@Banno

I don't think I've ever really directly tackled how an interpreter should handle an utterance like

(1) Sure, if I reprehend any thing in this world it is the use of my oracular tongue, and a nice derangement of epitaphs!

and I think I can say why.

Let's say the utterance (1) presents a problem for an interpreter.

The question is who or what solves that problem?

It is easy enough to explain this feat on the hearer’s part: the hearer realizes that the ‘standard’ interpretation cannot be the intended interpretation; through ignorance, inadvertence, or design the speaker has used a word similar in sound to the word that would have ‘correctly’ expressed his meaning. The absurdity or inappropriateness of what the speaker would have meant had his words been taken in the ‘standard’ way alerts the hearer to trickery or error; the similarity in sound tips him off to the right interpretation. Of course there are many other ways the hearer might catch on; similarity of sound is not essential to the malaprop. Nor for that matter does the general case require that the speaker use a real word: most of ‘The Jabberwock’ is intelligible on first hearing. — p. 252

So there's what I've alluded to as how we would analyze (1) if we notice its problems, and it's relatively straightforward: System 1 comes up short ("realizes that the ‘standard’ interpretation cannot be the intended interpretation") and asks for help from System 2 which goes through all these analytical steps. If System 2 is involved, it's natural enough to say the interpreter, this person, consciously, solves the problem.

But if you don't notice and still land on the intended interpretation? Then the utterance has just been handled by System 1 for you without bothering to tell you it corrected an error in the utterance. Who solved the problem then? Or what? As you like. You can say "I did" or "my System 1 did" or, if you have a theory, you might say, "Thank you, Darwin", or any of a number of other things. I'm not sure there's an obvious right way to talk about this. I've suggested that we should just expect some robustness built into our language use as it is in any communication technology, and that means an allowance for errors and a capacity to correct them without fuss.

But what does Davidson say?

It's clear he's not interested in the straightforward problem solving described above (and paragraph after next he'll distinguish what he's after from error that is not "philosophically interesting"), nor does he seem much interested in whatever actually goes on in speech perception. Let's put it this way: not the conscious reasoning of System 2; not the unconscious processing of System 1. What does that leave?

It leaves no psychology at all, that's for sure, which is the point. What it does leave is the theory of meaning taken as unrelated to psychology entirely. We know what we're talking about there, for Davidson, but any sort of formal model of the semantics of a language will do, some Tarski or Carnap kind of thing.

It seems unimportant, so far as understanding is concerned, who makes a mistake, or whether there is one. — ibid.

Davidson is going to abstract the formal symbol system people use from the historical and psychological facts of their using it, which is no big deal, but he's going to do it a particular way: the interpretation of (1) is captured along with the utterance itself. But what's the status of that interpretation?

The coupling between the historical psychological facts of using a language and that language as a formal system is a little loose, at least in one direction: sentences may be given a non-standard interpretation, have a meaning that is not their literal meaning. But within the system itself, there is no such looseness; when Davidson captures (1) and its intended interpretation (whether you describe that as being expressed in a meta-language or in the interpreter's language) they are captured together as if quite tightly coupled, and then this version of the formal system is attributed to the interpreter.

Which, if you explained what you were doing to the interpreter, they would not accept, because the coupling was loose on their end, not tight:

"You assigned the meaning of 'epithet' to this utterance of 'epitaph'."
"Yeah."
"So in your language, 'epitaph' is a synonym for 'epithet'."
"No, of course not. That's why I had to work it out."

There's a similar story if the error correction was carried out by System 1, except the interpreter will protest that they didn't even realize Mrs. what's her name had said 'epitaph'.

The interpretation is the result of someone or something solving the problem presented by the defective utterance, but it will be captured by Davidson simply as an interpretation, slotted into a bit of model theory in the usual way with no trace of its historical psychological origins. That procedure might be fine for aggregating language use within a population, but then attributing this "passing theory" to a member of that population isn't self-justifying.

I was being picturesque.

Clarification, please — bongo fury

[ "abstract" == "fantasy" ] is picturesque.

