IMPLICATURE,EXPLICATURE AND TRUTH-CONDITIONAL SEMANTICS

The following essay will be a summary of R.Carston's article of the same title (the first 23 pages)

As Wilson and Sperber, Carston starts off by questioning whether we can in fact draw a coherent distinction between what is said explicitly and what is said implicitly as Grice assumed we could.

It is a pre-requisite for Grice that we have access to the explicit since we cannot work out the implicit without first knowing what is said (the explicit). The two together exhaust the "propositional significance" of the utterance.

It has generally been believed that the explicature is

obtainable without problem by accessing the conventional sense of the linguistic form, together with filling in elliptic material, assigning reference and disambiguating where necessary.

Carston questions this with the following example

1) A: How is Jane feeling after her first year at

University?

B: She didn't get enough units and couldn't continue.

Speaker A is likely to interpret B's response as "meaning"

2) Jane didn't pass enough University course units to

qualify for admission to second year study, and,

as a result, Jane cannot continue with University

study. Jane is not at all happy about this.

Carston asks how much of this interpretation is explicit and how much is implicit. Leaving aside problems of reference assignment (she) and disambiguation (get, units) which we will for now treat as explicit, then, "Jane isn't feeling happy", is surely implicit since it doesn't get explicitly mentioned in 1B).

But what about "to qualify for admission to second year study" or "with University study" or "as a result". If we were to say that they too are part of the implicit meaning for the same reason as above (not explicitly mentioned), then what then has been explicitly said?

The answer would be "Jane didn't pass enough University courses and Jane cannot continue [something]". But this string of words doesn't constitute a propositional form (it can't be given a truth-value) and nor could the implicit material which would be "with University study".

NB. NBR has pointed out to me that even if we can't always assign reference or disambiguate, it may not follow that we can't arrive at a proposition. We could still have general propositions such as " x f(x). I do not pursue this point here)

Carston wants to find a definition for explicatures (and following from it implicatures) which doesn't fall foul of problems such as above (non-propositionality and obsolete explicatures contained within the implicature) She suggests that we must allow the explicature to be more than the linguistic sense plus disambiguation, reference assignment and the filling in of ellipted material. Her evidence is set out below.

PARTICULARIZED IMPLICATURES

In order to begin her work, Carston looks at the Grician idea of CI. A CI for him has a number of properties.

Firstly, it has to be "cancellable". Consider

1a) Smith doesn't seem to have a girlfriend these days

1b) He's been paying a lot of visits to NY lately.

Here, 1b) seems to implicate 1c)

1c) Smith may have a girlfriend in NY

But if the context had been different (2a), (1b) would have a different CI. Consider

2a) I believe that Smith is looking for a new job.

1b) He's been paying a lot of visits to NY lately.

Here, the CI is sooner 2c).

2c) Perhaps he's found one in NY

In brief, the CI (1c) is cancellable. (However note NBR who says that this is not cancellability since the CI doesn't get changed in the same context. Perhaps the term "context cancellable" would be better).

Secondly, the CI must be calculable by reference to the conversational maxims. If it is calculable by any other means such as logical entailment or deduction, then it is not a CI (which is confusingly sometimes called non-demonstrative for this reason). Consider

3a) Have you read Susan's book?

3b) I don't read autobiographies.

From 3b), we can calculate 3c and go on to deduce 3d.

3c) Susan's book is an autobiography (implicated premise)

3d) The speaker of 3b hasn't read Susan's book (implicated

conclusion)

3c is a CI (a hypothesis) while 3d is a demonstrable logical deduction.(and therefore not a CI).

This example lets Carston demonstrate a third property of CI. Note that 3c is not entailed by or entails 3b.

NB. Carston allows the "conclusion" to be an implicature.

NBR would, I believe, only allow the implicated premise.

She calls this the criterion of FUNCTIONAL INDEPENDENCE of explicatures. This criterion allows us to rule on our earlier dilemma. In 1b in the first section, we couldn't decide what was the explicature and what was the implicature.

Carston wants to "expand" the explicature to a proposition which isn't entailed by the implicature (or else it would be redundant). This proposition would be all of 2 except the last sentence.

