[There may be some minor typos resulting from the translation to ascii --Rick]
By Joe Shear.
Notes on Evans's Identity and Predication
The paper is divided into four parts. Summary and Discussion of parts I
and II follow. Parts III and IV are forthcoming (this evening). Introduction:
Quine's notorious 'indeterminacy of translation' thesis states that 'manuals
for translating one language into another can be set up in divergent ways,
all compatible with the totality of speech dispositions, yet incompatible
with one another' (WO, 27). The indeterminacy thesis was intended not only
to characterize translation, but also (and more importantly) concepts of
the theory of meaning (reference, denotation,extension, and the related
notions of satisfaction and truth). Evans turns to, and shall reject, a
Quinean argument for the indeterminacy in the theory of meaning in section
I. But before doing that, Evans spends the Introduction distinguishing and
spelling out the relations between TRANSLATING and THEORIZING ABOUT MEANING.
The considerations of the paper are at the level of the theory of meaning.
A translation manual must be finite as well. But the constraints involved with the demand for finiteness are less stringent. The translator must do the best he can to construct a coherent translation that makes sense of the data. Often this will involve mappings of single lexical items ("bird" to "l'oiseau"), but even when the translation focuses on the single parts of a sentence, the concatenation of those parts to form a sentence need not be deductive. The translator is not subject to the demands of semantic explanation.
Evans closes the introduction by noting the connection between the indeterminacy of translation and the indeterminacy of the theory of meaning:
"For, after all, although not every translational manual must provide the basis for a theory of meaning, every theory of meaning must provide the basis for a translation manual, since it must, in effect, pair sentences of the jungle language with equivalent sentences of the theory's language. And in so doing, the semanticist must draw upon exactly the same data, of native assent, dissent and unsolicited utterance, which are availabe to the translator." The theorist of meaning is a radical interpreter in the same boat as the translator.The theory of meaning can be set up in divergent ways, assigning different semantical properties to the parts of sentences. But again, these assignments, however they run, must deductively entail the meanings of the sentences of which they are parts. [[This section seems to presuppose a lot of phil-languuage background, and given my lack of it, the following (rather specualtive) discussion attempts to work through some of the assumptions and connections in this introduction. Is the explanatory constraint on the theory of meaning that it must capture the semantic contribution of sentence parts to explain sentence meanings motivated by the observed productive practice of comptetent speakers (as I imply above)? That is, is the observed competence the transcedental fact, and the conception of semantic explanation (and its putative explananda) the condition of the possibility of that fact? If so, perhaps there are alternative modes of explaining this fact that need not committ the theorist of meaning to the explanatory strategy. That is, perhaps my ability to understand unencountered sentences rests on an ability other than my acquaintance with the parts.
More problematically, if the alleged fact is this - that as a competent speaker I can understand unencountered sentences composed of familiar parts in ALL possible sentential contexts (generality constraint) - perhaps there is room to deny the fact (the explananda) . If so, given that "the semanticist aims to uncover a structure in the language that mirrors the competence speakers of the language have actually acquired", perhaps that structure, semantic explanation, and thus the theory of meaning, must be rethought. The way I have set things up above, the motivation for the brand of semantic explanation advocated by the post Fregean era is an observed fact about linguistic/communicative practice. But Cussins' (1992) diagnosis of the dominance of this mode of semantic explanation makes it seem that I have gotten things how might he say? "back to front". That is, the conception of semantic explanation, and its accompanying conception of communicative/thinking practice, follows not from a fact culled from 'observing practice', but rather from certain assumptions bound up with the logical tradition, certain assumptions about what semantic explanation SHOULD be, independant of bumbling semantic practice. I won't go into that here, because we have that article coming up. But at least this much about logical motivations is at the surface of Evans' introduction.The theorist of meaning must explain the inferential relations (their validity) of sentences composed of concepts (expressions), concepts that can enter into many positions within sentences. One's grasp of concepts (their truth conditions) underwrites the ability to carry out valid inferences. Thus semantic explanation is the assignments of semantical properties to concepts which are the components of sentences that are connected when engaged in inferential activity. What, then, is the link between the productive competence of the speakers (communicative practice) and the inferential relations (formal logic practice) such that the theorist of meaning, in giving the semantic explanation, can explain both? The link seems to be the grasp of concepts, or what Evans will later call identity conditions of objects/properties.]]