Why are you pestering this thread when there's a perfectly good Platonist to argue with in another thread? Why should I have to deal with him there and you here, when you two have so much to talk about?

I hope that I have given it to you. — Tristan L

No.

If I don't take predicates as Properties that have independent existence, I don't have to take vacuous predicates as Properties that themselves have the Property of having no instances.

Vacuous singular terms (Santa Claus, the Bermuda Triangle, the present king of France) aren't going to do it either.

You might eventually recognize that in giving this argument you're standing right next to natural numbers and sets, and those actually do represent some kind of trouble for me, but it's trouble I already know about.

Well, it's been fun!

Exactly.

Anyway, proved my point. You called reference a fantasy; implication of choosing a word like that is usually that this should influence what we think about it, perhaps we should choose to rethink our reliance on it in our theories in favor of something more substantial, blah blah blah ---

but of course it turns out this is a picturesque way of describing anything abstract and reference is keeping pretty respectable company, or it least company none of us is ditching anytime soon.

Because the operation is a fantasy. — bongo fury

What's the intended force of this though?

Are you distinguishing reference from something we do with language that is not a fantasy?

If you are, what's that? If you aren't, why should we care?

manage their book "life." — tim wood

There is no evidence this is something I can manage. I read much less than I'd like, but in practice what that really means is that I read and don't read in waves. The book buying follows that pattern. Often I acquire books in chunks. Ran out of shelves so long ago I stopped caring almost as long ago. I do still do some deliberate and some incidental rearranging of what's on a shelf or in a box or stacked on a table based on my interest at the moment.

I read a goodish amount online, papers and such, but I still don't do ebooks and doubt I ever will. They're both just technologies and paper suits the way I use books much better.

When I was young, I was not above a little underlining or a star in the margin. Now, never. The only thing I still do, and that pretty rarely, is make my own index inside the back cover or thereabouts. (If there's a quote I know I'll want later, I jot a key phrase from it and the page number. There's usually only a few.)

Have hardly used a library since I left school. I have sometimes joked that I'm not even sure I can read a book that I don't own.

For me the comparative ease of getting books these days is staggering. I remember when I was kid ordering books from the university bookstore: I would dig through those huge volumes of Books in Print to find all the info needed to fill out a little card for the nice woman at the desk and then a month or so later I would get a postcard saying my book had arrived. There were out-of-print books I used to look for for years on end, checking the used bookstores in whatever town I was in just in case.

This is more properly formulated as — Tristan L

If this translation system works so well, I can just as well translate your Platonish into something else, like regular English.

From the beginning I've said that translating everything I say or anyone else says into Platonish proves nothing at all. Maybe some people find it persuasive but it should be clear by now that I don't.

You provided no argument that besides clouds there's something called "cloudhood". You've shown nothing that the term "cloudhood" would add to a discussion of clouds except phrasing that pleases you more. You've given me no reason, in any of your posts, for me to consider Platonism anything more than your preferred way of talking.

Do you have anything that might persuade me?

It's not a convention that what one says by way of a statement is an assertion — Banno

Or you could say, as LW might in different words, that it's the context in which you state that leads people to treat your statement as an assertion; you acting in a play and them being in the audience is not that context.

Sometime check out a beautiful little book by Ruth Krauss called A Hole Is To Dig. It's definitions offered by a kindergarten class made into a little picture book. Besides the title, we find "A hole is when you step in it you go down," and "There's a difference between pretending you're a lion and pretending you're really a lion."

Yeah. It's already obvious that kids are generalizing. ("I breaked it." Or Google the wug experiment if you don't know it.) Chomsky's idea was that if kids are not just parroting what they've heard, but using the sample to construct a system of rules, and then that system is what they actually use to produce speech, that would provide a way of overruling the erroneous examples a simpler theory might expect them to regurgitate.

Subsequent research clearly backs up the basic idea. No linguist today would imagine kids just repeat what they've heard. All the fighting since has been about the nature and extent of the rules, what sorts of things might just be memorized, etc.

That's my understanding.

I dunno. The way we all encountered that sentence, not just in the context of philosophical discussion, but even set off specially by itself as an example for us to consider, it's all terribly artificial.

