One Word, Many Concepts (DRAFT:
Polysemy runs deep. But if polysemy runs deep, meanings do not map contexts to extensions--and sentences do not express propositions, or mappings from contexts to propositions--not even if each sense of a polysemous expression maps contexts (of using the expression with that sense) to extensions or proposition-constituents. A word like 'window', which has more than one sense, can be used neutrally without using it to express any particular sense.
(Topoi 40:5-25, 2021)
This paper--like I-Languages and T-sentences
Meanings via Syntactic Structures (Syntactic
Structures after 60 Years, edited by N. Hornstein et.al.,
De Gruyter: Mouton 2017)
This short essay was prompted by teaching Syntactic Structures, in an undergraduate course, and paying attention to the (often ignored) remarks about meaning.
Semantic Internalism (The Cambridge Companion to Chomsky,
edited by Jim McGilvray, CUP 2017)
This essay discusses some of Chomsky's views about meaning, contrasting them with some of Putnam's claims in "The Meaning of 'Meaning' ".
I-Languages and T-sentences
This paper, about the relevance of Liar Paradoxes for truth conditional semantics, and the paper below are companions. Bottom line for this one: sentences of a human language don't have truth conditions. No sentence of a human language is true. The previous sentence isn't true, and neither is this one. Snow is white isn't true, and neither is 'Snow is white.' is true if and only if snow is white.
Framing Event Variables
Slides for the talk, at a conference in Erfurt, can be found here. This paper is about the relevance of puzzles concerning event individuation for semantics. Bottom line: event analyses of 'Alvin chased Thedore' are good; truth-theoretic constuals of such analyses are bad. Together with the paper above, and Meaning Before Truth listed below, the larger conclusion is that Davidsonian conceptions of meaning are in big trouble. Even bracketing concerns about specific constructions, and focusing on cases that are supposed to motivate truth conditional semantics, foundational problems quickly emerge if you focus on truth, predication, or reference.
Concepts, Meanings, and Truth: First Nature,
Second Nature, and Hard Work (Mind and Language 25: 247-78, 2010)
The idea is that lexical expressions of a human I-language let children use available concepts to introduce formally distinct "I-concepts," which can then be combined via operations invoked by phrasal syntax. So while "prelexical" concepts may not exhibit the kind of systematicity required for truth, I-concepts do. But various empirical considerations suggest that I-concepts are massively monadic, and that the relevant "I-operations" are fundamentally conjunctive. This, I claim, makes it implausible that I-concepts are true of language-independent things. Meanings can be viewed as instructions to assemble concepts that make it possible for humans to have truth-evaluable thoughts. But forming such concepts requires independent cognitive work, not just a language with a compositional semantics. This paper, which abstracts from the technical details, forms a pair with Minimal Semantic Instructions (listed under Compositional Semantics)
Meaning Before Truth (Contextualism in
Philosophy, edited by G. Preyer and G. Peters, OUP 2005).
This paper extends the line of thought in "The Character of Natural Language Semantics." A running theme is that Chomsky offers a conception of semantics that lets us preserve what is right about truth-conditional semantics--and this has less to do with truth than the usual rhetoric suggests--while also preserving late-Wittgensteinian/Austinian insights about the relation between truth, meaning, and context. There are three main sections: one about the relevance of negative facts (and nativism) for semantics, and why this tells against both "deflationary" conceptions of meaning and Quine-Davidson "interpretability" conceptions; one that reviews some familiar reasons for rejecting the hypothesis that names denote things in the environment; and one that concedes externalism about truth, while noting that externalism about linguistic meaning does not follow. The paper ends with a brief tour of some alternatives, and some familiar reasons for rejecting the hypothesis that predicates are satisfied by things in the environment. A handout elaborates this line of thought (in a handouty way).
Types: Two is Better than Too Many.
(New Frontiers in Artificial Intelligence, edited by M. Sakamoto et. al., Springer LNCS/LNAI 12331, 2020. LENLS-16 conference proceedings.)
