A Corpus-based Multi-genre Study Of Ditransitive Construction In English

Previous studies have explored ditransitive construction (DC) from informational perspective and examined the recipient (R) and theme (T) in terms of informational or discourse-pragmatic variables with natural texts as databases. It is found that the variables exert strong influence on the ordering of R and T and hence can well explain and predict the choice of DCs. Pursuing the line of research further, the present study concerns itself with genre variation and brings it under rigorous examination and in-depth analysis. The aim is to explore whether the influence of informational variables on constituent ordering and hence on constructional choice varies from genre to genre. The study is of great significance in that it can not only thoroughly unravel their influence in each genre, but may also invalidate certain conclusions drawn by previous studies which fail to take the factor of genre variation into consideration.The database is the four sub-corpora in the British National Corpus (BNC), namely, Spoken English (E), Fiction (F), Newspaper (N), and Academic Journals (A). The most typical ditransitive verb give is selected for a case study. A sample of 200 tokens with the verb used in DC is randomly selected from each genre for examination. In each sample, Rs and Ts in tokens of double object construction (DOC) and prepositional dative construction (PDC) are analyzed in terms of accessibility, definiteness, pronominality, and length from the perspectives of the average value and relative value. The findings are presented by statistics for further qualitative research. Furthermore, in examination of each informational variable, comparison and contrast are made to unravel its constraint on constituent ordering and hence on constructional choice. Specifically, inter-constructional contrast is made between R and T in each construction variant in terms of each variable, cross-construction contrast made between DOC and PDC in terms of the informational distinctions of R and T within them, and cross-genre comparison and contrast made among different genres in terms of the informational constraint. The study reveals that:For DOC, the contrast of R and T in informational status is remarkable and the informational constraint is hardly violated. This is the case in all the genres. For PDC, however, these genres vary from one to another. Specific case concerning each variable is as follows. (1) Regarding accessibility, its constraint is not violated in S, violated by a tiny minority of the data in F, and by one third of the data in N and A. (2) In respect of definiteness, its constraint is violated by a small minority of the data in S, but by over one third of the data in the other genres. (3) In respect of pronominality, its constraint is not violated in F, N and A, but violated by a small proportion of the data in S. (4) Involving length, its constraint is hardly violated in F, but violated by a minority of the data in the other genres.Moreover, for the four genres under research, i.e. S, F, N and A, the frequencies of the constructional choices predicted correctly are:(1) 97.2%,92.5%,75.1%, and 68.2% respectively when examined from the absolute accessibility of R and T, and 100%,98.6%,90.8%, and 78.9% respectively when examined from their relative accessibility; (2) 93.5%,92.1%,71.9%, and 57.7% respectively when examined from their absolute definiteness, and 96.5%,93.3%,83.6%, and 65.3% respectively when examined from the relative definiteness; (3) 86.7%,84.6%,62.1%, and 58.8% respectively when examined from their absolute pronominality, and 97.4%,100%, 98.7%, and 98.4% respectively when examined from the relative pronominality; (4) 94.9%,97.6%,91.5%, and 86.4% respectively when examined from the relative length.Therefore, genre variation should be taken into consideration in explaining and predicting the choice of DCs from informational perspective. Furthermore, the variations in prediction accuracy when examined from different variables not only demonstrate the constraint of genre and each variable on constituent ordering and hence on construction more clearly, but also make the use of DC predicted more accurately.