EDITORS: Markus Werning, Wolfram Hinzen, Edouard Machery TITLE: The Oxford Handbook of Compositionality PUBLISHER: Oxford University Press YEAR: 2012
Michael T. Putnam, Penn State University
One of the most contentious topics in modern semantic inquiry -- and one that has drawn considerable debate -- is the concept of compositionality. In modern linguistic thought, the notion that the meaning of an expression is generally determined by the syntax and meaning of its (sub)components was first proposed by Frege, and, in more recent times, has been extended to studies related to what constraints are at play between syntactic and semantic systems, the underlying structure of representational systems and neural architectures. With the recent recommitment to attempt to connect linguistics more directly with other academic disciplines (e.g. language and mind, philosophy, and areas in cognitive science), this handbook is a much needed and timely addition to the literature. The scope and content of this handbook extends far beyond simply being a reference work that will only be consulted and cited as a trusted reference in term papers, dissertations, and journal articles. Rather, the contributions that Werning, Hinzen and Machery have compiled here provide a cutting-edge snapshot of our growing knowledge (including some contentious debates) of what the scientific community understands regarding the concept of compositionality, and exactly in what domains this knowledge is incomplete or in need of reconsideration and revision.
This volume consists of seven parts, with the focus of each subsection being dedicated to a primary topic pertaining to composition. In Part 1 (“History and overview”), the contributions are directed towards a clearer understanding of the history of compositionality in linguistics and philosophy. Theo Janssen starts the volume with his contribution entitled, “Compositionality: Its historical context”. After a detailed rundown of linguists’ and philosophers’ thoughts on the notion of compositionality throughout the 19th and 20th centuries, Janssen arrives at the conclusion that the wide support for compositionality is not for principled reasons, but for practical reasons. Marcus Kracht’s chapter (“Compositionality in Montague Grammar”) discusses how compositionality has traditionally been modeled in Montague Grammar. The third contribution in this section written by Zoltán Gendler Szabó (“The case for compositionality”), adopts a more philosophical point of view, discussing the difficulty of formulating a concrete definition of compositionality. In particular, Szabó raises interesting questions such as the following, “Are we concerned with ‘meanings’ that are being combined, or is it syntactic constituents? Are we talking about the meaning that constituents have individually, or that they have when taken together?” (p. 2). Szabó instantiates the hypothesis that once we have fixed the individual meanings of syntactic constituents in a given expression, and have fixed a particular syntactic structure, that no other components exist that will contribute to and determine meaning. The final paper in this section, Thomas Eide Zimmermann’s “Compositionality problems and how to solve them”, illustrates the role that compositionality has played as a constraint on semantic analysis in a number of case studies (e.g., quantified NPs in object position and intensional arguments).
The primary concern of Part II (“Compositionality in language”) highlights areas where philosophers and linguists have traditionally differed in fundamental ways in their understanding and application of compositionality. Pauline Jacobson (“Direct compositionality”) advances the argument that the mapping between syntactic structure and semantic meaning does not require any ‘hidden’ level of structural representation, i.e., Logical Form (LF), and can be mediated through a more direct mapping between some form of combinatorial syntactic grammar and a compositional semantics. The contribution by Paul Pietroski (“Semantic monadicity with conceptual polyadicity”) focuses on the connection between thought and linguistic expression. Pietroski discusses the disconnection between lexical meanings, which are held to be uniformly monadic, and arguments, which saturate these meanings. According to Pietroski, a simple conjoining operation (rather than a combinatorial operation) is responsible for maintaining strictly monadic relations. Francis Jeffry Pelletier (“Holism and compositionality”) and François Recanati (“Compositionality, flexibility, and context dependence”) address potential counterarguments against compositionality, when they discuss holism and context dependency respectively and how/if a concept of compositionality can exist in relation to these contrasting notions. Dag Westerståhl (“Compositionality in Kaplan style semantics”) addresses how compositionality can best be modeled in Kaplan style semantics, a theoretical approach that allows certain kinds of semantic values to be determined by extra-linguistic contexts. Introducing graded notions of compositionality (e.g., strictly, contextually, and weakly compositional readings), Westerståhl demonstrates how problematic phenomena such as indexicals, unarticulated constituents, modulation, and pragmatic intrusion can be modeled to conform to (some version) of compositionality. Lastly, Sebastian Löbner (“Sub-compositionality”) writes on the issue that some syntactic construction types do not correspond to a single way of combining the semantic types assigned to their constituents. Löbner sketches out a novel approach to deal with problems that arise from sub-compositionality, such as the elimination of a homomorphy of composition operations in the syntactic and semantic domains.