In this paper and the one below, I discuss the motivations and prospects for the very spare semantic typology employed in Conjoining Meanings.
Semantic Typology and Composition (The Science of Meaning,
edited by B. Rabern and D. Ball).
It is often said that expressions of a human language include (i) truth-evaluable sentences of a basic semantic type <t>, (ii) entity designators of a basic semantic type <e>, and (iii) unsaturated expressions whose semantic types are characterized by the recursive principle "if <α> and <β> are types, so is <α, β>." I think this hypothesis is wrong in three respects.
Minimal Semantic Instructions (in the Oxford Handbook of Linguistic
Minimalism, edited by Cedric Boeckx, 2011).
This is an attempt work out, for a range of basic constructions, the idea of meanings as "instructions to assemble conjunctive concepts." This paper, mainly devoted to technical details and minimalist reasoning, forms a pair with Concepts, Meanings, and Truth: First Nature, Second Nature, and Hard Work (listed under Semantic Internalism). And with regard to the syntactic details, I draw on the paper below.
Interrogatives, Instructions, and
I-languages: an I-Semantics for Questions, coauthored with
Terje Lohndal (Linguistic
Analysis 37:459-510, 2011).
The basic idea is simple: an "instructionist" conception of meaning, along lines developed in the paper above, can easily accommodate an attractive internalist version of the old force/content distinction; and there are interesting implications for the syntax/semantics of relative clauses and "sentential" expressions. I never intended to have views about--much less write a paper about--interrogatives. But my co-author was persuasive.
Language and Value,
edited by J. Yi and E. Lepore, ProtoSociology 31:
The meaning of a noun phrase like 'brown cow', or 'cow that ate grass', is somehow conjunctive. But conjunctive in what sense? Are the meanings of other phrases--e.g, 'ate quickly', 'ate grass', and 'at noon'--similarly conjunctive? I suggest a possible answer, in the context of a broader conception of natural language semantics.
Verbs, Complex Events: Analyticity without Synonymy
(in Chomsky and His Critics, edited [heroically] by Louise Antony and Norbert Hornstein, Blackwell 2003)
You may need to "Rotate View, Clockwise" to get the .pdf file to appear properly.
This paper was written in 1998, and so may be past its use-by date. Updated versions of various bits of the paper appear elsewhere; see note 1.
More Truth in Advertising: I'm not criticizing Chomsky; though I am being critical, and Chomsky does figure prominently.
The idea, as the subtitle suggests, is that there are analytic truths--even if the notion of synonymy is suspect. The trick involves (can you guess?) combining, in the right way, a neo-Davidsonian event semantics with a Minimalist syntax. Blatant Advertising: get hold of the entire book if only for Chomsky's replies; for anyone interested Chomsky's conception of meaning (and his semantic internalism), see especially his replies to Egan, Rey, Ludlow, Horwich, and Pietroski.
Explaining That (Journal of Philosophy 97: 665-62,
How can a speaker can explain that P without explaining the fact that P, or explain the fact that P without explaining that P, even when it is true (and so a fact) that P? Or in formal mode: what is the semantic contribution of 'explain' such that 'She explained that P' can be true, while 'She explained the fact that P' is false (or vice versa), even when 'P' is true? The proposed answer is that 'explained' is a semantically monadic predicate, satisfied by events of explaining. But 'the fact that P' (a determiner phrase) and 'that P' (a complementizer phrase) get associated with different thematic roles, corresponding to the distinction between a thing explained and the content of a speech act.
Sentence Like This Exhibit A Scope Ambiguity? coauthored
with Norbert Hornstein
(In Belief and Meaning, edited by W. Hinzen and H. Rott, Hansel-Hohenhausen 2002)
The answer is 'no'. Instances of 'every F likes some G' may not, after all, be examples of scope ambiguity.
