| GENERAL SYNOPSIS
The book under review wants to develop a formal framework of human language understanding from a common-sense viewpoint. This method and assumption are very different from those well-established approaches, which generally are static with a characterization of strings and some surface structure and fixed interpretation. On the contrary, this theory does not characterize the surface structure of a sentence, but tries to assign an interpretation to a string of words in a left to right fashion. In other words, this new model takes information from words, pragmatic processes and general rules, and derives partial tree structures that represent the underspecified content of the string, and finally gets some representation in logical form.
We can call such process dynamic, because structures grow from some initial skeletal form of interpretation to some more richly structured object from which a full interpretation of the sentence can be derived by logical means. In this model, the progressive product of partial tree structures is a very important concept, it reflects three properties central of natural language processing:
1. The parsing process is goal-directed, the goal being to establish some logical form as representation of a propositional content for some sentence.
2. The information needed to establish such interpretation is progressively accumulated across the left-right sequence until the goal is achieved.
3. The specification of interpretation provided by a sequence of words may be less than the interpretation assigned to that sequence, and in such cases interaction with pragmatic actions is needed to update the lexical specifications provided.
Another important notion is tree growth, which involves two primary concepts: growth of tree structure by addition of nodes and fixing of tree relations, growth of decorations on a node with progressive replacement of requirements by annotations.
This book sets out an original model of the dynamics of language processing, which can be used to explain the structural properties of language in a simple and elegant way. The model is introduced both informally and formally, and is applied to many languages with different features of the syntactic structures. Following the short history of this theory and method, we can see that this book is an
interdisciplinary product of linguistics, applied logic, and philosophy of language. Alongside formal definitions, the authors use step-by-step derivations and detailed lexical definitions to illustrate this new form of syntactic analysis and to show how the model can be applied to a broad range of languages. In this way, the reader gets a sense of the rich potential the framework provides for general application to linguistic analysis.
Chapter 1 (Towards a Syntactic Model of Interpretation): What is the notion 'Knowing a language'? The authors begin themselves book from the common-sense view of this banal notion. According to the authors, most linguists do not construct themselves theories following the general view. This book is to be an implicit plea to return to the common-sense view. Using 5 subchapters titled 'Natural Language as Formal Language?', 'Underspecification in Language Processing', 'The Representational Theory of Mind', 'Pronominal Anaphora: Semantic Problems', 'The Anaphora Solution - Towards a Representational Account', the authors presents the fundamental difference between dynamic and traditional syntax, and skeletal structure of this book. Some main topics, which will be touched in this book, are also summarized in this chapter, such as the difficult anaphora and long-distance dependency problems in language processing/parsing. In many points, the authors emphasize that the process of understanding a natural language string should be defined as a process of incrementally constructing representations. The view undoubtedly is a departure from the general assumption that natural language systems are formal languages whose syntactic properties are defined over strings of words or sentences with semantic interpretation defined relative to their syntactic structure.
Chapter 2 (The General Framework): As the mentioned above, the authors aim at defining a formal model of the left-to-right process of natural language understanding. For this reason, this chapter is the core of the book, presenting the important aspects of dynamic syntax. The so-called left-to-right process can be classified as a two-step process: First, a tree structure is induced from the linear sequence of words, with words introducing tree structure and lambda terms as annotations for nodes. Secondly, with the final tree reflecting the interpretation of the sentence, the non-terminal nodes are consecutively annotated - with 'reductions' - through a bottom-up process, that leads to a propositional formula as annotation at the root node. The heart of this process is the concept of goal-directed tree growth. The subchapter 'Preliminary Sketch' gives the reader a
concise and clear description of the left-to-right parsing process. The central point may be found in the part 'The Data Structures of the Parsing Model', because the essential goal of the dynamic syntax is implementation in a computer which processes language understanding in a the left-to-right fashion, thus the formal
model and data structures are the most necessary components. Here you can find the strict and formal definitions of these notion and structures, for example, 'language', 'basic/partial tree structures', which will be used in the following chapters. Anyone, who wants to continue reading this book, will have to carefully learn these formal definitions.
Chapter 3 (The Dynamics of Tree Building): In the second chapter, the data structures used to represent the results of parsing a natural language string have been introduced, as well as the language defined to describe these structures. Obviously, these descriptions and structures are static. How to use these static and declarative structures in dynamic parsing is main topic of the current
chapter. Here the procedural structure of the space of decorated trees is introduced. To make the abstract process and description more easily understood, first the authors introduce the rules that license transitions from tree to tree by setting out three examples, step by step. These rules license the development of a logical form represented by a tree. There are three types of rule: introducing
requirements for appropriately annotated nodes, regulating the incorporation of lexical information in the tree, and propagating and compiling the information upward in a tree. It is worth noting that this system differs from other, more familiar or traditional systems. The new system and method defines possible transitions from one tree to another without reference to any independent set of axioms that constitute the grammar of the language, in other words, the
transition rules themselves incorporate the grammar. In the 3.2 part, the parsing process above described is formally defined. According to this model, natural languages are no longer inference systems, but devices for creating expressions over which inference can be defined. If you have a background of computational linguistics and know other formalism, the summary of this chapter is very
useful. Here, the authors describe the differences and similarities between their system and many other formalisms, for example, Categorical Grammar, HPSG (Head-driven Phrase Structure Grammar), LFG (Lexical Function Grammar), DRT (Discourse Representation Theory), Minimalist Programme, and Left-Associative Grammar. In a nutshell, chapters 2-3 define a modular system which across a left-right sequence of words, induces a structure from which some lambda term is
compiled representing an interpretation of the string.
