« previous post | next post »
"Just because you collect them doesn't make them collectibles," says a wife to a dopey-looking husband in the cartoon that Bob Ladd peeled off his New Yorker calendar last December 23. And she's right. There's a difference in meaning between the -ible suffix, which attaches mainly to roots with a Latin origin, and the word possible: what is possible for you to X may not meet the standard for being Xible. As so often, the cartoonist (Barbara Smaller in this case) draws on not just a familiarity with life and relationships but also a pretty good implicit analytical knowledge of semantics and derivational morphology.
The -able spelling sometimes (but not always) stands for a different suffix that attaches to almost any word (even Anglo-Saxon roots) and is quite a bit closer to meaning "possible". Thus the conditions for being an eatable substance are not as strict as those for edibility. You may bite bits off your fingernails and swallow them, which is enough to prove they are eatable; but it isn't enough to convince me that they are edible.
January 8, 2010 @ 12:47 pm
· Filed by Geoffrey K. Pullum under Humor, Semantics, Words words words
Permalink
Mark Eli Kalderon said,
January 8, 2010 @ 1:04 pm
"The -able spelling sometimes (but not always) stands for a different suffix that attaches to almost any word (even Anglo-Saxon roots) and is quite a bit closer to meaning "possible"."
OK, here's one case where -able does not mean possible. I think it is possible to desire something that is undesirable, in which case -able in "desirable" is working like -ible in collectible.
Jonathan Badger said,
January 8, 2010 @ 1:23 pm
But isn't the UK spelling for collectible "collectable"?
Skullturf Q. Beavispants said,
January 8, 2010 @ 1:25 pm
Interesting. I've thought about the difference between "eatable" and "edible" before (and note that "eatable" seems to be a fairly uncommon word), but I'm not sure if I noticed that something a bit more general was going on with the suffix "-ible". For example, "credible" means something stronger than just "it is possible for someone to believe it" — if that were all "credible" meant, then pretty well anything and everything would be credible.
Bill Walderman said,
January 8, 2010 @ 1:26 pm
There are apparently some -ble words that can be spelled either with an "a" or an "i." The OED online gives as alternative spellings includible/includable (but not "excludible"), disgestible/disgestable, diffusible/diffusable, extractable/extractible (but not subtractible or intractible). Of course, with vowel reduction the pronunciation is the same.
Cheryl Thornett said,
January 8, 2010 @ 1:37 pm
I think I would reverse the implications of edible and eatable. An edible substance won't make you ill, but you wouldn't necessarily enjoy eating it. I believe my sister and I were discussing collard or mustard greens when we reached this conclusion many years ago.
Boris said,
January 8, 2010 @ 1:59 pm
I see no difference between edible and eatable, except the sophistication of the speaker or the audience (if someone says "eatable" I assume they don't know about the existence of "edible" or are not comfortable with using it. I might use it when speaking to someone who might meet the above criteria)
Benjamin Zimmer said,
January 8, 2010 @ 2:28 pm
And then there are horribles and terribles…
Faldone said,
January 8, 2010 @ 2:37 pm
The way I learned it, -ible was used when the root was derived from a Latin verb of 2nd, 3rd, or 4th conjugation and -able was used when the root was derived from a Latin verb of 1st conjugation or any non-Latin verb. I suspect any meaning differences between the two suffixes are purely idiosyncratic.
Simon Cauchi said,
January 8, 2010 @ 3:10 pm
I don't believe there's any difference in meaning between the suffixes -able and -ible. The choice is determined by the Latin conjugation of the (Romance-derived) verb. Thus:
-ible for verbs with infinitives in -ere (long e, e.g. ridere, risible, and short e, e.g. legere, legible; fallere, fallible)
-able for verbs with infinitives in -are (laudare, laudable; plicare, pliable).
And -able for verbs with non-Romance roots, e.g. washable.
Of course there are countless exceptions, e.g. capable, from capere, and sometimes -able is attached to words that aren't even verbs, e.g. knowledgeable.
The wife in the cartoon is of course quite right, and would be even if she had spelt the word (as it's quite often spelt) "collectables".
mollymooly said,
January 8, 2010 @ 3:22 pm
@Jonathan Badger But isn't the UK spelling for collectible "collectable"?
Not exactly. Soon after the internet came along, I read that factoid somewhere and was surprised. It's since been repeated by various spellcheckers and suchlike. I suspect it's one of the cases where:
* BrEng allows both X and Y
* USEng usage favours X.
* USEng reader sees Y in BrEng text
* surmises validly that BrEng is less sniffy about Y than USEng
* surmises invalidly that BrEng is more sniffy about X than USEng
I'm sure the reverse also happens.
Forrest said,
January 8, 2010 @ 3:28 pm
Jared Diamond mentions a small island in the South Pacific, called Tikopia. The island is about two square miles, and has kept a more or less constant population of about 1,200 people since time immemorial. Producing enough food has sometimes been challenging … this is a recurring theme in Diamond's books, at least in Guns Germs and Steel, and in Collapse, where this comes from.
