Donnellanian Implicature

Speranza

Why Donnellan chose to refute Strawson (and Russell) -- in their "On 
referring" and "On denoting" respectively -- is an interesting step in the 
history of analytic philosophy alla Grice.

Indeed, for Grice, the debate around what Whitehead and Russell call 'the 
iota operator' and which roughly translates as "the", runs along the lines
of  his 'Logic & Conversation'.

It is solved by posing, precisely,a distinction between utterer's  meaning
- the conversational implicature-level (what a neo-traditionalist like 
Strawson, an 'informalist' or 'neo-traditionalist' is attempting elucidate, but 
failing ) - and the truth-conditional semantic component (the Russellian 
component of the 'modernist' or 'formalist'). We cannot speak of 
truth-functionality here since the iota operator is NOT a truth-functional  operator.

Indeed, in the second William James Lecture, among the formal  devices
mentioned by Grice is the 'iota operator' ('(iota x)') and its  natural-language
counterpart, 'the'.

He goes into detail on these matters, not so much in the William James 
lectures, but in his "'Presupposition & Conversational Implicature" (where  he
refutes Strawson's solution involving a truth-value gap), "Vacuous Names"
and  "Definite descriptions in Russell and in the vernacular".

Grice develops a conversational-pragmatic treatment of

i. The king of France is bald.

in terms, not of truth-value gaps as Strawson would have it, but standard 
truth-conditional semantics (the Russellian tripartite analysis of the 
above) PLUS conversational implicature - notably by appeal to the 
conversational maxim falling under the category of Manner (or Perspicuity):  "Frame
whatever you say in the form most suitable for any repl that would be  regarded
as appropriate".

As if a conversation would proceed:

ii. A: The king of France is bald.
    B: He is not!
    A: Well, he doesn't have any hair!
    B: That's not the king of France. He is the  PRESIDENT of France: De
Gaulle.

Strawson's original example was

iii. The king (or president) of France is _wise_.

but Grice preferred to stick with Russell's original example, since it was 
trickier.

Russell played with

iv. The king of France is bald, but he wears a wig.

This inspired Dummett, who expanded on the scenario.

v. For that matter, we might just as well claim that the Queen of England, 
Elizabeth I, was bald, and wore a wig. (I mean, how can we verify the 
past?)

Grice's treatment, which also avails of the notion of "common ground 
status" something like "mutual knowledge" -- only that it can be false  -- as
discussed by philosophers more in connection with 'presupposition'  (notably
the ontological commitment of existential presupposition) rather than 
'definite description,' though.

But surely Grice is trying to show that Strawson's 'presupposition' does 
not exist, and it's a mere conversational implicature (Indeed, Strawson used 
'imply' instead of 'presuppose' in "On referring").

In "Vacuous Names", an irreverent tribute to Quine repr. in "Words and 
Objections", Grice considers the
issue, developing a formal system for the  treatment of conversational
pragmatics, which he calls system Q (later  re-labelled system G by Grice's
disciple, George Myro).

Grice makes some interesting remarks re: the pragmatics of  descriptions,
as they may trigger the use of names.

This he does in the tenth section, entitled, 'Names & Definite 
Descriptions'.

In 'Definite Descriptions in Russell & in The Vernacular' he goes  on to
defend Russell explicitly. Grice implicates Russell did not speak the 
vernacular as well as HE did!

Bealer, who went to Reed, has dwelt on these matters in his "Quality and 
Concept". Bealer relies on the notion of a pragmatically complete vs 
semantically incomplete symbol:

Bealer writes:

"On the  picture that emerges [from Grice's work], although definite
descriptions  usually DO HAVE A REFERENCE, REFERRING, unlike NAMING, is a
PRAGMATIC  relation, not a semantic relation. Thus, definite descriptions, while 
pragmatically complete symbols (they typically refer in conversational 
contexts), are semantically INCOMPLETE: their being co-referential in 
conversational context does not make them alike in any kind of genuine  semantic
meaning. Even if definite descriptions were taken as semantically  COMPLETE
symbols, one would intuitively not want to say that they NAME  anything (only
names name).  They would be more like predicates and  sentences: they would
express something, and what they MEAN would be what  they express. But what
about REFERRING? True enough, if definite  descriptions were semantically
COMPLETE symbols, referring would seem to be  a semantic relation. But it would be
only a DERIVED relation, defined as  follows. E refers to x iff
1. If E is a definite description, x = whatever E  EXPRESSES.
2. If E is a name, x = whatever E NAMES.
This would be  all there is to the commonsense theory of REFERENCE since
predicates and  sentences intuitively do NOT refer. On this account, then,
there is still  only one fundamental kind of meaning, and it partitions into
naming and  expressing. ... [In this account] Our model structures would need
to be  enriched with a PRIMITIVE DEFINITE DESCRIPTION OPERATION and
appropriate  predication and relativised predication operations.

From a historical point of view, we should add further keywords here: not 
just GRICE and DONNELLAN, but KRIPKE and BARCAN MARCUS, since the whole
picture  should provide a detailed historical development of these views.

Cheers,

Speranza