Obviously this is true for our reactions to Davidson's examples. What you'd want is research.

What everyone is inclined to talk about is how they'd analyze a word substitution, but then draw conclusions about how we naturally hear and understand them. Waste of time.

Davidson's reliance on examples from fiction, humor, poetry, and very self-conscious philosophical discussion, does not inspire confidence.

Just to gesture back at the context of this paper again...

Chomsky somewhere said his entire life's work is organized around two problems:
1. Plato's Problem: how do we know so much, given so little evidence?
2. Orwell's Problem: why do we know so little, given so much evidence?

The observation that leads to Chomsky proposing the idea of linguistic competence is that children land, to greater and lesser degrees, on the same language, no matter the sample of the language they were exposed to (unless it's really degenerate), no two samples being the same, and despite the inevitable presence of errors in that sample.

The existence of speech errors was one of the motivators for the idea of competence in the first place, so it's just really odd for Davidson to come along thirty years later and say, "But what about speech errors?"

I'm just thinking we need to be careful here, and that phrase might have been conflating three different things.

There's no reason to avoid talking in terms of algorithms if you know for sure what you're getting into and what you're claiming.

A guy moving a stack of 4x4s by picking one up, carrying it across the yard, putting it down, and then coming back for the next one -- sure, he's got an algorithm. Linguistics models are full of algorithms too.

An algorithmic process — Banno

That looks like an infelicitous phrase to me.

Algorithm is very high-level, perhaps as high as you can get before purpose.

Below there's implementation, and separately there's execution.

The rules described in an algorithm don't necessarily appear in an implementation in an identifiable way, but are virtually present in it, structurally. Certainly if you go down far enough, they're nowhere to be seen.

That's to say, speaking of algorithms is always descriptive, descriptive at an awfully high level, and capable of directly attaching to the purpose of the algorithm. Describing something biological as implementing an algorithm is either picturesque or it's just to propose a model.

whether it was the underlined words that needed to be swapped, or the previous occurrences — unenlightened

That occurred to me, but there's also the "assigning a new role to 'epitaph'" at the end of the passage.

(Just for jollies, I emailed Lepore yesterday to ask about this. I don't think I'd be breaking confidence to quote in full the response I received this morning: "Thanks.")

But if one reads the passage out of context - without already knowing the direction of the paper, then one could not tell ... — unenlightened

I'm okay saying that a lot of this is above my paygrade. TGW used to say that a lot of philosophy is just bad linguistics; that only means we ought to be careful distinguishing the sorts of things linguistics is good at figuring out, or could conceivably figure out, from philosophical conclusions we might draw from those results. That said --

I can't think of a reason not to assume that speech production and speech perception grew up together, however that happened. It makes sense that such a combined system would find its way to Postel's law: be conservative in what you send and liberal in what you receive. Both are relatively slow and expensive though -- how conservative must I be? how liberal? So we'd expect them to move toward each other a bit, maybe quite a bit: I need only send as much as you require; you need only require as much as I send. And there again it's clear that we're looking at a system that has to adjust both sides systematically to find some workable equilibrium. That we have machinery that found such an equilibrium seems clear enough to me.

And that machinery allows for speech errors, context dependence of various sorts, and so on.

I think there's a hope, we might say, among philosophers that we could talk about language use, and reason about it, roughly the way we talk about any other acquired skill. Children aren't born with language, and later on they have it, so "acquisition" has occurred. We think about how a skill is built up through consciously labored over practice and ends with reliable and unconscious habit, and how it's possible to sort of run the tape backward and show what learnable specific steps are involved in playing the violin, writing code, driving a car, cooking dinner or eating it with utensils.

Only language is way harder than those. Teaching my kids how to drive, I can readily break down the various things I do and give them specific advice about what to do and how to do it. That takes some effort and attention, but it's not that hard, and I get to stop before I would have to tell them how to make their foot push down on something or how to make their hands hold onto something. To explain language use, to really explain it, we seem to need to go all the way down. There is no comfortable stopping point. And we cannot just slow down speaking or understanding and observe the steps we take when we're doing it because those steps turn out mostly never to have been conscious and are incapable of being made conscious. Attention to how you speak isn't quite worthless -- there are clues you can pick up -- but it's nowhere near as useful as attention to what you do when you're driving is for explaining that.

