After Noam Chomsky (Reference Chomsky1995) put forward the idea of Minimalist Program, a more concrete formalization of it called Minimalist Grammar (MG) was proposed by Edward Stabler (Reference Stabler1997). The book under review, Minimalist Parsing (henceforth MP), edited by Robert C. Berwick and Stabler, is a collection contributing to both theoretical and applied issues of the MG framework. This review not only gives a brief introduction to the book but also positions it from a historical perspective and provides insights into how the fields of theoretical linguistics (TL) and computational linguistics (CL) can be combined.
A brief introduction to the volume
To start with, a few statistics might be informative: the volume comprises seven chapters, including the editors’ introductory chapter. Nine authors are involved in contributing to the volume. More than half of them are from their universities’ departments of linguistics, while others come from the discipline of computer science or neuroscience. At the same time, almost all of them once have had educational backgrounds in science (mostly computer science) and engineering, or received interdisciplinary training. Two of them are Stabler’s former PhD students, and one of them adopts his MG framework in their doctoral thesis.
Lying at the heart of all seven chapters is Stabler’s foundational work in MG. Although all the chapters are related to the same theme, the whole volume is somehow loose and overarching as explained in the introduction (p. 1). Therefore, MP also embraces theoretical descriptions (Ch. 2) and a neurolinguistic study (Ch. 7), rather than exclusively focusing on computational parsing. In that vein, it could equally have been titled The Application and Verification of Minimalist Grammar, to help the reader grasp the gist. Note that MG is just one formulation of Minimalist Program based on Stabler’s understanding, as explained in Collins and Stabler (Reference Collins and Stabler2016), and one formalization can also diverge into different implementations, which is why the reader finds different varieties in the various chapters. This is worth noting in case some generative linguists might argue that Stabler has somehow misunderstood the original thoughts of Chomsky.
Chapter 1, by Berwick and Stabler, plays two separate roles. On the one hand, it summarizes the following chapters (Section 1.2). On the other hand, it describes the theoretical foundation in common, that is, MG (Section 1.3), so that the reader could translate between different varieties proposed by the contributors in the volume. This foundation is a successor of Stabler (Reference Stabler1997, Reference Stabler2001, Reference Stabler2011). The essence of MG can be summarized as follows: MG is a beautifully and simply defined formalism with four types of features (categories, selectors, licensees, and licensors) in set-theoretic terms and just one operation (merge, internal, or external) in its original version. The features are grouped into two pairs, and the attraction between the corresponding features in each pair serves as the motivation of external merge (or simply ‘merge’) and internal merge (move). The expressive power of MG is mildly context-sensitive, that is, somewhere between context-free and context-sensitive grammars, which is generally considered to be closest to human language. The syntactic structures in MG can be both presented in tree forms with an arrow indicating the head and in linear forms with operations and brackets, just like an arithmetic expression. In his 1997 paper, Stabler also insightfully distinguishes between the notion of syntactic structure, which is a relational network among the parts of a sentence or what a parser assigns to a sentence, and formal grammar, which is a set-theoretic characterization of a language, that is, the set of all grammatical sentences in the technical sense, or a boundary between the sets of well-formed and non-well-formed strings. Many syntacticians barely differentiate between these in their textbooks (see e.g., Carnie Reference Carnie2013, which is nevertheless one of the best textbooks of syntax).
There are three noteworthy points about Chapter 1. First, the authors depict MG without the tool of feature structure, even though sentence structures in MG could no doubt be transcribed in that form, which is standard for modern syntactic theories. Second, the features that he sets as the propellant of merge are somehow close to the notion of valency in dependency grammar, which Lucien Tesnière (Reference Tesnière2015) describes as a bond (Fr. crochet), and which is one of the reasons why dependency grammarians argue that Minimalist syntax is evolving into a version of dependency grammar (Osborne, Putnam, and Gross Reference Osborne, Putnam and Gross2011). Last but not least, it needs to be clarified, as many others have done previously, that notions such as derivation, merge, and move are just metaphors. These concepts are but resorting to a dynamic-like terminology to describe static structural relations. Only the parsing process is involved in real-time computation. If there is an MG treebank, all such terms will simply be represented by relations between entities.
