Jump to Content Jump to Main Navigation

Linguistics Compositionality
by
Jeff Pelletier

Introduction

“Compositionality” is used in two different senses, and sometimes the literature seems to be antagonistic because the discussants are employing the different senses. In one sense—let’s call it the “ontological sense”—some complex thing is compositional if it is identical with its parts (with due consideration to the way the parts are arranged). In another sense—let’s call it the “functional sense”—something is compositional if it is a complex thing with some property that can be defined in terms of a function of the same property of its parts (with due consideration to the way the parts are combined). In formal semantics and philosophy of language, the complex things are (usually) syntactically complex items of language, and the property of interest is (usually) meaning. So the question of whether some complex thing is compositional is normally understood as asking whether the meaning of some complex piece of language is a function of the meanings of its parts together with consideration as to how those parts are syntactically combined. Many of the writings about compositionality have occurred in the philosophical literature, and there the contrast has often been with meaning holism. In the “formal” literature the question has been about the conditions on a language (and associated meaning function) that will guarantee that there is some compositional semantics. Within the linguistics literature, much of the writing has been to show how some apparently noncompositional construction can be given a compositional treatment or to argue that it cannot be. There has also been a separate discussion about the effects of context on meaning and how that interacts with communication. Many of these works have broached the topic of whether and how context can be accommodated compositionally.

General Overviews and Definitions

Most of the works that deploy the notion of compositionality in any way at all also make a statement as to what the author believes compositionality to be. But these statements are not usually accompanied by an analysis of terms used in their explanation. This section contains some works that analyze the notion of compositionality. Partee 2004 (originally published in 1984) is one of the earliest works on compositionality, introducing most of the issues taken up by later writers. Pelletier 2004 discusses the relevance of compositionality to semantics and surveys the various considerations that have been prominent in discussions of the notion. Dowty 2007 is by the author of one of the first textbooks to employ the classical version of compositionality as put forward by Montague 1974 (cited under Problem of Logic). Hodges 1998 argues that many statements of compositionality are in fact statements of some other (perhaps related) notion; notably Hodges’s discussion makes the Montague 1974 (cited under Problem of Logic) version be “not compositionality” and be the cause of what “lies behind the bid to describe [various claims about linguistics] as ‘problems for compositionality.’” Janssen 1997 also provides much historical detail concerning the development of compositionality. Jacobson 2011 describes a particularly “pure” version of compositionality. Pagin and Westerståhl 2010 offers a detailed exposition of the various terms employed in discussions of compositionality, while Sandu and Hintikka 2001 also discusses different interpretations of compositionality in the literature. Szabó 2007 is a large descriptive survey of work on compositionality together with some recommendations on which of the interpretations is best.

  • Dowty, David. 2007. Compositionality as an empirical problem. In Direct compositionality. Edited by Chris Barker and Pauline Jacobson, 23–101. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    The early portions of this article describe the variety of formulations that have been offered for compositionality along with an attempt to adjudicate among them.

    Find this resource:

  • Hodges, Wilfrid. 1998. Compositionality is not the problem. Logic and Logical Philosophy 6:7–33.

    Save Citation »Export Citation »E-mail Citation »

    Hodges analyzes the terms involved in defining “compositionality,” investigating the historical usage of the term and the general idea behind the term. This is a fundamental paper in the new understanding of compositionality.

    Find this resource:

  • Jacobson, Pauline. 2011. Direct compositionality. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 109–128. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    A clear statement of the notion of direct compositionality—the idea that there should be no “movement rules” in the syntax and that each syntactic rule will give rise to a unique semantic interpretive rule.

    Find this resource:

  • Janssen, Theo. 1997. Compositionality. In Handbook of logic and language. Edited by Johan van Benthem and Alice ter Meulen, 417–473. Amsterdam: Elsevier.

    Save Citation »Export Citation »E-mail Citation »

    This work includes many topics from Janssen’s 1983 underground-classic dissertation. It discusses various formal statements of compositionality and its relation to the “context principle.”

    Find this resource:

  • Pagin, Peter, and Dag Westerståhl. 2010. Compositionality I: Definitions and variants. Philosophy Compass 5:250–264.

    DOI: 10.1111/j.1747-9991.2009.00228.xSave Citation »Export Citation »E-mail Citation »

    This portion of a two-part survey article explains the various terms used in definitions of compositionality and provides a variety of formulations of the notion. The logical relationships among the different formulations are shown.

    Find this resource:

  • Partee, Barbara H. 2004. Compositionality. In Compositionality in formal semantics: Selected papers. By Barbara H. Partee, 153–181. Malden, MA: Blackwell.

    DOI: 10.1002/9780470751305.ch7Save Citation »Export Citation »E-mail Citation »

    This article points out that there are various interpretations of the notion of compositionality and that some of the popular versions obviate the need for an “intermediate” level of representation, and it discusses some putative violations of the principle. Originally printed in Fred Landman and Frank Veltman, eds., Varieties of Formal Semantics: Proceedings of the Fourth Amsterdam Colloquium, September 1982 (Dordrecht, The Netherlands: Foris, 1984), 281–312.

    Find this resource:

  • Pelletier, Francis Jeffry. 2004. The principle of semantic compositionality. In Semantics: A reader. Edited by Steven Davis and Brendan S. Gillon, 133–156. New York: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Contains a discussion of the terms used in most statements of compositionality, such as “meaning,” “function,” “whole,” “part,” and the like. The position taken is that these terms are vague in their meaning and have given rise to differing accounts and criticisms of compositionality. Originally printed in Topoi 13 (1994): 11–24; the book chapter contains additions.

    Find this resource:

  • Sandu, Gabriel, and Jaakko Hintikka. 2001. Aspects of compositionality. Journal of Logic, Language, and Information 10:49–61.

    DOI: 10.1023/A:1026530709583Save Citation »Export Citation »E-mail Citation »

    Introduces three senses of “compositionality” and shows how they are related by applying Hodges’s theorem (Hodges 2001, cited under Problem of Logic) to the Hintikka-Sandu version of logics of imperfect information (within “independence friendly languages”).

    Find this resource:

  • Szabó, Zoltán. 2007. Compositionality. In Stanford encyclopedia of philosophy. Edited by Edward N. Zalta.

    Save Citation »Export Citation »E-mail Citation »

    This thorough discussion of the philosophical aspects of compositionality includes the author’s earlier views of compositionality as supervenience, which is one of the standard viewpoints concerning the topic.

    Find this resource:

Historical Antecedents

Many scholars trace the conscious employment of a notion of compositionality to Gottlob Frege (b. 1848–d. 1925). However, there are also scholars who doubt this attribution. When discussing this principle in the context of Frege, it is important to pay attention to his midcareer introduction of the sense-reference distinction. While it is widely thought that Frege held the principle of compositionality for reference, it is argued by some (Janssen 2001, Janssen 2011, Pelletier 2001) that it is less clear that he held it for sense. Carnap 1956 first attributed the principle to Frege, and Michael Dummett (Dummett 1981a, Dummett 1981b) found places where it might plausibly be thought to occur in Frege. Besides studying Frege for this notion, there have been attempts to find it employed in earlier works, especially in the non-Western tradition. Hodges 2006 and Pagin and Westerståhl 2010 investigate such places. In the modern era, the term seems to have been employed for the first time in something like the modern sense in Katz and Fodor 1963.

  • Carnap, Rudolf. 1956. Meaning and necessity: A study in semantics and modal logic. 2d ed. Chicago: Univ. of Chicago Press.

    Save Citation »Export Citation »E-mail Citation »

    Carnap was the first to call compositionality “Frege’s principle” and to attribute compositionality for both reference and sense to Frege (p. 121). First published in 1947.

    Find this resource:

  • Dummett, Michael. 1981a. Frege: Philosophy of language. 2d ed. Cambridge, MA: Harvard Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Dummett (p. 152) identifies compositionality of sense as the first of Frege’s important theses and cites numerous places in Frege’s writings where this is suggested. However, Dummett also identifies Frege’s “context principle” as his basic thesis. So the apparent tension between compositionality and contextuality is a basic interpretative crux for understanding Frege. Much of this book is concerned with that issue.

    Find this resource:

  • Dummett, Michael. 1981b. The interpretation of Frege’s philosophy. London: Duckworth.

    Save Citation »Export Citation »E-mail Citation »

    This work touches on much of the same issues concerning compositionality as Dummett 1981a but in a somewhat wider arena, where a larger portion of Frege’s philosophical interests are considered.

    Find this resource:

  • Hodges, Wilfrid A. 2006. Two doors to open. In Mathematical problems from applied logic: Logics for the XXIst century. Vol. 1. Edited by Dov M. Gabbay, Sergei S. Goncharov, and Michael Zakharyaschev, 277–316. New York: Springer.

    Save Citation »Export Citation »E-mail Citation »

    In Part 2 of this article Hodges traces the notions involved with the “infinity of language” as they appear in works by medieval Arabic linguists and commentators. He also remarks on some unpublished work of Brendan Gillon that finds similar sentiments in the Sanskrit linguist Patañjali in his commentary on P\=a\d{n}ini.

    Find this resource:

  • Janssen, Theo. 2001. Frege, contextuality, and compositionality. Journal of Logic, Language, and Information 10:115–136.

    DOI: 10.1023/A:1026542332224Save Citation »Export Citation »E-mail Citation »

    Discusses the relative importance of compositionality and contextuality in Frege’s writings, taking a chronological approach to investigate the development of Frege’s thoughts on the two concepts. Janssen concludes that Frege embraced contextuality throughout his career and that the various places where one is tempted to read compositionality (such as those places where he appears to employ the argument from understandability) can be otherwise explained.

    Find this resource:

  • Janssen, Theo. 2011. Compositionality: Its historic context. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 19–46. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Discusses the development of compositionality and contextuality from the mid-19th century, through Frege, to its modern employment. Although both principles were discussed in the early days, it was contextuality that was accepted. Janssen concludes with the view that compositionality nowadays is accepted for “practical reasons,” not for principled ones.

    Find this resource:

  • Katz, Jerrold J., and Jerry A. Fodor. 1963. The structure of a semantic theory. Language 39:170–210.

    DOI: 10.2307/411200Save Citation »Export Citation »E-mail Citation »

    This article seems to be the first to use the term “compositionality” in more or less the modern sense—even though the authors seem to be more concerned with lexical composition than composition in larger portions of syntax.

    Find this resource:

  • Pagin, Peter, and Dag Westerståhl. 2010. Compositionality I: Definitions and variants. Philosophy Compass 5:250–264.

    DOI: 10.1111/j.1747-9991.2009.00228.xSave Citation »Export Citation »E-mail Citation »

    Part 1 of this two-part article discusses “particular semantic analyses that are in fact compositional” that were given by early theorists. As examples the authors mention Plato, Śabara, Jaimini, Peter Abelard, John Buridan, Peter of Ailly, and Frege. However, it seems that the pre-Frege thinkers did not really have an idea of compositionality in quite the modern manner.

    Find this resource:

  • Pelletier, Francis Jeffry. 2001. Did Frege believe Frege’s principle? Journal of Logic, Language, and Information 10:87–114.

    DOI: 10.1023/A:1026594023292Save Citation »Export Citation »E-mail Citation »

    This paper argues, by looking at a wide variety of Frege’s texts, that he held neither the principle of compositionality nor the principle of contextuality in any firm manner. It also evaluates the various attempts made by scholars who claim that Frege held both principles.

    Find this resource:

Collections

Only a small number of collections focus on compositionality, all of which have been put together in the early 21st century. The most thorough is Werning, et al. 2011, and it is expected to be the standard reference for topics involved in compositionality, although it is perhaps geared more toward philosophical issues than linguistic issues. Pagin and Westerståhl 2001 contains many papers that are regarded as foundational in the various aspects of the topic. Unfortunately, Werning, et al. 2005 and Werning, et al. 2006 are often quite difficult to obtain, but many of the authors of papers in these volumes (as well as in Pagin and Westerståhl 2001) have written contributions to Werning, et al. 2011. Barker and Jacobson 2007 describes a particular direction in linguistics that employs compositionality.

  • Barker, Chris, and Pauline Jacobson, eds. 2007. Direct compositionality. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    This book examines the hypothesis of “direct compositionality,” which requires that semantic interpretation proceed in tandem with syntactic combination. This was the dominant view in formal semantics of the 1970s and 1980s, and the present volume is an attempt at reviving it by considering a number of different phenomena in various languages.

    Find this resource:

  • Pagin, Peter, and Dag Westerståhl, eds. 2001. Special issue: Compositionality; Current issues. Journal of Logic, Language, and Information 10.1.

    Save Citation »Export Citation »E-mail Citation »

    This special issue contains six articles plus an introduction, and all of them have attained a special place in the area of compositionality. The individual articles will be cited in other sections of this article.

    Find this resource:

  • Werning, Markus, Edouard Machery, and Gerhard Schurz, eds. 2005. The compositionality of meaning and content. Vol. 1, Foundational issues. Frankfurt: Ontos Verlag.

    Save Citation »Export Citation »E-mail Citation »

    Volumes 1 and 2 of this two-volume set are the result of a very successful two-part conference. Volume 1 reports the papers that were presented in the first of the two parts, which was held in Düsseldorf.

    Find this resource:

  • Werning, Markus, Edouard Machery, and Gerhard Schurz, eds. 2006. The compositionality of meaning and content. Vol. 2, Applications to linguistics, psychology, and neuroscience. Frankfurt: Ontos Verlag.

    Save Citation »Export Citation »E-mail Citation »

    Volume 2 of this two-volume set reports the papers from the second portion of a two-part conference, which was held in Paris. These conferences and the resulting papers are, to some considerable degree, responsible for the modern interest in compositionality.

    Find this resource:

  • Werning, Markus, Wolfram Hinzen, and Edouard Machery, eds. 2011. The Oxford handbook of compositionality. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    This is the result of a very intensive search for accounts of the role of compositionality in very many areas of philosophy and formal semantics. The authors of the articles are well-known in the area, and most of them have contributed to the other volumes mentioned in this section.

