Effects Of Packet Nouns On Information Density In Argumentative Discourse

Based on the achievements of cognitive linguistics and language information,the author proposes the hypothesis: packet nouns as a means of information packagehave the function of uniforming the discourse information density, which bridges theprevious studies on packet nouns and Uniform Information Density (UID) and testifiesUID at discourse level. The idea of UID is that speakers prefer utterances thatdistribute information uniformly across the signal (information density) within thebounds defined by grammar. Previous studies have testified UID as a general principleat lexical and syntactic level, however little research has been conducted to examineUID at discourse level. Here this study investigates packet nouns only inargumentative discourse in which the information package phenomenon is the mostrepresentative.Packet nouns can package discourse information, thus lower information density.The higher the information density (namely the more information units) is, the morepeople should be likely to produce packet nouns and package information into packetnouns(can be seem as information nodes), then come to next information layer, formthe discourse information flow as Given…..New……..Given…..New………Given……..New. Here, the empirical study suggests that the more information unitsthere are, the more packet nouns exists, and vice versa. Hence, there should be stableand equilibrial correlation between packet nouns and information units.Specifically, it centers upon the following questions:1. What is the relationship between “packet nouns” and “information units”?2. How do “packet nouns” effect information density in English argumentativediscourse?The author divided information units (IU) to statistically analysis the correlationbetween PN and IU by adopting quantitative principle of language information basedon question logic advocated by Lv (2000).The author collects60argumentative texts from the well-know biweekly magazine National Review online to establish a small-sized and specialized corpus inwhich the PN and IU manually annotated. What needs to be especially pointed out isthat the data used to compile the corpus are the authentic and vivid language in actualcommunication and written by different author on different topics. The data meetNwogue’s criterions as representation, reputation and accessibility.The Regressive Analysis, Unit Root and Cointegration Test are conducted toanswer the research questions above. The major findings are as follows:The result of regressive analysis shows t-Statistic is99.5above its critical value,which suggests an argument PN and IU are positive correlated at a-level0.01.Thechange of PN can illustrate the87.2%(R2=0.872) change of IU. Regressioncoefficient is11.7, which indicates the optimal discourse information density is1PNcorrespond11.7IU.At1%level、5%level and10%level, unit root is found between those twovariables，which indicates they are unstable sequence. The result of first-orderdifference parameters shows unit root is not found, which accounts for PN and IU areboth first-order integrated process I(1), and fits the terms of Johansen’s cointegrationBoth trace test and Maximum Eigenvalue test denotes rejection of the hypothesisof no cointegration at the0.05level. The normalized cointegrating coefficients showsthere exists a positive stable and balanced relation and between PN and IU. The resultof the regression analysis as table4.2is not a spurious regression. Adding or deductingthe amount of IU will cause the corresponding change of PN.The results of the study reflect language iconicity in a broad sense. That is tosay linguistic sign resemble cognitive style similar to the real world with the cognitiveoperation in certain aspects to certain extent On the other hand, the present studyinvestigates the human being’ s unconscious laws based on the differentiation of thelanguage information and meaning. Combined with the related theory, languageinformation as cue inducing underlying cognitive process was discussed. Classicalcategory theory can be used to explain language，otherwise meaning is a dynamicprocess induced by one or more sememes of language information.There are both great theoretical and practical significance in the study. This thesis provides new evidence on the hypothesis of cognitive language that language isnot an autonomous cognitive faculty, at the same time, this study provides a new andunexplored language informative perspective to Discourse Analysis. Then, we find thisthesis is proved feasible to introduce corpus linguistic methods to mental languageresearch on the premise of making distinction between language information andmeaning. Practically, the study is of certain directive significance to Englishargumentative writing. Consciously balancing information density using packet nounsin English argumentative writing will help to transmit and understand the discourseinformation.