Diamond asserts that the Tikopians classify plants into three categories: edible, only eaten during the great storm, and not eatable. He didn't use those exact words, obviously, and this is from memory.
So this distinction has been attested, and in the wild.
Army1987 said,
January 8, 2010 @ 5:39 pm
Skullturf Q. Beavispants might have a valid point, indeed the -able equivalent "believable" is much more commonly found in the negative "unbelievable".
Colligo, -is… "Colligible"? nooo, that's too weird…
Collecto, -as (the frequentative of the above)… "Collectable"? yeah, that sounds right to me, and so it does to Firefox's spell checker. "Collectible"… sound right to Firefox too, though I like the former more.
Richard said,
January 8, 2010 @ 6:57 pm
"Just because you collect them doesn't make them collectibles"
Actually, there is another layer of complexity, I think. We should distinguish between adjectival and nominal/substantival uses, since the latter use – seen here – seems to me to be the result of a second process of derivation.
Put simply, not every thing that is 'collectible' (or 'collectable', however you want to spell it) in the sense 'able to be collected' counts as 'a collectible'. In other words, to qualify as a referent of the noun, something has to meet a test that in this case is stronger than (and more generally therefore can, though need not, be different from) the test for the applicability of the adjective: the noun has separate semantics from the adjective, albeit semantics that are closely related.
J. Goard said,
January 9, 2010 @ 1:54 am
You ought to mention that this is basically true for every derivational affix in every language. Langacker's discussion of -er is what instigated my eventual rejection of compositional semantics in favor of usage-based models, where semantics is conceptualization and new forms emerge via frequency-sensitive analogy, not compositional algorithms.
To be a complainer, you have to complain excessively, or at inappropriate times. To be a drinker, you have to drink a certain kind of liquid more than some baseline. To be a teacher, you have to teach professionally or in some authoritative capacity, not just teach your little brother how to ride a bike. But to be a murderer, you've only gotta murder once! The point is that we can't even know what the suffix does unless we employ rich world knowledge about the significance of complaining, drinking different things, teaching, murdering, etc.
I'd be happy if someone could give me an example of a derivational suffix that doesn't clearly fit this phenomenon. I've yet to encounter one.
Nabil Hathout said,
January 9, 2010 @ 2:40 am
The difference between collectible and collectable could be that the first is derived from the noun collection and the later from the verb collect.
Clarissa said,
January 9, 2010 @ 3:57 am
When I was a kid and we used to go antiquing in the southern US, my mom told me that the reason some shops had signs advertising "COLLECTABLES" was to prevent lawsuits or accusations of false advertising due to all the worthless junk they were selling. When I got older I realized that probably nowhere near that much thought had gone into most of the signs, but I have actually seen a couple of signs that said "COLLECTABLES AND COLLECTIBLES" … so I wonder.
Mr Punch said,
January 9, 2010 @ 8:50 am
I think Nabil Hathout has it right, basically. A collectible is something created to be part of a collection — either as part of a series (an annual Christmas plate) or as a commemorative (a coronation tea cup) — these happen to be objects actually in my house, oddly. A regular-issue coin is certainly collectable, but it's not a collectible (although special issues such as the recent state-by-state US quarter series might be).
mollymooly said,
January 9, 2010 @ 10:01 am
There's a well known alleged instance of the fallacy of equivocation in JS Mill, who allegedly uses "desirable" in the sense "capable of being desired". Utilitarianism, Chapter 4: Of what sort of Proof the Principle of Utility is Susceptible:
Dan T. said,
January 9, 2010 @ 11:39 am
If you're amiable, does that mean that somebody can amy you?
Simon Cauchi said,
January 9, 2010 @ 2:54 pm
@ Dan T. Only if you both can toler that sort of word play.
m said,
January 9, 2010 @ 3:29 pm
Where does that leave Michael Pollan's suggestion that we need to "distinguish real food from edible food-like substances" ?
Graeme said,
January 10, 2010 @ 9:20 am
Someone recently (he blushes) told me I was 'f*ckable'. No suggestion that 'f*ckible' would be a more exalted description.
But when the flattery was repeated in an email, it was Cupertinoed into 'fungible'.
Barbara Partee said,
January 10, 2010 @ 9:27 pm
Wasn't this discussed in whichever article it was where Chomsky argued against Paul Chapin's dissertation (which I guess Chomsky had supervised?)? (This is off the top of my head — I just drove home from Baltimore and am too tired to try to do "research" right now.) Chapin proposed transformations for -able words (maybe -ible too, I really don't remember), deriving e.g. 'X is breakable' from something like 'X is able to be broken' (I may have it wrong in detail), and Chomsky noted lots of exceptions like 'admirable' (much stronger), 'laughable' (stronger and needs a preposition 'laugh at'), etc., and that was one of the things that led him to the "lexicalist" hypothesis, wasn't it, i.e. the hypothesis that most word-formation is not sufficiently regular to be described via transformational rules. And this "semi-regularity" and "semi-productivity" led Dowty to propose that one of the main differences between syntactic rules with compositional semantics and lexical rules is that a lexical rule that might say e.g. that you can derive an -able adjective from a transitive verb and its default choice of meaning will be "able to be Verbed", but that (i) these are rules for expanding the lexicon, not rules that are "part of the grammar", (ii) the default choice of meaning may be overridden if the hearer can be expected to be able to guess what the speaker intends, and (iii) once a word is already added to the language, its meaning can evolve and change in all the ways word meanings do, so well-established words can very well have meanings different from the default.