    Find this resource:

Arguments For

Arguments in favor of compositionality take one of two forms: (a) it is the only coherent, or anyway the best, way to organize a linguistic theory; and (b) it is the only way to account for the way language can be employed. The former style of argument is often seen as “a matter of aesthetic taste in theory-construction,” but the latter style is usually taken to be quite telling. This latter style turns on a feeling of “unboundedness” in aspects of people’s use of language: that people can understand an unbounded number of distinct sentences, that people can employ novel sentences to express thoughts that are new to them, that children learn the meanings of an unbounded number of sentences, and that people can produce sentences that they have never heard. These four arguments have respectively been called “the argument from understandability,” “the argument from creativity,” “the argument from learnability,” and “the argument from productivity.” The conclusion of any of these arguments is that semantic theory must be compositional, since that is the only (or the only known) way that people could have such abilities. Almost every work on compositionality mentions one or another of these arguments, so we will restrict our attention to those that discuss the argumentation in more detail.

In Favor of Unboundedness Arguments

This style of argument is usually traced to Frege 1963. Davidson 1965 and Davidson 1967 brought these considerations to the attention of modern philosophers. Grandy 1990 added an important consideration to these arguments: that learning an infinity of things was not essential since, if the amount learned was finite but sufficiently large, the arguments retain their force.

  • Davidson, Donald. 1965. Theories of meaning and learnable languages. In Proceedings of the 1964 International Congress for Logic, Methodology, and Philosophy of Science. Edited by Yehoshua Bar-Hillel, 383–394. Amsterdam: North-Holland.

    Save Citation »Export Citation »E-mail Citation »

    Here Davidson emphasizes the idea that any adequate theory of language must show how it is possible for people to have learned language, and he emphasizes the role that compositionality would play.

    Find this resource:

  • Davidson, Donald. 1967. Truth and meaning. Synthèse 17:304–323.

    DOI: 10.1007/BF00485035Save Citation »Export Citation »E-mail Citation »

    Although the argument from creativity can be found in Gottlob Frege, the version that emphasizes learnability and understandability was brought to philosophers’ attention by this very influential article by Davidson as well as Davison 1965.

    Find this resource:

  • Frege, Gottlob. 1963. Compound thoughts. Mind 72:1–17.

    DOI: 10.1093/mind/LXXII.285.1Save Citation »Export Citation »E-mail Citation »

    The opening two paragraphs of this work are usually taken to be the first clear statement of (what is now called) the arguments from productivity and understandability. Originally published in 1923 as “Logische Untersuchungen: Dritter Teil; Gedankengefüge,” Beiträge zur Philosophie des Deutschen Idealismus 3:36–51.

    Find this resource:

  • Grandy, Richard E. 1990. Understanding and the principle of compositionality. Philosophical Perspectives 4:557–572.

    DOI: 10.2307/2214205Save Citation »Export Citation »E-mail Citation »

    An evaluation of the force of the argument from understanding. Notable in the article is the now commonly cited fact that one need not employ the notion of infinity in these arguments—even a very large finite number will make the point quite well.

    Find this resource:

Opposed to Unboundedness Arguments

Although almost every writer pays homage to the unboundedness arguments, it has frequently been noted that they are “inferences to the best explanation.” That is, the unboundedness is acknowledged, and then it is pointed out that the argument only says that compositionality is one way this could be accomplished. This means that there might be other explanations that could be better. Some other writers have complained that compositionality by itself does not really give enough restrictions to account for the unboundedness. Pelletier 2004 claims that the author’s notion of semantic groundedness could also explain the unboundedness phenomena. And Pelletier 2012 claims that there are many forms of semantic atomism that are not compositional that can explain them also—it is atomism, not compositionality, that carries the explanatory force. Werning 2005 points to noncompositional systems that are nonetheless productive in the sense required by the productivity argument. And Schiffer 1986 claims that considerations of belief statements show that compositionality is false, even though there is unboundedness in natural languages. Robbins 2002 argues more radically that the facts of conceptual combination tell against compositionality of language and of mind. Pagin and Westerståhl 2010 wishes to add restrictions on the sort of computability that is associated with the composition function. Szabó 2010 claims that the classic unboundness arguments do not carry over to the consideration of semantic content in context, and other types of argument are required.

  • Pagin, Peter, and Dag Westerståhl. 2010. Compositionality II: Arguments and problems. Philosophy Compass 5:265–282.

    DOI: 10.1111/j.1747-9991.2009.00229.xSave Citation »Export Citation »E-mail Citation »

    This second part of a two-part article (see Pagin and Westerståhl 2010, cited under General Overviews and Definitions) discusses the arguments from learnability, novelty, productivity, systematicity, synonymy, intersubjectivity, and communication. It also includes a discussion of the role of computability in defining an appropriate compositional function.

    Find this resource:

  • Patterson, Douglas. 2005. Learnability and compositionality. Mind and Language 20:326–352.

    DOI: 10.1111/j.0268-1064.2005.00288.xSave Citation »Export Citation »E-mail Citation »

    Argues that while considerations of learnability might require compositionality, they actively tell against Jerry A. Fodor’s notion of “reverse compositionality.”

    Find this resource:

  • Pelletier, Francis Jeffry. 2004. The principle of semantic compositionality. In Semantics: A reader. Edited by Steven Davis and Brendan S. Gillon, 133–156. New York: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Elaborates the arguments from learnability, understandability, and creativity. Points out that these are abductive arguments for “compositionality being the best conclusion” and claims that the learnability (etc.) facts can be explained as well by Pelletier’s theory of “semantic groundedness.” Originally printed in Topoi 13 (1994): 11–24.

    Find this resource:

  • Pelletier, Francis Jeffry. 2012. Holism and compositionality. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 149–174. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    In this article the arguments concerning learnability, productivity, understandability, and creativity are claimed to be telling against theories of meaning holism. But if meaning holism is contrasted with meaning atomism, then this forms an argument for atomism. And although compositionality is a form of meaning atomism, it is claimed to be just one of a variety of types.

    Find this resource:

  • Robbins, Philip. 2002. How to blunt the sword of compositionality. Noûs 36:313–334.

    DOI: 10.1111/1468-0068.00373Save Citation »Export Citation »E-mail Citation »

    Considers the argument that the facts about conceptual combination do not support compositionality. Since Robbins takes concepts to be the meanings of linguistic items, he concludes that natural languages are also not compositional. Also considered are “two-factored” theories of meaning with wide and narrow semantics. Robbins gives a discussion about compositionality for both aspects.

    Find this resource:

  • Schiffer, Stephen. 1986. Compositional semantics and language understanding. In Philosophical grounds of rationality: Intentions, categories, ends. Edited by Richard E. Warner and Richard Grandy, 175–208. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    This work contains an argument to the conclusion that no natural language has a compositional semantics. The argument proceeds by arguing that compositionality implies that belief statements are “relational” (between a believer and a proposition-mental representation, sentence, etc.) and then arguing that this latter is false. Hence natural languages do not have a compositional semantics. Thus the understandability requirement must have a different footing than compositionality.

    Find this resource:

  • Szabó, Zoltán. 2010. The determination of content. Philosophical Studies 148:253–272.

    DOI: 10.1007/s11098-008-9296-zSave Citation »Export Citation »E-mail Citation »

    The topic here is the semantic content of an expression in a context of utterance. It is claimed that the traditional productivity and systematicity arguments cannot establish that a complex expression is compositionally determined from its structure and parts in this setting, and a new argument is given. See also Systematicity.

    Find this resource:

  • Werning, Markus. 2005. Right and wrong reasons for compositionality. In The compositionality of meaning and content. Vol. 1, Foundational issues. Edited by Markus Werning, Edouard Machery, and Gerhard Schurz, 285–309. Frankfurt: Ontos Verlag.

    Save Citation »Export Citation »E-mail Citation »

    Evaluates the plausibility of the various “unboundedness of language” arguments as reasons to embrace semantic compositionality, particularly productivity, systematicity, and inferentiality (see Systematicity). Werning’s conclusion is negative: productivity can occur in noncompositional languages, systematicity is not relevant to compositionality but instead to semantic categories, and there is no proof that inferentiality can occur only in compositional languages.

    Find this resource:

Formal Properties

A number of works have analyzed the logical or mathematical features of some definition or other of compositionality. Some of these works prove claims about the class of languages to which a compositional semantics can be given. A recent trend in the formal analysis of compositionality started with the work of Wilfrid Hodges. That sort of material is the focus of the subsection Problem of Logic. A separate issue is the question of whether compositionality is “formally vacuous.”

Potential Formal Vacuity

Compositionality would be formally vacuous if it were the case that any language whatsoever can be given a compositional semantics. For then it would be only a matter of taste or methodological preference—and not an empirical claim—that a given language was compositional. Although the topic had been discussed in many articles, such as Partee 2004 (cited under General Overviews and Definitions), it attained more prominence with the publication of Zadrozny 1994. This article spawned responses such as Westerståhl 1998, Kazmi and Pelletier 1998, and Dever 1999.

  • Dever, Josh. 1999. Compositionality as methodology. Linguistics and Philosophy 22:311–326.

    DOI: 10.1023/A:1005410301126Save Citation »Export Citation »E-mail Citation »

    A detailed response to the claim of Zadrozny 1994 to construct a compositional semantics from any semantics at all. Dever shows that it always relies on what he calls “occult meanings” and that the “ordinary, manifest meanings” never play a role.

    Find this resource:

  • Kazmi, Ali, and Francis Jeffry Pelletier. 1998. Is compositionality formally vacuous? Linguistics and Philosophy 21:629–633.

    DOI: 10.1023/A:1005388721969Save Citation »Export Citation »E-mail Citation »

    A response to Zadrozny 1994, which complains that the technical apparatus being employed uses the string, in addition to the meaning, when doing the conversion to a compositional semantics. Also argues that we have not really been given a compositional semantics for the same meanings.

    Find this resource:

  • Westerståhl, Dag. 1998. On mathematical proofs of the vacuity of compositionality. Linguistics and Philosophy 21:635–643.

    DOI: 10.1023/A:1005401829598Save Citation »Export Citation »E-mail Citation »

    Discusses the various ways theorists have attempted to prove that compositionality is vacuous in the sense of there always being a compositional semantics for any grammar whatsoever. Westerståhl considers the attempts by Theo Janssen, Herman Hendriks, Johan van Benthem, and Wlodek Zadrozny in this regard and finds implausible presuppositions in each.

    Find this resource:

  • Zadrozny, Wlodek. 1994. From compositional to systematic semantics. Linguistics and Philosophy 17:329–342.

    DOI: 10.1007/BF00985572Save Citation »Export Citation »E-mail Citation »

    This paper famously claims to prove that any semantics can be turned into a compositional semantics that maintains the same meaning relations. It has garnered many replies.

    Find this resource:

Problem of Logic

The metalogical characterization of compositionality started with Montague 1974 and was discussed in the 1980s by various logicians. The publication of Pagin and Westerståhl 2001 (cited under Collections) marked a new beginning in the formal studies, including especially Hodges 2001 and Hendriks 2001. Follow-up papers to Hodges 2001 include Pagin 2003 and Westerståhl 2004 as well as Hodges 2011.

  • Hendriks, Herman. 2001. Compositionality and model-theoretic interpretation. Journal of Logic, Language, and Information 10:29–48.

    DOI: 10.1023/A:1026592526654Save Citation »Export Citation »E-mail Citation »

    Hendriks’s reformulation of Richard Montague’s theory (Montague 1974) used a many-sorted algebra rather than Montague’s single-sorted one. Hendriks argues against Montague’s requirement that syntax and semantics be “similar” algebras and urges the more liberal requirement that syntax be “interpretable” into semantics. Hendriks claims this is a clearer formulation of the roles played by intermediate languages, by “model-theoretic interpretation” of a grammar, and by “meaning postulates.”

    Find this resource:

  • Hodges, Wilfrid. 2001. Formal features of compositionality. Journal of Logic, Language, and Information 10:7–28.

    DOI: 10.1023/A:1026502210492Save Citation »Export Citation »E-mail Citation »

    Hodges initiated a new direction in studying compositionality, focusing attention on synonymy and its properties rather than meaning. Laying down various restrictions on synonymy (such as full abstraction, cofinality, or the “Husserl property”), the resulting theory of semantic properties can be analyzed. Hodges’s innovation has moved the discussions of compositionality in a new direction and to a higher level of clarity.

    Find this resource:

  • Hodges, Wilfrid. 2011. Formalising the relationship between meaning and syntax. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 245–261. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Hodges axiomatizes the notion of a constituent and proves that, under some weak conditions, a semantics defined on sentences always generates a compositional semantics on all expressions. The paper also contains a discussion of some of the historical antecedents to modern discussions, including those coming from the Arabic tradition.

    Find this resource:

  • Montague, Richard. 1974. Universal grammar. In Formal philosophy: Selected papers of Richard Montague. By Richard Montague, 222–246. New Haven, CT: Yale Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    The classic place where compositionality is described and joined to the formal semantics for natural language. Compositionality here is described as a meaning algebra being a homomorphic image of the syntactic algebra. Compositionality became one of the main themes in linguistic semantics through the discussion of Montague’s works in Partee 2004 (cited under Works That Discuss Many Linguistic Constructions). Originally printed in Theoria 36 (1970): 373–398.

    Find this resource:

  • Pagin, Peter. 2003. Communication and strong compositionality. Journal of Philosophical Logic 32:287–322.

    DOI: 10.1023/A:1024258529030Save Citation »Export Citation »E-mail Citation »

    Investigates various formal notions of compositionality and inverse compositionality within the framework laid out in Hodges 2001. Argues that they call for a notion of structured proposition to account for synonymy and for communicative success.

    Find this resource:

  • Westerståhl, Dag. 2004. On the compositional extension problem. Journal of Philosophical Logic 33:549–582.

    DOI: 10.1023/B:LOGI.0000046069.01088.0bSave Citation »Export Citation »E-mail Citation »

    Considers the problem of extending a compositional semantics that is partial (only covers a portion of the language) to a complete compositional semantics. Illustrates several ways this might occur in practice and shows that there is a “weak” condition (namely, that the set of meaningful terms is closed under subterms) from which it can be guaranteed that there is a compositional extension.

