Reading in a Foreign Language
Volume 15, Number 1, April 2003
Metadiscourse and ESP reading comprehension:
Belinda Crawford Camiciottoli
Recent trends in the study of written texts reflect a growing interest in interaction between readers and writers. Several studies have focused on metadiscourse as an important interactive feature that is believed to facilitate the reading process. While several authors have studied metadiscourse from the descriptive and contrastive perspectives, there is a lack of experimental work on this topic. This paper describes exploratory classroom research with a group of Italian university students to gain further insight into the effect of metadiscourse on ESP reading comprehension. Two groups of students read selected extracts from two versions of the same text differing according to quantity and type of metadiscourse. Each group then took a reading comprehension test and their mean scores were compared. The findings suggest that a more pronounced use of metadiscourse may be associated with improved comprehension in some cases. A post-reading questionnaire showed that students had substantially no awareness of metadiscourse. The results provide useful indications for further research and also highlight the need for targeted instruction on metadiscourse in ESP reading courses.
The notion of reading as an interactive process of bottom-up, top-down and metacognitive skills (Dubin and Bycina, 1991; Shih, 1992; Vacca et al., 1995) is now well consolidated in both first language (L1) and second language (L2) instructional frameworks. This approach is particularly effective in teaching reading skills for academic or special purposes. In addition to decoding meaning from print with bottom-up skills, successful readers implement top-down skills to activate their prior knowledge of content and use textual cues to help them cope with new information. Parallel to this interactive process between reader and content, there is also another important type of interaction: the one between reader and writer. This "dialogue" is known as metadiscourse, defined by Vande Kopple (1997: 2) as "discourse that people use not to expand referential material, but to help their readers connect, organize, interpret, evaluate and develop attitudes towards that material."
Several studies have discussed the positive effects of the presence metadiscourse in texts. With reference to Halliday's (1985b) metafunctional theory of language, on the interpersonal level, Schiffrin (1980: 231, as cited in Hyland, 2000: 109) and Crismore (1989) both point out that metadiscourse allows written texts to take on some features of spoken language (e.g., personal pronouns to establish an "I-you" relationship), and thus become more "reader-friendly". On the textual level, Crismore & Farnsworth (1990) and Crismore (1989) note that the discourse structuring functions of metadiscourse guide readers through a text and help them to organize content as they read, thus fostering global comprehension. Crismore further suggests that metadiscourse can promote critical thinking as readers are able to formulate their own opinions and compare them to those of the writer.
Other benefits of metadiscourse derive from its use of explanatory and persuasive elements (e.g., code glosses, attitude markers, evidentials) which attest to its key rhetorical function (Crismore, 1989; Hyland, 1999; Hyland 2000). In fact, writers use these devices to produce a desired effect, depending on their underlying purposes and perception of readers' expectations. For example, in expert to non-expert communication (e.g., textbooks) metadiscourse helps to present information in a clear, convincing and interesting way in an effort to promote acceptance and understanding, as well as reader-writer solidarity. It is also an important persuasive resource used to influence readers' reactions to texts according to the values and established conventions of a given discourse community.
In L2 instructional contexts, it has been posited that an awareness of metadiscourse is particularly useful in helping non-native speakers of English with the difficult task of grasping the writer's stance when reading challenging authentic materials. Bruce (1989: 2) suggests that this ability enables non-native learners to better follow the writer's line of reasoning in argumentative texts. Vande Kopple (1997: 14) observes that specific instruction on metadiscourse can be useful to help L2 readers learn to distinguish factual content from the writer's commentary.