All that to say, I'm not sure any model of how we speak or how we understand speech that a philosopher could come up is much worth thinking about. Psycholinguistics will do its best and, if we're interested, that's what we should pay attention to.

If the existence of speech errors and their success tells us that some philosopher's view of meaning is faulty, I guess I can go along with that, but it's not like we should have expected it to work anyway if it wasn't grounded in science to start with.

So here's the question for me. Is this paper more like Jerry Fodor storming into the biology department to tell everyone there that natural selection is a bogus, circular concept? Or is it a contribution to formal semantics intended to be a piece of (something that aspires to be a) science? Or is it a piece of philosophy drawing conclusions from what we've learned from the work of linguistics? It's not the last one, for sure; it might be the second, and I think it's widely taken to be; but it looks more and more to me like the first.

The only thing in it I still find interesting is the argument from "getting away with it", which strikes me as of genuine philosophical interest, and a move I'd consider, no doubt too harshly, as typically Davidsonian.

The Flaming Lips — Jack Cummins

Tell everybody
Waitin' for Superman
That they should try to
Hold on
Best they can
He hasn't dropped them
Forgot them
Or anything
It's just too heavy
For Superman
To lift

theory — Janus

It's just shoptalk.

"Theory of meaning for a language" is just Davidson-speak for a semantics for that language.

Not expected to be something you're aware of unless something goes wrong that needs attention to fix.

I think "efficiency" is a key term that is missing here, as in efficiently using words to communicate, and that includes being able to interpret similarly sounding, shaped, and typed words in the improper context, as meaning the words that the they sound like in the proper context, so that they don't have to be repeated. — Harry Hindu

You may have something here. We regularly produce speech errors (I haven't found a solid source on the frequency). Why? Why isn't our speech production better at its job?

I would guess the answer is it's too slow and too expensive. Perfect is the enemy of good.

Given that speech errors are predictable, why don't we detect them more often?

Well, to what end? The choices seem to be:

• guess, hopefully well
• consciously work through the options and try to deduce the correct word, weighing evidence, etc.
• in conversation, you could produce speech of your own, asking for clarification and then you have to deal with the speech produced in response

The fastest and cheapest on this list is clearly the first, and we do have plenty of reason to think we are good at the guessing. (As Kahneman says, System 1 is a machine for jumping to conclusions, so it's raring to go.) So in essence you can just expect the guessing solution to swallow the error detection pass entirely.

Do you think it was deliberate? — unenlightened

No. But I'm inclined to doubt it was Davidson's mistake. It's just amusing.

PhilPapers links to another version (mostly) available through Google Books; it has the same transposition. The paper seems to have appeared simultaneously here (in what looks like a collection devoted to Grice) and in Lepore's collection devoted to Davidson (both 1986). Unfortunately the latter cannot be previewed in Google Books, so I can't check that one.

Someone made a mistake (later taken up by this reprint in The Essential Davidson) but it's unclear who.

But this is surely the death knell for any complete description. — unenlightened

Complete description of what?

Someone who grasps the fact that Mrs Malaprop means ‘epithet’ when she says ‘epitaph’ must give ‘epithet’ all the powers ‘epitaph’ has for many other people. Only a full recursive theory can do justice to these powers. These remarks do not depend on supposing Mrs Malaprop will always make this ‘mistake’; once is enough to summon up a passing theory assigning a new role to ‘epitaph’. — p. 262, my underlining

Here's a curiosity. I only noticed yesterday that the underlined words should be swapped around.

What are the rules for what can or cannot be thought of (understood, interpreted) as a variant? — unenlightened

While I take your point, what you quoted refers to a single and specific rule change that is widely made by players for speed chess: in all versions of the official rules (USCF, FIDE, whatever) it is simply illegal to leave your king in check, and the king is thus never capturable; it is very common to play speed chess (or "blitz") dropping this rule so that a game may end with the capture of a king.

