| Ever since the 1970s,researchers have paid extensive attention to the study of academic discourse.However,the majority of previous researchers only focused on the study of certain separate and independent linguistic features,e.g.first person pronouns,references and metatextual expressions,without a systematic framework.Based on this research background,the present study attempts to analyze person manifestation in research articles from the new perspective of voice and under the guidance of a relatively comprehensive analytical framework.Based on previous studies relevant to academic voice,a new analytical framework consisting of eight linguistic features is constructed.Taking discipline as the only variable,the present study is contrastive in nature.Comparisons and investigations are based on three self-built corpora(sub-corpus H,S and N),which contain 767,797 words in total for 90 research articles distributed evenly over three disciplines,i.e.Linguistics,Economics and Physics.Both quantitative and qualitative methods are adopted.The former is employed to find out the occurrence frequencies of eight selected voice features and to compare the similarities and differences in their distribution patterns,while the latter is applied to study what functions these voice features perform in the construction of academic discourse and then author identity.The statistical results demonstrate that there are both similarities and differences in the distributions of different voice features in three sub-corpora.The similarities are as follows:(1)All three sub-corpora possess more First Person Plural tokens than First Person Singular tokens.For the use of First Person Singular,in all three sub-corpora the Researcher role is the dominant role,the Writer role ranks the second,and the Arguer role occurs less frequently and nearly no Evaluator role is identified.For the use of First Person Plural,in sub-corpus H&S the Inclusive use occurs more than the Exclusive use,and in three sub-corpora the Metonymic use occurs relatively less frequently;(2)Indefinite Pronoun occurs in relatively low frequency in all three sub-corpora,among which One 3 occurs the most frequently in sub-corpus H&S,and One5 and One6 occur less frequently in all three sub-corpora;(3)In all three sub-corpora Let-Imperative occurs the least frequently and in particular this feature is absent from most RAs.The differences are as follows:(1)First Person Singular occurs more frequently in sub-corpus S than in H,the least frequently in N(almost none).First Person Plural occurs slightly more frequently in sub-corpus N than in S,the least frequently in H.The occurrence frequency of Arguer role in sub-corpus S is higher than that in H;(2)Indefinite Pronoun occurs more frequently in sub-corpus N than in S,the least frequently in H.One1 and One2 in Physics show relatively strong presence compared to Linguistics and Economics;(3)Metatext occurs more frequently in sub-corpus N than in S,the least frequently in H;(4)Bibliographical Reference occurs more frequently in sub-corpus H than N,and S ranks the last.Partly Integral Reference and Semi-Integral Reference occur most frequently in sub-corpus H&S and Non-Integral Reference has the highest occurrence frequency in sub-corpus N,while Fully Integral Reference occurs the least frequently in all three sub-corpora;(5)Negation and Adversative occur more frequently in sub-corpus H than S,and N ranks the last.Descriptive Use of Negation occupies the highest occurrence frequency in all three sub-corpora,Refutative Use and Delimitative Use occur the most frequently in sub-corpus H,Corrective Use occurs the most frequently in sub-corpus S.The present study is significant in both theoretical and practical perspectives.Theoretically it provides a new perspective to study the genre of research article and develops a relatively comprehensive framework to analyze voice features in academic discourse.Practically perspective it offers great pedagogical implications for academic writing. |