METaDISCOURSE IN lINgUISTICS MaSTER’S ThESES IN ENglISh l1 aND l2

Recent investigations of academic discourse have revealed considerable variations in the interpersonal uses of language: academic texts representing different genres, disciplines and languages/cultures show differences in the ways writers create interaction in text. Studies of English native (L1) and non-native/interlanguage (L2) texts have also shown variability in the use of interpersonal strategies. Many contrastive studies favour the Research article genre, whereas other genres seem to be under-investigated. This paper focuses on the Master’s thesis genre in the discipline of Linguistics1 in different cultural/educational contexts: British and Lithuanian universities. It describes patterns of metadiscourse in the MA thesis genre; it compares metadiscourse strategies in English texts by L1 and L2 writers; it considers the role of institutional practices and individual writer style in the way writers manage their discourse.


Introduction
Recent investigations of academic discourse have revealed considerable variations in the interpersonal uses of language: academic texts representing different genres, disciplines and languages/cultures show differences in the ways writers create interaction in text.Studies of English native (L1) and non-native/interlanguage (L2) texts have also shown variability in the use of interpersonal strategies.Many contrastive studies favour the Research article genre, whereas other genres seem to be under-investigated.This paper focuses on the Master's thesis genre in the discipline of Linguistics1 in different cultural/educational contexts: British and Lithuanian universities.It describes patterns of metadiscourse in the MA thesis genre; it compares metadiscourse strategies in English texts by L1 and L2 writers; it considers the role of institutional practices and individual writer style in the way writers manage their discourse.

A new model of metadiscourse
In discourse literature definitions of metadiscourse have varied from broad ones, such as "writing about writing" (Williams 1981, 211) or "discourse about discourse or communication about communication" (Vande Kopple 1985, 83), to more specific ones, such as "writing about the evolving text rather than referring to the subject matter" (Swales 2004, 121).the models of metadiscourse have also varied: earlier models have grouped metadiscourse categories into 'interpersonal' and 'textual' (Crismore et al 1993), whereas later models have distinguished between 'interactional' and 'interactive' (Hyland andtse 2004) or 'intra-textual' and'inter-textual' (Ifantidou 2005) markers.Some scholars (Mauranen 1993(Mauranen , 2008) have used the terms 'text reflexivity' or 'discourse reflexivity' to refer to metadiscoursal uses of language.
In this study, I define metadiscourse as the language used to express the author's explicit awareness and management of the discourse-as-process; which includes awareness and management of the organization of the text, of the participants of the discourse process: the reader and the self, and of the author's attitude towards the discourse process.My model includes three major categories: text-organising, participant-oriented and evaluative metadiscourse.Text-organising metadiscourse can help the writer to signpost the structure of the unfolding text and to signal the structural links between the various parts of the developing argument; participant-oriented metadiscourse can help the writer to make himself/herself visible and also engage the reader in the discourse process; and evaluative metadiscourse can help the writer to position himself/herself in the text and indicate explicitly his/her attitude towards the ongoing discourse.A taxonomy of metadiscourse is given in Table 1. at the rhetorical level, the use of metadiscourse has been proved to be an important strategy (Ifantidou 2005).It helps the writer to guide the reader through the unfolding text and engage in the discourse process; it makes the text more interactive and therefore accessible to the reader.the reader's task then of understanding and interpreting the message becomes easier.Metadiscourse can be said to be an indicator of author explicitness and, consequently, an indicator of text quality.

Research of metadiscourse in academic genres
although metadiscourse is a relatively new area of linguistics, there has been a vast array of studies of both spoken and written texts, representing different genres, disciplines and languages/cultures.The scope of this paper only allows us to list a few studies to illustrate the range of academic genres in which metadicourse studies have been carried out: they have included coursebooks (Moreno 2003;Hyland 2005), science popularizations (Crismore and Farnsworth 1990;Hyland 2005), research articles (Mauranen 1993;Valero-Garces 1996;Bäcklund 1998;abdi 2002;Breivega et al 2002;Hyland 2005;Pisanski Peterlin 2005), doctoral theses (Swales 1990;Bunton 1999), undergraduate essays (Crismore et al 1993;Barton 1995;Krause and o'Brien 1999;Ädel 2003).Master's theses have been explored by telenius (1994), Hewings and Hewings (2002), Hyland and tse (2004).My research draws on Hyland and tse's (2004) work, but, since their model is different from the one I have created, I will not be able to make a valid comparison of the results of my study with the results of their study.

