A Corpus-based Study On The Variation And Change Of English Collective Classifier Constructions

All human languages have linguistic devices decoding people's perception about quantity.Although not a classifier language,English has an open inventory of words that are functionally similar to classifiers,as is indicated in binominal phrases like a jug of wine,a loaf of bread,and a box of candy.These phrases can be grouped as a quantifying structure a NOUN₁ of NOUN₂,and can be termed as English classifier constructions.Within the theoretical framework of construction grammar,this thesis focuses on a subset of English classifier constructions,namely,the English collective classifier constructions?ECCC?and aims at investigating its synchronic variations and diachronic changes.The synchronic part tries to accomplish the categorization of ECCCs based on their similarities and differences.To be more specific,taking Corpus of Contemporary American English?COCA?as the source of data,27 ECCCs are selected as the objects for study from the top 6000 a N₁ of N₂ binominals.Then,for each ECCC,100 random concordance samples are retrieved,and N₂ in those samples are semi-automatically tagged in three dimensions,namely,semantic,pragmatic,and genre dimensions,with the assistance of Wmatrix.After that,multiple correspondence analysis and cluster analysis are applied to categorize these constructions in terms of similarities in the aforementioned three dimensions.At the same time,collostructional analysis is introduced to discern the subtle differences between highly similar ECCCs.The diachronic part aims to explore the feature of the diachronic change of ECCCs through a case study of a bunch of.On the basis of data collected in Corpus of Historical American English?COHA?,corpus-based approach is applied to reveal the overall development trend of a bunch of and to divide the change of a bunch of into different periods with the help of variable-based neighbor clustering.Then multiple correspondence analysis is introduced to explore the semantic,pragmatic,and genre changes affecting the schematic part of the string.Finally,by analyzing the development of a bunch of in terms of its changes in productivity,compositionality,and schematicity,the author tries to interpret the feature of the diachronic change of ECCCs and provide a construction grammar account.In the synchronic part,results indicate that the categorization of ECCCs through data-driven approach is quite different from traditional ones.Firstly,by multiple correspondence analysis and cluster analysis,the 27 ECCCs involved can be categorized into four groups as followings:1)human gathering ECCCs:a gang of,a mob of,a bunch of,a horde of,a band of,a legion of,a group of,a team of,a panel of,a troop of,a crowd of,a squad of,a throng of;2)species ECCCs:a flock of,a herd of;3)immaterial concept ECCCs:a flood of,a host of,a couple of,a body of;4)arrangement ECCCs:a bundle of,a cluster of,a clutch of,a handful of,a fleet of,a pack of,a heap of,a stack of.Secondly,distinctive collexeme analysis illustrates that there exist subtle differences between those seemingly similar ECCCs.For instance,compared with a mob of,a gang of shows more preference for nouns denoting negative meanings;a team of distinguishes itself from a group of in that it tends to collocate with nouns with specific semantic meaning,especially collocates with the meaning of certain occupations,such as researchers,scientists,and doctors;as for a flock of and a herd of,the main difference lies in that a flock of tends to be frequently associated with poultry whereas a herd of is prone to modify herbivores with a large body size.In the diachronic part,the case study of a bunch of reveals that the development of ECCCs can be considered as a part of the process of constructionalization.Firstly,via corpus data and variable-based neighbor clustering,it is apparent that the overall frequency of a bunch of from the 1830s onwards has been increasing and can be divided into five periods,i.e.the 1830s-the 1900s,the 1910s-the 1960s,the 1970s-the 1980s,the1990s,and the 2000s.Secondly,the 1910s-the 1960s period acts as a catalyst for semantic,pragmatic,and genre expansion of a bunch of.During this period,the“group”meaning with negative evaluation outweighs the original overwhelming“bundle”meaning with neutral evaluation,and is later overtaken by the“large quantity”meaning,which has been increasing from the 1970s onwards.In the same period,the text types of a bunch of apparently diversify with an obvious increase of the occurrence of the string in genres of newspapers and magazines.Thirdly,multiple correspondence analysis demonstrates that there exist statistically significant correlations between some of the meaning dimensions,i.e.in the semantic,pragmatic and genre properties of the noun in the string a bunch of NOUN.For example,when the noun refers to people,the string usually has a negative semantic prosody.Lastly,the conclusion part summarizes that the diachronic changes of ECCCs is a recursive process involving constructional changes and constructionalizations.In that case,the changes affecting the string a bunch of NOUN from the 1830s onwards is best considered as a part of constructionalization which involves a neoanalysis in both meaning and form.Theoretically speaking,this thesis extends the discussion of applying construction grammar to explore the synchronic variations and diachronic changes of ECCCs,and therefore makes contributions to cross-linguistic and typological study on quantifying structures.Methodologically speaking,the thesis integrates corpus linguistics and quantitative analysis methods with data visualization,and shed light on the means of quantifying language variations and changes via multivariate analysis.Practically speaking,the thesis may provide some clues for English learners to have a better mastery of ECCCs.