and J. C. Eisterhold. 1983. 'Schema theory and ESL reading pedagogy.'
TESOL Quarterly 17/4:553-73.
Cervantes, R. 1983. 'Say It Again, Sam: The Effect of Exact Repetition on Listening
Comprehension.' University of Hawaii at Manoa. Manuscript.
Chaudron, C. 1983. 'Simplification of input topic reinstatements and their effects on L2
learners' recognition and recall." TESOL Quarterly 17/3:437-58.
Chaudron C. 1985. 'A method for examining the input/intake distinction' in S. M. Gass
and C. Madden (eds.): Input and Second Language Acquisition. Rowley, MA:
Newbury House.
Chaudron C. (forthcoming). 'Second Language Classrooms: Research on Teaching and
Learning.'
Clark, H. H. and E. V. Clark. 1977. Psychology and Language. New York: Harcourt

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

Brace Jovanovich.
Clements, P. 1979. The effects of staging on recall from prose' in R. O. Freedle (ed.):
New Directions in Discourse Processing. Norwood, NJL Ablex.
Cook, J. R. S. 1975. 'A Communicative Approach to the Analysis of Extended Mono-
logue Discourse and Its Relevance to the Development of Teaching Materials for ESP.'
M. Litt thesis, University of Edinburgh, Scotland.
van Dijk, T. 1977. 'Semantic macro structures and knowledge frames in discourse com-
prehension' in P. Carpenter and M. Just (eds.): Cognitive Processes in Comprehension:
Proceedings of the Twelfth Carnegie-Mellon Symposium on Cognition. Hillsdale, NJJ
Lawrence Eribaum.
Dudley-Evans, A. and T. F. Johns. 1981. 'A team teaching approach to lecture compre-
hension for overseas students' in The Teaching of Listening Comprehension. London:
The British Council.
Hasher, L., and R. T. Zacks. 1979. 'Automatic and effortful processes in memory.'
Journal of Experimental Psychology: General 108/3:356-88.
Henning, G., N. Gary and J. O. Gary. 1981. 'Listening Recall—A Listening Compre-
hension Test for Low Proficiency Learners.' Paper presented at AILA, Lund,
9 August, 1981.
HUler, J. H., G. A. Fisher, and W. Kaess. 1969. 'A computer investigation of verbal
characteristics of effective classroom lecturing.' American Educational Research
Journal 6:661-75.
Kelch, K. 1985. 'Modified input as an aid to comprehension.' Studies in Second
Language Acquisition 7/1:81-9.
Long, M. H, 1985. 'Input and second language acquisition theory' in S. M. Gass and
C. Madden (eds.): Input and Second Language Acquisition. Rowley, MA: Newbury
House.
Meyer, B. J. F. 1975. The Organization of Prose and its Effects on Memory. Amsterdam:
North-Holland.
Meyer, B. J. F. 1984. Text dimensions and cognitive processing' in H. MandL N. L.
Stein, and T. Trabasso (eds.): Learning and Comprehension of Text. Hillsdale, NJu
Lawrence Eribaum.
Meyer, B. J. F., D. M. Brandt, and G. J. Bluth. 1980. 'Use of the top-level structure in
text key to reading comprehension of ninth-grade students.' Reading Research
Quarterly 16/1:72-103.
Meyer, B. J. F. and G. E. Rice. 1982. The interaction of reader strategies and the
organization of text' Text 2:155-92.
Murphy.D. andC.Candlin. 1979.'Enguieeringlecturediscourseandlisteningcompre-
hension.' Practical Papers in Language Education. VoL 2. Lancaster, England:
University of Lancaster.
 CRAIG CHAUDRON AND JACK RICHARDS 127

Richards, J. C. 1983. 'Listening comprehension: approach, design, procedure.' TESOL

Quarterly 17/2:219-40.
Rumelhart, D. E. 1977. Toward an interactive model of reading' in S. Domic (ed.):
Attention and Performance, VI. Hillsdale, N J J Lawrence Erlbaum.
Sawa, G. H. 1985. 'Modified Input as an Aid to Comprehension.' University of Hawaii at
Manoa. Manuscript
Young, L. and B. Fitzgerald. 1982. Listening and Learning. Rowley, MA: Newbury
House.
Yuan, D-z. 1982. 'Chinese Scientists' Difficulties in Comprehending English Science
Lectures.' M A in TESL thesis, University of California at Los Angeles, Los Angeles.

APPENDDCA

Downloaded from http://applij.oxfordjournals.org/ at Russian Archive on December 4, 2013

Micro-Markers
Segmentation Temporal Causal Contrast Emphasis
Well At that time So Both Of course
OK And Then But You can see
Now After this Because Only You see
And For the moment On the other hand Actually
Right Eventually Obviously
All right Unbelievably
As you know
In fact
Naturally

Macro-Markers
What I'm going to talk about today is something Another interesting development was
you probably know something about already— You probably know that
What [had] happened (then/after that] was [that] The surprising thing is
We'll see that As you may have heard
That/this is why Now where are we
To begin with This is how it came about
The problem [herej was that You can imagine what happened next
This/that was how In this way
The next thing was It's really very interesting that
This meant that This is not the end of the story
One of the problems was Our story doesn 't finish there
Here was a big problem And that's all we 'II talk about today
What we've come to by now was that