Study On Chinese Donkey Sentences And Negative Sentences Based On MMCCG

Logical semantics is committed to achieving natural language formalism.With the field of big data and big knowledge blossomed in computer and artificial intelligence,contemporary logical semantics has met with new challenges and opportunities.During this period,Multi-modal combinatory categorial grammar(MMCCG)sparked by Steedman,Baldridge and others has been the latest innovations to solve these challenges.The field of natural language processing fall into two approaches:empiricism and rationalism.Empiricism is a big data approach based on statistics,while rationalism focuses on logical formalism via a big knowledge approach.Traditional logical semantics seeks for rationalism.These approaches(schemes)have their own limitations along with upgrading of computer and artificial intelligence from time to time.It is reasonable to explore another approach,and MMCCG can act as a “glue”allowing for a combination of the two different approaches.Up to now,in terms of generative power and explanatory adequacy,MMCCG has provided a perfect logical semantic tool for closing to natural language and combining rationalism and empiricism together.MMCCG is concerned with a wide range of truthoriented text,case-study description,natural language combination stemmed from rich rules,universal irregular rules and even makes use of non-computational combinatory rules and rule schemes.The logical semantic mechanism presented by MMCCG fully suits for the spirit of computational processing within natural language,and embodies the essence of big data and big knowledge.In order to reveal the analytical power of MMCCG in Chinese context,this thesis aims to do a specific formal natural language processing about Chinese donkey sentences and negative sentences with the framework of MMCCG.This thesis is broadly divided into four parts.First and foremost,a detailed introduction to MMCCG from AB-calculus,lambek calculus and standard CCG to MMCCG.In particular,making an account for MMCCG properties in different stages,technical advantages and disadvantages,the evolvement of MMCCG.Part I also clarifies what technical means MMCCG should inherit in various stages and what kinds of bottlenecks MMCCG has broke through.Secondly,the thesis reviews the basic content and methodological features of the MMCCG in detail,and makes a more complete and systematic introduction and investigation to MMCCG.Thirdly,the thesis analyses the treatment of Chinese donkey sentences which rest on the classification of the “Chinese donkey sentences” examined by linguists at home and abroad.Three kinds of Chinese donkey sentences can be got: the conditional donkey sentences,the relative donkey sentences and the generic implicit donkey sentences.In MMCCG,different types of Chinese donkey sentences can show a syntax-semantics correspondence,labelled lexical entries for derivation.Part III also introduces the generalized skolem terms to deal with quantifier scope ambiguity in Chinese donkey sentences.Part IV demonstrates the MMCCG treatment of Chinese negative sentences.What's more,on the basis of the basic types of Chinese negative sentences,Part IV sorts out typical cases of Chinese negative sentences and does derivation analyses to prove that MMCCG has the function of analyzing Chinese negative sentences.Finally,it is certainly feasible to use MMCCG for natural language processing in Chinese context,and MMCCG has strong expressive power.The expressive power has broadened the application domain of MMCCG.