The Generative Mechanism Of Chinese Network Blended Words From The Aspect Of Connectionist Model

Information age has greatly changed people’s lives and provided a convenient communication platform.Network blended words are an important part of network language and a product of “Economy Principle” in language evolution.In recent years,linguists have begun to pay more attention to this particular phenomenon,but related works are still limited due to its short history.As for linguistics,traditional research methods can no longer meet the needs of linguistic research,and the combination of computational methods and linguistic research is the only way under such circumstance.Connectionist model is one of main methods in computational phonology.It uses the Parallel Distributed Processing Model(shortened as PDP model)to re-examine language structure,and aims to reveal the hidden characteristics of natural language phenomena.So,with the help of connectionist model,this thesis focuses on the generative mechanism of Chinese network blended words,and analyzes those words from three parallel layers of phonology,semantics and morphology by constructing a PDP model.This thesis takes twelve typical Chinese network blended words as research targets,analyzes their phonological,semantic,and morphological representations from inputs to outputs,and uses Beijing Language and Culture University Corpus Center(shorted as BCC)as data source to construct a PDP model for Chinese network blended words.The model points out the connections among phonological,semantic and morphological systems,as well as the influence of context on Chinese network blended words.The major findings are as follows.Firstly,Chinese network blended words mainly follow “head + tail” regular fanqie method or “head + word” blending with zero-initial sound in phonological level.Secondly,semantic meanings of Chinese network blended words are determined by at least one of two source characters.When the input is a verb phrase,it is a subordinate endocentric construction.When the input is a noun phrase,it can be either a subordinate endocentric or exocentric construction.When the input belongs to interjection,it is usually a coordinate endocentric construction.When the input is the mixture of different parts of speech,it followsexocentric construction.Meanwhile,when Chinese network blended words are used in the middle of sentences,they can indicate emphasis,and when they are used at the end of sentences,they can not only indicate emphasis but also shorten the distance between the user and the receiver.Furthermore,there is no direct morphological relationship between input and output for most Chinese network blended words;they only keep consistency in part of speech.In addition,context also has a great influence on Chinese network blended words.Linguistic context largely restricts the environment in which Chinese network blended words are used,while non-linguistic context has a lot to do with the generation of Chinese network blended words.In short,the application of PDP model on Chinese network blended words makes the research findings more scientific and reliable.It can enrich people’s understanding on this linguistic phenomenon and provide more inspirations to the development of word formation in the future.