Rationale and aims of the study
The use of metadiscourse strategies shows to what extent student writers are aware of discourse-as-process and how they manage it.In other words, (1) how they structure their text, (2) how they engage readers and themselves and (3) how they evaluate their own writing.In broader terms, metadiscourse strategies in Ma theses show how independent and confident students are as researchers and academic writers -as novice members of the academic discourse community; how they perceive their audience and themselves, what sort of interpersonal relations they establish; how they persuade the expert members of the discourse community (examiners/reviewers) to accept their ideas.
The aims of this study are manifold: (1) to develop a methodological framework for analysing metadiscourse in the Master's thesis genre; (2) to describe patterns of metadiscourse in the Ma thesis genre in the discipline of Linguistics; (3) to compare the use of metadiscourse in native and non-native/interlanguage English Ma theses from British and Lithuanian universities.

data and methods
the data used for the study consists of 40 Master's theses in Linguistics (see footnote 1): 20 texts in English as the mother tongue from 2 British universities (L1 corpus) and 20 texts in English as a foreign language (Lithuanian-English interlanguage or L2 corpus) from 2 Lithuanian universities.Total corpus size is 537 975 words.
To describe the patterns of metadiscourse in the Ma thesis genre, which involves functional analysis, I had to deal with the problem of fuzziness of metadiscourse: since it is a vague category, it is not always easy to decide whether the 'candidate' item is metadiscoursal or propositional (non-metadiscoursal). the criterion of reflexivity is helpful in this respect.Example (1) below illustrates the kind of difficulty that the analyst faces in trying to decide whether the candidate item 'this study' is an instance of reflexive language use or not, i.e., whether it refers to itself (the current text) or to the event which is part of the world outside the text.In this case, I decided that 'this study' is reflexive (and metadiscoursal) since it can be substituted by 'this text / thesis'.Example ( 2) is an unambiguous instance of reflexive/metadiscoursal use of 'this study' (= this text).Example (3) is a clear instance of propositional use of 'this study' (= experiment, project, test).
The UG principle that was investigated in (1) this study was the subjacency principle.(NS-2) this study deals with previous research into language attrition… (NS-2) To compare the use of metadiscourse in the native (L1) and interlanguage (L2) English texts, I used the Log Likelihood Calculator2 , which indicated significant differences in the frequencies of categories in the two corpora.In this study, I use the widely-accepted terms 'overuse' and 'underuse' in a pure quantitative sense -to refer to higher and lower frequencies of metadiscourse markers in the L2 corpus relative to the L1 corpus.I do not, however, imply that the terms 'overuse' or 'underuse' have the connotation of 'deviant from the norm'.Both native and interlanguage English texts are treated as equally valid representatives of the genre.

Results and discussion
The scope of the article only allows us to present a quantitative overview of metadiscourse use in the Master's thesis genre, to highlight major variations at linguistic/cultural level in the two corpora, and to briefly consider the role of institutional practices and individual writer style in creating academic discourse.

Patterns of metadiscourse in the Master 's Thesis genre
Table 2 gives the results of the functional analysis: it shows the distribution of various categories of metadiscourse in the Master's thesis genre in the discipline of Linguistics.
We can see that the MA thesis genre is characterized by highly extensive use of textoriented metadiscourse (184.52),rather limited use of participant-oriented metadiscourse (23.60), and quite sparse use of evaluative metadiscourse (8.79).The total frequency of metadicourse markers is 216.92 per 10,000 words.What do these results suggest about the students' awareness of discourse structure, their awareness of audience and themselves, and their attitude towards their text?Does this pattern illustrate the optimal use of metadiscourse and therefore the optimal degree of awareness and management of discourse in the Ma thesis genre?Are these typical characteristic features of the MA thesis genre?these questions can be answered affirmatively only if and when we assume that the corpus used for the study is sufficient to make generalisations at the genre-descriptive level.