Figuring out whether a given expression with multiple quantifiers is semantically ambiguous is hard.
Second-Order Monadicity (Philosophical
Perspectives 17: 259-298, 2003).
The first part of this paper reviews some developments regarding the apparent mismatch between the logical and grammatical forms of quantificational constructions like 'Pat kicked every bottle'. I suggest that (even given quantifier-raising) many current theories still posit an undesirable mismatch. But all is well if we can treat determiners (words like 'every', 'no', and 'most') as second-order monadic predicates without treating them as predicates satisfied by ordered pairs of sets. Drawing on George Boolos's construal of second-order quantification as plural quantification, I argue that we can and should view determiners as predicates satisfied (plurally) by ordered pairs each of which associates an entity with a truth-value (t or f). The idea is 'every' is satisfied by some pairs iff every one of them associates its entity with t. It turns out that this provides a kind of explanation for the "conservativity" of determiners. And it lets us say that concatenation signifies predicate-conjunction even in phrases like 'every bottle' and 'no brown dog'.
To Be a
Value of a Plural Variable, You Don't Have to Be Plural
(You Just Have to Be)
This is something between a handout and a paper. It focusses on an idea, acquired from George Boolos, discussed in the papers immediately above and below. For purposes of giving a compositional semantic theory for a natural language, we can and should allow for genuinely plural variables; where a genuinely plural variable is one that has more than one value relative to each assignment of values to variables.
Working Papers in Linguistics, 15: 157-90, 2006)
This speculative paper is an attempt to say why Frege's Theorem might bear, in interesting ways, on several issues in linguistics.
Concatenation (in Logical Form, edited by G.
Preyer and G. Peters, OUP 2002).
Explores the idea that concatenating natural language expressions corresponds to predicate-conjunction, as opposed to function-application. The proposal is developed in more detail in Events and Semantic Architecture (OUP 2005). But the paper gives the main idea, in the context of questions about how natural language syntax is related to Logical Form.
and Concatenates (Philosophical
Issues 16:221-45, 2006).
This paper presents a slightly modified version of the compositional semantics proposed in Events and Semantic Architecture.
Some readers may find this shorter version, which ignores issues about vagueness and causal constructions, easier to digest. The emphasis is on the treatments of plurality and quantification, and I assume at least some familiarity with more standard approaches. Space constraints caused the final document to be considerably shorter than drafts with homophonous titles. The paper above (Systematicity via Monadicity) is a kind of companion piece, showing how to locate the proposed conception of semantic composition in the context of more general attempts to simplify (or "minimize") theories of linguistic competence, with the aim of isolating the distinctively human aspects of the human language faculty. There are points of contact with recent suggestions by Elizabeth Spelke and her colleagues; see also the BBS paper by Peter Carruthers, my colleague in philosophy at Maryland.
via Monadicity (Croatian
Journal of Philosophy 7:343-374, 2007)
This is the written version of a conference presentation in Dubrovnik (Fall 2006). I argue that a "Conjunctivist" conception of semantic composition, of the sort articulated in some of the papers above, helps explain many otherwise puzzling features of natural language. More speculatively, a Conjunctivist language faculty might also help explain why human thought is as systematic as it is.
Semantic Monadicity with Conceptual
Polyadicity (In the Oxford
Handbook of Compositionality, M. Werning, W. Hinzen,
and E. Machery, eds., 2012).
Another paper in the same vein.
Language and Conceptual Reanalysis (In Towards a Biolinguistic
Understanding of Grammar: Essays on Interfaces, edited by A. DiSciullo,
John Benjamin 2012).
Like the paper above, but more detailed, and drawing some connections to Frege's notion of fruitful definitions.
and Numerosity (under construction)
Here is a video of a 2014 talk in the Defining Cognitive Science series at Simon Fraser
My thanks to my hosts, especially Endre Begby. In the talk, I discuss some
of the findings reported in the papers below. There are also
pictures of my collaborators.
