Damon Horowitz
Papers
Hermeneutic Contextualism and the Knowledge Predicate
(Journal article -- under review) [pdf]
In this paper, I propose a novel account of utterance meaning, Hermeneutic Contextualism (HC), which holds that the truth-conditional content of an utterance is interpreter-sensitive. Focusing on the discussion of the knowledge predicate in epistemology, I show that our assessments of the truth of a knowledge attribution legitimately vary depending upon our interpretive context -- i.e., different ascriptions of truth-conditional content to a single utterance are acceptable on different occasions of assessment -- in a way that previous Invariantist, Contextualist, and Truth-Relativist views cannot easily account for. I present HC as an analytic version of the hermeneutic approach to meaning, extending the best aspects of Contextualism and Relativism by encouraging us to understand content as being more like an appearance of an utterance than an intrinsic property. I conclude by describing the potential applicability of HC to other domains of language use, and discussing general difficulties with the underlying assumptions and methodology surrounding the theoretical notion of utterance content.
Must we be so literal?
(Journal article -- under review) [pdf]
The
relation between metaphorical meaning and literal meaning is relevant to
current semantics/pragmatics debates which seek to clarify the notion of 'what
is said'. Under some Contextualist proposals, the very words of a metaphorical
utterance undergo a shift in meaning and express novel metaphorical meanings --
a radical departure from the traditional conception of 'what is said' as
determined by words' fixed conventional literal meanings. An increasingly
popular Literalist strategy for attempting to defeat this proposal argues that
there are normative constraints on our use of language which reveal an
underlying dependence of metaphorical speech upon literal speech, providing
reason to exclude metaphorical meaning from 'what is said'. In the present paper, I
contend that this argument misconceives the role of normative considerations in
a theory of utterance content: it
extends the reach of semantics beyond its constructive definition in linguistics,
and thereby overestimates the authority of conventional literal meaning. By clarifying the proper domains of
semantics and pragmatics, we may see how the Literalist errs in taking literal
meaning to be explanatorily central in our understanding of utterance
truth-conditions -- and thus why the Literalist argument is unsuccessful at
defeating the Contextualist proposal.
Davidson's Intentions and Reimer's Conventions
(Journal article -- under review) [pdf]
In both folk and scientific linguistics, we think of language as a system of conventions, with words bearing prior shared literal meanings. These seem to be what we learn when we learn a language, and what our communication depends upon. Without them, it is not clear how we would get by linguistically.
In "A Nice Derangement of Epitaphs", Donald Davidson famously denies the importance of just such conventional literal meanings. Focusing on the phenomenon of malapropisms, but extending his conclusions to cover ordinary uses of language as well, Davidson claims that interpretation does not require prior shared meanings.
Marga Reimer rejects this view of malaprop interpretation. She objects that Davidson has misunderstood the importance of conventions for communication; and therefore, we should not accept Davidson's conclusions about the dispensability of conventions for utterance interpretation in general.
I argue that Reimer's objections to Davidson cannot be sustained -- for they crucially misconstrue the type of creative and cooperative playing along which Davidson emphasizes in his characterization of communication. I clarify Davidson's position on this question, and in so doing elaborate upon the way in which language use is analogous to participation in a game. With these clarifications, we find that Davidson's radical conclusions remain important to reckon with.
Metaphor and Mercurial Content
(Journal article -- under review) [pdf]
What is it for an utterance to have truth-conditional content? While Literalists and Contextualists attempt to adjudicate the proper domains of semantics and pragmatics, they do not adequately answer this fundamental question. I propose that both camps' concern with the determinants of meaning, and in particular with the crucial issue of indeterminacy thereof, might be better addressed by exploring a hermeneutic approach to utterance content, recognizing its inherently interpreter-sensitive character. I focus here on metaphor as an initial testbed for this proposal.
PRESENTATIONS
What Metaphor Tells Us About Truth
(Philosophy Department, University of Maryland, College Park, February 2007)
Must we be so literal?