Barbara Partee said,
January 10, 2010 @ 9:32 pm
p.s. to J.Goard — don't reject compositional semantics for the non-idiomatic parts of syntax — there compositionality is fundamental and we'd be lost without it. But do indeed reject it for derivational morphology of the sort discussed in this thread, as Dowty advocated already in the late 1970's, and as I think all formal semanticists have been in agreement about ever since. (But as Dowty argued, and Jackendoff had argued in slightly different terms even earlier, even semi-productive rules for derivational morphology, with compositional interpretations (sometimes several alternative choices) as merely default meaning assignments can be very helpful for the learner and for being understood when you coin new words on old patterns.)
J. Goard said,
January 10, 2010 @ 11:19 pm
Barbara,
I wonder if you think "derivational morphology of the sort discussed in this thread" is all derivational morphology, universally. As I said before, I've never encountered in the languges I've studied any example that seems straightforwardly compositional, without requiring substantial "world knowledge" concerning the root.
With syntax, it sure seems like a similar phenomenon applies to even the most common constructions, like [ADJ N] and [P N]. For example, the idea that "red X" refers to the intersection of the set of red things and the set of things that are X seems very problematic, both because the range of hues which count as red varies considerably with nouns, and because the part/amount of the object which is typically red is related to fairly rich world knowledge about the object.
I certainly don't write off formal compositional semantics, but I do see it losing ground. I'm well aware of Jackendoff's early and recent work (less so on Dowty), but I think he still doesn't quite follow the reasoning as far as it can go. I'm sure you're aware of the connectionist/usage-based position that frequency-sensitive analogical learning and parsing naturally produce "non-idiomatic" constructions as a special, limiting case, without requiring special machinery for the partly idiomatic.
Thanks for your reply. I do need to review the older semantics and philosophy literature, for sure.
Barbara Partee said,
January 11, 2010 @ 1:57 am
@J.Goard,
Thanks for the reply. I don't know if what I said applies to all derivational morphology; I suspect maybe so, but I haven't worked on that question so I would hesitate to make too strong a claim. As for Adj-N, sure, there are lots of idiomatic combinations, and fixed collocations that maybe aren't strictly idioms, like the senses of red in 'red wine' or 'red hair' — Hans Kamp and I wrote about that in our article on prototype theory and compositionality in 1995 in the journal Cognition. Probably all short phrases can pick up conventionalized meanings that end up being non-compositional.
I don't think of compositionality as having lost ground; maybe there were some overblown expectations at the beginning, and maybe some people thought we formal semanticists were claiming more than we actually were (and/or maybe we were claiming more than we should have.) But the robustness of language-specific intuitions about possible and impossible interpretations for novel instances of complex recursive constructions remains the strongest argument for compositionality, I think.
I'm not expert on frequency-based work; I'm sure that's an important factor. (At one extreme, I think it's known that in some kinds of aphasia, productive syntax is lost but lots of fixed phrases and even longer memorized things like the Lord's Prayer are retained.) But I would strongly suspect that usage and frequency would be more important for close collocations, like particular pairs of Adj N, or for (in general non-compositional) noun-noun compounds, or Adv-Verb pairs, etc, and less important for the interpretation of conditional sentences or comparative constructions or Wh-questions or other major sentence-building constructions. Probably both kinds of things are very real, and the challenge is how to build a theory that integrates them appropriately. They may be operating 'on different planes' — I don't exactly know how to think about it.
J. Goard said,
January 11, 2010 @ 9:15 am
Your take on the history of this question is certainly enlightening. A clarification about my interest in compositionality, though:
The most interesting thing doesn't really seem to be "conventionalized meaning". It's not linguistic convention, I think, that makes someone who murders once a murderer, but not someone who complains once a complainer. Nor that a pencil in a drawer has more of its volume spatially enclosed than a pencil in a pencil sharpener. Nor indeed (so that it's clear we're not entirely hijacking the comments) that something describable merely need be possible to describe, while something huggable must have some characteristic making it likely to be hugged. No, the interesting thing is that language users can compute the meaning quite easily, but that this must be from a rich perceptual-functional-experiential semantics for murder, complain, pencil sharpener, hug, etc., not from something Montagovian like (in the case of murder) the set of all triplets -x,y,t- such that x murders y at time t. This is where I think Langacker's view of semantics as conceptualization is pretty strongly supported.
Syntax is definitely a more contentious ground, but construction grammars already seem to have plausible theories of linguistic competence, acquisition, and production/comprehension of novel utterances, and (especially in the last few years) strong links to very plausible evolutionary hypotheses. I think we need to know a lot more about what the general human faculty of analogy can do when presented with rich learning environments, before this issue will be satisfactorily resolved.