    Find this resource:

Nonstandard Interpretations and Alternatives

Over the years a variety of somewhat different understandings of what compositionality applies to and how it is to be understood has been put forward. For example, there is the deflationary understanding, strong compositionality, and the fuzzy theory of compositionality (see Deflationary Theory, Strong Compositionality, Fuzzy Theory). Additionally, there have been slight alterations and additions to compositionality, such as reverse compositionality, semantic groundedness, and systematicity (see Reverse Compositionality, Semantic Groundedness, Systematicity). Striking against all these has been the simpler syntax hypothesis (see Simpler Syntax Hypothesis).

Deflationary Theory

A deflationary theory of compositionality tries to show that compositionality “automatically follows” from other principles and hence is not an independent constraint on anything. This understanding of compositionality is associated with Horwich 1998. It was strongly attacked in Fodor and Lepore 2001 but was in turn defended in Horwich 2001. Robbins 2001 investigates what remains of the arguments in favor of compositionality if we adopt the deflationary understanding. Rauti 2009 asks whether the deflationary principle can be derived from considerations brought forth from related fields.

  • Fodor, Jerry A., and Ernie Lepore. 2001. Why compositionality won’t go away: Reflections on Horwich’s “deflationary” theory. Ratio 14:350–368.

    DOI: 10.1111/1467-9329.00169Save Citation »Export Citation »E-mail Citation »

    Fodor and Lepore argue, against Horwich’s deflationary theory, that Horwich confuses metaphysical theories about meaning with epistemological theories of what makes it true that a language user understands an expression. Compositionality is deflatable when one abstracts from its role in explaining productivity, systematicity, and the like, but that is the main rationale for compositionality, and so it should not be abstracted from.

    Find this resource:

  • Horwich, Paul. 1998. Meaning. Oxford: Clarendon.

    DOI: 10.1093/019823824X.001.0001Save Citation »Export Citation »E-mail Citation »

    Gathers a number of articles by Horwich on deflationary theories of meaning, particularly on his deflationary theory of compositionality of meaning. According to his theory, compositionality turns out to be “automatically satisfied” as part of a theory of understanding a complex expression, imposing no constraint on how the meaning properties of words are constituted. Complex-expression understanding is similar to a “translation manual.”

    Find this resource:

  • Horwich, Paul. 2001. Deflating compositionality. Ratio 14:369–385.

    DOI: 10.1111/1467-9329.00170Save Citation »Export Citation »E-mail Citation »

    This is a defense of Horwich’s deflationary theory of compositionality against the attack of Fodor and Lepore 2001. One crux of this defense is that a certain “uniformity assumption” is false, namely the assumption that if meanings of lexical items are determined by a given class of properties (e.g., their inferential roles), then so too are the meanings of complexes made up of these items.

    Find this resource:

  • Rauti, Antonio. 2009. Can we derive the principle of compositionality (if we deflate understanding)? Dialectica 63:157–174.

    DOI: 10.1111/j.1746-8361.2009.01185.xSave Citation »Export Citation »E-mail Citation »

    Rauti considers Horwich’s claim that a kind of compositionality principle “automatically follows” from a deflationary account of understanding a complex expression. But he claims that Horwich does not present a detailed derivation, and so Rauti provides one. But the account is circular, and Rauti argues that any other such derivation is also likely to be circular.

    Find this resource:

  • Robbins, Philip. 2001. What compositionality still can do. Philosophical Quarterly 51:328–336.

    DOI: 10.1111/1467-9213.00232Save Citation »Export Citation »E-mail Citation »

    Argues that both a “strong” and a “weak” version of deflationism about meaning are credible. In particular, Robbins argues that weak deflation (which is held to be more plausible) cannot be maintained in conformity with deflationist commitments to explaining meanings of complex expressions in compositional terms (by appeal to facts about the lexicon).

    Find this resource:

Reverse Compositionality

Reverse compositionality, as introduced by Jerry A. Fodor, is the view that, in a complex expression, each part “contributes the whole of its meaning” to the complex (Fodor 1998a, Fodor 1998b). As stated, it seems to imply that the meaning of a word is derivable from the meaning of any single complex containing it (and syntactic rule’s semantic effect). Many writers have thought this implausible (Robbins 2005, Johnson 2006, Patterson 2005), but the redescription of the point of reverse compositionality’s role in communication in Pagin 2003 is quite interesting.

  • Fodor, Jerry A. 1998a. There are no recognitional concepts; not even RED. In In critical condition: Polemical essays on cognitive science and the philosophy of mind. By Jerry A. Fodor, 34–48. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    Here Fodor introduces the notion of reverse compositionality, arguing that concepts cannot be prototypes, stereotypes, exemplars, and so forth. Also reverse compositionality shows that a concept cannot express more than what the concept contributes to a complex concept in which it occurs. So a complex concept cannot contain “emergent features” that are not already present in component concepts. Originally printed in Philosophical Issues 9 (1998): 1–14.

    Find this resource:

  • Fodor, Jerry A. 1998b. There are no recognitional concepts—not even RED, Part 2: The plot thickens. In In critical condition: Polemical essays on cognitive science and the philosophy of mind. By Jerry A. Fodor, 49–62. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    In this article Fodor argues that reverse compositionality prohibits recognitional concepts (e.g., RED). In general, one cannot understand a complex concept without understanding the simpler ones that comprise it. Finally, what is said here about concepts goes for the semantics of natural language. (Importantly, reverse compositionality also features in Fodor and Lepore 2001, cited under Deflationary Theory.)

    Find this resource:

  • Johnson, Kent. 2006. On the nature of reverse compositionality. Erkenntnis 64:37–60.

    DOI: 10.1007/s10670-005-0362-zSave Citation »Export Citation »E-mail Citation »

    Johnson thinks there is a fundamental flaw in reverse compositionality, but what that flaw is depends on whether the principle is viewed as a metaphysical or an epistemological thesis. Johnson also gives many “empirically plausible” counterexamples to the principle that deserve attention by anyone wishing to formulate some principle similar to reverse compositionality.

    Find this resource:

  • Pagin, Peter. 2003. Communication and strong compositionality. Journal of Philosophical Logic 32:287–322.

    DOI: 10.1023/A:1024258529030Save Citation »Export Citation »E-mail Citation »

    Compositionality must be a function that is effectively computable and also be cognitively symmetric in the sense that both a producer and an interpreter can employ it. This calls for an inverse notion of compositionality. Compositionality and inverse compositionality together are called “strong compositionality.” Time is spent on the question of whether English is inversely compositional.

    Find this resource:

  • Patterson, Douglas. 2005. Learnability and compositionality. Mind and Language 20:326–352.

    DOI: 10.1111/j.0268-1064.2005.00288.xSave Citation »Export Citation »E-mail Citation »

    Argues that the learnability argument for compositionality is actively opposed to reverse compositionality. Furthermore, some facts about ignorance of meaning in a natural language show that the natural languages cannot be reverse compositional. Also arguments in favor of complex meanings having a structure are criticized.

    Find this resource:

  • Robbins, Philip. 2005. The myth of reverse compositionality. Philosophical Studies 125:251–275.

    DOI: 10.1007/s11098-005-7785-xSave Citation »Export Citation »E-mail Citation »

    Considers arguments given (by Jerry A. Fodor) in favor of reverse compositionality and finds them wanting. Also gives two arguments against reverse compositionality.

    Find this resource:

Semantic Groundedness

Semantic groundedness is explained in Pelletier 2004 as a kind of semantic atomism. All terms in the language do have some specific value, and no meaning is “holistically” explained. But this is not, strictly speaking, the same as compositionality, because in the simple atomistic view not everything needs its meaning to be defined in terms of the meanings of its syntactic parts.

  • Pelletier, Francis Jeffry. 2004. The principle of semantic compositionality. In Semantics: A reader. Edited by Steven Davis and Brendan S. Gillon, 133–156. New York: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Semantic groundedness is defined as a sort of atomism (with a computability constraint) and is urged as a way to capture the desirable features of compositionality and yet violate the specific aspects of compositionality that are found to be false. Originally printed in Topoi 13 (1994): 11–24.

    Find this resource:

Simpler Syntax Hypothesis

The idea behind the simpler syntax hypothesis explained in Culicover and Jackendoff 2006 (and many other articles by the authors separately) is that “syntax should trump semantics.” If there is a choice between a very pleasing simple syntax that is not semantically compositional and a very complex and syntactically unmotivated syntax that is semantically compositional, one should always choose the simple syntax. This is a methodology that is opposed by most linguists who are concerned with maintaining compositionality: they will “syntactically complicate” an apparently simple and straightforward syntactic analysis of some construction so that the resulting syntax allows for a compositional semantic account.

  • Culicover, Peter W., and Ray Jackendoff. 2006. The simpler syntax hypothesis. Trends in Cognitive Science 10:413–418.

    DOI: 10.1016/j.tics.2006.07.007Save Citation »Export Citation »E-mail Citation »

    The position in this work poses a challenge to compositionalism in semantics. The idea is that “the simplest syntax” that will account for all the standard syntactic features of a natural language (English in the authors’ example) will be noncompositional in its semantics. So one needs to choose: adequate syntax or inflate the syntax merely to save compositionality?

    Find this resource:

Strong Compositionality

Strong compositionality is introduced as an explanation of the notion that is in common use in linguistics (although possibly not what is commonly understood in philosophy of language). It should be noted that this use is different from the “strong compositionality” of Pagin 2003 (cited under Communication). The notion in Larson and Segal 1995 is what Peter Pagin called “compositionality,” whereas Richard Larson and Gabriel Segal’s “compositionality” is a still weaker principle to the effect that meaning depends on total structure and meanings of simple expressions.

  • Larson, Richard, and Gabriel Segal. 1995. Knowledge of meaning: An introduction to semantic theory. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    The authors of this textbook introduce constraints on semantic theory called “strictly local” and “purely interpretive.” Together these are called “strong compositionality,” and this notion is widely appealed to by many linguistics-oriented theorists. This definition is in pages 78–82; see also pages 11–16 and 54–56.

    Find this resource:

Systematicity

Systematicity is the idea that one does not come to know one sentence at a time but rather learns a group of different sentences all at once. A standard example is that if you know that “John loves Mary” is a well-formed sentence, then you will also know that “Mary loves John” also is a well-formed sentence. This notion was introduced in Fodor 1987, but the main work that attracted attention to it was Fodor and Pylyshyn 1988, where it was employed in an argument that connectionist theories cannot have systematicity and hence were inadequate models of the mind. This conclusion was widely attacked by neural network theorists and their allies, such as in Chalmers 1993, Hadley 1994, and Aizawa 2003. Others have complained that systematicity is not in fact a feature of language or language learning, for instance, in Johnson 2004 and Pullum and Scholz 2007.

  • Aizawa, Kenneth. 2003. The systematicity arguments. Boston: Kluwer.

    DOI: 10.1007/978-1-4615-0275-3Save Citation »Export Citation »E-mail Citation »

    This book investigates the logic of the “argument from productivity” and the “argument from systematicity.” The starting point for this book is the systematicity objection to connectionism in Fodor and Pylyshyn 1988.

    Find this resource:

  • Chalmers, David J. 1993. Connectionism and compositionality: Why Fodor and Pylyshyn were wrong. Philosophical Psychology 6:305–319.

    DOI: 10.1080/09515089308573094Save Citation »Export Citation »E-mail Citation »

    Challenges Fodor and Pylyshyn 1988 on both theoretical and empirical grounds. Chalmers claims that Jerry A. Fodor and Zenon W. Pylyshyn have conflated localist and distributed representations. The empirical evidence consists of examples where distributed representations support direct structure-sensitive operations but not in a classicalist way.

    Find this resource:

  • Fodor, Jerry A. 1987. Psychosemantics: The problem of meaning in the philosophy of mind. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    Systematicity is introduced and explained on pages 147–150.

    Find this resource:

  • Fodor, Jerry A., and Zenon W. Pylyshyn. 1988. Connectionism and cognitive architecture: A critical analysis. Cognition 28:3–71.

    DOI: 10.1016/0010-0277(88)90031-5Save Citation »Export Citation »E-mail Citation »

    The famous “systematicity objection” to connectionist models of cognition is first brought out in this piece. The discussion sparked a vast literature throughout the 1990s concerning the role of representations within connectionism.

    Find this resource:

  • Hadley, Robert F. 1994. Systematicity in connectionist language learning. Mind and Language 9:247–272.

    DOI: 10.1111/j.1468-0017.1994.tb00225.xSave Citation »Export Citation »E-mail Citation »

    Takes Jerry A. Fodor and Zenon W. Pylyshyn’s systematicity as plausible. Hadley thinks that certain connectionist architectures appear to have answered certain understandings of the challenge but also thinks that there are ways of understanding systematicity that have not been answered by these connectionist models. (The same issue contains replies by various connectionists, and the next issue contains a response by Hadley.)

    Find this resource:

  • Johnson, Kent. 2004. On the systematicity of language and thought. Journal of Philosophy 101:111–139.

    Save Citation »Export Citation »E-mail Citation »

    Objects to systematicity on the grounds that there is no formulation of the notion that will correctly describe natural language. Numerous violations of systematicity in English are cited.

    Find this resource:

  • Pullum, Geoff, and Barbara Scholz. 2007. Systematicity and natural language syntax. Croatian Journal of Philosophy 7:375–402.

    Save Citation »Export Citation »E-mail Citation »

    Gives various definitions of “systematicity,” showing that each seems to miss capturing the intent of the proponents of systematicity. These various definitions are formally shown to capture certain classes of languages and methodologies, none of which are what the proponents wish. In the end the authors despair of finding any definition of the concept that accords with natural language.

    Find this resource:

Fuzzy Theory

Fuzzy theory takes the attitude that sentences are not merely true or false but rather can take any of the real-number values between zero (completely false) and one (completely true). This gives rise to a rather different notion of compositionality, as explained in Zadeh 1983.

  • Zadeh, L. A. 1983. A fuzzy-set-theoretic approach to the compositionality of meaning: Propositions, dispositions, and canonical forms. Journal of Semantics 2:253–272.

    Save Citation »Export Citation »E-mail Citation »

    Claims that compositionality as traditionally conceived does not have a very wide applicability to natural language. But it can be modified in a fuzzy-logic context to be that the meaning of a proposition is composed from the meaning of a collection of fuzzy relations that form an “explanatory database” that is associated with the proposition.