Back to the paper.
I. In this section, Evans wants to pull apart Quine's marriage of predicational structure and Quine's 'apparatus of individuation'. He shall then offer what he takes to be the primary role for predicational structure in semantic explanation: the explanation of how the truth conditions of sentences depend upon their parts. Granting that there is indeterminacy in the theory of meaning, Evans wants to take issue with Quinean argument used to establish this indeterminacy. Such a taking issue, if successful, will establish that Quine's indeterminacy is not as gross as is typically thought :
[The apparatus of individuation as described in a Quinean inscription: "Our individuating of terms of divided reference, in Englsh, is bound up with a cluster of interrelated grammatical particles and constructions: plural endings, pronouns, numerals, the "is" of identity, and its adaptions "same" and other". It is the cluster of interrelated devices in which quantification becomes central when the regimentation of symbolic logic is imposed." For Quine's considerations above, especially premise (iii), see p.33 of Ontological Relativity.]
The remainder of the paper is an exploration of the idea that discerning
predicational structure in a sentence is to explain how its truth conditions
depend on their parts. What kind of explantion is afforded and what makes
a sentence be one that demands such explanation? A way in to these questions
is provided by a kind of language (languageth featureth) in which predication
is noticably absent in its theory of meaning.
II.
But the theorist can find no context in a pure feature domain for there are no identity conditions for objects. To explain the semantics of 'Red Water' via a predicational construction would be allow for sentences such as "There is water here which is not red." to bear truth conditions. But this rich of a semantics is not warranted by the assent to spatial overlap of compound features. What would warrant such a semantics? That is, what constitutes the satisfaction of identity conditions such that a compound (GF, where G is a rabbit and F is a feature) could count as predicate construction?
"What is required for many of these is that the F feature be distributed in a characteristic way in relation to the boundaries of a single object whose presence prompts assent to the queried G term." Senstivity to the boundedness of the object warrants explanation by the predicate semantics because it allows for the possibility of contradiction. With a bounded region of water, [Red Water] and ~[Red Water] yields a contradiction. Either the red feature is distributed across some space of the bounded region or not; can't have it both ways. This is the connection fundamental connection between identity and predication. This connection is 'a matter embedded much deeper than Quine's talk of jiggling with the translation of the individuative appatatus would lead us to believe."
Section II ends with two rich quotes that I will reproduce in full:
and...
"To say that an expression has a particular divided reference makes sense only in the context of the explanation of compound sentences. To decide that a term divides its reference over rabbits is to decide in which it occurs involves predication of rabbits. And to decide that a set of sentences involves predication is to identify the way those sentences assent conditions are generated from their parts as depending on the identity conditions of rabbits, and so systematic mastery of those sentences requires mastery of the identity conditions of rabbits."
The defense of the semantical theory against Quinean alternatives takes up the rest of the paper. Those notes are forthcoming. I'll close this section with a quote from Frege found in fn 8 of Evans' Varieties of Reference (p.11): "Where inferences are to be drawn...it is essential that the same expression should occur in two propositions and should have exactly the same meaning in both cases. It must therefore have a meaning of its own, independant of the other parts of the proposition." (Philosophical and Mathematical Correspondence)
Before moving on to section III, there is one quick matter in section
II
I would like to remark on. The matter arises in section 2 of II (pg.36):
(i) the relations among predication, identity, and identity predicate:
Evans claims that we (the radical interpreter) can "recognize"
predication
without identifying an identity predicate in the alien language.
However, when we do recognize an identity predicate, we do so in virtue
of the
inferential relations among the identity sentence and other predicative
sentences. Thus it seems that, on Evans' view, the apparatus of the
predicative structure -- and the inferential relations that it involves
-- is explanitorily prior to the identity predicate.
But recall that IDENTITY, where identity is construed as characteristic
of an expression that divides its reference, makes sense only in
the explanation of the truth conditions of compound sentences. My worry
is this: if we can recognize predication without identifying an identity
predicate, that means that we have already observed a sensitivity on the
part of the natives to withold contradictory predicates to the bounded
things identified. And if we have observed this sensitivity, that means
we have already identified (at least) a scheme that
includes the identity predicate, for it depends on identity conditions of
objects. But we have not yet necessarily identified THE identity
predicate, however certain we are that it lurks in the background as part
of the semantic explanation when the appropriate context and behavior is
exhibited. Which is Evans' point, so I guess my worry is not a worry after
all.
