The Routledge Handbook of Semantics

Nick Riemer

• Publisher: Routledge
• Year: 2015
• Print ISBN: 9780415661737
• eBook ISBN: 9781317412441
• Topic: Languages & Linguistics
• Subtopic: Linguistic Semantics
• Index: Languages & Linguistics

The Routledge Handbook of Semantics provides a broad and state-of-the-art survey of this field, covering semantic research at both word and sentence level. It presents a synoptic view of the most important areas of semantic investigation, including contemporary methodologies and debates, and indicating possible future directions in the field.

Written by experts from around the world, the 29 chapters cover key issues and approaches within the following areas:

• meaning and conceptualisation;
• meaning and context;
• lexical semantics;
• semantics of specific phenomena;
• development, change and variation.

The Routledge Handbook of Semantics is essential reading for researchers and postgraduate students working in this area.

Part I

Foundational issues

---

1

(Descriptive) Externalism in semantics

Steven Gross

---

1 Introduction

Semantics is the study of meaning—in some sense. In what sense? According to a common view, semantics concerns inter alia the relation between words and the world—in particular, their intentional (or representational, aboutness) relations. When a competent user utters “Schnee ist weiss” to make an assertion, she makes a claim about how the world is. What in part enables her to represent the world as being this way is that “Schnee” refers to snow, something satisfies “ist weiss” just in case it’s white, and so “Schnee ist weiss” is true just in case snow is white. As David Lewis (1970: 18) famously put it: “semantics with no treatment of truth-conditions is not semantics.” Similar sentiments are found in leading semantics textbooks:

To know the meaning of a sentence is to know its truth-conditions. If I say to you [“There is a bag of potatoes in my pantry”] you may not know whether what I said is true. What you do know, however, is what the world would have to be like for it to be true . . . . A theory of meaning, then, pairs sentences with their truth-conditions.

(Heim and Kratzer 1998: 1)

The study of the relation of symbols to what they stand for [“refer to,” “denote,” “their informational content”] must indeed be a part of an account of meaning. For otherwise how could we understand the fundamental fact that configurations of symbols carry information about all the diverse aspects of our experience?

(Chierchia and McGonnell-Ginet 1990: 53)

Characterizing these semantic properties is no easy matter. Language users understand many more expressions than they could store individually in memory. So much of the hard work involves compositionally characterizing the semantic properties of a potentially infinite number of complex expressions as a function of those of their constituents and their mode of combination. Moreover, the subtleties of these semantic properties are often masked by our own competence with the terms. “Snow” is a mass term, apparently denoting some sort of undifferentiated stuff, unlike a count noun such as “snowballs”; and “snow is white” expresses a generic claim that is not falsified by a bit of yellow snow. How best to (compositionally) characterize the semantics of mass nouns and of generics is much disputed.

Our focus, however, is not these complexities, but the underlying word-world, or externalist, conception of semantics. A central debate in the foundations of semantics concerns whether it’s correct. Internalist opponents maintain that semantics rather concerns, or ought to concern, only non-intentional relations among linguistic items and (other) mental structures (see Chapter 2). On their view, semantics lays out what concepts or thoughts expressions directly activate or express, without recourse to intentional relations to things external to the mind/brain. (The qualification “directly” is meant to put to one side priming, association, and inference via world knowledge.) For example, the word “Schnee” might activate one’s concept snow; one might utter “Schnee ist weiss” to express one’s belief that snow is white. Whether these concepts and thoughts represent aspects of the world—and whether such representational properties have a place in empirical inquiry into the mind-brain—is another matter, not a question for linguistic semantics, which concerns itself only with the semantic properties of linguistic items (words, phrases, sentences, etc.).