It starts to look like even chess cannot be specified exactly; — unenlightened

You could think of it a little like how you specify things in general. In some cases, the specification is shockingly precise, far beyond what you might think of as the game's rules: the heights of the pieces, colors, all the details of the time control, scoring the game, touch-move, endless physical conditions like lighting, height of the table, the chairs (and for chunks of this stuff you can "thank" RJF below). World Championship matches are like this. Club play isn't. A casual game in the student lounge isn't. Weekend tournaments mostly aren't, but as the stakes and talent involved rise, things get more precise.

Is that a surprise to anyone? I've coached elementary school chess and they did some weird shit -- they were still, at least by and large, playing chess, and were certainly trying to play chess. The high schoolers I coached had very different issues. As a former weekend tournament player, frequent casual and blitz player, and coach at noticeably different levels, it never once occurred to me say that it was only coincidence or family resemblance that led me to call all these activities "chess".

I seem to remember endless negotiations about the fine details of grandmaster matches - between Fischer & Spassky and Fischer and Karpov. Making up the rules? — unenlightened

The explanation is just Fischer. An interesting but terribly sad story.

I have a mouth; caves have mouths, rivers have mouths. We all know what a mouth is - what is a mouth? — unenlightened

Indo-Europeans seem to have been pretty adept at analogy. We might even be able to figure out which one came first, but maybe not. (If indeed these are all the same word, which you can't know just by looking at the modern usage. Still, I suspect they are, and I'm not looking.)

This doesn't look a family resemblance concept to me, because it's so obvious how to extend it. (I think I extrapolated the pattern myself just in the last week, but I don't remember what I was talking about.) Jars have mouths. Canyons have mouths. Gun barrels and cannon have muzzles -- and that's related too!

But I suspect none of this addresses the point you were trying to make, if it's the one you made earlier in the thread: that practice comes first, theory after and always imperfectly. (All models are wrong, as the saying goes.) That's true enough, but, on the one hand, psycholinguists are trying to figure out the actual mechanisms that produce and consume speech, and, on the other hand, speakers of natural languages (let alone chess players) in the modern world are themselves aware of some theorization of their practice. Ordinary people argue about the right way to say things, and not just under the influence of school grammar (as @Dawnstorm helpfully reminded us) but because of their knowledge of the shades of meaning that distinguish words, elevation, tone, connotation, all that. Should they all just shut up and talk?

Sorry @Hirnstoff but you lost me 2 minutes in this time.

I know your heart is in the right place, but when it comes to the flat earth it is probably not the case that Larry wants to "know the truth"; Larry wants to make the truth.

Cloudhood need not be added to the discussion, for the meteorologist has already brought it in when defining what a cloud is. The only difference between him or her and the philosopher is that the latter is highly aware of cloudhood itself whereas the former only has a diffuse and subconscious awareness of it (unless she or he is also a philosopher). — Tristan L

I see. The difference is that you know more about cloudhood than meteorologists do, even if they know more about clouds than you do. If meteorologists discovered that cloud formation actually occurs in a way quite different from what they thought, that in a sense clouds aren't quite the sort of thing we thought they were, would your knowledge of cloudhood also change? Would you need to know they had made this discovery for your knowledge to change? What if the discovery was that several sorts of things previously just called "clouds" were actually very different, so that the world "cloud" was now considered old-fashioned and misleading by meteorologists? What then?

The accepted transcription of this seems to be

With the kids sing out the future
Maybe kids don't need the masters

but the Pillows' English can be hard to parse. Anyway, listen to The Pillows and rediscover hope:

A piece of apparent language - a 'move' - is presented which goes against those criteria. — Banno

In what way? If there are semantic rules, what are they like? You seem to assume, and impute to me and other "conventionalists", a view of the lexicon as telling you how a word must be used on pain of not making sense. But this is obviously crazy because usage changes over time. You want to say people can be creative, can break the rules and still be understood because language is not some locked-down closed system with strictly prescribed ways of using words. Well, no one thinks that, not even us defenders of rules and conventions. But that doesn't mean words don't have meanings or that language doesn't have rules.

