• Ni Luh Putu Yesy Anggreni

  Author

• I Putu Eka Indrawan

  Author

Academic writing is characterized by certain linguistic rules which are different from those of informal speech; however usually acquired by students in an implicit and inconsistent way. In order to fulfill the requirement of empirical precision, this paper provides a corpus-based computational approach for the discovery of linguistic regularities in academic writing within various fields. Relying on register theory and genre studies, the study analyzed a specially created corpus of 200 research articles in 10 disciplines, which could be called "hard" and "soft" sciences, analyzing their argumentative segments through automated corpus-based analysis. There were five linguistic phenomena selected and measured in terms of 1,000 words: nominalizations, hedges, linkers, passives, and complex noun phrases. The quantitative data show unambiguous divergence between disciplines in terms of linguistic features: hard sciences contain more nominalizations (42 vs. 33), passives (28 vs. 16), and complex noun phrases (35 vs. 24) in order to facilitate concise, objective reporting of empirical data. In turn, soft sciences focus on the use of hedges (27 vs. 18) and linking adverbials (19 vs. 12) for cautious, well-signalized interpretation. The largest differences in absolute values have been shown by passives (+12) and complex noun phrases (+11). Linking adverbials (-7) form the most stable base in both areas. These results illustrate the specific epistemological and rhetorical needs of each academic community. The research comes to the conclusion that corpus analysis allows providing the empirical basis for substitution of general writing tips with EAP instructions for specific communities.

• PDF