In this chapter, we examine some of the prominent arguments for and against externalism, as well as the dispute’s upshot for existing work in the field. Note that the question is not what the word “semantics” means, but rather what are, or would be, fruitful paths for work in semantics to take. Also, the question is not whether characterizing words’ intentional relations to the world exhausts semantics, but rather whether it is, or should be, a part of semantics. This allows that semantics might concern both words’ non-intentional relations to other mental structures and their intentional relations to the world, so that work championed by typical externalists and work championed by typical internalists could turn out to be compatible. As we frame things, however, the theses of externalism and internalism are opposed to one another, even if work championed by their proponents is not—since we include in internalism the rejection of intentional relations so far as the scientific study of linguistic semantics is concerned. Finally, regarding the upshot for extant work, we will need to consider whether, if internalist considerations win the day, this requires us to jettison work in truth-conditional semantics or rather just to reconstrue it in internalist terms.

2 Varieties of externalism and internalism

The “externalist vs. internalist” label is also applied to other disagreements concerning the nature of semantics. This is no accident, as the various externalisms sometimes come packaged together, likewise for the various internalisms. But the disagreements are, or seem to be, dissociable. So our first order of business is to clarify our main topic further by distinguishing it from these other externalist/internalist debates.

Let’s first distinguish descriptive and foundational semantics (Stalnaker 1997; cf. Kaplan 1989 on semantics vs. metasemantics). Descriptive semantics asks what semantic properties expressions have. Foundational semantics asks in virtue of what they have them. A candidate descriptive semantic claim, in this case externalist, might be: “Hund” means dog. Foundational semantics assumes that this descriptive fact (granting it is one) obtains in virtue of something more basic. It assumes that a dependence relation obtains between semantic properties and more fundamental non-semantic properties (the supervenience base), and it inquires into that relation. A candidate foundational semantic claim, again from an externalist, might be: The descriptive semantic fact that “Hund” means dog obtains in virtue of a convention among Germans, involving attitudes and expectations they have towards one another (cf. Lewis 1975). Note that the foundational question is not about causes (though answers may invoke them), but about constitution. Phylogeny, glossogeny, and ontogeny might bear on how it came to pass that “Hund” means dog, but it’s a further question whether elements of the causal story are also among the supervenience base for “Hund” now having that semantic property. (Compare: this is a water molecule in virtue of the relations among the hydrogen and oxygen atoms that constitute it, however they came to be so arranged.)

Distinct externalist/internalist debates correspond to the descriptive/foundational distinction. Our focus is the descriptive externalist/internalist debate: whether intentional relations to aspects of the world are among the semantic properties of linguistic items. The foundational externalist/internalist debate concerns what makes it the case that those properties obtain—in particular, whether the supervenience base for semantic properties includes only properties internal to language users or rather also includes relations to things external to them. It is this second debate that philosophers of language usually have in mind when they talk of semantic externalism and internalism. A typical foundational externalist allows that neurophysiological twins can have otherwise identical lexical items with different semantic properties; what makes the difference must thus reside in some relation to something external. Hilary Putnam (1975), for example, argues that what a speaker’s natural kind terms refer to depends in part on the speaker’s natural and social environments. A typical foundational internalist denies this.

Putnam argues for foundational externalism by arguing that a descriptively externalist semantic property—the reference of natural kind terms—does not supervene on an internalist supervenience base. His view is thus both foundationally and descriptively externalist. But it would seem that (pending substantive argument to the contrary) a descriptive externalist may be either a foundational internalist or a foundational externalist—similarly for descriptive internalists. The debates thus cross-cut. Various descriptive externalists (e.g. Segal 2000) have challenged Putnam’s arguments for foundational externalism, maintaining instead that speakers’ intrinsic properties fix the reference of their terms. Likewise, it’s a possible position (though proponents are not readily found) to deny, with descriptive internalists, that linguistic items stand in representational relations to worldly items, but also hold, with foundational externalists, that the non-intentional semantic properties expressions do have are determined in part by (non-intentional) relations to things in the world. For example, one might hold that linguistic items are computational mental structures without representational properties, but also that computational structures are themselves individuated in part in virtue of the mind/brain’s relation to its environment (cf. Bontly 1998, Ludlow 2003a).