Linguistic-cultural variability
The quantitative results are much more interesting when we consider the linguistic-cultural differences between the two corpora.The results of the contrastive analysis are given in table 3. the figure next to each metadiscourse category is the Log Likelihood value (see 2).Significant differences are in bold and underlined.We can see that, overall, the difference between L1 and L2 Ma theses in the frequency of metadiscourse use is insignificant (-2.30).this is quite an important finding as it supports the above claim that the pattern of metadiscourse use established by this research may be seen as the optimal pattern of metadiscourse typical of the Ma Thesis genre in the discipline of Linguistics, regardless of the cultural-educational context in which MA texts have been produced.As there is no significant overall difference between the L1 and L2 texts, we can assume that MA students in Linguistics, whether L1 or L2 writers of English, use metadiscourse devices with approximately the same overall frequency.The requirements of the genre seem to be more important than the mother-tongue/culture specificity.However, we can notice significant differences (underlined) between the two corpora in the use of specific metadiscourse categories: text-connectives, endophoric markers, reader-oriented markers and emphatic markers.The overuse of text-connectives (+62.37)suggests that Lithuanian MA student writers are (too) highly aware of text coherence and use a variety of transition markers to manage the unfolding text.one possible cause for this is an (over-)excessive emphasis on connectives in academic writing instruction at secondary or undergraduate level.The underuse of endophoric markers (-43.83) suggests that Lithuanian students have (too) low awareness of the ongoing discourse structure and the constituent parts of the Ma paper.often, vague markers are used (such as below, above) instead of explicit markers (such as In Section X… In Chapter Y…). a possible cause for this is that the MA paper is the first major text that students write (bigger than the essay or the Bachellor's paper) and therefore they do not have the skill/habit of referring to specific parts of a bigger text or to appendices.The underuse of reader-oriented markers suggests that non-native student writers have (too) low awareness of their audience and use (too) few engaging devices to create writer-reader interaction.a cause of this could be the general implicitness in communication of Lithuanian students, whether written or spoken (although this is just a speculation, there is no research evidence to support this claim).Finally, the underuse of evaluative markers, emphatics in particular, suggests that Lithuanian students hardly evaluate their texts, which might be caused by the general lack of confidence, reluctance to express opinion, poor/no tradition of critical evaluation in the Lithuanian culture.the students may believe that it is the reviewer's task to evaluate their paper and not theirs.

Institutional and individual differences
alongside with linguistic-cultural differences, the analysis has revealed differences in the institutional practices in creating academic discourse: Master's theses in Linguistics from different universities within the same culture/country may show different patterns in the use of metadiscourse.one such difference is illustrated in Figure 1.

Figure 1. Participant-oriented metadiscourse in texts from two UK universities (normalised frequency, occurrences per 10 000 words).
We can clearly see that students in University a use participant-oriented markers, in particular reader-oriented markers, much more extensively than students in University B from the same country/ culture (in this case UK).This may be caused by a few factors: differences in the formal requirements for Ma paper writing set by the different departments; differences in the student-staff relationship and the perception of power-relations; degree of formality in communication or general degree of interactivity between the novice members and professional members of the academic discourse community.
another important factor in shaping academic discourse, alongside with institutional traditions and requirements, is the individual writer style.The study has revealed that different students within the same university department may use very different amount of metadiscourse. Figure 2 is an illustration of such a difference.
We can see that Student 2 has used much more metadiscourse than Student 3 in all categories: text-organising (236/115), participant-oriented (75/13) and evaluative (14/2).one can assume that the two students had different personal characteristics, such as self perception, confidence or self-esteem.they must have had different awareness of themselves as scholars or different attitude to their supervisor and other professional members of the discourse community.