Metadiscourse has been investigated from a descriptive standpoint and has been shown to be a prominent feature of various types academic discourse. These include school textbooks (Crismore, 1989; Crismore and Farnsworth 1990), university textbooks (Hyland, 2000; Bondi, 1999) and doctoral dissertations (Bunton, 1999). It has also been studied comparatively in order to understand differences in usage across cultures (Mauranen, 1993; Valero-Garces, 1996). Some work has focused on metadiscourse in student writing. Intaraprawat & Steffensen (1995) analyzed ESL university students' essays and concluded that good writers used a greater variety of metadiscourse than poor writers. Steffensen & Cheng (1996) conducted an experiment to investigate the effect of targeted instruction on metadiscourse on the writing abilities of native-speaker university students. An experimental group that had been taught the form, function and purpose of metadiscourse learned to use it effectively and produce compositions that earned significantly higher scores than those of a control group, which had received no instruction on metadiscourse. However, little experimental work has been done on the effects of metadiscourse on reading comprehension. Two studies have been undertaken with native speakers of English. Crismore (1989) attempted to determine whether including informational and attitudinal metadiscourse in passages of social studies textbooks would influence reading retention (among other factors) with sixth graders. She found that there was some improvement in retention after reading passages with both types of metadiscourse, but only with certain participant subgroups. In a later study with ninth graders, Crismore and Vande Kopple (1997) investigated the effects of hedging (e.g., metadiscursive devices that express the writer's commitment to the truth value of the proposition being made) on reading retention. Experimental groups read passages from both social studies and science textbooks containing various types of hedging, while a control group read corresponding passages in which all hedging had been removed. Even if other factors (e.g., gender, subject matter and type/position of hedging) were also found to have an influence on retention scores, the authors propose that, in general, students learned more from reading science and social studies passages with hedging included.
Although the findings of these studies do not provide clear evidence that the presence of metadiscourse in a text improves comprehension, they do suggest that it has a facilitating role, and is therefore a topic that merits further study. The purpose of this action research was to gain more insight into the influence of metadiscourse on reading comprehension levels in an L2 setting: a group of Italian ESP students at the Faculty of Economics of the University of Florence (Italy). The research question is: Are these L2 readers able to understand a text containing more metadiscourse better than one with less? The study was undertaken in two phases:
The two texts utilised for this study come from the works of the great nineteenth-century British economist Alfred Marshall. Groenenwegen (1995) notes that Marshall's theories have had a major influence on economic thought. His works are considered classics and are still widely read by students of economics. Principles of Economics was first published in 1890 in the form of a main treatise, while an abridged version, Elements of Economics of Industry, was later published in 1892. The availability of two authentic versions of the same text by the same author presents a rather unique opportunity to investigate metadiscourse without having to artificially manipulate texts. Marshall himself indicated that in Elements of Economics of Industry, he had sought to adapt Principles of Economics to the needs of less expert student readers, primarily by simplifying the more complex theoretical concepts and by eliminating what he considered to be minor points. However, an analysis of the two texts by Del Lungo (1998) points out that are also important metadiscursive differences. More specifically, she found that the abridged version contains less metadiscourse. This study builds on these findings, with an underlying hypothesis that the long version (Principles of Economics) is actually more comprehensible to L2 readers than the abridged version (Elements of Economics in Industry) due to the presence of more metadiscourse.
Three corresponding passages of Principles of Economics and Elements of Economics of Industry (hereinafter referred to as Principles and Elements, respectively) were selected for the experiment. The extracts, ranging from 120 to 254 words, were taken from three different corresponding parts of the two texts (Books I, IV and V) in order to obtain a sufficiently representative sample. The extracts from Book I (1a and 1b) introduce the relationship between the incentive to earn money and the level of individual wealth. The extracts from Book IV (2a and 2b) explain how human population growth is related to the production of wealth. The extracts from Book V (3a and 3b) illustrate concept of economic utility. Minor adjustments were made to render the passages more suitable for reading as isolated paragraphs. In the extract from Book V of Principles, I removed a lengthy footnote to the main text.
The extracts were then analysed in terms of the quality and quantity of metadiscourse found in them (see Appendix A). The analysis was based on Hyland's taxonomy (Hyland, 2000: 111), which identifies various functional categories of metadiscourse found in university level textbooks as follows:
Table 1 illustrates the metadiscursive differences between the two text treatments. Three of Hyland's categories do not appear in the extracts (evidentials, relation markers, and attitude markers), which is not surprising since they tend to be more frequent in strongly persuasive genres, such as research articles (Hyland, 1999). As can be seen, there is some overlapping of categories (e.g., frame markers can also include person markers, hedges and boosters).