III. Evans has advanced a proposal about the structure of semantic
explanation, a proposal that anchors the predicative scheme by the
manifest sensitivity to the identity conditions of an object, i.e. its
boundaries that set it apart from other things. But Quine has offered
alternatives --e.gg. rabbit stages and rabbit parts -- to the Evansionian
(object-dependant) link between identity and predication that Evans'
wishes to defuse.
Take the expression "rabbit" and treat the expression as a
singular term
designating the universal rabithood. Can this proposal capture the
predicative scheme? Evans thinks not. "A white rabbit" would translate
into
'Rabbithood has a white instance here", but "non-white rabbit"
into "
Rabithhood has no white instance.", and these are not synonymous. The
universal rabithood cannot generate the appropriate truth conditions.
Another alternative is to treat 'rabbit' as denoting "rabbit fusion".
Recall that the satisfaction condition for the whiteness feature for
Evans is the distribution of whiteness within the boundary possessed by
the rabbit. What is the satisfaction for the feature over rabbit fusion?
"iff x has a white part" is too weak because of white-feeted brown
rabbits and 'iff x has a white, rabbit-sized part' is too weak because of
the collection of white tails of brown rabbits that form a contigous
mass.
How about having rabbit divide over rabbit parts? We must change the
satisfaction condition of whiteness such that things that are only parts
of the white rabbit satisfy the predicate. But if the satisfaction
condition, then, is '...iff x is part of white rabbit' , that would lead
to
"White hut" only being satisfied if it was part of white rabbit.
White
should be able to couple with other terms without being wedded to
rabbits, but it can not on this scheme.
The more major difficulty is that changing the satisfaction condition
of whiteness from 'rabbit' to 'part of a white rabbit' results in
"different parts of the same rabbit are indistunguishable by the
predicates of the language." This somehow contradicts some Quinean
comittment about an 'absolute and objective' criterion for what two-place
predicate to count as the identity predicate. [I don't really understand
this. It just doesn't sound right given Quine's comittment to
indeterminacy. In fact, I don't really understand much of the next few
paragraphs, so I am simply moving on to the more comprehensible section
4
of IV.]
Evans wants to get in one last push for his proposal by showing how
predicates involving demonstrative singular terms are also sentences
sensitive to identity. Say "This rabbit is F" to some English
speakers
when there is no rabbit in the presentational context, or there is a mass
of rabbits (with no one discernable), and the speakers are puzzled. Does
not this warrant the spotting ofa genuine singular term?
Quine resists for alternative hypotheses are equally as elgible. Evans
thinks this is mistaken, and it rests on two 'false assumptions'. The
first is that all acts of demostrative reference involve pointing, which
is notoriously indeterminate. The second is Quine's assumption that when
you point to rabbit, you are pointing to a rabbit stage or part. Evans
says you are rather pointing to many rabbit parts. Thus if 'this' is to
act as demonstrative s. term, fixed principles for dividing everything into
parts are presupposed. [Doesn't this beg the question against Quine in
favor of the demands of Evans' conception of semantic explanation?] Evans
wonders how sentences with demonstrative singular terms can be accounted
for accepting any gross indeterminacy of the reference of those terms.
IV.
Evans concludes by noting that the general pattern of difficulty with
Quine's semantic hypotheses is that the reference of rabbit is cut
either too course or too fine in comparison to the 'orthodox' theory. The
proposals that cut the reference too fine attribute unwarranted
dispositions. This contradicts the aforementioned constraint on
semantic explanation: that the attribution of semantic properties for
which there is no evidence is prohibited. The courser theories don't
account for the evidence that suggests a sensitivity to the identity
and distinctness of rabbits. When meaning is (plausibly) grounded in
identity and distinctness, and not the apparatus of individuation,
semantics are not so indeterminate after all. Three cheers for the
tradition of orthodox semantic explanation. [Chase?, I don't hear ya].