Such a criteria (FUNCTIONAL INDEPENDENCE) is the main one since ambiguity and reference assignment are always [context] cancellable whether we have an explicature or an implicature. Carston believes that such a criteria can be shown to be the consequence of a more deeply-rooted human principle (RELEVANCE) which I have already covered (see W & S essay).

GENERALIZED IMPLICATURES

Returning to Grice's concepts, he would probably agree that generalized implicatures are "carried by saying that P in all contexts except where there is something specific present to block/cancel that GCI. Hence

1a) She gave him the key AND he opened the door.

generally implicates

1b) She gave him the key AND THEN he opened the door.

I will here take it for granted that "and" isn't merely semantically ambiguous. (Ambiguities are language specific and unamenable to pragmatic explanation)

But given that in the Grician model, it is necessary to disamgbiguate and assign reference (via the [pragmatic] maxims) in order to access the explicature, why are we so sure that our reading 1b is pragmatically implicit and not explicit? The co-reference of "him" and "he" are no more given in the semantics than the temporal sequentiality of the first and second verb are.

Bearing in mind Carston's earlier remarks that the explicature must not be entailed by the implicature, it is clear that she would want 1b as the explicature (otherwise if 1b was taken as the implicature, it WOULD entail the explicature 1a).

Where "and" is said to indicate causal relations as opposed to temporal ones (as above), the debate is even more heated. Consider

2a) Mr.Jones has been insulted AND he's going to resign.

2b) Mr.Jones has been insulted AND AS A RESULT he's going to

resign.

Whereas ellipsis, ambiguities and reference assignment are signalled by the grammar which tell us there is additional work to be done in order to reach the explicit proposition, there is no such signal telling us to postulate a causal relation (or any other) in 2a. Carston dubs this criteria as the LINGUISTIC [or GRAMMATICAL] DIRECTION PRINCIPLE RULE. However, in the above example (2a & b), this principle is at odds with FUNCTIONAL INDEPENDENCE criteria.

In order to decide which criteria/ principle might be favoured, Carston wishes to look a little more closely at explicatures.

EXPLICATURES

In the Grician framework, the explicature must be propositional (and hence truth-conditional) for only then can its implicatures be worked out. (A certain part of the explicature might be non-truth-conditional [conventional implicature] but it is not propositional alone). This means that in addition to conventional meaning, just those pragmatic aspects of an utterance needed in order to arrive at a proposition were counted as part of the explicature (i.e. ellipsis, disambiguation and reference assignment). All other pragmatic aspects and inferences were counted as implicit.

On this account, even 1b would not be part of the explicit meaning since the two conjuncts could be given a truth-value irrespective of the order of events. If we were to refer to this idea of truth-evaluability ( fully propositional) as THE MINIMAL TRUTH-EVALUABILITY CRITERIA, then we could observe that this criteria and the LINGUISTIC DIRECTION PRINCIPLE generally coincide, the first being a psychological theory and the second its linguistic counterpart.

But earlier on, we saw that the LINGUISTIC DIRECTION PRINCIPLE was at odds with the FUNCTIONAL INDEPENDENCE CRITERIA. If it happens that we discard the LINGUISTIC DIRECTION PRINCIPLE, then we are in effect discarding the psychological MINIMAL TRUTH-EVALUABILITY CRITERIA as well which was equivalent (in one respect) to the LINGUISTIC DIRECTION PRINCIPLE.

This is exactly what Carston wants to do, for she believes that the MINIMAL TRUTH-EVALUABILITY CRITERIA is not a plausable psychological theory of utterance interpretation. If she is correct, then, there is a difference between the domains of grammar and truth-conditional semantics.

To highlight why Carston comes to this conclusion, she uses examples where the utterance has a truth-value (i.e. is a proposition) after disambiguation, reference assignment and the filling in of ellipted material but yet the proposition (in her opinion) is still too underspecified to be taken as the explicature of the utterence e.g.

1) The park is some distance from where I live

Carston believes that if we were to accept 1 as the explicature (in its entirety), then the speaker would be disregarding the principles of relevence and informativeness (since she argues that parks are generally some distance from where people live). Therefore, whatever the psychological implicature (a la Grice) of 1 is, becomes the explicature for her. This is something like

2) The park is furthur from my house than you thought

It is noticable that 2 entails 1 (Functional Independence therefore says that 2 is explicit)

I am unconvinced by this particular example which extends the explicature of 1 to 2. "Some" is here probably stressed and therefore a synonym for "rather a long" and so IS informative. Also, although I agree with relevance and informativeness principles, I don't see why they must operate on the explicit level. The relevance and informativeness can be on the implicit level.