Table 2 summarizes the metadiscursive differences between the two text treatments in terms of frequency. The analysis confirms Del Lungo's previous findings that there is the greater amount and variety of metadiscourse in Principles as compared to Elements. On a general level, it is also in line with Hyland's (1999) study of metadiscourse in university textbooks, showing that textual metadiscourse is more prominent than interpersonal metadiscourse (14 vs. 9 items, respectively).1
Table 2: Quantitative analysis of metadiscourse in Principles vs. Elements
Participants and administration procedures
The test and questionnaire were administered to two large sections (ranging from 100-120 students) of the ESP reading course at the Faculty of Economics. From those sections, two groups of 55 participants were randomly selected (approximately every other test). Students can enroll in the course starting only from the third year of their degree program. Therefore, they come to the course with quite a lot of content-specific background knowledge acquired from their native language coursework. The English proficiency level of the students ranged from intermediate to upper intermediate. The experiment was conducted towards the end of the semester-long course to prepare students for a multiple-choice examination on economics texts. Thus, the students had already been introduced to the multiple-choice format, as well as the English vocabulary and type of questions used in the test. In fact, the reading course focused on global reading strategies (e.g., identifying the main points, inferring meanings and recognising discourse functions) and vocabulary building.
Group 1 read the three extracts from Principles and then took a reading comprehension test and completed a questionnaire. Group 2 instead read the three Elements extracts, took the same reading comprehension test and completed the same questionnaire. Both the test and the questionnaire were administered during regular class periods with no time limit.
A reading comprehension test was developed for the text treatments of this experiment (see Appendix B), consisting of four multiple-choice questions, each with four options. The first question was of a global nature referring to the content of all three extracts. The second, third and fourth questions focused instead on the individual content of extracts #1, #2, and #3, respectively. Each question was weighted 1 point.
A five-item post-reading questionnaire (see Appendix C) was also designed specifically for this study. Unlike the reading comprehension test, the questionnaire was formulated in Italian (the native language of the students) in an attempt to avoid any failure to understand or correctly interpret the questions. Two questions involved students' evaluation of the levels of difficulty and comprehension of the two texts, while the other three were designed to measure students' awareness of metadiscourse. Each item offered a closed response format with four options. Scores ranged from four points for the option indicating the highest level of understanding and awareness of metadiscourse to one point for the lowest.
For the reading comprehension test, mean scores were calculated globally (minimum 0/maximum 4) and for also each individual question (1 point for the correct response vs. 0 points for an incorrect response). It seemed useful to compare responses to individual items since they were not all of the same nature (e.g., global vs. individual focus) and since selected extracts contained varying types of metadiscourse. The mean score of each questionnaire item was also calculated.
All the means were then compared by using a two-tailed t-test, with a .05 level of significance required to reject the null hypothesis that there would be no statistically significant differences between the two treatment groups.
Table 3 illustrates the results of the t-test analyses for the reading comprehension test. Out of a maximum score of 4, the mean scores were 2.98 for the Principles group and 2.71 for the Elements group. The difference between the two means was not statistically significant. However, at the level of individual questions, two significant differences were found. In the Principles group, students had significantly higher scores on question #1 (p = .01) that asked students to identify the main function of the three extracts and question #4 (p = .02) that asked students to determine the main point of extract #3.
Table 3: Reading comprehension test: comparison of mean scores (both global and for individual items) assessed by means of the independent samples t-test
N = 55 in both sample groups
df = 108
t critical two-tail = 1.98
The results of the analysis of the questionnaire are shown in Table 4. For question #1, the mean scores of 1.89 (Principles) and 2.00 (Elements) indicate that both texts were considered rather difficult. Yet, as shown by question #2, both groups were fairly certain of having understood the main points of at least some of the extracts, with mean scores of 2.98 (Principles) and 3.05 (Elements) for question #2. There were no significant differences between the mean scores relating to the levels of difficulty and understanding of the two texts. For questions 3, 4 and 5 which inquired about the presence of various features of metadiscourse found in the two texts, the mean scores for both groups were rather non-committal ranging from 2.27 to 3.20, corresponding to the options I don't know and somewhat. The only significant difference between the means of the two groups was found for question #5, which inquired about the presence of framing devices. However, in this case, the Elements treatment group had a significantly higher mean score (p =.04), indicating that the students perceived more of this type of metadiscourse in the Elements extracts than in the Principles extracts.