Chapter 4 (Linked Tree Structures): The authors give a characterization of the process of building up interpretation for relative clause sequences, with a preliminary indication of how a cross-language typology for these structures might be developed and how information may be copied from one tree structure to
another. Setting out this typology will involves surveying restrictive and non-restrictive relatives, head-initial, head-final and head-internal relatives. For demonstrating the cross-linguistic validity of this method, the examples from some languages are given.
Chapter 5 (Wh questions: A General Perspective): in this chapter the authors investigate the range of wh questions across languages in the framework of dynamic syntax they have build in the previous chapters. Many questions on wh problems are discussed and defined: the semantic diversity, scopal properties of wh expressions, wh-initial vs wh-in-situ structures, expletive wh structures. One important difference between this approach and all others is that it completely
lacks the concept of the wh expression corresponding to an operator binding some discrete position and variable.
Chapter 6 (Crossover Phenomena): the focus of this chapter is the interaction between the processes involved in establishing the interpretation of left-dislocated wh expressions and anaphoric expressions - the phenomenon of crossover. What we can see in these analyses of relative clauses and wh questions is that natural
typologies emerge as individual language systems which are analyzed as making available different underspecified descriptions, with consequent variation in the dynamics of their resolution.
Chapter 7 (Quantification Preliminaries) outlines ways in which quantification might be reflected in this model.
Chapter 8 (Reflections on Language Design): the authors step back to consider the general issues raised by this dynamic view for studying the relevance of the results on general linguistic theory. On this view, the well-formedness of a string is: if, using all its words in a left-to-right order, it creates at least one logical form. The syntax of natural language will thus be described as incrementally building
up structures representing the interpretation. For the readers, the first part of this chapter provides an overall perspective of the whole book in general language. Other subchapters are: 'Underspecification and the Formal Language Metaphor, English is not a formal language', 'Well-formedness and Availability of Interpretations', 'Universals and Language Variation', 'On Knowledge of Language'. The final chapter (The Formal Framework) provides a more explicit and formal statement of the rules and assumptions set out in Chapter 2 and 3. The declarative and procedural part of this model is described by Propositional Dynamic Logic (PDL). In addition to the mentioned nine chapters, there is a 12-page bibliography and 9-page 'general index'. As a book with a rich logical notation, the authors also add a useful 'symbol index' in the end of this book. This book has beautiful aspects, but it is not an easy thing to attain this result for such content with complex symbols and tree structures. There is a small mistake in Bibliography. The authors list the Jiang (1995) and Yiang (1995) as two different items of same PhD thesis of Jiang Yan; the latter is the result of an erroneous
understanding of a Chinese surname.
In the preface of this book, the authors say that they want to convince their readers that the time has come to shift from the static perspective of formalisms based on the familiar classical logics to a more dynamic formalism where the emphasis is on the process and the incremental development of structure. In fact, we can see such shift from some others works, for example, Left Associative Grammar (Hausser, 2001), Axiomatic Grammar (Milward, 1991), Dynamic Dependency Grammar (Milward, 1994), State Transition Grammar (Tugwell, 1998). In other words, the researchers now pay attention to the dynamic process
of language understanding. Although we have had many formal theories and systems for natural language processing, the final goal of language processing by computer is still a long way off. The model proposed by the authors of this book undoubtedly is a useful exploration for natural language processing in particular, and the study of human intelligence in general. According to the principle of
this model, it seems a better simulation of the mechanism of human language processing than others. Because there are essential differences between the structures of computers and humankind, we do not know whether we should make the systems following the mechanisms of machines or humankind. Fortunately, many experiments and studies on language processing have proved that language processing is a knowledge-based mechanism. Unfortunately, we still do not clearly know what precise meanings are, how to use this knowledge during the
procedure of language understanding, and how to structure and store this knowledge in the computer. We have a right to reserve the implications of such mechanism in a system, but the knowledge processing elements is certainly necessary to any system of language processing. In the eyes of the authors, at least, ''so-called 'knowledge' of language is simply the capacity to construct
appropriate decorated structures as interpretations of stimuli of a particular sort''.
We remark that in this book the authors use IF-THEN-ELSE fashion as conditions of lexical action (transition), it is not difficult to view that as a knowledge-driven action. Using IF-THEN-ELSE as the method of knowledge representation is not rare in the technique of artificial intelligence (Expert System based on rule). Feng Zhiwei has mentioned the usefulness of this rule's representation in natural language processing (Feng, 1992:116-123).
In this book, there are many examples from different languages, in other words, the model is cross-linguistically valid.
This book is very helpful and useful to anyone, who wants to know how to simulate the capability of human language processing from the common-sense view in a logical fashion. More precisely, it is useful to all interested in the modeling of natural languages based on logic in particular, language understanding and linguistic theory in general.
FENG Zhiwei (1992) Zhongwen Xinxichuli yu Hanyu yanjiu (Chinese Information Processing and The study of Chinese). Beijing: Commercial Press.
Hausser, R. (2001) Foundations of Computational Linguistics. The second edition. Berlin: Springer.
Milward, D. (1991) Axiomatic Grammar, Non-Constituent Coordination and Incremental Interpretation . PhD. Thesis, Computer Laboratory,
University of Cambridge.
Milward, D. (1994) Dynamic Dependency Grammar, Linguistics and Philosophy 17, 561-605.
Tugwell, D. (1998). Dynamic syntax. Ph.D Dissertation. Edinburgh University.