(Colloquium presentation, American Philosophical Association, Eastern Division, December 2006)
Science and Relativism
(Guest lecture, Steinhardt, New York University, October 2006)
Hermeneutic Contextualism: Meaning and Relativism in Knowledge
(Philosophy Department, Stanford University, May 2006)
See my Dissertation section, read the introduction online in [pdf].)
Remarks on Flowcharts and Psychological Explanation
(Berkeley/Stanford Graduate Student Conference, March 2003)
Works in Progress
Josef Stern and the Crisis of Literalism
(Journal article -- in progress) [pdf]
Davidson on Metaphors and Malaprops
(Journal article -- draft complete) [pdf] (very large chapter-sized version)
In his later work, Donald Davidson advances two controversial views about two apparently exceptional linguistic phenomena: metaphors and malaprops. Davidson claims that the value and power of a metaphor as a communicative device derives from its experiential impact upon a hearer, not from any "metaphorical meaning" which it allegedly conveys; and he claims that speaker intentions rightfully and ineliminably dominate conventional literal meanings in interpreters' determinations of the content of malapropic utterances.
But these important views require more careful articulation than is provided in Davidson's original writings. As they currently stand, they seem inconsistent on fundamental questions about literal meaning and communication. What is the explanatory role of conventional literal meaning in language? What is it to communicate a cognitive content? And how do "fringe" uses of language relate to ordinary uses? The project of the present paper is to develop Davidson's ideas, while disposing of some of the complications which arise from his less carefully considered peripheral claims, in such a way as to resolve these apparent inconsistencies and to suggest Davidsonian answers to the fundamental questions.
I propose that the key to satisfying the apparently conflicting desiderata which Davidson articulates is to understand utterance meaning as a product of both speaker and hearer contributions. By finding a balance between these factors, we can explain how it is that the specific content which is ascribed to an utterance may legitimately vary across interpretive contexts; and thereby accommodate Davidson's demands for explanatory value in a notion of utterance meaning, while still capturing the flexibility of meaning which metaphors and malaprops require. I conclude by discussing the applicability of this approach to ordinary (non-metaphorical and non-malapropic) language as well.
Distributed Meaning
(Journal article -- in progress)
While many contemporary philosophers of language have some sympathy for semantic holism, it is often accompanied by defeatism about the prospects for explanation and generalization under such a conception. I believe that this defeatism is unfounded, and arises only from the lack of clear and detailed examples of what more holistic explanation may look like.
In this paper, I offer a technical introduction to contemporary paradigms of associative computation, including network-based and statistical models of language processing, demonstrating how a range of operations (from unsupervised clustering to transductive inference in classification) are performed. I sketch how a holistic model of word and sentence meaning based upon associative computation can be used to explain word meaning acquisition, sense discrimination, effective communication, folk psychological generalizations, and various semantic phenomena (truth evaluations, indirect discourse reports, etc.) by reliance upon a technically-articulated notion of sufficient similarity. I characterize the types of explanations we should expect in this mode, and specifically, the way in which these explanations support more general theses of Particularism.
My immediate goal in this project is to show how distributed notions of meaning can provide a viable alternative to traditional notions of discrete literal meaning across a wide range of theoretical tasks. My larger goal is to explain the relationship between holistic phenomena and local explanation -- as elaborated in the following paper.
Local Normativity: Moral Particularism and Semantic Particularism
(Journal article -- in progress)
Moral Particularism is a position in contemporary ethical theory which questions the status of moral principles: Do legitimate principles exist, are they knowable, do they play any role in moral deliberation? Most versions are motivated by the difficulties we face in specifying moral principles -- proposed principles seem plagued by exceptions, they often conflict with each other, and they are never sufficiently elaborated to uniquely specify correct action when applied to the complexities of a specific situation. Particularists emphasize that our moral conduct tends to be guided by a sort of moral intuition or perception of right action in specific situations, rather than by rational deduction from normatively guiding principles.
