5. Results and discussion

Tables 4 to 7 present the semantic sequences identified in each discipline; Table 8 summarizes the analysis and makes a comparison of the distribution of semantic sequences across the four disciplines.

Table 4 Semantic sequence analysis of V that in the subcorpus of Linguistics

Table 5 Semantic sequence analysis of V that in the subcorpus of Education

Table 6 Semantic sequence analysis of V that in the subcorpus of Physics

Table 7 Semantic sequence analysis of V that in the subcorpus of Mechanics

Table 8 Comparison of semantic sequences across the four disciplines

In the following we discuss both observations about disciplinary epistemology which are newly revealed by semantic sequences and those which are consistent with previous investigation, aiming to further support our argument that semantics sequences can be used to investigate and characterize disciplinary epistemology.

1)Disciplinary epistemology that is newly revealed by semantic sequences

Based on the analyses, we can notice two distinctive features concerning disciplinarity that are newly revealed by semantic sequences. The first is that the most frequent semantic sequence in all four disciplines examined is Evidence + Interpret +that + Interpretation, which highlights that knowledge construction in all disciplines is largely evidence-based. This contradicts with traditional view that research in humanities and social sciences is subjective and not evidencebased. Noun groups that function as the grammatical subject in instances instantiating this sequence are mostly what are termed textual nouns(e.g. table, figure)or research nouns(e.g. data, analysis), as noted earlier. Since such nouns are not conscious participants, it is not plausible to say that it is the entity indicated by such nouns that is projecting an idea or thought. Rather, it is the writer/author who interprets external evidence. The purpose of using this sequence in academic writing is, on the one hand, to delegate responsibility because the role of the writer/author is hidden, and on the other hand, to present an observation more objectively and convincingly because the observation is made not based on intuitive speculation, but on external evidence.

Further, Table 8 shows that the sequence Evidence + Interpret + that + Interpretation occurs relatively more frequently in disciplines of pure sciences(e.g. Linguistics and Physics)than in disciplines of applied sciences(e.g. Education and Mechanics). This, on the one hand, offers empirical evidence to support the distinction made between pure sciences and applied sciences(Becher,1994). On the other hand, since interpretation is nevertheless subjective and since disciplines in applied sciences do not involve interpretation as frequent as those in pure sciences do, it is arguable that applied disciplines relatively prefer more objective ways to construct knowledge than those in pure sciences.

The second observation relates to research methodology used in soft sciences and hard sciences. Table 8 shows that the sequences Research-participant +Think +that +Thought and Research-participant + State +that +Statement were only identified in soft sciences(i.e. Linguistics and Education respectively)but not in hard sciences(e.g. Physics and Mechanics), which suggests a methodological difference in knowledge construction and, further, epistemological difference. That is, knowledge in soft sciences may sometimes be constructed by methods such as questionnaire surveys or interviews which usually involve research-participants, whereas knowledge in hard sciences is predominantly constructed based on experimental evidence(cf. Charles, 2006a). This can be generally discussed in terms of in/dependence on research-participant of knowledge construction between soft sciences and hard sciences. This distinction has not been recognized by previous research, which in turn suggests that semantic sequences may not only be an alternative, but more importantly, a more useful way to investigate disciplinary epistemology.

2)Disciplinary epistemology that is consistent with previous investigation

Apart from offering new insights into disciplinary epistemology, some observations are also consistent with those drawn in previous investigation. For example, it is observed that disciplines in soft sciences prefer to build knowledge by reporting others' study whereas disciplines in hard sciences by reporting the researcher's own study, which is evidenced with the finding that the sequence Researcher-other+Find+that+Finding occurs more frequently in Linguistics and Education whereas the sequence Researcher-self + Find +that +Finding is used more frequently in Physics and Mechanics, as shown in Table 8. The distinction can then be drawn between“othersourced” and “self-sourced”knowledge accumulation in academic research (see also Charles, 2006a).

This is consistent with Becher and Trowler's(2001:36)discussion of the cumulative nature of knowledge construction in hard sciences and that of the reiterative nature in soft sciences. That is, disciplines like Physics tend to construct knowledge by using the experimental results that were obtained from the researcher's own experiments and his/her work is seen as“contributing to a reservoir of shared knowledge and expertise that will facilitate the answering of fundamental questions about the empirical world”(Groom,2007:20), whereas disciplines in soft fields tend to proceed recursively or reiteratively, “by putting forward the views of others in order to take up a position in relation to them”(Charles,2006a:493)and researchers feel that“they are obliged to revisit and reinterpret material already studied by previous researchers”(Groom,2005:270). In a nutshell, knowledge is constructed through accumulating experimental results obtained by a series of experiments in hard fields whereas knowledge is constructed through reinterpreting the same issues or phenomena in soft fields. This clearly relates to disciplinary epistemology and ideology, which in turn provides some underlying rationale for Becher and his associates'general distinction between“soft” and “hard”sciences for the characterisation of academic disciplines.

Moreover, according to Becher and Trowler (2001:36), disciplinary cultures can be differentiated in terms of im/personalizedness, which can to some extent be corroborated with the current investigation. As can be seen in Table 8, both sequences“Researcher-other + Argue +that + Argument” and “Researcher-self +Argue +that +Argument”, occur more frequently in soft sciences than in hard sciences. As discussed in systemic functional linguistics (Halliday &Matthiessen,2004), verbs of ARGUE (e.g. argue, suggest)typically realize verbal processes which require a human subject to perform the action of saying/arguing. Then, the higher frequency of the “Argue”sequence in soft sciences(i.e. Linguistics and Education)indicates that human subjects, either the author(s)or other researcher(s), are more frequently presented when making an argument, which inevitably makes the constructed knowledge more subjective and personalized. In contrast, the lower frequency of the “Argue”sequence in hard sciences(i.e. Physics and Mechanics)indicates that human subjects are less frequently presented and the propositional responsibility is obscured, thus resulting in seemingly more objective and impersonal way of knowledge construction.