These externalist/internalist debates—descriptive and foundational—both concern the semantic properties of linguistic items. It’s worth noting that parallel debates arise concerning the properties of non-linguistic items, particularly non-linguistic mental structures: what, if any, representational properties do they have, and in virtue of what do they have them? In the introduction, we said in effect that linguistic descriptive internalism needn’t address whether the concepts and thoughts that they take linguistic items to activate or express are themselves representational. Having drawn the descriptive/foundational distinction, we can now add that linguistic descriptive internalism also needn’t address foundational questions concerning what makes concepts and thoughts have whatever properties they do. (For a seminal extension of Putnam’s linguistic foundational externalism to mental content, see Burge 1979; cf. also Burge 2003 on Chomsky’s internalism more generally.)

Though one can contrast linguistic items with concepts and thoughts, it doesn’t follow that linguistic items are not mental structures (of some other kind). This brings us to yet another debate to which the label “externalist vs. internalist” gets applied—now concerning what words are, as opposed to what semantic properties they have.

Chomsky (1986) famously distinguishes two conceptions of language. An I-language—or “internalized” language—is the computational system, realized in an individual’s mind/brain, that generates the structures specifically implicated in linguistic behavior. An E-language is an “externalized” language that is understood independently of the properties of the mind/brain. Chomsky introduces the notion of E-language in discussing proposed technical constructs in the scientific study of language. But it also naturally comprises pre-theoretic conceptions of languages—English, German, etc.—as existing external to and independently of any particular person.

Chomsky argues that I-languages are the proper target of scientific inquiry: E-languages, among other things, have insufficiently clear identity conditions, while a focus on I-languages opens fruitful lines of inquiry concerning acquisition, the causal basis of linguistic behavior, the interface of linguistic competence with other cognitive capacities, cross-linguistic commonalities and variation, etc. (See Wiggins (1997) for a defense of E-language, and Heck (2006) in reply; see also Stainton (2006) for an overview of Chomsky’s views on these and related matters.) But debates about the proper object of linguistic inquiry—I-languages or E-languages (or both)—again seem to cross-cut our primary concern, at least pending substantive argument. Various authors (Larson and Segal 1995; Ludlow 2003b, 2011) explicitly present themselves as providing a descriptive externalist semantics (a characterization of intentional word-world relations) for the structures generated by an I-language. Others (Brandom 1994) embrace a social, E-language conception of language while eschewing representational properties in their theorizing.

Finally, the externalist/internalist debate is sometimes formulated as one concerning what kind of thing a meaning is: whether it is something that can be external to the mind/brain (the referent of a term) or something internal to it (a concept). But descriptive externalism need not commit one to external meaning entities. Some descriptive externalists—notably Donald Davidson (1984, cf. also Lepore and Ludwig 2005)—eschew meaning entities altogether as obscure and explanatorily otiose. The claim is not that the putative referents of words—tables, rocks, Donald Davidson—don’t exist, but rather that there are no entities that are meanings; so, in particular, those entities aren’t meanings. Moreover, a meaning-entity internalist can embrace descriptive externalism. Elbourne (2011), who casts the externalist/internalist debate in meaning-entity terms, favors meaning-entity internalism (meanings are concepts), but maintains that meaning-entity externalists and internalists agree that the meanings of words—whatever they are—are what enable words to “hook on” to the world. (Incidentally, instead of identifying meanings either with referents or with concepts, one might suggest that they are the combination of the two, with concepts determining reference. Cf. Frege (1918), albeit this use of “concept” aligns more with his “Sinn” (sense). His use of the term “Begriff” (concept) corresponds more closely, albeit imperfectly, to our “property.” We should note also that some take concepts to be non-spatial abstracta, and so in that sense neither internal nor external.)

These other ...

Table of contents

• Cover Page
• Half Title Page
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• List of figures
• List of tables
• List of contributors
• Acknowledgements
• Introduction: Semantics – a theory in search of an object
• Part I Foundational issues
• Part II Approaches
• Part III Meaning and conceptualization
• Part IV Meaning and context
• Part V Lexical semantics
• Part VI Semantics of specific phenomena
• Part VII Extensions
• Index