Meaning of 'Most': semantics, numerosity, and psychology:
Paul Pietroski, Jeff Lidz, Justin Halberda, and Tim Hunter
(Mind and Language, 24:554-85, 2009). The title is descriptive. We offer experimental evidence in support of a certain view about how the meaning of the English determiner 'most' is related to various psychological capacities potentially relevant to human capacities for counting and quantifying. In this first installment of an ongoing project, we offer experimental evidence that adult speakers of English do indeed understand sentences like 'Most of the dots are blue' in terms of cardinality comparison (as opposed to, say, one-to-one correspondence). We also make some tentative suggestions about how the meaning of 'most' is related to potential verification procedures and the "analog magnitude system" that humans share with other animals.
Seeing What you Mean, Mostly.
Authors: Paul Pietroski, Jeff Lidz, Justin Halberda, Tim Hunter, and Darko Odic
(Syntax and Semantics: Experiments at the Interfaces, edited by J. Runner, 37:187-224, 2011).
Another paper in the same vein, stressing that while our proposal is not a form of verificationism, meanings are related to verification strategies in empirically testable ways--at least with regard to "logical" vocabulary.
Poverty of Stimulus ArgumentsVocabulary Matters. This short essay appeared in a 2015 MIT Working Papers in Linguistics volume (edited by Angel Gallego and Dennis Ott),
The Language Faculty coauthored with
Stephen Crain, in The Handbook for Philosophy of Cognitive
Science (edited by E.Margolis, S. Laurence, and S. Stich,
OUP 2011). An essay on the language faculty, in keeping with the
papers below, but also discussing some new material.
Think of the Children (Australasian Journal of Philosophy 86:657-669, 2009). This was a critical notice of Michael Devitt's book, Ignorance of Language. Michael's reply, which you might want to look at, appeared in the same issue.
Brass Tacks in Linguistic Theory
coauthored with Stephen Crain and Andrea Gualmini
(In The Innate Mind: structure and contents, edited by S. Laurence, P. Carruthers, and S. Stich, 175-197, Oxford University Press, 2005).
Yes, still arguing for innate constraints on linguistic meanings. Here, we discuss in more detail some of the individual phenomena addressed in other papers. And we're not replying to anyone in particular.
Innate Ideas coauthored with
Stephen Crain, in The Cambridge Companion to Chomsky
(edited by James McGilvray, 164-180, Cambridge Univ. Press
2005). You may need to "Rotate View, Clockwise" to get the .pdf
file to appear properly.
A more general discussion of innateness and universal grammar, in the context of Chomsky's version of rationalism.
Some of the examples mentioned here are discussed in more detail in the other papers.
Why Language Acquisition is a Snap
coauthored with Stephen Crain (Linguistic Review, 19:
Presents additional empirical arguments for universal grammar in reply to a target article by Pullum and Scholz. The main arguments concerns a cluster of semantic phenomenon concerning downward entailment, negative polarity, and the "pragmatic" implicature associated with disjunctive claims.
Nature, Nurture, and Universal Grammar
coauthored with Stephen Crain (Linguistics and Philosophy
24: 139-86, 2001).
Discusses the logic of "poverty of stimulus" arguments and some specific empirical premises, concerning both adults and children, in reply to recent empiricist conceptions of language acquisition--with particular focus on Cowie's book What's Within.
Twentieth Century Papers
Actions, Adjuncts, and Agency
(Mind 107: 73-111, 1998)
Experiencing the Facts: critical notice of John McDowell's Mind and World (Canadian Journal of Philosophy 26: 613-36, 1996)
A Defense of Derangement (Canadian Journal of Philosophy 24: 95-118, 1994)
Prima Facie Obligations, Ceteris
Paribus Laws in Moral Theory (Ethics 103: 489-515,
Intentionality and Teleological Error (Pacific Philosophical Quarterly 73: 267-82, 1992)