Another Carston example is

3) He ran to the edge of the cliff and jumped

which for Carston by FUNCTIONAL INDEPENDENCE has as its explicature

3a) He ran to the edge of the cliff and jumped off it

She argues that the verb "jump" is intransitive and the additional PP in 3a isn't indicated by the grammar. Therefore, just as in this example we think it quite natural to do some extra work on the explicature, even if "and" in 2a of the previous section doesn't indicate that additional work need be done, we can still read into it causality and then postulate 2b as the explicature since it entails 2a.

We have seen then that according to Carston, the Grician implicature is often an explicature in her model. The explicature is expounded as it were at the expense of the implicature. What sets the upper limits to this expansion?

We know the criterion Carston uses to trigger this expansion (Functional Independence). It is the same idea that interestingly sets the upper-bounds on what the explicature is. Grician implicatures become Carstonian explicatures where there is a relationship of entailment.

Consequently, where there is no such entailment (i.e. where the logical forms are independant, the two propositions are deemed to be different...one is then the implicature and the other the explicature.

On reflecttion, we would not expect Grician implicatures to semantically contradict their explicatures (cases of irony and sarcasm aside). Of course, this is compatible with some Grician implicatures entailing their explicatures. Where this is so, they are for Carston explicatures.

However, NBR has pointed out a problem for Carston with regards to Relevance Theory and GCI's.

Relevance Theory should, according to NBR only apply to PCI's since relevance is by definition, relevant to a particular context. Carston must handle GCI's by another method OR dismiss them as GCI's. She chooses the latter solution. Her CRITERION OF FUNCTIONAL INDEPENDENCE effectively "converts" GCI's into explicatures. But is this correct in the case of all GCI's. Let us consider another sort of GCI's not considered by Carston (numerical GCI's).

A) I've got four pens utterance

B) I've got at least four pens explicature

C) I've got no more than four pens GCI

D) I've got exactly four pens Propositional

significance

If the above is taken as correct, it is immediately noticable that the implicature does not necessarily entail the explicature (and hence the CRITERION OF FUNCTIONAL INDEPENDENCE does not apply). Carston, according to NBR, has mistaken the propositional significance for the GCI.

TRUTH-CONDITIONAL SEMANTICS

In the title of her article, Carston includes the words "truth-conditional semantics". In her section (same title as this one), she goes on to look at the effect of her adopted FUNCTIONAL INDEPENDENCE criteria on our view of truth-conditional semantics.

Grice was interested in attributing as much as possible to the pragmatics in order to leave a natural language semantics identical to formal logic (and, or, if, neg).

He also wanted to distinguish the truth-conditional aspects of an utterance from the non-truth conditional ones. e.g. P & Q are taken to have an identical truth value to P but Q. This shows us that not all grammatically given information is truth-conditional.

Now, if Grice was right, then there should be no explanation of the following

1) It's better to meet the love of your life and get

married than to get married and meet the love of

your life

This is equivalent to saying "better P & Q than P & Q" according to a semantic theory of "and" not supplemented by pragmatics. Yet we feel that 1 says more than this apparent contradiction. This prompted Gazdar to say

"There is increasing evidence that the semantic component of the theory (of meaning ) must sometimes have access to the pragmatic properties of constituent clauses when assigning the truth conditions of compound sentences"

Carston concludes that

"This evidence indicates that the semantic component is not autonomous with respect to the pragmatic component".

NB. For NBR, the literal contradiction of 1 and similar sentences is the trigger for pragmatic re-analysis and metalinguistic negation. Carston is saying that the semantics in 1 is OPERATING on the pragmatic level. These views aren't compatible.

Carston then adheres to two kinds of semantics. There is the usual formal logical semantics and there is the pragmatic semantics seen in examples such as 1. She names the first one LINGUISTIC and the second one TRUTH-CONDITIONAL. LINGUISTIC semantics is autonomous with respect to pragmatics and is the input to pragmatic processes. The two together (linguistic semantics and pragmatics) make propositional forms which are the input for truth-conditional semantics.

Such a view is a step away from the old formuala that pragmatics = semantics - truth conditions.