    Find this resource:

Arguments Against

A number of linguistic constructions have been thought to pose extreme difficulties for semantic compositionality. (Some more general considerations are raised under Meaning Holism, Context in General, and Communication). Authors can be divided into those who think the difficulties are unavoidable and hence that compositionality is false and those who work at finding a “compositional solution” to the construction. The case of adjectives (see Adjectives) is the most commonly cited: “red” in “red wine,” “red wagon,” “red hair,” “red watermelon,” and “red apple” means quite different things. This must be the effect of the noun it is modifying, yet that modified noun is not a part of the meaning of “red,” and so compositionality is false. The “synonymy objection” works like this. If expressions A and B are synonymous, then compositionality demands that any complex structures must mean the same if they are otherwise identical except that one contains A and the other contains B. Yet some have claimed this to be violated. Similarly, a number of cases involving anaphora, genitives, verb phrase operators, and relative clauses have been brought forward as counterexamples to compositionality. Since the direct quotation of a linguistic item changes its use from expressing meaning to naming itself, it has seemed to many that quotation must also violate compositionality. Yet many have worked on trying to give a compositional account of this phenomenon. Idioms are normally defined as violations of compositionality (i.e., where the meaning of the phrase is not determined by the literal meaning of the parts and their combination). Yet some theorists have been trying to account for this phenomenon within the framework of compositionality. Finally, cognitive grammarians, who trace linguistic meaning to preexisting mental operations and functions, have argued that this itself will show that compositionality is wrong.

Adjectives

Issues concerning adjectives are probably the most commonly cited (alleged) counterexample to semantic compositionality, except possibly for concerns over nonlinguistic context. The argument that adjectives violate compositionality often takes this form: “red” in “red wine,” “red wagon,” “red hair,” “red watermelon,” “red apple” means quite different things. This must be the effect of the noun it is modifying, yet that is not a part of the meaning of “red.”

Adjectives in Context

The argument that the meaning of adjectives (and common noun phrases containing adjectives) are not compositionally describable can be found in both Travis 1997 and Lahav 1989, although they give rather different accounts of the phenomenon. This challenge has been criticized in Siebel 2000, Szabó 2001, Reimer 2002, Predelli 2005, Rothschild and Segal 2009, and Kennedy and McNally 2010.

  • Kennedy, Christopher, and Louise McNally. 2010. Color, context, and compositionality. Synthèse 174:79–98.

    DOI: 10.1007/s11229-009-9685-7Save Citation »Export Citation »E-mail Citation »

    This paper provides a linguistic perspective on the issue of color adjectives, which the authors believe defuses the challenge of Travis 1997. The authors provide empirical arguments that color adjectives are in fact ambiguous between gradable and nongradable interpretations and that this simple ambiguity, together with independently motivated options concerning scalar dimension within the gradable reading, accounts for the Travis facts.

    Find this resource:

  • Lahav, Ran. 1989. Against compositionality: The case of adjectives. Philosophical Studies 57:261–279.

    DOI: 10.1007/BF00372697Save Citation »Export Citation »E-mail Citation »

    Lahav argues that the propositional content of an adjective varies with the noun it modifies. On this basis Lahav claims that adjectives violate compositionality and that there is no “easy fix” to compositionality that accommodates this type of variation and at the same time explains productivity. (Lahav also generalizes this phenomenon to verbs.) This article has spawned many responses.

    Find this resource:

  • Predelli, Stefano. 2005. Painted leaves, context, and semantic analysis. Linguistics and Philosophy 28:351–374.

    DOI: 10.1007/s10988-004-6136-2Save Citation »Export Citation »E-mail Citation »

    According to contextualism, utterances of nonelliptical, nonambiguous, and nonindexical sentences may be associated with contrasting truth conditions. In this essay Predelli grants the contextualist analysis of the sentences in question and the contextualist assessment of the truth conditions for the corresponding utterances. But he argues that the resulting situation is not incompatible with the traditional approach to semantics.

    Find this resource:

  • Reimer, Marga. 2002. Do adjectives conform to compositionality? Philosophical Perspectives 16:183–198.

    Save Citation »Export Citation »E-mail Citation »

    Reimer distinguishes “literal meaning” and “contextual meaning.” Since “adj + noun” means “<noun> which is <adj>,” these phrases are obviously compositional in their literal meaning. Reimer says that the terms are “contextual” in the same way that indexicals are: they mean the same across contexts, but like an indexical, adjectives designate different things in different contexts. (Adjectives are thus “semantically underdetermined.”)

    Find this resource:

  • Rothschild, Daniel, and Gabriel Segal. 2009. Indexical predicates. Mind and Language 24:467–493.

    DOI: 10.1111/j.1468-0017.2009.01371.xSave Citation »Export Citation »E-mail Citation »

    The authors discuss the challenge to truth-conditional semantics presented by apparent shifts in extension of predicates such as “red,” as outlined by Ran Lahav. They propose an explicit indexical semantics for “red” and argue that their account is preferable to the alternatives on conceptual and empirical grounds.

    Find this resource:

  • Siebel, Mark. 2000. Red watermelons and large elephants: A case against compositionality? Theoria 15:263–280.

    Save Citation »Export Citation »E-mail Citation »

    Explains the argument in Lahav 1989 using adjectives but then provides a vigorous explanation and defense of the point of view of Jerry A. Fodor (something that is not done very often in the literature) with, however, two provisos that make Siebel think that more needs to be added to Fodor’s theory.

    Find this resource:

  • Szabó, Zoltán. 2001. Adjectives in context. In Perspectives on semantics, pragmatics, and discourse: A Festschrift for Ferenc Kiefer. Edited by István Kenesei and Robert M. Harnish, 119–146. Amsterdam: John Benjamins.

    Save Citation »Export Citation »E-mail Citation »

    Discusses a variety of not-so-challenging challenges to compositionality before taking on Charles Travis’s examples of painted leaves. Szabó canvasses a number of possible solutions from a compositional viewpoint and shows their similarity to older views in ethics about the meaning of “good.” Szabó adopts an intensional language to develop this intuition and then postulates a “hidden variable” that gets values in context in the adjective.

    Find this resource:

  • Travis, Charles. 1997. Pragmatics. In A companion to the philosophy of language. Edited by Bob Hale and Crispin Wright, 87–106. Oxford: Blackwell.

    Save Citation »Export Citation »E-mail Citation »

    Travis argues that structurally isomorphic sentences containing color adjectives can shift truth-value from context to context depending on how they are used and in the absence of effects of vagueness or ambiguity/polysemy and concludes that a deterministic mapping from structures to truth conditions is impossible. There are a number of examples that have become standard in the literature.

    Find this resource:

Adjectives and Syncategorematicity

Since the problem for compositionality seems to be that adjectives require information about the particular (type of?) noun it is modifying in order to give it the required meaning, one response denies that adjectives have any “independent” meaning but rather asserts that they are functional terms that operate on nouns or common noun phrases. This idea goes back to Aristotle, who remarked that “healthy” meant different things when applied to food, lifestyle, doctors, and so forth but nonetheless had a “central meaning” that involved the noun being modified, and it has featured in many semantic treatments, especially in the Middle Ages. But in more modern (formalistic) times it can be found in Katz 1964 and then again in Parsons 1972 and Kamp 1975 (most modern references are to Kamp 1975). Although none of these authors were especially concerned with compositionality, this treatment can be applied to the issues raised by Travis 1997 and Lahav 1989 (cited under Adjectives in Context; see also Szabó 2001 and Reimer 2002, cited under Adjectives in Context, for an application of this idea to compositionality).

  • Kamp, Hans. 1975. Two theories of adjectives. In Formal semantics of natural language. Edited by Edward L. Keenan, 123–155. Cambridge, UK: Cambridge Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Kamp, like Katz 1964, considers adjectives to be noun modifiers in the sense that the adjective’s meaning is determined in part by the noun being modified, so that “large mouse” amounts to “large for a mouse.” But he also allows for further contextual factors to be relevant: “Smith is a remarkable violinist” might be true in one context but false in another.

    Find this resource:

  • Katz, Jerrold J. 1964. Semantic theory and the meaning of “good.” Journal of Philosophy 61:739–766.

    DOI: 10.2307/2023019Save Citation »Export Citation »E-mail Citation »

    Katz is working in the semantic theory developed in Katz and Fodor 1963, cited under Historical Antecedents. His goal in the paper is to show how “an adequate semantic theory” can give a rigorous (and compositional) account of the meaning of philosophically interesting words, such as “good.” His development of a semantic meaning for “good” in terms of a “syncategorematic account” is noteworthy.

    Find this resource:

  • Parsons, Terence. 1972. Some problems concerning the logic of grammatical modifiers. In Semantics of natural language. Edited by Donald Davidson and Gilbert Harman, 127–141. Dordrecht, The Netherlands: Reidel.

    Save Citation »Export Citation »E-mail Citation »

    This is a more careful description of the “syncategorematic account” of adjectives offered in Katz 1964. Here we are introduced to different categories of such adjectives, such as “subsective” (e.g., “red”), “alienating” (e.g., “fake”), and “privative” (such as “alleged”).

    Find this resource:

Ambiguity and Synonymy

If the meaning of a complex is entirely determined by the meanings of its parts and the way the parts are combined, then an ambiguous complex must either have ambiguous parts or be able to be put together in two different ways. But Pelletier 1994 challenges this. Likewise, if two complexes are put together in the same syntactic way and out of pair-wise synonymous parts, then they must mean the same. Pelletier 2000 challenges this also. Werning 2005 considers whether these are reasonable challenges.

  • Pelletier, Francis Jeffry. 1994. Semantic compositionality: The argument from synonymy. In Philosophy and the cognitive sciences. Edited by Roberto Casati, Barry Smith, and Graham White, 208–214. Vienna: Hölder-Pichler-Tempsky.

    Save Citation »Export Citation »E-mail Citation »

    This argument uses Mates-style violations, where synonymies are distinguished by means of a person not having the same intensional attitude toward each of the synonymies. It is argued that this constitutes a violation of compositionality.

    Find this resource:

  • Pelletier, Francis Jeffry. 2000. Semantic compositionality: Free algebras and the argument from ambiguity. In Formalizing the dynamics of information. Edited by Martina Faller, Stefan Kaufmann, and Marc Pauly, 207–218. Stanford, CA: Center for the Study of Language and Information.

    Save Citation »Export Citation »E-mail Citation »

    If the syntax of a language is freely generated, then there can be no ambiguities other than lexical or (surface-) structural ones. And from this it follows that the language has a compositional semantics. But it is argued (a) that it is question-begging to assume this and (b) that there are examples that violate this stricture on ambiguity.

    Find this resource:

  • Werning, Markus. 2005. Right and wrong reasons for compositionality. In The compositionality of meaning and content. Vol. 1, Foundational issues. Edited by Markus Werning, Edouard Machery, and Gerhard Schurz, 285–309. Frankfurt: Ontos Verlag.

    Save Citation »Export Citation »E-mail Citation »

    Discusses the constraints that unique readability and presence versus lack of synonymy have on whether compositionality is a “vacuous constraint” and other consequences of compositionality.

    Find this resource:

Anaphora, Genitives, and Verb Phrase Operators

These are some examples of linguistic constructions that have provoked questions about compositionality. Landman and Moerdijk 1983 is an early investigation into compositionality in Montague grammar focusing on the feature of Richard Montague’s analysis that requires an index on pronouns. Siegel 1987 looks at pairs of verb phrases (VPs) that are not related by domination and yet that one semantically wants one of the two VPs to have scope over the other. Partee 2004 investigates the very rich domain of genitive constructions. When one says “John’s coat,” for example, there are many different relations that might hold between John and the coat. This seems like a prima facie counterexample to compositionality, and Barbara H. Partee has investigated the phenomenon deeply. Pylkkänen 2008 looks at “adjuncts” to VPs and tries to give a compositional analysis of this sort of phenomenon.

  • Landman, Fred, and Ieke Moerdijk. 1983. Compositionality and the analysis of anaphora. Linguistics and Philosophy 6:89–114.

    DOI: 10.1007/BF00868091Save Citation »Export Citation »E-mail Citation »

    Discusses the effect of compositionality on the form of grammatical rules within a Montague grammar containing a strong version of the compositionality principle. The examples under discussion concern anaphora that requires indexed pronouns and a version that employs Tanya Reinhart’s non-c-command condition.

    Find this resource:

  • Partee, Barbara H. 2004. Genitives: A case study. In Compositionality in formal semantics: Selected papers. By Barbara H. Partee, 182–189. Malden, MA: Blackwell.

    DOI: 10.1002/9780470751305.ch8Save Citation »Export Citation »E-mail Citation »

    It is claimed that, although the problems with genitive constructions can be solved by positing a free variable and saying that the genitive always expresses one argument of the relation, further problems arise in that the resulting theory seems “inelegant,” according to Partee. Originally printed in Johan van Benthem and Alice ter Meulen, eds., Handbook of Logic and Language (Amsterdam: Elsevier, 1997), 464–470.

    Find this resource:

  • Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    This monograph offers a compositional theory of verbal argument structure in natural languages that focuses on how “noncore” arguments of the verb (arguments that are not introduced by verbal roots themselves) are introduced into argument structures.

    Find this resource:

  • Siegel, Muffy E. A. 1987. Compositionality, case, and the scope of auxiliaries. Linguistics and Philosophy 10:53–75.

    DOI: 10.1007/BF00603392Save Citation »Export Citation »E-mail Citation »

    Siegel discusses cases where a syntactic element of one verb phrase has to be interpreted as having scope over a separate verb phrase that it does not dominate. This is a prima facie failure of compositionality.

    Find this resource:

Cognitive Grammar

Cognitive grammar is a linguistic theory that takes as a starting point that psychological states and functions should serve as the basis for linguistic theory. Thus factors such as how people understand new or unusual situations are employed to explain how sentences that describe such situations should be linguistically expressed. Pretty much all varieties of cognitive grammar are opposed to compositionality, usually on the grounds that the semantics of complexes involves “emergent properties.” This seems to be a complaint that the “ontological compositionality” version is false. But it is not yet proved, at least not by this reason, that the “functional compositionality” version is false. Ronald W. Langacker is one of the original writers on the theory of cognitive grammar, and Langacker 1987 is widely cited in discussions of compositionality within cognitive grammar. Coulson and Fauconnier 1999 uses the types of materials that fall under the adjectives category (see Adjectives) but gives a distinctively cognitive grammar account, which denies that these can be given a compositional description.

