4. Semantic sequence analysis

This section discusses the analytic process and gives examples for the sequences identified. As this is a corpus study which emphasizes the co-selection of lexical and grammatical choices in meaning-making, we looked at all the elements involved in instances instantiating the pattern V that, i.e., the noun group realizing the grammatical subject, the node verb, and the complement that-clause. We then proposed an analytic framework for analyzing extracted instances and, further, for identifying semantic sequences(Figure 2). Since the semantics of the complement thatclause is largely dependent on the semantics of the corresponding node verb, the following discussion focuses primarily on the noun group realizing the grammatical subject.

We found it useful to make a broad distinction between instances whose grammatical subjects indicate either Researcher or Research-participant and those which indicate what is discussed under the umbrella term Evidence. Semantic elements such as Researcher/Research-participant and Evidence are used because all the texts under scrutiny are research articles; as such it is reasona ble to propose research-related terms for the subsequent semantic sequence analysis.

Let us first look at instances whose grammatical subjects indicate external evidence. Such instances often have textual nouns(e.g. table, figure)or research nouns(e.g. data, analysis)as the grammatical subjects(see also Charles,2006a:501-502); for example,

1 This figure shows that there is a strong increase... (Physics)

2 FE analysis indicates that plastic strain is localised to just below... (Mechanics)

The grammatical subjects in Examples 1 and 2 refer to external evidence; the verb (show, indicate)does not indicate a process of reporting, but rather, indicate a process of interpreting external evidence; and the complement that-clause presents the projected interpretation. Accordingly, instances like these can be reasonably analyzed as Evidence +Interpret +that +Interpretation. This might suggest that, in addition to reporting, to present the writer's interpretation of external evidence may be another important function of V that. This observation complements previous investigation in that, while studies (e. g. Charles,2006a,2006b)have extensively explored the “reporting”function of V that, the function of “presenting interpretation”has barely been discussed.

The next to be discussed is those instances whose grammatical subjects usually have human references(i.e. either Researcher or Research-participant), which is more complex. A primary distinction worth drawing is the one between noun groups indicating Researcher-self and those indicating Researcher-other. This is because whether knowledge is constructed based on one's own study or other's study is one factor that serves to characterize disciplinary cultures (Becher, 1994; Becher & Trowler,2001). Typical examples illustrating this distinction are instances which have or do not have a first person pronouns as the grammatical subject; for instance,

3 Bazerman (1985)argues that people with highly developed reading skills ... (Education)

4 I would argue that it is difficult to determine clear differences... (Linguistics)

“Bazerman”in Example 3 denotes other researcher whereas“I”in Example 4 refers to the researcher himself/herself; instances like these are analyzed as Researcher-other +Argue +that+ Argument and Researcher-self + Argue + that + Argument respectively. This distinction can also be supported with the observation that instances can be used to report research findings obtained from either one's own investigation or somebody else's, as exemplified in 5 and 6 which instantiate Researcher-self +Find +that +Finding and Researcher-other+Find+that +Finding respectively.

5 We also find that the lower density borosilicate glass has... (Physics)

6 Huang found that the second row of cylinders had little... (Mechanics)

At this point, it needs to be noted that noun groups analyzed as Researcher do not necessarily have to have a human reference; instances where noun groups such as previous research, numerous studies and some it clauses can also realize the role of Researcher. For example,

7 The present study demonstrates that axial load pattern updating can be performed... (Mechanics)

8 Yet other studies find that digit span backward specifically distinguishes... (Education)

It is clear that, although the surface realizations of grammatical subjects in Examples 7 and 8 are different from those in Examples 5 and 6, they have the same function and, accordingly, instantiate the same sequences(i.e. Researcher-self + Find + that + Finding and Researcherother + Find + that + Finding respectively)as Examples 5 and 6 do. This in turn suggests that semantic sequences would be more useful for describing phraseological patterns than other approaches are, because semantic sequences capture not only regularities of formal surface realizations, but also the underlying functional regularities(cf. Groom,2009).

Furthermore, it is important to note that in some cases the grammatical subject may neither indicate a researcher nor an evidence, but may indicate a research-participant. For example,

9 Two participants from each group indicated that instead of have problems understanding... (Linguistics)

10 Nineteen participants reported that they were students. (Education)

Instances like these are used to report a statement made by a participant involved in a research activity; and consequently, they are analyzed as the sequence Research-participant +State + that +Statement. Given the experimental or empirical nature of natural sciences, it is not surprising that this sequence is only found in Linguistics and Education in the present study. This may be suggestive of the disciplinary epistemology of social sciences and humanities in general;for example, interviews may be conducted or questionnaires may be used to perform a social science research, which will be discussed in more detail in the first subsection of Section 5.

Lastly, it is necessary to discuss briefly the analysis of instances starting with an introductory it. This type of clause may either be used to report one's own study or somebody else's. For example,

11 Instead of fixing the problems of some students in an“ad-hoc”manner (Ivanic &Lea,2006:11), it has been argued that UK universities need to teach writing as“an integral, ongoing part of disciplinary learning for all students”(Mitchell & Evison, 2006:71-72). (Linguistics)

12 It has been argued that dual task studies provide the most compelling evidence for the specific processes involved in the task of interest because they can be used to isolate the roles of the different components of working memory. (Education)

The difficulty of analyzing such instances lies in that, while they have similar surface reali zations(e.g. it has been argued that...), their functions may be different:Example 11 is used to report other's study whereas Example 12 to report the writer's own argument. One way to analyze such instances is to look at its wider immediate context (e.g. presence of references)to determine whether a given instance is used to report somebody else's or the researcher's own research. This, though time-consuming, to a great extent validates the analysis.

The above discussion has illustrated the analytic process and exemplified the semantic sequences identified. It is perhaps worth noting that, although studies have proposed meaning groups to categorize the node verb occurring in V that (e. g. Charles,2006a; Francis et al., 1996), it is not practical to simply use meaning groups for semantic sequence analysis because a verb may fall into different meaning groups. For example,

13 The Sunk Cost Effect suggests that designers fixate more when they devote... (Mechanics)

14 The authors also suggested that the H2 formed could be used to increase ... (Mechanics)

Suggest in Example 13 indicates the process of interpreting what“The Sunk Cost Effect”means whereas in Example 14 is used to report an argument. This means that the meaning of suggest in Example 13 is different from that in Example 14, and, consequently, suggest falls into different meaning groups, depending on its phraseological pattern. Charles (2006b:319)has similarly observed that“the verb observe can appear in the FIND group with the meaning of‘becoming aware'or‘noticing'and in the ARGUE group when it refers to the associated language activity”. This, on the one hand, supports the argument that has often been made by corpus researchers that a word in isolation only has meaning potential, but does not have specific meaning; its meaning is dependent upon its phraseological behaviour(e.g. Hanks,2013; Hunston&Francis,2000; Sinclair,1991,2004). On the other hand, this indicates the importance of looking at the immediate context of a lexical item when performing semantic sequence analysis, which is what we did in the present study.

---