Moving on, Chapters 2 and 3, both formulated by the same pair of authors, Sandiway Fong and Jason Ginsburg, respectively, focus on characterizing a modified version of MG and address five specific topics: including multiple agree, expletives, thematization, wh-movement, and relative clauses within their own framework. Kristine M. Yu adds prosodic information to MG and offers a case study of Samoan prosodic constituency structure in Chapter 4. In Chapter 5, Gregory M. Kobele employs a proof-theoretic approach to syntactic derivation in a similar way to MG. He tackles the issue of ellipsis specifically. Tim Hunter’s Chapter 6 is the most relevant one with respect to the central topic of parsing. He proposes a left-corner parsing strategy to make up for what a top-down parser cannot do. The last chapter by Jixing Li and John Hale integrates insights from neurolinguistics. It discusses how humans parse, as opposed to machines, from the perspective of Minimalism. However, this chapter is a little abrupt, since nothing specific about the Minimalist framework is presented and the analytical conventions employed are in fact in the form of a Government and Binding version of generative syntax. The reason for this may be attributed to the fact that all the articles in this volume are from a workshop on Minimalist Parsing held at MIT in 2015, and the themes are therefore not rigorously restricted.
The position of MP from a historical point of view
Where do we find the position for MP from a historical point of view? To explain this, something needs to be said about a puzzle which can be formulated as ‘formal syntax is not formal’ (FSNF). It is generally acknowledged that the later versions of generative grammars are rarely formalized (see Müller Reference Müller2018, Section 1.2), far away from the starting point where Chomsky originally set out, while at the same time, practitioners are still firmly occupying the name of formal syntax. Thus, most of the works since the 1970s have attracted such critiques. However, Edward Stabler, the second editor of this book, is among the last scholars who have devoted themselves to bucking this trend. He is a strong advocate of the formalization approach of generative linguistics. As early as in 1992, he published a monograph on the formalization and computational implementation of Government and Binding version of generative grammar. His seminal paper of 1997 laid the foundation for the MG, which forms the backbone of this book, and gave rise to a large amount of research, including doctoral theses supervised by Stabler himself (such as Harkema Reference Harkema2001; Kobele Reference Kobele2006), and a number of papers presented at the International Conference on Logical Aspects of Computational Linguistics, one of the top conferences in the field (extending from Cornell Reference Cornell1999 to Stanojević Reference Stanojević2016). This research has progressed so far that the properties of MG, such as its generative power and the complexity of parsing them, are already well studied (p. 12). Sadly, however, all of these might be unfamiliar even to those who have pursued Chomsky’s tenets in the linguistic field, since most of the work was presented in the field of CL. Later, in 2016, he co-authored a paper with Chris Collins and published it in an influential linguistic journal, Syntax, which officially announced the advent of the age of formalization of Minimalism. They brought what had been extensively studied in CL, insights of which come from TL, back into the view of theoretical linguists. MP is then a natural follow-up of that paper. Once you formalize a theory, you naturally turn to consider how to implement it. This time MP is published by OUP under the subcategory of linguistics, which theoretical linguists are more familiar with, rather than in a computational linguistics series by Springer. In light of this, the book is in a position to bring into the view of mainstream linguistic communities what computational linguists are doing and how far they have gone. Stabler has thereby set up a good example of bridging the gap between TL and CL to solve the FSNF Puzzle.
Conclusion
To summarize, MP on the one hand is targeted at computational linguists who work under the specific formalism of Minimalism. On the other hand, it can also serve as a thread which leads a linguist into Edward Stabler’s sustained efforts in the formalization of mainstream generative grammar. The editors have provided abundant and detailed literature in the introductory chapter so that interested readers could easily follow it to learn more about MG. Furthermore, just as the editors say in the preface, the efforts made in MP are also expected to ‘inspire a new generation of researchers to continue to explore how linguistics and computation can be made to fit together’ (p. ix). More gap bridgers like Stabler and his colleagues are expected to appear, for example, by training linguistics students to think computationally and teaching them about linguistic formalism as their working language in the curriculum. To this end, a (real) formal syntax textbook is also called for. Phonologists are already on their way to increased use of more formal linguistics (Bale and Reiss Reference Bale and Reiss2018), and this book helps pave the way for syntacticians to follow up.
Acknowledgments
I would like to express my gratitude to the review editor for scrutinizing the content, for commenting, and especially for improving the language. Any potential faults are mine.