Table 4: Post-reading questionnaire: comparison of mean scores on individual items
N = 55 in both sample groups
df = 108
t critical two-tail = 1.98
The statistical analysis of the mean scores did not produce conclusive evidence that the Principles extracts containing more metadiscourse were more comprehensible than the Elements extracts with less. However, the fact that there were significant differences between the two groups for two of the four items provides some food for thought. In question #1, the Principles group more clearly understood the main discourse function of the three extracts. A possible explanation could be found in the more extensive use of frame markers. By signalling the author's intention, they essentially introduce the concept before proceeding to exemplify it. Thus, this discourse pattern (introduction of concept followed by an example) becomes evident to readers. In question #4, unlike Elements extract 3b, Principles extract 3a contains an introductory paragraph in which the author uses frame markers to guide and prepare the reader for the following discussion. Again, the frame markers preview the concepts of 'distant return' and 'balance of efforts and sacrifices', which are then basically reiterated in the example that follows. This provides readers with repetition and reinforcement of the content. It is also worth noting that this particular passage makes the most use person markers, which tends to support the idea that this form of reader-writer solidarity promotes comprehension (Crismore, 1989). Furthermore, in the concluding sentence of the passage, Principles extract 3a contains a hedge (This illustration may serve ) that could function to mitigate the writer's authorial stance, thus rendering in more "reader-friendly". As Crismore and Vande Kopple (1997) found in their study, hedges generally had a positive affect also on readers' attitudes towards reading a given text. It stands to reason that readers who have a more positive attitude towards a text are also likely to understand more of what they read.
The analysis of the questionnaire revealed that both groups of students perceived the texts to be rather difficult. As regards lexis, even though some words were probably unfamiliar, there were also many Italian or Latin cognates which would presumably not present difficulties for native speakers of Italian. It should also be remembered that these students have a consolidated background in economics, and therefore probably did not encounter particular problems with the concepts presented. For this reason, this perception of difficulty was likely to be more related to the stylistic features of the two texts, rather than the content. Marshall's sentences tend to be quite long and elaborately structured with many embedded clauses. The average sentence length of the two Marshall texts was 42 words per sentence. This can be compared to 23.8 words in average modern-day educated and scientific writing and 24.9 in the average academic writing in all disciplines, according to a study on average sentence length conducted by Peck MacDonald (1990). This longer length also contributes to the relatively high level of lexical density found in the texts, which determines the density of information and thus, the complexity of a text. Halliday (1985a) measures lexical density in terms of the average number of lexical items per clause. According to these parameters, the lexical density of the Marshall texts yields a relatively high average score of 9.9, based on 100-word samples from each of the six extracts. In addition, several extracts (e.g., 2a/2b and 3a/3b) present low frequency lexical items (e.g., sustenance, worthily (2a); forethought, duty (2b); behalf, efforts (3a); implements, elapse (3b)). There is also some use of nominalisation. (e.g., " a means to the sustenance of man " = a means to sustain man (2a); "...the labour of making them being counted " = the labour to make them is counted (3a/3b). These are all factors which tend to render a written text more difficult to understand. Moreover, the dated and literary use of some conjunctions (e.g., "The production of wealth is but a means " and " for if the poorer man spends money ") may have caused problems. In spite of this perceived difficulty, both groups seemed to be rather confident in their ability to understand the main points of the extracts, as indicated by the medium-high scores on questionnaire item #2 (2.98 for Principles and 3.05 for Elements). However, it is interesting to note that, unlike the Principles group whose two corresponding scores matched exactly (2.98 global mean comprehension score vs. 2.98 mean perceived ability score), the Elements group somewhat overestimated their ability (2.71 global mean comprehension score vs. 3.05 perceived ability score). At the level of speculation, it could be that the presence of metadiscourse assists readers, even if on a subconscious level, in evaluating their own understanding of texts, as previously suggested by Crismore (1989).