  • Coulson, Seana, and Giles Fauconnier. 1999. Fake guns and stone lions: Conceptual blending and privative adjectives. In Cognition and function in language. Edited by Barbara Fox, Dan Jurafsky, and Laura A. Michaelis, 143–158. Stanford, CA: Center for the Study of Language and Information.

    Save Citation »Export Citation »E-mail Citation »

    This variety of cognitive grammar denies that there is compositionality, because the complexes that are formed by “conceptual blending” and by methods that construct conceptual content give rise to semantic properties not predictable from those of the parts. Hence there will be “more” in the complexes than in the parts. The present article focuses on privative adjectives, such as “fake” and “stone” (when applied to lions).

    Find this resource:

  • Langacker, Ronald W. 1987. Composition. In Foundations of cognitive grammar. Vol. 1, Theoretical prerequisites. By Ronald W. Langacker, 448–480. Stanford, CA: Stanford Univ. Press, 1987.

    Save Citation »Export Citation »E-mail Citation »

    Composition, as understood here, is the relation between “component structures” and the “composite structure” that derives from them. The composite structure results when two or more component structures blend or combine in some way. The given reason for denying compositionality is that the resulting composite structures can have “more meaning” than would be given by the “building block model” of combination—which is identified with compositionality.

    Find this resource:

Conditionals, Quantifiers, and Related Constructions

Quantifiers in natural language give rise to difficulties for compositional semantics in three directions. First, as discussed in Cooper 1983, in sentences with multiple quantifiers there are (semantically) different “scopes” that the quantifiers can take, and these different scopes generate different meanings. But there seems to be but one syntactic analysis of such sentences. Robin Cooper introduces a syntactic-semantic mechanism to deal with this problem. Second, not every language handles quantifiers in the same way as English. Bittner 1995 describes how Eskimo syntax works to prevent the standard methods of compositional semantic analysis in English for quantifiers. And third, the quantifier “any” poses special problems because of its “ambiguous” nature that is resolved by sentential position. Hintikka 1980 shows that one standard explanation of the meaning of “any”—as captured in the author’s game-theoretic semantics—forces noncompositionality on such a quantifier. Higginbotham 2003 is a revised version of an earlier paper that challenged compositional analyses of quantified conditional sentences on the grounds that they seem to require a semantic understanding of the quantifier before one could adequately understand the connective being employed. But this violates compositionality, because the connective is not a part of the quantifier phrase. This challenging paper has gathered responses, such as Pelletier 1994 and Leslie 2009, which offer differing types of solutions to the apparent violation of compositionality.

  • Bittner, Maria. 1995. Quantification in Eskimo: A challenge for compositional semantics. In Quantification in natural languages. Vol. 1. Edited by Emmon Bach, Eloise Jelinek, Angelika Kratzer, and Barbara H. Partee, 59–80. Dordrecht, The Netherlands: Kluwer.

    DOI: 10.1007/978-94-011-0321-3Save Citation »Export Citation »E-mail Citation »

    (West Greenlandic) Eskimo is a heavily polysynthetic language in which distinctions such as those in English between sentence adverbial quantification and nominal quantification exist, but the surface constituent structure is very different from its English equivalents. These facts about a natural language can be seen as challenging the hypothesis that the truth conditions of sentences can be derived systematically by applying compositional semantic rules to independently motivated structures.

    Find this resource:

  • Cooper, Robin. 1983. Quantification and syntactic theory. Dordrecht, The Netherlands: Reidel.

    Save Citation »Export Citation »E-mail Citation »

    Cooper introduced the technique known as “Cooper storage” to handle sentences that have no syntactic (or lexical) ambiguity that corresponds to a semantic ambiguity of quantifier scope, where there is no corresponding syntactic ambiguity. Cooper proposes that in the semantic evaluation one “store” the quantifiers, then interpret the matrix, and then “remove” the quantifiers from the store. This (quasi?) compositional maneuver was adopted widely.

    Find this resource:

  • Higginbotham, James. 2003. Conditionals and compositionality. Philosophical Perspectives 17:181–194.

    DOI: 10.1111/j.1520-8583.2003.00008.xSave Citation »Export Citation »E-mail Citation »

    This is a revision and update of a 1986 paper by Higginbotham that has been widely circulated underground but was very difficult to obtain in its published form (which appeared in the proceedings of a meeting of the Congress of Logic, Methodology, and Philosophy of Science). It famously describes cases where the interpretation of a lexical item (“unless” and “if”) depends on material that occurs in other clauses. This forms a counterexample to compositionality.

    Find this resource:

  • Hintikka, Jaakko. 1980. On the any-thesis and the methodology of linguistics. Linguistics and Philosophy 4:101–122.

    DOI: 10.1007/BF00351815Save Citation »Export Citation »E-mail Citation »

    Within the game-theoretic approach to semantics favored by Hintikka, the any-thesis says that “any” is unacceptable if exchanging it for “every” yields an equivalent sentence. Hintikka claims that this shows that the set of sentences of English are not recursively enumerable. The any-thesis also violates semantic compositionality, because one needs to check whether a putative meaning is or is not the same as that of a different sentence.

    Find this resource:

  • Leslie, Sarah-Jane. 2009. “If,” “unless,” and quantification. In Compositionality, context, and semantic values: Essays in honour of Ernie Lepore. Edited by Robert J. Stainton and Christopher Viger, 3–30. Berlin: Springer.

    DOI: 10.1007/978-1-4020-8310-5Save Citation »Export Citation »E-mail Citation »

    A response to Higginbotham 2003 arguing that the steps James Higginbotham takes in trying to salvage compositionality from his own examples are unsatisfactory as accounts of “if” and “unless” in natural language. An account that takes them as restricting quantifier domains is argued to give a better analysis.

    Find this resource:

  • Pelletier, F. J. 1994. On an argument against semantic compositionality. In Logic and philosophy of science in Uppsala. Edited by Dag Prawiz and Dag Westerståhl, 599–610. Dordrecht, The Netherlands: Kluwer.

    Save Citation »Export Citation »E-mail Citation »

    This is a response to James Higginbotham’s examples of the noncompositionality of “unless.” It argues that there are two ways to avoid all examples of the Higginbotham sort: a syntactic strategy and a semantic strategy.

    Find this resource:

Direct Quotation

Direct quotation is normally seen as a matter of changing the semantic value of an occurrence of an expression to become the expression itself. This change from using the expression to mentioning it seems at first glance to be an instance of noncompositionality, since the semantic value of the expression appears to play absolutely no role in computing the semantic value of the resulting quoted expression. Furthermore, it seems to require having an infinite set of semantic primitives (values of everything that can be quoted). Yet some authors have challenged that attitude. Davidson 1979 put the problem on the semantic agenda, arguing for a “paratactic” theory of quotation. Parsons 1982 is concerned with whether a Fregean-style theory could work as an alternative explanation. Washington 1992 offers a critique of these views and posits instead that quotation is punctuation. Various authors have argued against the Fregean account on the grounds that it is not “semantically innocent,” while others have tried to merge the Fregean and Davidsonian accounts (e.g., Pietroski 1999, Reimer 1996). Botterell and Stainton 2005 argues against the Davidson account, while Lepore 1999 defends it. Potts 2007 offers a somewhat different account of quotation and compositionality.

  • Botterell, Andrew, and Robert J. Stainton. 2005. Quotation: Compositionality and innocence without demonstration. Crítica 37.110: 3–33.

    Save Citation »Export Citation »E-mail Citation »

    The paper discusses both direct and indirect quotation. With respect to direct quotation, the authors ask whether a semantic theory can be both compositional and innocent and argue against a Davidsonian paratactic account and in favor of pure quotation as a kind of identity function.

    Find this resource:

  • Davidson, Donald. 1979. Quotation. Theory and Decision 11:27–40.

    DOI: 10.1007/BF00126690Save Citation »Export Citation »E-mail Citation »

    Argues against “naming theories” and “description theories” of quotation. Significant for claiming that a satisfactory account of the conditions under which an arbitrary sentence containing a quotation is true must be given. Since there is an infinite number of quotation mark names, it follows that there has to be some recursively specifiable method of creating the names or else the language would be unlearnable.

    Find this resource:

  • Lepore, Ernest. 1999. The scope and limits of quotation. In The philosophy of Donald Davidson. Edited by Lewis Edwin Hahn, 691–714. Chicago: Open Court.

    Save Citation »Export Citation »E-mail Citation »

    Argues against a “disquotational semantics” account of quotation on the grounds that the class of quotable items is not recursively definable and hence a semantic theory would be impossible. Lepore claims that whatever is tokened between quotation marks in a sentence is in no sense a constituent of the sentence.

    Find this resource:

  • Parsons, Terence. 1982. What do quotation marks name? Frege’s theories of quotations and that-clauses. Philosophical Studies 42:315–328.

    DOI: 10.1007/BF00714364Save Citation »Export Citation »E-mail Citation »

    Parsons is concerned with giving a true-to-Frege account of quotation, despite Gottlob Frege’s very meager assertions about quotation. Quotation marks are functional names. Then the appropriate Fregean question is, “What are the sense and reference of such a function-name?”

    Find this resource:

  • Pietroski, Paul M. 1999. Compositional quotation (without parataxis). In Philosophy and linguistics. Edited by Kumiko Murasugi and Robert Stainton, 245–258. Boulder, CO: Westview.

    Save Citation »Export Citation »E-mail Citation »

    Pietroski works to give a “broadly Davidsonian” account of how direct quotation works using interpreted logical forms. An interesting part of the account is an attempt to separate the principle of compositionality from the issue of substitutivity. In doing this, interpreted logical forms are treated as ordered pairs of a linguistic form with its (Fregean-like) sense.

    Find this resource:

  • Potts, Christopher. 2007. The dimensions of quotation. In Direct compositionality. Edited by Chris Barker and Pauline Jacobson, 405–431. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Potts analyzes expressions as triples and allows the quotation function to apply to these triples. This blends nicely with a “recursive” semantics, as Potts shows. But it is not strictly compositional.

    Find this resource:

  • Reimer, Marga. 1996. Quotation marks: Demonstratives or demonstrations? Analysis 56:131–141.

    DOI: 10.1111/j.0003-2638.1996.00131.xSave Citation »Export Citation »E-mail Citation »

    Argues that the best theory of quotation will combine the Fregean identity theory with the Davidsonian paratactic-demonstrative theory. Gottlob Frege is seen as correct in viewing the referring expression as the quotationally embedded expression; Donald Davidson is seen as correct in thinking that the referring expression was a demonstrative expression (but Davidson wrongly identified the referring expression with the quotation marks).

    Find this resource:

  • Washington, Corey. 1992. The identity theory of quotation. Journal of Philosophy 89:582–605.

    DOI: 10.2307/2941057Save Citation »Export Citation »E-mail Citation »

    The (Fregean) identity theory of quotation alleges that the correlation between expressions used and the expressions thereby mentioned is rule governed. Washington contrasts this with views of Willard Van Orman Quine, Peter Geach, and Donald Davidson, who take quotation as a kind of either naming, describing, or demonstrating. An interesting part of the view is that quotation marks are seen as neither mentioning expressions nor parts of mentioning expressions but rather as punctuation.

    Find this resource:

Idioms

Idioms are often defined as phrases whose meaning cannot be predicted from the words constituting them and the way those words are combined. This means that almost by definition they are noncompositional. Nonetheless, there are works that challenge this attitude toward idioms. Nunberg, et al. 1994 claims that once we distinguish different varieties of idioms, we will find that each of the differing types allows for a compositional treatment. Westerståhl 2002 argues that idioms can always be embedded in compositional languages and proposes three ways of doing so.

  • Nunberg, Geoffrey, Ivan A. Sag, and Thomas Wasow. 1994. Idioms. Language 70:491–538.

    DOI: 10.2307/416483Save Citation »Export Citation »E-mail Citation »

    Argues that the usual feeling that idiomaticity entails noncompositionality is due to a failure to distinguish several dimensions of idiomaticity, such as conventionality and figuration. The authors distinguish “idiomatically combining expressions” (which have conventional meanings in their parts) from “idiomatic phrases” (which do not distribute their meanings to their components.

    Find this resource:

  • Westerståhl, Dag. 2002. On the compositionality of idioms: An abstract approach. In Words, proofs, and diagrams. Edited by Dave Barker-Plummer, David Beaver, Johan van Benthem, and Patrick Scotto di Luzio, 241–272. Stanford, CA: Center for the Study of Language and Information.

    Save Citation »Export Citation »E-mail Citation »

    Possible compositional methods of treating idioms: (1) extend the atomic expressions by a holophrastic reading of the idiom, (2) extend the syntax so that literal and idiomatic readings of an idiom become outcomes of different syntactic operations, and (3) make syntactic parts of the idiom be homonyms of their occurrences in its literal reading and add them to the set of atomic expressions.

    Find this resource:

Relative Clauses

A question of the 1970s was whether relative clauses should be analyzed as noun + sentence (N + S) or as noun phrase + sentence (NP + S). That is, should the analysis be [npthe [nman [relwho is in the corner]]]? Or should it be [npthe [nman]][relwho is in the corner]? Considerations of compositionality led Barbara H. Partee (Partee 1975) to prefer the N + S solution. Stechow 1980 worries that it is quite difficult to make the N + S solution work correctly but thinks that there are other advantages to it. Bach and Cooper 1978 is concerned that the N + S solution seems to work only for a certain type of language.

  • Bach, Emmon, and Robin Cooper. 1978. The NP-S analysis of relative clauses and compositional semantics. Linguistics and Philosophy 2:145–150.

    DOI: 10.1007/BF00365132Save Citation »Export Citation »E-mail Citation »

    Bach and Cooper are following up on the suggestions in Partee 1975. But they find that other languages, such as Hittite, which they investigate here, do not have the same sort of structures as English. This seems to throw into question the entire procedure of allowing considerations of compositionality to guide syntactic analyses. But a generalized quantifier analysis can allow for the NP + S analysis to be compositional.