In this paper I propose that heightened attention to the use of language in specifying moral principles can show how and why it should be that our ethical generalizations seem at such an ineliminable distance from our particular ethical judgments. I seek to substantiate the claims of Moral Particularism by elaborating a thesis of Semantic Particularism about language: there is no single purely principled way to compute meanings in a situation; rather, ordinary speakers use their best pragmatic judgment in their present situation to see meaning what is called for. This view is consistent with Radical Contextualism more broadly (and my hermeneutic variant), which explains how it is that our linguistic generalizations (e.g., about the literal meanings of words) may have social and classificatory utility even if they lack "psychological reality" and fail to predict or determine the meanings of our linguistic utterances.
My strategy here is to focus on several analogies between Moral Particularism and Semantic Particularism. For instance, the difficulties faced in specifying moral principles correspond to the difficulties encountered in developing analytic theories of concepts (i.e., as candidate discrete literal meanings for words); and the diversity of ways in which we elaborate an ethical principle in different contexts of moral judgment correspond to the diversity of ways in which we elaborate the significance of an utterance in different contexts of interpretive evaluation. I conclude by assessing whether it is possible to maintain my broader thesis that in both ethical and linguistic cases, the normative factors which are operative may be analyzed as pragmatic and instrumental, as opposed to inherently moral or semantic.
Metaphorical Content and Its Discontents
(Journal article -- in progress)
Four preparatory squibs: on paraphrase [pdf], on likenesses [pdf], on transformation [pdf], and on truth [pdf].
Do children's first meaningful words have meanings?
(Journal article -- draft complete)
There is a popular and intuitive approach to early language acquisition which considers word learning to be a naming game. It traces back at least to Augustine, finding modern articulation in Brown's "original word game", and continuing in contemporary language acquisition research (e.g., Semantic Feature Hypothesis, Lexical Contrast Theory, etc.). In this approach, the learner's goal is to figure out the correct meaning for a word. The learner forms hypotheses about possible word meanings based upon prior conceptual knowledge and observation of referential word uses in social interactions. These hypotheses may be mistaken, resulting in word use errors such as overextensions in production or incorrect responses in comprehension. Eventually, the hypotheses are refined by feedback from other speakers, in the form of rejections, corrections, and acceptances, until the learner has acquired the conventional adult meaning for the word.
This paper examines the viability of an alternative approach. In particular, I consider whether an approach which restricts itself to functional characterizations of children's abilities using only an associationist descriptive vocabulary is able to give a satisfactory account of early word acquisition. This minimalist approach deliberately avoids reliance upon the mentalistic and intentional vocabulary commonly used to describe adult language, in order to avoid unsupported attribution of adult mental constructs to children.
The specific question I consider here is on whether there is stage of language development in which this alternative approach proves equally or better equipped than the traditional approach to describe the known data. My hypothesis is that early word acquisition for 1-2 year-olds is such a stage. It is not necessary, or plausible, to attribute knowledge of literal meanings, externalist reference, adult-like categories, hypothesis formation and testing, and pragmatic reasoning to 1-year-old children. The alternative approach I describe offers a simpler and more cognitively realistic explanation for children's early successes at producing and comprehending words in their earliest forays into meaningful linguistic behavior.
Review of Ray Jackendoff's Foundations of Language
(Book review -- draft complete)
In Foundations
of Language, Jackendoff articulates his approach to linguistics, and
presents his theory of the general nature of linguistic structure.
Following Chomsky (in rough outline), Jackendoff's approach is strongly mentalistic
and internalist: he takes linguistic structure to be
"psychologically real", characterizing the individual language user's
knowledge of his language.
Jackendoff further asserts that the externalist study of reference and truth is incompatible with his internalist approach to language. His response to this conflict is to discard the popular philosophical notions of reference and truth. However, adopting the approach which Jackendoff suggests requires more fundamental changes to our ordinary understanding of communication than Jackendoff himself may be willing to accept.