Suppose the rules of your lexicon are overwhelmingly permissives, with a handful prohibitives here and there. They would be rules like "You can use 'ball' when you want to talk about an object, usually round and small enough to hold, often used in play or sport, or anything with a similar shape." Such a rule does not tell you this is the only possible use of 'ball'. Children will tend to learn this rule first, and learn an additional rule from Disney movies. ("You may also use 'ball' to talk about a large fancy party with dancing.")

There is, on the conventionalist side, very good reason to think the lexicon is permissive in this way. The arbitrariness of the sign comes from there being an indeterminate number of equilibria available to solve the sort of coordination problems language use solves. A convention is the one people land on somehow, but another one could have done as well. Words are things that can serve a given purpose, but that says nothing about whether they could serve some other purpose as well, and of course lots of words have multiple uses. (Or, in a different sense, almost all of them do, the whole point of most words being their multiple applicability. Most words aren't names after all.)

If you think the defense of conventions, rules and regularities means defending the idea that people cannot use words in new ways, you're barking up the wrong tree.

In fact there's plenty of evidence, near as I can tell, that top-down constraints play a huge role here -- the phrasal, sentence, and conversational context. We take a speaker to have uttered a word that would make sense in the context as we understand it, rather than whatever mispronunciation they actually produced. All of that "correction" happens below the level of our awareness.

As it turns out, a lot of research in the field of "speech perception" nowadays is driven by the desire to have computers that can understand human speech.

The field itself is old though, and there are large experiments dating back to 1900 showing that people generally do not detect minor mispronunciations.

That video is excellent!

When setting up the board put the knights where you'd normally put the rooks, and put the rooks where you'd normally put the knights, and then play the game according to initial postion rather than according to the look of the pieces. You'd have a piece that looks like a knight but behaves like a rook. That's pretty much what a malapropism is: it's a mishap about appearance, and it works because of the arbitrariness of the sign. As long as your knight-looking piece moves like a rook, it's a rook in all but looks. The biggest challenge is habit: if you're used to playing chess with a knight-looking knight and rook-looking rook, you might confuse the pieces based on habit. That's an additional challange, but it doesn't really ruin the game. Same rules and same pieces; just a mismatch in the "lexicon". — Dawnstorm

That's actually really nice!

(Players actually find themselves doing this out of necessity too, like at a club where there's one set that's short a piece, or if you promote but the set doesn't have an extra queen, so you use a pair of pawns on one square, or an upside down rook.)

@Banno won't go for it, because in the land this analogy comes from, moves in the game are analogous to utterances (chess <---> language-game).

As helpful as I find the chess analogy, it looks to me like it's getting in the way, though.

It's so clear to me that chess has semantic rules -- violation of which is usually, usually but not always, can't believe I left this out! a mistake -- in addition to the syntactic rules printed in the rulebook: I remember sitting with a friend trying to reconstruct a game I'd just played but didn't have a complete score for because the end of the game had been played at speed; I was trying to fill in the rest of the score and we were stuck on a move my opponent had made I just could not remember. Finally I got it. Turns out I couldn't remember what he had played because it was a move that didn't make any sense, didn't fit with the flow of the game at all.

I've also played blitz games where in the post-mortem we discovered that one of us had made an illegal move. It has in point of fact happened even in high-level tournament play, and is probably not that uncommon in scholastic chess. Often blitz is played with the onus of recognizing the illegal move on the other player, and it is extremely common for blitz to be played where making a move that leaves your king in check counts as a move so your opponent can -- if he notices! -- end the game by capturing your king, none of which is possible according to the standard rules, so you could also just think of this as a variant.

I suppose this is an argument that you can, and people do, carry on with something that everyone involved thinks of as a game of chess even when the rules of syntax have been broken. But only little kids make lots of illegal moves or just do random non-rule-bound stuff on the board.

So then, do you want to say that a malapropism is merely "unconventional" and not a trangression against any actual rule? — Janus

I don't have anything in particular to say about malapropisms at the moment. They're not in themselves important to Davidson's argument, near as I can tell. What matters is that they are speech errors, and what matters more is the case where someone gets away with it.