Part III (“Compositionality in formal semantics”) consists of three papers, whose purpose, which in some regards is similar to the discussions and debates found in Part I, is to arrive at a working historical and contemporary definition of compositionality and how it can and should be best applied in formal semantic analyses. Wilfrid Hodges (“Formalizing the relationship between meaning and syntax”) treats both historical and formal aspects of compositionality in semantics. From a historical perspective, Hodges traces the concept of compositionality back to the Aristotelian theory of meaning and Arab philosophers such as Ibn Sina and Al-Farabi. From a formal perspective, Hodges returns to the so-called Extension Problem: Given that one has already assigned meanings to some subset of expressions in a language by some meaning function, what are the conditions under which one can extend this original meaning function to the rest of the language? Hodges discusses how his (2001) extension theorem addresses these issues. Gabriel Sandu’s contribution, “Compositionality and the Context Principle”, continues Hodges’s discussion of the Extension Problem. The prime aim of Sandu’s chapter focuses on the relation between the principle of compositionality and the Principle of Contextuality. This third section closes with Tim Fernando’s paper entitled “Compositionality in discourse from a logical perspective”, where he extends the discussion of compositionality to include discourse-level phenomena. Fernando provides an overview of a number of formal semantic theories of discourse, including model-theoretic treatments such as Discourse Representation Theory (DRT; Kamp and Reyle, 1993) and competing proof-theoretic accounts.
The fourth subsection of volume (“Lexical decomposition”) moves toward the question of the meaning of lexical constituents themselves. Dieter Wunderlich’s “Lexical decomposition in grammar” delivers a survey of ideas originating in generative semantics and how they have been implemented in various theoretical approaches, eventually arriving at Lexical Decomposition Grammar (which stands in contrast to the strictly atomist position of Fodor & Lepore 1998, 1999). Similar to Wunderlich, Heidi Harley argues against the atomist position and in favor of some version of lexical decomposition in her chapter, “Lexical decomposition in modern syntactic theory”. Harley approaches this problem from modern minimalist syntax, and shows clear ways where modern theory has adopted core conceptual notions found in Generative Semantics, as well as areas of divergence. Wolfram Hinzen’s piece, “Syntax in the atom”, re-opens the atomist vs. lexical decomposition debate, approaching this debate in similar fashion to Pietroski’s earlier contribution in Part II, from the perspectives of the atomicity of human thought and its relation to lexical decomposition. In conclusion, Hinzen accepts that the only logical approach to amend these concerns is through a revision of the architecture of the grammar. James Pustejovsky’s “Co-compositionality in grammar” closes out this section on lexical decomposition. Pustejovsky takes a closer look at ‘co-composition’, a term he uses to define the process of how compositional processes determined by the phrasal structure of an expression are supplemented by additional interpretative mechanisms at the interface between the mental lexicon and syntactic structure.
Part V (“The composition of mind”) shifts the focus of the volume to theoretical challenges involving the principle of compositionality and its relation to psychology. In “Emergency!!! -- Challenges to a compositional understanding of noun-noun combinations” Edward Wisniewski and Jing Wu explore how speakers interpret novel noun-noun compounds (e.g. ‘zebra football’) with a particular focus on speakers of English and Chinese. Wisniewski and Wu show that speakers attribute ‘emergency properties’ that are not attributed to the individual members of the compound. James Hampton and Martin Jönsson continue with this theme in their contribution, “Typicality and compositionality: The logic of combining vague concepts”. Hampton and Jönsson introduce and endorse the prototype theory of concepts and its relation to the principle of compositionality. Lila Gleitman, Andrew Connolly, and Sharon Lee Armstrong challenge the core notions held in prototype theory in their chapter, “Can prototype representations support composition and decomposition?” Jesse Prinz’s “Regaining composure: a defense of prototype compositionality” also enters into this ongoing debate regarding the validity of prototype theory in defense of the psychological research on concept combination. In closing, Edouard Machery and Lisa Lederer present a critical overview of three influential models of concept combination in their chapter, “Simple heuristics for concept combination”.