The outcome of the questionnaire analysis showed that the students' awareness of metadiscourse is indeed scarce. This was indicated by the generally indecisive responses to the questions about the presence of metadiscourse. Moreover, the prompting effect of the questions themselves apparently had no impact on students. This was true even for question #4 which, by means of a note, specifically alerted students to the use of personal pronouns to create a dialogue between author and reader. Furthermore, the significantly higher score in the Elements group for question #5 inquiring about the presence of introductory passages (i.e., metadiscourse containing frame markers, code glosses and person markers) is quite paradoxical, since the Elements extracts actually have fewer of these passages than the Principles extracts. Apparently, these students have difficulty in conceiving of reading as interaction between reader and writer, and perhaps approach reading as a fact-finding mission only.
On a general level, the results of this study lend further support the idea that metadiscourse can have a positive influence on comprehension. The greater presence of some types of metadiscourse (e.g., frame markers, person markers and hedges) could be linked to the better performance of the Principles group in some of the comprehension questions. However, this interpretation needs to be couched with caution. As previously mentioned, this action research was undertaken as an exploratory initiative. It was conducted under conditions imposed by the instructional setting, and therefore presents some limitations. Although the general English language proficiency level of the students was homogeneous, it would be important to take account of possible differences in individual reading levels by means of a pre-test. Moreover, a larger-scale study with more participants, longer text treatments and more test and questionnaire items would provide more data, and therefore a more reliable picture. Nevertheless, these findings indicate that metadiscourse is a topic that deserves attention in L2 reading research, and perhaps most importantly, identifies some specific directions for further research. In fact, it would seem that certain types of metadiscourse may be more facilitating than others during reading. It would be worthwhile to set up more articulated experiments that isolate these different forms of metadiscourse in order to determine their effect on comprehension. In short, this study is best viewed as a springboard for more refined experimental work on specific aspects of metadiscourse under more controlled conditions to filter out potentially influential variables.
On a practical level, these findings may be used to determine instructional actions to be undertaken in this or similar teaching contexts. In fact, the most important contribution of this study is its classroom applications. Since students seem to have little awareness of metadiscourse and the interactional aspects of reading in general, specific instruction should be integrated into the ESP reading course to help students become more successful readers. This is a particularly crucial aspect in academic fields in which most students have scientific backgrounds and scarce knowledge of linguistic notions. From the textual viewpoint, students can be asked to identify instances of frame marker previews and then predict content. Attention to logical connectives will help students analyze the writer's line of reasoning and rhetorical strategies. Tracing endophoric markers can help students understand the macrostructure of a text and also encourage them to retain and build on newly acquired knowledge. Students can also be encouraged to notice code glosses as the writer's way of helping them to understand important new concepts. On the interpersonal level, students can look for hedges, boosters, and first person pronouns and reflect on why the writer has chosen to use these features. Attitude markers can prompt students to contribute their own ideas and thus critically react to the text. Although it may not be appropriate to burden students with the terminology of metadiscourse, teachers can nonetheless exploit the concepts when working with students in this way. For example, a series of simple questions (e.g., Where does the writer tell you what is coming next, Where does the writer mention other parts of the text, How does the writer tell you that he/she is not completely sure, etc.) could be substituted. Once this type of instruction has been undertaken, it would be interesting to investigate further the impact of metadiscourse on comprehension to determine potential gains. This type of research would not only heighten our understanding of the reading process on a general level, but would also lead to more effective teaching methodologies and better criteria for the selection of materials for ESP reading instruction.
1. Apparently, the primary aim of textbook writers to guide readers and clarify content meaning has not changed much over the last hundred years or so.
Bondi, M. (1999). English across genres: Language variation in the discourse of economics. Modena: Edizioni Il Fiorino.
Bruce, N. J. (1989, August). The role of metadiscourse, speech acts and the language of abstraction in a top-down approach to teaching English for academic purposes. Paper presented at the European Languages for Special Purposes Symposium, Budapest, Hungary (ERIC Document Reproduction Service No. ED331299).