    Find this resource:

  • Partee, Barbara H. 1975. Montague grammar and transformational grammar. Linguistic Inquiry 6:203–300.

    Save Citation »Export Citation »E-mail Citation »

    Partee was the first to use the consideration of semantic compositionality as a constraint on syntactic analyses; the question of N + S versus NP + S was a test case mentioned in this article.

    Find this resource:

  • Stechow, Arnim von. 1980. Modification of noun phrases: A challenge for compositional semantics. Theoretical Linguistics 7:57–110.

    DOI: 10.1515/thli.1980.7.1-3.57Save Citation »Export Citation »E-mail Citation »

    This early work uses the tool of compositionality to favor the noun-sentence analysis of relative clauses as opposed to the noun phrase–sentence analysis. But much work is needed for it to come out right. However, some side benefits are supposed to be that the restrictive versus nonrestrictive relative clause distinction comes out as a semantic distinction, with no syntactic correlate, and that there is a unified semantics for “and.”

    Find this resource:

Works That Discuss Many Linguistic Constructions

The works in this section discuss a number of the linguistic constructions that have been implicated in noncompositionality (including those mentioned individually in Arguments Against). Partee 2004 discusses generics, adjectives, adjuncts, compounds, and genitives; Partee 1995 discusses also the topics of context dependence and point of view. Pelletier 2004 contains a discussion of generics, superlatives, quantifier scope, ambiguity, and other constructions. Pagin and Westerståhl 2010 discusses the more general issues of ambiguity, synonymy, idioms, and quotation.

  • Dowty, David. 2007. Compositionality as an empirical problem. In Direct compositionality. Edited by Chris Barker and Pauline Jacobson, 23–101. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    In section 2 of this article Dowty confronts a very wide range of “trouble cases” for a compositional analysis arising in quite different areas of the syntax. This is the most thorough discussion of the topic of compositionality within syntax, and Dowty canvasses a range of solutions in quite a few variants of Richard Montague and categorial grammars.

    Find this resource:

  • Pagin, Peter, and Dag Westerståhl. 2010. Compositionality II: Arguments and problems. Philosophy Compass 5:265–282.

    DOI: 10.1111/j.1747-9991.2009.00229.xSave Citation »Export Citation »E-mail Citation »

    The final section of this second part of the authors’ two-part article discusses the issues of ambiguity, synonymy, idioms, and quotation.

    Find this resource:

  • Partee, Barbara H. 1995. Lexical semantics and compositionality. In An invitation to cognitive science. 2d ed. Vol. 1, Language. Edited by Lila R. Gleitman and Mark Liberman, 311–360. Cambridge, MA: MIT Press.

    Save Citation »Export Citation »E-mail Citation »

    Discusses “ambiguous” adjective forms: a “compound pattern” versus a “modifier pattern” as well as adjectives that incorporate either a restrictive or a nonrestrictive relative-clause meaning. Also discusses the case of adjectives, vagueness, context dependence, point of view, compounds versus modifiers, and more. Emphasizes that the correct analysis of any linguistic (semantic) phenomenon requires evaluating hypotheses about lexical meanings, syntactic structure, and modes of semantic composition.

    Find this resource:

  • Partee, Barbara H. 2004. Compositionality. In Compositionality in formal semantics: Selected papers. By Barbara H. Partee, 153–181. Malden, MA: Blackwell.

    DOI: 10.1002/9780470751305.ch7Save Citation »Export Citation »E-mail Citation »

    In this broad-ranging article Partee considers challenges to compositionality from a variety of linguistic constructions: generics, dependent plurals, “any,” adjectives, “occasional,” free adjuncts, compounds, and some genitives. Originally published in Fred Landman and Frank Veltman, eds., Varieties of Formal Semantics: Proceedings of the Fourth Amsterdam Colloquium, September 1982 (Dordrecht, The Netherlands: Foris, 1984), 281–312.

    Find this resource:

  • Pelletier, Francis Jeffry. 2004. The principle of semantic compositionality. In Semantics: A reader. Edited by Steven Davis and Brendan S. Gillon, 133–156. New York: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    In addition to its discussion of synonymy and ambiguity as problematic for compositionality, it also discusses a number of claims in the literature that were supposed to be counterexamples to compositionality, such as generics, “displaced” superlatives, adjectives, quantifier scope, nonrestrictive relative clauses, and others. Originally printed in Topoi 13 (1994): 11–24.

    Find this resource:

Noun-Noun Compounds

Noun-noun (N-N) compounds, such as “child prodigy,” “child murderer,” and “axe murderer,” present a challenge to compositionality in that, while it seems that there is but one syntactic rule used to form any of these complexes, there is more than one semantic rule that is being employed. It should be noted that this is quite similar to the challenge that adjectives pose to compositionality (see Adjectives). It is also quite similar to some of the work discussed in the section Anaphora, Genitives, and Verb Phrase Operators. Levi 1978 and Ryder 1994 provide linguistic resources on the topic, each from a markedly different point of view. Weiskopf 2007 argues for treating such compounding as “context sensitive,” while Sainsbury 2001 argues for a “context invariant” meaning for them. Wisniewski and Wu 2011, on the other hand, presents evidence that such combinations are in fact failures of compositionality. Downing 1977 argues that there is no finite list of interpretations of the semantic force of N-N compounding, thereby implicitly arguing that they form an inescapable counterexample to semantic compositionality.

  • Downing, Pamela. 1977. On the creation and use of English compound nouns. Language 53:810–842.

    DOI: 10.2307/412913Save Citation »Export Citation »E-mail Citation »

    Downing uses a number of experimental tasks in which subjects are asked to create or interpret novel N-N compounds. Her main conclusion is that there is no finite number of interpretations for the ways nouns can be compounded. Along the way there are a number of important observations made about ways to distinguish compounds from modifications.

    Find this resource:

  • Levi, Judith N. 1978. The syntax and semantics of complex nominals. New York: Academic.

    Save Citation »Export Citation »E-mail Citation »

    Levi writes in a generative-semantics framework; nonetheless, the solution(s) canvassed are strikingly similar to the ones that are currently proposed in other frameworks. The basic position Levi promotes is that N-N structures are ambiguous and their ambiguity is accounted for by their being derived from twelve underlying logical structures, each of which contains a different specific relationship between head and modifier.

    Find this resource:

  • Ryder, Mary Ellen. 1994. Ordered chaos: The interpretation of English noun-noun compounds. Berkeley: Univ. of California Press.

    Save Citation »Export Citation »E-mail Citation »

    Ryder writes in a cognitive grammar framework but has an extensive summary of previous treatments. She separates speaker from hearer strategies: a speaker chooses a head noun and then a modifying noun so as to highlight a particular feature. The hearer attends to the form chosen by the speaker. The result is said to be a number of “schemata.” Reported experiments claim to substantiate this model.

    Find this resource:

  • Sainsbury, R. M. 2001. Two ways to smoke a cigarette. Ratio 14:386–406.

    DOI: 10.1111/1467-9329.00171Save Citation »Export Citation »E-mail Citation »

    Sainsbury gives a “unified compositional account” of a number of different types of counterexamples to compositionality, of which N-N compounding is one. The general thrust of the unified account is that expressions will have a context-invariant “unspecific meaning,” which does not supply any information as to how the compounds are related. This unspecific meaning can be instantiated in many different ways.

    Find this resource:

  • Weiskopf, Daniel A. 2007. Compound nominals, context, and compositionality. Synthèse 156:161–204.

    DOI: 10.1007/s11229-005-3489-1Save Citation »Export Citation »E-mail Citation »

    Since noun-noun combination is quite productive, it presents a serious challenge to the learnability-understandability arguments for compositionality. Weiskopf develops a theory that treats such compounds as fully compositional but context-sensitive in that there is a “hidden” indexical element. In this Weiskopf is (consciously) following the treatment of adjectives in Szabó 2001, cited under Adjectives in Context.

    Find this resource:

  • Wisniewski, Edward, and Jing Wu. 2011. Emergency!!!! Challenges to a compositional understanding of noun-noun combinations. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 403–417. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Presents evidence that people interpret novel noun-noun combinations in ways that are not a straightforward function of the meanings of their constituents but rather arise from an interaction of the meanings of their constituents, resulting in an “emergent feature” that is not represented in either constituent. One might say that this notion of (failure of) compositionality is the “ontological sense” mentioned in the Introduction.

    Find this resource:

Conceptual Combination

The linguistic phenomena characterized as noun-noun compounds (see Noun-Noun Compounds) are matched by a more psychological way of describing them that goes under the name “conceptual combination”—the concept “pet fish” seems not to be compositionally generated from the component concepts “pet” and “fish.” The psychology literature on this latter matter is massive and will not be delineated here; only a few articles of this sort will be mentioned. Ran and Duimering 2010 evaluates ten of the major efforts by psychologists to understand how conceptual combination works. Although this is done from the “cognitive grammar” point of view and the authors consider it a strike against a study if it does not fit well into cognitive grammar, it is still a nice resource for discovery of the breadth of the field. Gagné and Shoben 1997 discusses one of the major models for this, and in Gagné 2002 the model gets a lot more empirical backing together with a model for its employment. A different model is advocated in Hampton 1991. Wisniewski 1996 argues against all of these “relational” models on the grounds that there “is more to combination” than just finding a relation that links constituents.

  • Gagné, Christina L. 2002. Lexical and relational influences on the processing of novel compounds. Brain and Language 81:723–735.

    DOI: 10.1006/brln.2001.2559Save Citation »Export Citation »E-mail Citation »

    The general picture advocated by the CARIN model was employed. Here the goal was to examine the role of lexical and relation information by preceding the combinations with some different types of semantic priming materials. Even though the modifier and head noun are both retrieved independently before they are combined, only information associated with the modifier affects the availability of a particular relation.

    Find this resource:

  • Gagné, Christina L., and Edward J. Shoben. 1997. Influence of thematic relations on the comprehension of modifier-noun combinations. Journal of Experimental Psychology: Learning, Memory, and Cognition 23:71–87.

    DOI: 10.1037/0278-7393.23.1.71Save Citation »Export Citation »E-mail Citation »

    The CARIN model places emphasis on selection of a combination relation during conceptual combination. This requires knowledge about the relations used by the modifier during conceptual combination and the ease of interpreting a combination. A number of works by Gagné and her collaborators have investigated and classified the sorts of relations that can be employed.

    Find this resource:

  • Hampton, James. 1991. The combination of prototype concepts. In The psychology of word meanings. Edited by Paula J. Schwanenflugel, 91–116. Hillsdale, NJ: Lawrence Erlbaum Associates.

    Save Citation »Export Citation »E-mail Citation »

    Hampton proposes the “composite prototype model,” which assumes that concepts are represented schematically as sets of attributes connected by theory-driven relations. Attributes are assumed to have a “definingness degree” called importance. The proposal is that a conjunctive concept is represented semantically by a composite prototype that is formed as the union of the sets of attributes from both parent concepts.

    Find this resource:

  • Ran, Bing, and P. Robert Duimering. 2010. Conceptual combination: Models, theories, and controversies. International Journal of Cognitive Linguistics 1:65–90.

    Save Citation »Export Citation »E-mail Citation »

    This overview article provides a critical review of ten major models and evaluates them along four dimensions: the causal role of (cognitive grammar) schemata in the model, the role of “cognitive harmony and consistency” in the model, the “pragmatic orientation” in the model, and the explanatory scope of the model.

    Find this resource:

  • Wisniewski, Edward J. 1996. Construal and similarity in conceptual combination. Journal of Memory and Language 35:434–453.

    DOI: 10.1006/jmla.1996.0024Save Citation »Export Citation »E-mail Citation »

    Wisniewski argues against interpreting novel phrases by linking one constituent to another via a relation. Experiments show that there are some relation linkings going on but also that there are also other ways novel combinations are interpreted, and Wisniewski proposes a two-part model of “comparison process.”

    Find this resource:

Prototypes

The notion of conceptual combination mentioned in the section Conceptual Combination presumes some notion of a concept as the meaning of a linguistic item (or as correlated with the meaning), and that concept is quite often seen to be a prototype. (Separating works on conceptual combination from works on prototypes is rather arbitrary.) It might be mentioned that one difference between the linguistic and conceptual areas seems to be that the conceptual realm employs the “metaphysical compositionality” sense of the term while the linguistic realm employs the “functional compositionality” sense. The seminal work in conceptual combination and compositionality is Osherson and Smith 1981. Crucially, this work involves the mental construct of a prototype. Later work has discussed this along two dimensions: one involves refinements or changes to the notion of a prototype, while the other argues that prototypes are not concepts—interpreting concepts as the meaning of linguistic items. If prototypes do not combine compositionally, it is argued, it is not a difficulty for semantic compositionality if prototypes are not meanings. Along the first branch of this dichotomy are Braisby 1998, arguing that prototypes need to be understood “relationally”; Jylkkä 2011, distinguishing “extensional” from “intensional” understandings of prototypes; and Prinz 2011, arguing in favor of a “possibilist” understanding of prototypes. One detail in implementing the employment of prototypes involves describing a way to tell whether or not something in fact manifests the prototype. Kamp and Partee 1995 proposes replacing the traditional fuzzy logic with supervaluation theory as a measure of the degree to which something manifests the property that a prototype names. This involves a discussion of the notion of vagueness in the degree of manifestation. Osherson and Smith 1997 replies to this aspect of the authors’ definition of prototypes. Along the other dimension of discussion, where prototypes are argued not to be concepts (that is, argued not to be meanings of linguistic items), are Fodor and Lepore 1996 and Gleitman, et al. 2011.

  • Braisby, Nick. 1998. Compositionality and the modelling of complex concepts. Minds and Machines 8:479–508.

    DOI: 10.1023/A:1008310427321Save Citation »Export Citation »E-mail Citation »

    Argues that a “relational view of concepts” can accommodate a range of complex concepts, including the cases that have been argued to be noncompositional. The idea is that as the content of a concept varies systematically with perspective, compositionality should thus be considered to be sensitive to perspective.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 1996. The red herring and the pet fish: Why concepts still can’t be prototypes. Cognition 58:253–270.