Conclusions and implications
the functional analysis of metadiscourse in the MA thesis genre has produced a taxonomy of categories: text-organising, participant-oriented and evaluative markers.this taxonomy can be used in further investigations of the Master's thesis genre in other disciplines and languages.It can also be applied in the analysis of the Doctoral thesis genre.
the analysis of metadiscourse in a corpus of 40 MA theses in Linguistics has established the following pattern of distribution: extensive use of text-organising markers; limited use of participant-oriented markers; and sparse use of evaluative markers.Further analysis of a larger corpus of data is necessary to make generalizations in a more confident manner.The contrastive analysis of Ma theses by native and non-native/interlanguage (Lithuanian) speakers of English has shown that the overall frequency of metadiscourse is rather similar in L1 and L2 English texts.on the other hand, the contrastive analysis of the 40 MA theses has shown that there exist significant differences in the use of specific metadiscourse categories: interlanguage (L2) texts are characterized by (1) significant overuse of text-connectives, which might have been caused by an (over-)excessive emphasis on connectives in writing instruction at secondary or undergraduate level; (2) underuse of endophoric markers, which may be explained by writers' inexperience in structuring big texts; (3) underuse of reader-oriented markers, which may be a result of low audience-awareness and the general implicitness of communication; (4) underuse of evaluative markers, emphatics in particular, which might reflect the general lack of confidence, reluctance to express opinion or poor tradition of critical evaluation in the Lithuanian culture.
It should be emphasized that the overuse or underuse of metadiscourse in L2 texts is not treated here as a 'deviation' from a norm, but merely as a culture-based peculiarity of interlanguage texts.Both L1 and L2 theses are considered as equally valid representatives of the genre.This position follows the line of ethnorelativism which, according to Connor (1996, 162), "promotes empathy for different behaviours and cultures" and is contrary to ethnocentrism which regards one language/culture as being 'central' and results in negative stereotyping of other languages/cultures.In this study, L2/interlanguage features of Ma theses are regarded as features reflecting differences in the writing conventions that are learned in a particular culture.
Furthermore, the study has shown that variability is not just a matter of mother-tongue/ culture, it is also a matter of conventional practices typical of an educational institution: universities in the same country seem to have slightly different traditions of academic writing.and even further, the study has revealed the importance of the individual writer style: students at the same university department use different amount or type of metadiscourse to create interpersonal relations with their readers.This implies that individual writers, regardless of the input they receive and of the educational/cultural context in which they study, may perceive their intended reader and the genre which they produce in a rather idiosyncratic way determined largely by their personal characteristics.
As we have seen, the use of metadiscourse in the Master's thesis genre is determined by a number of factors: the linguistic-cultural background, the institutional tradition and the individual writer style.These factors need further investigation on larger scale.I would encourage descriptive studies of Ma theses from a range of cultural/linguistic backgrounds, including american, Canadian, australian etc. universities, with an aim of identifying the most typical patterns used by native and non-native/interlanguage writers.the findings of such large-scale descriptive studies could be analysed on academic writing courses by L2 as well as by L1 English students and creatively applied for their individual purposes.Metadiscourse, afterall, is that area of language use where the writer's individual perception of discourse-as-process is represented.

Figure 2 .
Figure 2. The use of metadiscourse strategies in two MA theses from the same university (normalised frequency, occurrences per 10 000 words).

Table 1 . A model of metadiscourse: categories and examples.
endophoric markersThis thesis… Next Chapter… The previous section…Table 1… Example 2… Appendix 3… The following paragraph… Here… discourse labels This paper addresses the issue… Thus far I have argued…My analysis will follow the claim… Two questions are asked here… code glosses In other words… that is… or… namely… i.e… also called… target language (TL)… mother tongue (= Lithuanian) participant-oriented writer-oriented markers I… My… We… Our… the author... inclusive markers inclusive 'we': We, Our, Let us; questions: What conclusions can be made?It-type directives: It should be pointed out…