The distribution of referring items, cohesive ties and cohesive chains

142

Finally, two splitting chains were not identified as such in one analysis;

nevertheless, the cohesive ties all appeared, but under the same presupposed item.

Figure 8 shows one of these instances, with the two different chains from the two analyses in parallel:

Chain 1 Chain 2

the Winter 2003 semester the Winter 2003 semester

the semester the semester

the first midterm the first midterm the first midterm the first midterm the second midterm,

the end of semester

the second midterm

Figure 8. Sample from the reliability analysis at Stage 2.

Occasional differences occurred with items like the computer which was ambiguous between exophoric and cohesive.

143

Cohesive ties and chains in 10 RAs Total number of

sentences: 2324 Total number of cohesive

chains (average/RA) 69

Total number of words: 66277 Total number of cohesive

ties (average/RA) 188 The percentage of

cohesively used referring items in the corpus:

35%

Total number of cohesive chains containing more than two cohesive ties

40% of total chains Table 18. The distribution of referring items, cohesive ties and cohesive chains

Table 18 also shows that 40% of the cohesive chains consisted of more than two cohesive ties, meaning that they contributed to the global cohesion of the text; while the remaining 60% established local, inter-sentential relationships. Again, this high ratio of short-distance references may be genre-specific: long-distance references are more difficult to track, and might make the RA less comprehensible.

Interestingly, the definite article – which made up 56% of the total number of cohesive items – was the only referring item that was used for long-distance reference (where the presupposed item is not in the immediately preceding sentence, but forms a non-mediated tie with an item earlier in the text). The definite article was also unique among demonstratives in taking part in cataphoric (forward pointing) reference. The demonstrative this, second in frequency in the corpus, was only used for referring to something in the immediately preceding sentence (this is also true of its plural form), the only exception being this study, this research. Figure 9 shows the frequency of the occurrences of the cohesive items. The most interesting finding shown in Figure 9 is that out of the personal pronouns only the plural (they, their) and the neutral (it, its) were used cohesively.

144

Figure 9. Frequencies of cohesively used referring items in the 10 analyzed RAs

Other pronominals, even those non-cohesively used were rare in the corpus. The demonstratives that do not appear in Figure 9, that, those, there, were often used in the RAs, but almost always non-cohesively. Comparatives were rare, and they were only used for short distance reference.

6.6.3 Patterns of chains within the subsections of RAs. This part of the paper describes how cohesive chains of reference are typically organized in the ten RAs. The research articles in the corpus were very similar in their organization. Independently of the length of the article, the following sections could be identified:

Abstract

Introduction (Sometimes the introduction appeared as part of the Background section)

Background / Context Methodology or Method Findings / Results / Discussion Summary / Conclusions

0% 10% 20% 30% 40% 50% 60%

its here their comparatives it they these this the

% of cohesive out of total no. of ties

145

The boundaries of these Sections were indicated in the Excel worksheets used for analysis by highlighting the first row of each Section. Furthermore, a dividing vertical line was drawn in front of the first referring item in each Section. This way, the new and extended cohesive chains became visible and the number of merges and splits in the chains could be counted for each section.

6.6.4 New and extended chains. When dividing the RA into subsections, two main types of referential chains can be identified: extended (which have their referent in some previous Section) and new. The ratio of new and extended chains in each Section shows the ratio of new referents introduced. Figure 10 shows the outcome of this analysis. In Figure 10 the rightmost two columns represent the ratio of the total number of new and extended chains in the RAs, as compared to Biber et al.’s (1998) results for given versus new referring expressions in academic prose. The similarity of the two columns hopefully provides some evidence to the validity of this part of the present study.

Figure 10. The ratio of new and extended chains per section in 10 RAs

146

As regards the sections in the papers, the Abstract cannot contain any extended chains, but it is closely related to the Introduction, as shown by the relatively high percentage of extended chains in the Introduction. The Background again contains more new chains, thus probably more new information, which is then referred to by various elements in the Method and Findings sections. This high ratio of extended chains in the Method section of RAs supports the observation by Swales (1990, p. 169) that in “soft”

sciences, the Method is very explicit, and contains a detailed description supported by a high number of anaphoric references and lexical repetition. As can be expected, the Summary and Conclusions section carry low new information, as shown by the lowest percentage of new chains, but has the highest portion of extended chains as compared to the other sections.

6.6.5 Chain complexity: merging and splitting. Merging and splitting of chains add to the complexity of the referential chain. When a chain splits, the information content of the referent is usually carried on in two or more chains, while the reader has to keep the original referent in mind for the comprehension of the text.

Merging is even more complex in that it requires the reader either to produce a summary of the previous text, or to select the relevant referents from various preceding text segments. These processes are much more complex than a simple recall of a previous item. This complexity is expected to be reflected in a greater number of errors in student writing.

In Figure 11, we can see a high number of merges and splits for the Introduction of the RA. This was expected as Introductions generally contain diverse information about previous research and the study presented in the RA, establishes connections and relationships between these two topics.

147

Figure 11. Chain complexity in the subsections of 10 RAs

The other section that is dense in splitting and merging chains is the presentation of the Findings of the paper. This is not surprising. This section is where new information is related to previous findings, to the background, to the methods used to arrive at those findings, and also, how the various outcomes are related to each other.

6.6.6 The density of ties and the abstract. Let me mention here an initial outcome of the comparison of the density of cohesive chains in RAs. An emergent research question during the analysis was whether the referents of the chains with the highest number of cohesive references carry any information about the main message of the text. In order to test this hypothesis, the referents of the six highest density chains in each RA were compared to the abstracts of the corresponding paper. All the abstracts contained at least five of these referents; therefore, it can be concluded that the density of chains is probably relevant for the identification of the main message of the text. This again proves that our analysis provides a valid description of cohesive reference, as it can select the main referents that are also perceived as such by expert writers.

148

6.6.7 Some unexpected inconsistencies in the references in the corpus. Here,