    DOI: 10.1016/0010-0277(95)00694-XSave Citation »Export Citation »E-mail Citation »

    Contrary to Kamp and Partee 1995 (and the view in Osherson and Smith 1981), these authors argue that the “standard argument” is correct: because prototypes are not compositional, they cannot be concepts, which are the meaning of natural language items.

    Find this resource:

  • Gleitman, Lila, Andrew Connolly, and Sharon L. Armstrong. 2011. Can prototype representations support composition and decomposition? In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 418–436. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    The authors try to undermine prototype theories by distinguishing “having a prototype” from “being a prototype” (people think robins are more prototypical birds than penguins, but they also think they are both birds). Another attack on prototypes concerns lexical combinations: people do not think that a prototypical green apple is a prototypical apple that is prototypically green.

    Find this resource:

  • Jylkkä, Jussi. 2011. Hybrid extensional prototype compositionality. Minds and Machines 21:41–56.

    DOI: 10.1007/s11023-010-9217-8Save Citation »Export Citation »E-mail Citation »

    Responds to the standard argument that prototypes cannot compose by distinguishing an intensional versus extensional approach to prototype compositionality. Both are argued to be faulty, but a hybrid extensional theory is advocated according to which the extension of a complex concept is a function of what triggers its constituent prototypes.

    Find this resource:

  • Kamp, Hans, and Barbara H. Partee. 1995. Prototype theory and compositionality. Cognition 57:129–191.

    DOI: 10.1016/0010-0277(94)00659-9Save Citation »Export Citation »E-mail Citation »

    A very detailed attempt to replace fuzzy logic with supervaluation theory as an account of the way prototypes can combine and how they can be used to provide truth conditions for statements. This paper distinguishes two kinds of gradedness, typicality and membership (in a concept), and argues that many of the objections to prototypes have to do with conflation of these two types of “fuzziness.”

    Find this resource:

  • Osherson, Daniel N., and Edward E. Smith. 1981. On the adequacy of prototype theory as a theory of concepts. Cognition 9:35–58.

    DOI: 10.1016/0010-0277(81)90013-5Save Citation »Export Citation »E-mail Citation »

    This is the classic paper in which the issue of compositionality in concept formation and prototypes is discussed, using “pet fish” as a leading example. The “standard objection” is brought forward: since prototypes are not compositional and since meaning is compositional, prototypes cannot be meanings.

    Find this resource:

  • Osherson, Daniel, and Edward E. Smith. 1997. On typicality and vagueness. Cognition 64:189–206.

    DOI: 10.1016/S0010-0277(97)00025-5Save Citation »Export Citation »E-mail Citation »

    This is a response to Kamp and Partee 1995 and Fodor and Lepore 1996 by the authors of the initial “standard argument” against prototypes being meanings. In this paper they agree with the two kinds of gradedness but argue that the typicality type is still not able to be captured. They accuse Jerry A. Fodor and Ernest Lepore of misunderstanding the aspect of prototypicality that they are concerned to elucidate.

    Find this resource:

  • Prinz, Jesse. 2011. Regaining composure: A defense of prototype compositionality. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 437–453. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Distinguishes the view that concepts necessarily combine compositionally from the view that they are capable of combining compositionally and claims that only the latter is plausible. Develops a theory that implements this idea.

    Find this resource:

Meaning Holism

To many theorists, the opposite view of semantic compositionality is meaning holism. The topic of meaning holism is heavily explored in the philosophical literature. Until the publication of Fodor and Lepore 1991, it was probably the conscious or unconscious choice of most philosophers of language, due to its pedigree, stemming from many different sources: Willard Van Orman Quine, Ludwig Wittgenstein, D. K. Lewis, and other influential writers. That article sparked many outraged responses, but the very next year, when it was incorporated into Fodor and Lepore 1992, the outrage became a tsunami of protest. This book occasioned many author-meets-critics sessions at conferences and in the journals. The list here includes only some of the important contributions. Even a cursory look at the literature will uncover numerous others. Pagin 1997 gave expression to the technical feeling that if holism entailed that a word’s meaning was its contribution to all the sentences in which it occurred, then it would have to be compositional. Acknowledging this feeling, Pelletier 2011 attempts to describe a notion of meaning holism that is not compatible with compositionality. Along the way it canvasses the standard objections that are leveled against meaning holism. (These objections can also be derived from Fodor and Lepore 1992, although it is perhaps not so concisely stated.) Montminy 2005 argues that there is a style of meaning holism that can counter the standard arguments against such holism to the effect that it is unable to account for productivity and systematicity in language. Robert B. Brandom (Brandom 2001) is well known for his view of “inferentialism,” which is a type of meaning holism. Fodor and Lepore 2001 responds to this. Warfield 1993 is an attempt to sidestep the argument by claiming that inferential roles do not have to define meaning (which appears to be something assumed by Jerry A. Fodor and Ernest Lepore). One place where meaning holism has gained acceptance is in scientific theories: the meaning of any theoretical term is its role within the theory. Fodor and Lepore 1992 has many arguments against this view, but Schurz 2005 claims that a noncompositional theory can avoid all of them.

  • Brandom, Robert B. 2001. Articulating reasons: An introduction to inferentialism. Cambridge, MA: Harvard Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Brandom recommends an inferentialist semantics according to which there are performances that have a (semantic?) content that is thereby given derivatively to a sentence as its meaning. These contents are to consist of their inferential properties. The ability to know and understand a language is the practical ability to keep track of these inferential commitments and entitlements.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 1991. Why meaning (probably) isn’t conceptual role. Mind and Language 6:328–343.

    DOI: 10.1111/j.1468-0017.1991.tb00260.xSave Citation »Export Citation »E-mail Citation »

    The first of the Fodor and Lepore contributions in this area. Officially, it finds an internal problem for (a class of) semantic theories, whereas it claims that holism and its problems are external. But the holism difficulties form a very strong background in the paper. The internal problem is that while meanings are compositional, inferential roles are not. Hence meanings cannot be due to inferential roles.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 1992. Holism: A shopper’s guide. Oxford: Blackwell.

    Save Citation »Export Citation »E-mail Citation »

    This is the locus classicus of argumentation against meaning holism. In this book the authors do not attempt to “prove” that natural language is compositional. Rather, their purpose is merely to “sweep away” the alternative view.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 2001. Brandom’s burdens: Compositionality and inferentialism. Philosophy and Phenomenological Research 63:465–481.

    DOI: 10.1111/j.1933-1592.2001.tb00119.xSave Citation »Export Citation »E-mail Citation »

    This is a reply, much in the spirit of Fodor and Lepore 1992, to the influential view of Robert B. Brandom, called “inferentialism,” according to which meaning is elucidated in terms of the inferences that are justified.

    Find this resource:

  • Montminy, Martin. 2005. A non-compositional inferential role theory. Erkenntnis 62:211–233.

    DOI: 10.1007/s10670-004-0899-2Save Citation »Export Citation »E-mail Citation »

    Argues that a “dispositional inferential role theory” can account for the arguments concerning productivity and systematicity in language without assuming compositionality.

    Find this resource:

  • Pagin, Peter. 1997. Is compositionality compatible with holism? Mind and Language 12:11–33.

    DOI: 10.1111/1468-0017.00034Save Citation »Export Citation »E-mail Citation »

    Gives both intuitive reasons and formal proof of the belief that (a certain kind of) meaning holism must be compatible with compositionality.

    Find this resource:

  • Pelletier, Francis Jeffry. 2011. Holism and compositionality. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Gives an account of the standard notions of meaning holism together with a summary of the standard objections to them. It goes on to develop a “nonindividualistic” account of semantic holism (based on ideas of Ferdinand de Saussure) that is not compatible with compositionality. Many of the objections to holism that are based on an “individualistic” understanding are avoided, but it is argued that there are nonetheless “public” versions of other “individualistic” objections that still hold.

    Find this resource:

  • Schurz, Gerhard. 2005. Semantic holism and (non-) compositionality in scientific theories. In The compositionality of meaning and content. Vol. 1, Foundational issues. Edited by Markus Werning, Edouard Machery, and Gerhard Schurz, 271–284. Frankfurt: Ontos Verlag.

    Save Citation »Export Citation »E-mail Citation »

    Schurz considers meaning holism in the realm of scientific theories, where the meaning of a term is its inferential role in a theory. He identifies six problems with such holism and argues that a Ramsey-Carnap-Lewis view avoids five of them but fails on the topic of compositionality. Schurz concludes that a theory can be holistic (and thus noncompositional) and nonetheless avoid the other standard objections to holism.

    Find this resource:

  • Warfield, Ted A. 1993. On a semantic argument against conceptual role semantics. Analysis 53:298–304.

    DOI: 10.2307/3328254Save Citation »Export Citation »E-mail Citation »

    An early and influential paper complaining about the style of argument employed by Jerry A. Fodor and Ernest Lepore, especially in Fodor and Lepore 1991. The claim is that the class of semantic theories under attack by Fodor and Lepore do not all think of inferential roles as defining meaning.

    Find this resource:

Context in General

This is a huge topic that cannot be adequately surveyed here. We therefore give five examples that indicate different positions along a spectrum of positions that are outlined in Recanati 2004. Borg 2004 defines what semantic minimalism is: semantic content is exhausted by the contributions of the syntactic constituents and their mode of composition. There is no role for any contextual information (except for a short list of indexicals). Cappelen and Lepore 2005 also argues for minimalism but by attacking all forms of contextualism. The most radical form of contextualism is that advocated in Travis 2008; Pagin and Pelletier 2007 defines a method for giving an account of some forms of contextualism.

  • Borg, Emma. 2004. Minimal semantics. Oxford: Oxford Univ. Press.

    DOI: 10.1093/0199270252.001.0001Save Citation »Export Citation »E-mail Citation »

    Minimalism provides an answer to two questions: (a) what counts as semantic content and (b) what work semantic content does. Semantic content is exhausted by syntactic constituents and their composition, so features of context cannot permeate semantic content unless there is an item in the syntax that dictates so. Semantic content guides pragmatic speech-act content and helps in communication.

    Find this resource:

  • Cappelen, Herman, and Ernest Lepore. 2005. Insensitive semantics: A defense of semantic minimalism and speech act pluralism. Malden, MA: Blackwell.

    Save Citation »Export Citation »E-mail Citation »

    This very influential book defends minimalism in semantics by contrasting it with “rabid contextualism” of the variety found in Travis 2008. The authors say that “no one” would believe rabid contextualism, but they proceed to argue that any so-called moderate contextualism inevitably slides into rabid contextualism. Hence only semantic minimalism is plausible.

    Find this resource:

  • Pagin, Peter, and Francis Jeffry Pelletier. 2007. Content, context, and composition. In Context-sensitivity and semantic minimalism: New essays on semantics and pragmatics. Edited by Gerhard Preyer and Georg Peter, 25–62. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    This is an account of a way to accommodate, in a compositional manner, some of the contextual information alleged by François Recanati (and Charles Travis) to be contextual and nonsemantic. In the terms of Cappelen and Lepore 2005, this counts as a moderate contextualism but one that does not lead to “rabid contextualism,” as Herman Cappelen and Ernest Lepore claim any moderate view would.

    Find this resource:

  • Recanati, François. 2004. Literal meaning. Cambridge, UK: Cambridge Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    This is a major work offering a “truth-conditional pragmatic” account of the truth of utterances in a context. Among the other features of this discussion, a number of contextualist positions are described that differ in “how much context” they allow in the interpretation of an utterance. Different locations along this spectrum give rise to different accounts of compositionality.

    Find this resource:

  • Travis, Charles. 2008. Occasion-sensitivity: Selected essays. Oxford: Oxford Univ. Press.

    DOI: 10.1093/acprof:oso/9780199230334.001.0001Save Citation »Export Citation »E-mail Citation »

    Travis has emphasized the importance of context for meaning since the late 20th century. A given linguistic form, even with its meaning fixed, may express an indefinite variety of thoughts. Travis falls on the “far contextual” end of the scale of contextuality that François Recanati describes. He is often the “opponent” in discussions about the importance of semantics (and compositionality).

    Find this resource:

Communication

Justifications of compositionality usually focus on understandability and productivity. However, the requirement of compositionality is different in the two cases. The understander receives a sentence that can be broken into its structured parts and then evaluated compositionally to determine the sentence’s meaning. But the producer instead has knowledge of what is desired to be conveyed but needs to discover which words plus structures will do this. This cannot be done by straightforward compositionality. Pagin 2003 argues that it requires what Peter Pagin calls “inverse compositionality.” Pagin 2005 applies the apparatus of compositionality applied to literal meaning to contextual meaning, claiming that with “unarticulated constituents” this can also be made compositional. Pagin 2011 considers the requirements of communication, arguing that compositionality by itself is not enough and that communication requires that linguistic computation be of a certain “minimalistic” variety. Another aspect of compositionality and communication concerns the intuitive feeling that compositionality mandates a kind of “realism” about meanings that goes against indeterminacy of translation, which is seen as a feature of holistic theories instead. Werning 2004 argues for realism in semantics, claiming that Wilfrid Hodges’s theorem requires it. Leitgeb 2005 responds that Hodges’s theorem is neutral on that point.

  • Leitgeb, Hannes. 2005. Hodges’ theorem does not account for determinacy of translation: A reply to Werning. Erkenntnis 62:411–425.

    DOI: 10.1007/s10670-004-1992-2Save Citation »Export Citation »E-mail Citation »

    Argues against Marcus Werning’s claim that Wilfrid Hodges’s theorem shows that Quine-style indeterminacy of translation is wrong. Indeed, Leitgeb’s claim is that there is no way to employ Hodges’s theorem so as to move away from an indeterminacy of meaning of lexical items, since Hodges’s theorem assures the existence of a unique extension only up to isomorphism. This will not guarantee the sort of “realism about meanings” that Werning argues for.

    Find this resource:

  • Pagin, Peter. 2003. Communication and strong compositionality. Journal of Philosophical Logic 32:287–322.

    DOI: 10.1023/A:1024258529030Save Citation »Export Citation »E-mail Citation »

    In order to account for both productivity and understandability, it is argued that the language must be not only compositional in the usual way but inversely compositional (akin to Jerry A. Fodor’s “reverse compositionality”).