Step one would be to figure out if there are patterns to speech errors being noticed or not. The ones that are noticed we may correct through conscious deduction of what the speaker was trying to say, if we can, and that's only kinda interesting, and I think not to Davidson's purpose at all. Or you say something, engage as you would if someone used a word you don't know.

For the ones that aren't noticed, we would separate cases: (1) unnoticed because accepted, that is we don't recognize that what the speaker said might not be what they intended; (2) unnoticed because corrected without our awareness, by lexical lookup that treated the utterance as an error. And in the latter case, does that error correction module itself make errors, correct statements that were fine as is? I think there's evidence of this, but I don't have an example handy.

Mostly I think it's a matter for the professionals, but if they asked me I'd suggest the above.

But Davidson takes as given the getting away with it part, and all his conclusions stem from that single datum, not from wherever that datum comes from. So that argument is fair game.

It's a thread with 400 something posts and nothing much of any cogency has been said so far — Janus

Hey! I resemble that remark!

He does try to make his case against conventions, rules, and regularities quite broad -- going beyond malapropisms to include not only all successful speech errors, but also throwing in "Jabberwocky", about which he comments, as I recall, that most of it can be understood on a first reading.

Of course "Jabberwocky" isn't all made-up words, and the made-up words obey English phonology, and it's grammatical.

What's more, just "getting the gist" is maybe a little less than we expect of comprehension.

And still more, it's not like he can point to a large body of speech as odd as "Jabberwocky".

Some natural questions then: (1) Which conventions, rules, and regularities are we keeping? (2) If people can talk this way all the time and get along perfectly fine, why don't they? Why don't we all encounter dozens or hundreds of words each day made up on the spot? (We do, I've read, average perhaps dozens of speech errors per day.)

So my resistance here is not just based on malapropisms, anymore than Davidson takes the case he's made only to apply to them.

For instance, where he says "or born with" referring to the linguistic competence that doesn't exist, that can't be a reference to anyone but Chomsky can it? And Chomsky's original claims for an inborn facility are all to do with syntax. Did Davidson really try to just throw in a denial of our inborn capacity for syntactic speech on the strength of his analysis of semantic speech errors? Yeah I think he did.

No they aren't. The substituted word is almost always the same part of speech, even the same number of syllables with the same prosody, and the resulting expression is grammatical.

The analogy in chess would be a move that, while legal, "doesn't make sense" according to some view of chess, but works for some specific reason.

Think of a great game of chess: every single move is in accordance with the rules, but if you asked me to explain what happened and why, I wouldn't just hand you a copy of the rulebook. The rules don't explain what happens in a game of chess, even though all of it is rule-governed and could not exist without this system of rules.

Am I still not quite making the idea clear? It's very intuitive to me, so I could be missing the mark.

I'd taken his argument as being against those who suppose that all there is to understanding language is understanding conventions. — Banno

Two points:

1.) I don't know what you're talking about here. It is consistent to hold of linguistic communication, that absolutely all of it is governed by conventions, rules and regularities, and that these conventions, rules, and regularities do not explain what people communicating using language are doing. I think a similar claim is true of chess, to me obviously true. I would say further that the rules of chess, again in a way I find obvious, don't just allow the feats of creativity we observe but enable them; and I could claim the same for language use.

1.a.) None of this is about other competencies a language user must have, how a linguistic agent is embedded in culture or society, etc., not directly anyway, but about the nature of rule-governed creativity.

1.b.) To hold such a position, I'd need an account of malapropisms as either unsuccessful or successful because of some particular convention, etc.

2.) I just don't see your Gödel reading in the text.

The problem we have been grappling with depends on the assumption that communication by speech requires that speaker and interpreter have learned or somehow acquired a common method or theory of interpretation — p. 265

I read that as saying communication by speech does not require any such thing. He's not about to claim, just a sentence or two hence, that linguistic competence is just not quite enough to explain linguistic communication -- maybe we need just a smidge of something else; no, he's going to claim there just is no such thing as linguistic competence, and though he describes it with his three principles, he really seems to intend them to be broad enough to take in anyone you can think of.