“Evolutionary and communicative success” is the central focus of the contributions found in Part VI. Michael Arbib (“Compositionality and beyond: Embodied meaning in language and protolanguage”) argues that natural languages have compositionality but that they need not intrinsically be compositional by default in their design. Arbib compares and contrasts ‘the compositional view’ with ‘the holophrastic view’ in describing the ontology of the protolanguage of homo sapiens, with the former championing the position that protolanguage consisted of “words, but not syntax” and the latter where protolanguage consisted of communicative acts that could not be further decomposed into smaller meaningful units. In this chapter, Arbib supports ‘the holophrastic view’, which plays an important role in Kenny Smith and Simon Kirby’s contribution, “Compositionality and linguistic evolution”. Smith and Kirby take issue with Pinker and Bloom’s (1990) approach to biological evolution and, in contrast, hypothesize that compositionality is a socially learned behavior. Peter Pagin’s “Communication and the complexity of semantics” challenges the traditional idea that compositionality is required for a grammar to be learnable. Pagin postulates that compositionality ensures that complex expressions can be learned and processed in a quick and efficient manner, and, as a result, compositionality reduces computational complexity. This discussion of evolution and communication success closes with Gerhard Schurz’s chapter, “Prototypes and their composition from an evolutionary point of view”, where he presents his case for why prototypes are an efficient way of representing natural classes of objects and concepts.
The final section of this volume, Part VII (“Neural models of compositional representation”), is dedicated to cutting-edge research involving neural networking and its relation to theories of compositionality. Terry Horgan’s programmatic article, “Connectionism, dynamical cognition, and non-classical compositional representation”, develops a dynamical-cognition framework to account for a non-traditional notion of compositionality. Horgan suggests that his framework might provide a foundation for cognitive science. Martina Penke (“The dual-mechanism debate”) takes on the task of addressing the controversy between classicism and connectionism in psycholinguistic modeling of morphological inflections. Terrence Stewart and Chris Eliasmith (“Compositionality and biologically plausible models”) compare recent proposals for the implementation of compositionality in local and distributed connectionist models. Alexander Maye and Andreas Engel (“Neural assembly models of compositionality”) present empirical evidence in favor of object-related neural synchrony in the cortex, as well as topologically structured cortical feature maps. The idea of developing a neuro-emulative semantic system is taken up in Markus Werning’s contribution, “Non-symbolic compositional representation and its neuronal foundation: toward an emulative semantics”. Although structurally similar to model-theoretic approaches to semantics (e.g. Discourse Representation Theory (DRT)), Werning employs the notion of set-theoretic constructions of neural emulations (but not of their denotations). The section and volume as a whole concludes with Giosuè Baggio, Michiel van Lambalgen, and Peter Hagoort’s chapter entitled “The processing consequences of compositionality”. In this chapter they speculate as to whether or not the notion of compositionality can be reduced to a testable principle from a processing perspective. Although they uphold compositionality in most cases, they point out that an account of processing from a compositional standpoint faces significant challenges when it must account for interactions between sentences and discourse context, perceptual cues and stored knowledge.
Reviewing a handbook represents a serious challenge, because in most cases, a handbook contains discussions and treatments of long-held axioms and conventions that are generally agreed upon as being more or less “canonical” in a given field of academic study. Although this handbook definitely exhibits these qualities to some extent -- especially in the first half of the volume -- the editors should be commended for also having the vision to create a forward-looking volume that highlights both current debates in the field as well as speculative theoretical questions that will likely shape and direct future research endeavors in the years to come. The editors did an excellent job of bringing together academics with various, diverse specializations (e.g. philosophy, linguistics, and cognitive science) to contribute to a volume with such a broad scope of coverage. As is the case with most of these handbooks, the price of these volumes, especially those which are not published in a paperback edition as well as a hardback, is far too pricey for the average consumer to buy. Aside from the unfortunate high price for this reference work, this volume and its contents will undoubtedly have a high impact in various fields of language science for years to come.
Fodor, J. & E. Lepore. 1998. The emptiness of the lexicon: Reflections on James Pustejovsky’s ‘The Generative Lexicon’. Linguistic Inquiry 29: 269-88.
Fodor, J. & E. Lepore. 1999. Impossible words? Linguistic Inquiry 30: 445-53.
Hodges, W. 2001. Formal features of compositionality. Journal of Logic, Language and Information 10: 7-28.
Kamp, H. & U. Reyle. 1993. From discourse to logic: Introduction to model-theoretic semantics of natural language, formal logic and discourse representation theory. Dordrecht: Kluwer.
Pinker, S. & P. Bloom. 1990. Natural language and natural selection. Behavioral and Brain Sciences 13.4: 707-84.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Michael T. Putnam is an Assistant Professor of German & Linguistics at Penn State University. His primary research interests lie in the fields of theoretical syntax, lexical semantics, contact linguistics, and bilingualism.