    Find this resource:

  • Pagin, Peter. 2005. Compositionality and context. In Contextualism in philosophy: Knowledge, meaning, and truth. Edited by Gerhard Preyer and Georg Peter, 303–348. Oxford: Clarendon.

    Save Citation »Export Citation »E-mail Citation »

    This is a discussion of how the concept of compositionality can be extended from context-invariant to context-dependent meaning and of how the compositionality of natural language might conflict with context dependence. Presents a compositional theory of the unarticulated constituent variety.

    Find this resource:

  • Pagin, Peter. 2011. Communication and the complexity of semantics. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 512–531. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Justifies compositionality by means of computational needs for communication. Pagin defines a kind of “minimal complexity” computation that certain types of compositional systems have and concludes that although it is not necessary to be compositional to be efficient, it is reasonable to conclude that natural language approximates this type of minimal-complexity compositionality.

    Find this resource:

  • Werning, Markus. 2004. Compositionality, context, categories, and the indeterminacy of translation. Erkenntnis 60:145–178.

    DOI: 10.1023/B:ERKE.0000012876.85940.b3Save Citation »Export Citation »E-mail Citation »

    Argues for “realism” in semantics by claiming that Wilfrid Hodges’s theorem shows that Quinean indeterminacy of translation is undermined by taking the observation sentences as stimulus meanings and showing that this meaning assignment is uniquely extendable to all the expressions that occur in observation statements.

    Find this resource:

Context and Contextual Variables

Context—in the sense of the “real world circumstance” in which an utterance is made—is often seen as providing a challenge to compositionality, and numerous examples are brought forth to show that the intuitive meanings (and truth values) of a sentence differ depending on the context of utterance. Although indexicals (“I,” “you,” “now,” etc.) are context dependent, these are not normally seen as counterexamples to compositionality. Another class comes from such terms as “local,” “enemy,” and “foreigner.” Often there is the appearance of some indexical being “missing” from the sentence, such as in “It is raining,” when the location is where the speaker is. Perry 2001 rehearses the author’s earlier “unarticulated constituents” solution to these problems. More telling are cases where context expands or restricts the meaning of terms that otherwise seem not to have any “hidden variable” (as “foreigner” and so forth arguably do). For example, quantifier phrases can be seen as nonuniformly doing this: “All the beer is gone” might be true when said in one house, even though there is beer in a neighbor’s house. Stanley 2000 proposes having a variable that gets its value from context but that is nonetheless in the logical form (semantics) of the sentence. This solution is a denial of the hidden-indexical framework of John Perry, since it has a variable in the syntax. His justification for placing it in the semantics became known as “the binding argument,” because the value of the variable could change within the sentence on account of higher-level quantifiers that bind it. Stanley and Szabó 2000 investigates further the details of Jason Stanley’s proposal, arguing that the variables to be bound have to attach to lexical nouns. Pelletier 2003 argues that there are other options open for the invocation of variables that, contra Stanley and Szabó 2000, do not imply noncompositionality. Rett 2006 claims that Stanley’s method cannot do both domain restriction and contextual effects in other parts of a sentence. Gauker 2010 argues against the Stanley 2000 “hidden variable” account and offers a slightly different “relational” account of lexical nouns. Recanati 2002 argues against Stanley and in favor of unarticulated constituents, giving Perry’s account a fuller set of methods for interpreting these constituents. Recanati 2011 argues that many of these kinds of “semantic flexibility” are due to context effects that do not challenge compositionality claims that other types of context effects do militate against compositionality.

  • Gauker, Christopher. 2010. Global domains versus hidden indexicals. Journal of Semantics 27:243–270.

    DOI: 10.1093/jos/ffq001Save Citation »Export Citation »E-mail Citation »

    Gives an alternative compositional account from Stanley 2000 of the logical form that is relevant to the way of getting contextual information into the representation of a sentence. Jason Stanley has an account where nouns had a “hidden indexical.” Gauker has nouns be relational, giving the same reading for the examples considered by Stanley. However, they differ on a sentence such as “Every student is happy and some student is not happy.”

    Find this resource:

  • Pelletier, Francis Jeffry. 2003. Context dependence and compositionality. Mind and Language 18:148–161.

    DOI: 10.1111/1468-0017.00218Save Citation »Export Citation »E-mail Citation »

    Argues against the claim in Stanley and Szabó 2000 that attaching contextual information directly to noun-phrase nodes of a logical-form tree is noncompositional. An evaluation of what the functional notion of compositionality allows for shows that this is a legitimate option.

    Find this resource:

  • Perry, John. 2001. Reference and reflexivity. Stanford, CA: Center for the Study of Language and Information.

    Save Citation »Export Citation »E-mail Citation »

    Perry is credited with introducing the notion of an unarticulated constituent in his 1986 paper “Thought without Representation” (Proceedings of the Aristotelian Society 60:S137–S151) and having refined it through many iterations, here in the chapter “Unarticulated Constituents.” An unarticulated constituent of a proposition is something that is just “too boring” to be vocalized by the sentence uttered in the context. Nonetheless, it is present in the proposition that was expressed.

    Find this resource:

  • Recanati, François. 2002. Unarticulated constituents. Linguistics and Philosophy 25:299–345.

    DOI: 10.1023/A:1015267930510Save Citation »Export Citation »E-mail Citation »

    This is a major response to Stanley 2000. Jason Stanley argues that there are no “unarticulated constituents,” contrary to what “truth-conditional pragmaticists” have claimed, but rather that all truth-conditional effects of context can be traced to logical form. Recanati claims that Stanley’s argument is fallacious—that there are unarticulated constituents and that truth-conditional pragmatics is correct. Much of Recanati’s position relies on his accounts of “saturation,” “enrichment,” and “modulation” of semantic values.

    Find this resource:

  • Recanati, François. 2011. Compositionality, flexibility, and context-dependence. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 175–191. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    Semantic flexibility is defined as the way a word’s meaning can be affected by other words in the same sentence. Recanati argues that this flexibility amounts to a kind of context sensitivity that does not pose a real problem for compositionality. However, another type of context sensitivity (“sense modulation”) does tell against compositionality.

    Find this resource:

  • Rett, Jessica. 2006. Context, compositionality, and calamity. Mind and Language 21:541–552.

    DOI: 10.1111/j.1468-0017.2006.00294.xSave Citation »Export Citation »E-mail Citation »

    Explores a variety of issues with the proposal of Stanley 2000, especially those in which there is domain-restricting information elsewhere in the sentence (such as in a prepositional phrase). Rett’s claim is that there is no coherent way that Jason Stanley can make room for both context information and for domain-restricting information elsewhere in the sentence.

    Find this resource:

  • Stanley, Jason. 2000. Context and logical form. Linguistics and Philosophy 23:391–434.

    DOI: 10.1023/A:1005599312747Save Citation »Export Citation »E-mail Citation »

    In Stanley’s account, context assigns some value to words in a sentence. When there is no explicit item in the sentence, we need to have a variable attached to some item or other that can accept the value put forth by context. Of interest is Stanley’s demonstration that this variable must actually be in the sentence’s representation, since it can be quantified over by explicit phrases in the sentence.

    Find this resource:

  • Stanley, Jason, and Zoltán Szabó. 2000. On quantifier domain restriction. Mind and Language 15:219–261.

    DOI: 10.1111/1468-0017.00130Save Citation »Export Citation »E-mail Citation »

    This influential paper argues that “domain restrictions,” for example, when a phrase such as “all the students” is interpreted as “all the students in Linguistics 201 this year,” must be part of the semantic interpretation of the resulting sentence and that the relevant restriction comes from associating a contextual variable with lexical nouns. Exhaustively surveying alternative accounts has left them with just one way to implement this idea.

    Find this resource:

Neural Issues

Many theorists feel that all mental activity, and in particular language-oriented processing, is a feature of the brain and hence is neurally controlled. But then, what should we say about the status of the principle of semantic compositionality? Is it also to be found in the neural substructure? Werning 2009 is concerned with accommodating compositionality in a neutrally plausible system. Werning 2011 discusses the notions of information flow in a brain being compositional. Engel and Maye 2011 adds the notion of temporal flow of information carried in language through the brain.

  • Engel, Andreas, and Alexander Maye. 2011. Neural assemblies, the binding problem, and neural synchrony. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 618–634. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    The authors focus on models in which it is the temporal structure of activity that carries information, review the capability of such systems to develop compositional structures, and propose synchronization of neuronal activity as an underlying mechanism.

    Find this resource:

  • Werning, Markus. 2009. The compositional brain: Neuronal foundations of conceptual representation. Paderborn, Germany: Mentis.

    Save Citation »Export Citation »E-mail Citation »

    Werning here attempts to meld the psycho-philosopho-linguistic conception of concepts as bearers of intentional content and meaning with a neural-oriented view of information flow in the brain. An underlying motif is the attempt to accommodate classical notions of compositionality within this framework.

    Find this resource:

  • Werning, Markus. 2011. Non-symbolic compositional representation and its neuronal foundation: Towards an emulative semantics. In The Oxford handbook of compositionality. Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery, 635–656. Oxford: Oxford Univ. Press.

    Save Citation »Export Citation »E-mail Citation »

    A neurobiologically motivated theory of meaning as internal representation is developed that holds on to compositionality but is nonsymbolic, being implemented by recurrent neural networks. The semantics to be developed is structurally analogous to model-theoretical semantics, but it regards meanings as set-theoretical constructions of the neural counterparts of denotations.

    Find this resource:

Jerry A. Fodor

Jerry A. Fodor is possibly the most famous scholar who makes unrelenting use of the notion of compositionality and its alleged features in philosophy of mind and other psychologically oriented areas. It would be remiss not to have a special section on Fodor, given his influence on the application of compositionality in the philosophy of mind. It should also be remarked that the original use of “compositionality” is in Katz and Fodor 1963, cited under Historical Antecedents. Fodor has been concerned with the topic for most of his academic life. Much of his output on this topic has been in collaboration with Ernest Lepore (a collection of their papers on compositionality is Fodor and Lepore 2002). Possibly their most basic paper is Fodor and Lepore 1991 (reprinted in Fodor and Lepore 2002), the argument of which is generalized in Fodor and Lepore 1992 to cover a wide range of views that all contend that the meaning of a term or phrase is the various connections it has with other terms and phrases, and its workings are manifest in many of Fodor’s other works. It is impossible to distinguish separate themes relevant to compositionality among the books, since they all seem to draw further conclusions from the presumed truth of compositionality. (Fodor and Lepore are famous for claiming such things as “Compositionality is, as they say in England, non-negotiable” and “So non-negotiable is compositionality that I’m not even going to tell you what it is.”) Many topics that are in train with compositionality are also discussed in Fodor’s works: Fodor and Pylyshyn 1988 introduces systematicity, Fodor and Lepore 1992 heavily discusses holism, Fodor 1998 argues for the concept of atomism, Fodor 2001 discusses “underived content,” and Fodor 2008 discusses reverse compositionality.

  • Fodor, Jerry A. 1998. Concepts: Where cognitive science went wrong. Oxford: Clarendon.

    Save Citation »Export Citation »E-mail Citation »

    In this book Fodor argues for an atomistic theory of concepts. He claims that other accounts of concepts are committed to inferential role theory, which falls prey to arguments that employ compositionality. In this account, concepts are argued not to be stereotypes, prototypes, abstractions from belief systems, or various other psychological items that have been proposed by such theorists as Ray Jackendoff and James Pustejovsky.

    Find this resource:

  • Fodor, Jerry A. 2001. Language, thought, and compositionality. Mind and Language 16:1–15.

    DOI: 10.1111/1468-0017.00153Save Citation »Export Citation »E-mail Citation »

    This is Fodor’s “millennial paper.” It discusses the issue of whether it is thought or language that has semantic content (“in the first instance”). Fodor claims that it is thought and that in fact language does not have content at all. He furthermore claims that this is an empirically supported position.

    Find this resource:

  • Fodor, Jerry A. 2008. LOT 2: The language of thought revisited. Oxford: Clarendon.

    Save Citation »Export Citation »E-mail Citation »

    This book offers a more detailed account of the language of thought hypothesis that was started in Fodor’s earlier works. The role of compositionality is reemphasized as crucial for a representational theory of mind—most of what we know about concepts follows from the compositionality of thoughts.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 1991. Why meaning (probably) isn’t conceptual role. Mind and Language 6:328–343.

    DOI: 10.1111/j.1468-0017.1991.tb00260.xSave Citation »Export Citation »E-mail Citation »

    Describes inferential-role semantics of the sort associated with “pragmatism” in philosophy of mind and language. Famously, it argues that while meanings are compositional, inferential roles are not. Hence meanings cannot be due to inferential roles.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 1992. Holism: A shopper’s guide. Oxford: Blackwell.

    Save Citation »Export Citation »E-mail Citation »

    This is the locus classicus of argumentation against meaning holism. In this book the authors do not attempt to “prove” that natural language is compositional. Their purpose is rather merely to “sweep away” the alternative view of meaning holism.

    Find this resource:

  • Fodor, Jerry A., and Ernest Lepore. 2002. The compositionality papers. Oxford: Clarendon.

    Save Citation »Export Citation »E-mail Citation »

    This collection of works by Fodor and Lepore has three parts: four papers on the nature of compositionality; two papers on the meaning of terms in the mental lexicon arguing for Fodor’s semantic atomism; and finally, three papers that critique two versions of “meaning pragmaticism,” those of Robert B. Brandom and Paul Churchland.

    Find this resource:

  • Fodor, Jerry A., and Zenon W. Pylyshyn. 1988. Connectionism and cognitive architecture: A critical analysis. Cognition 28:3–71.

    DOI: 10.1016/0010-0277(88)90031-5Save Citation »Export Citation »E-mail Citation »

    The famous “systematicity objection” to connectionist models of cognition is first brought out in this piece. The discussion sparked a vast literature throughout the 1990s concerning the role of representations within connectionism.

    Find this resource:

LAST MODIFIED: 10/28/2011

DOI: 10.1093/OBO/9780199772810-0044

back to top

Article

Up

Down