Bunton, D. (1999). The use of higher level metatext in Ph.D theses. English for Specific Purposes, 18 supplement, 41-56.
Crismore, A. (1989) Talking with readers: Metadiscourse as rhetorical act. New York: Peter Lang Publishers.
Crismore, A. & Farnsworth, R. (1990). Metadiscourse in popular and professional science discourse. In W. Nash, (Ed.), The writing scholar (pp. 119-136). Newbury Park: Sage Publications.
Crismore, A. & Vande Kopple, W. J. (1997). Hedges and readers: effects on attitudes and learning. In S. Markkanen & H. Schroeder (Eds.), Hedging and discourse: Approaches to the analysis of a pragmatic phenomenon in academic texts (pp. 83-114). Berlin: Walter de Gruyter and Co.
Del Lungo, G. (1998, April). Metadiscourse in Marshall's Elements of economics of industry. Paper presented at the International Conference on Institutional Linguistics, Bologna, Italy.
Dubin, F. & Bycina, D. (1991). Academic reading and the ESL/EFL teacher. In M. Celce-Murcia (Ed.), Teaching English as a second or foreign language (pp. 195-215). Boston: Heinle and Heinle Publishers.
Groenenweg, P. D. (1995). A soaring eagle: Alfred Marshall 1842-1924. Aldershot, UK: Edward Elgar Publishers.
Halliday, M. A. K. (1985a). Spoken and written language. Oxford: Oxford University Press.
Halliday, M. A. K. (1985b). An introduction to functional grammar. London: Edward Arnold.
Hyland. K. (1999). Talking to students: Metadiscourse in introductory textbooks. English for Specific Purposes, 18, 3-26.
Hyland, K. (2000). Disciplinary discourses: Social interactions in academic writing. Harlow, UK: Longman.
Intaraprawat, P. & Steffensen, M. S. (1995). The use of metadiscourse in good and poor ESL essays. Journal of Second Language Writing, 4, 253-272.
Marshall, A. (1932). Elements of economics of industry. London: MacMillan.
Marshall, A. (1961). Principles of economics. London: MacMillan.
Mauranen, A. (1993). Contrastive rhetoric: Metatext in Finnish-English economics texts. English for Specific Purposes, 12, 3-22.
Peck MacDonald, S. (1990). The literary argument and its discursive conventions. In G. Cooper & S. Greenbaum (Eds.), The writing scholar (pp. 31-62). Newbury Park: Sage Publications.
Schiffrin, D. (1980). Metatalk: Organisational and evaluative brackets in discourse. Social Inquiry: Language and Social Interaction 50, 199-236.
Shih, M. (1992). Beyond comprehension exercises in the ESL academic reading class. TESOL Quarterly, 26, 289-311.
Steffensen M.S. & Cheng, X. (1996). Metadiscourse and text pragmatics: How students write after learning about metadiscourse. In: L. F. Bouton (Ed.), Pragmatics and language learning. Monograph Series, vol. 7 (pp. 153-171). (ERIC Document Reproduction Service No. ED400709).
Vacca J. L., Vacca, R. T. & Grove, M. K. (1995). Reading and learning to read. New York: HarperCollins College Publishers.
Valero-Garces, C. (1996). Contrastive rhetoric: Metatext in Spanish-English economics texts. English for Specific Purposes 15, 279-94.
Vande Kopple, W.J. (1997, March). Refining and applying views of metadiscourse. Paper presented at the 48th Annual Meeting of the Conference on College Composition and Communication, Phoenix, AZ. (ERIC Document Reproduction Service No. ED411529).
Comparative analysis of metadiscourse (in italics) in Marshall's Principles of Economics vs. Elements of Economics in Industry
Reading Comprehension Test (correct response in italics)
Instructions: Choose the best alternative and write your answer in the boxes provided below.
Questionnaire (translated version)
Instructions: Please check the option that applies to you.
About the Author
Belinda Crawford Camiciottoli is an EAP/ESP lecturer at the University of Florence (Italy) Faculty of Economics. She has published articles in the areas of L2 reading instruction, modality and spoken academic discourse. In 2002, she was a Morley Scholar at the English Language Institute of the University of Michigan.