Davidson wrote considerably more — Banno

Yeah, that's totally fair. I haven't gone back and re-read the earlier stuff for this discussion (and never read much later stuff) so I'm not in any position to judge his project. Insofar as I have in this thread, that's overstepping on my part.

This paper, by itself, feels pretty thin to me on that front, but I'm open to being shown I'm missing something.

I've never tried to work through it in any practical way!

I think I might have an example of how it could work. The approach ends up being inherently comparative and differentiating. (I'd like to say inferential, but I'm really not sure.) I can imagine figuring out that "very" is an intensifier by lining up

"A is tall"
"B is tall"
not "A is very tall"
"B is very tall"

You might also find that "taller than" is never such that both of these can appear

"A is taller than B"
"B is taller than A"

And then further that

"B is taller than A" goes with {"B is tall" and not "A is tall"} or with {not "A is very tall" "B is very tall"}

Anyway, that was kind of what I had in mind as playing semantic sudoku. It's a lot of cross-referencing and looking for consistency among subsets, minimal to maximal, adding and dropping members, etc.

Agreed pretty much all around, except I'd be more inclined to say "following the rules", or if I wanted to be really careful, "acting in accordance with the rules", rather than "knowing the rules". An awful lot of the linguistic machinery we operate is below our level of awareness -- some of it might always be, but at least in use it is: we don't consciously work out what the appropriate rule is and then consciously refer to it as we apply it and check that we've applied it properly. We can do a lot of that sort of thing, and will when there's trouble, but mostly the rules take care of themselves without us paying them any attention. Not once we've learned them, at any rate, and though learning requires a lot of conscious effort, it eventually results in reliable habits that require no awareness.

Until we clear up whether or not the use was intended or not, no communication has happened. — Harry Hindu

That seems fair, and an interesting point, that communication is not just the delivery of a semantic payload but confirmation of that delivery. But absent specific cues, we often just assume we've communicated successfully, don't we?

That is, as the audience, I'm not sure; to a third party, until the audience is sure, there's at most incomplete or partial communication; but the speaker is still entitled their presumption of success.

By virtue of the T-sentences "This is a nice soup latrine" and "This is a nice soup tureen" should be synonymous, so I should be able to write "soup latrine" in my attempt to tansliterate my holistic interpretation, too, right? — Dawnstorm

I believe in Tarski's original version, which was intended for formalized not natural languages, the LHS within quotation marks is in the object language while the rest is in the meta-language. (Of most importance to Tarksi is that "true" here is part of the meta-language; the truth predicate for a language cannot be defined within that language at all.)

I think the Davidsonian adaptation here is that the LHS is in the speaker's language and the rest is in the interpreter's. Thus, just because you use "latrine" as a synonym for "tureen" -- if that's what you're doing -- that doesn't mean I have to, and I'll continue saying "tureen" when I mean tureen.

How do T-sentences deal with word meaning? — Dawnstorm

I think you have to imagine cataloging if not all the sentences then at least all the types of sentences in which a given word could occur, together with their actual or just stipulated truth values, and then working out the "value" of the word like a sudoku puzzle.

If the language is consistent, there will be systematic relations among sentences containing a given word -- that is certain sentences being false will require certain other sentences being true and others still also being false, and so on. Thus the meaning of a word is the "contribution" it makes to the truth or falsity of sentences it could appear in.

Does that make sense yet? It's an approach that was really designed for mathematics, and I think if you get a sort of whiff of David Hilbert's formalism, that's the right impression.

We only question the proper use of some words when words were used but communication didn't occur. — Harry Hindu

No, that's clearly not right. I might wonder whether you've misused a word if I understood what you said but am very surprised to hear you say it, especially if a substitution would yield a sentence I think you more likely to say.

Should we say communication has or hasn't occurred here? Evidently, as the audience, I'm not sure.

"Did I hear you right? Are you saying we should tell our customers when Gimbels has an item we don't, or is offering the same item at a lower price?"
"That's exactly what I'm saying!"