Eggcorn Forum

Discussions about eggcorns and related topics

You are not logged in.

Announcement

Registrations are temporarily closed as we're receiving a steady stream of registration spam.

Anyone who wishes to register, please email me at chris dot waigl at gmail dot com with the desired username and a valid email address, and I will register you manually.

Thanks for your understanding.

Chris -- 2011-03-08

  • Forum index
  •  » Soapbox
  •  » morphology and the recognition of sound-symbolic quasi-morphemes

#1 2009-05-04 05:38:14

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

morphology and the recognition of sound-symbolic quasi-morphemes

(Continuing a discussion kem and I started over on the thread about crams/crumps/craps/crimps << cramps my style. Only the last 3 paragraphs way down below are new; the rest is an attempt to summarize.)
.
kem quoted me (dt):

Find how many words meaning compression or restriction begin with ‘cr’ and how many begin with some other random onset cluster, and what percentage of words beginning with ‘cr’ fit that meaning to some degree.

kem commented:

So you believe that finding words beginning with “cr-” that don’t bear the hypothesized meaning is relevant? Then I’m confused. Why did you think that the list of “cr-” words that don’t seem to imply restriction or compression, the one I gave in the earlier post, was not a meaningful addition to the discussion?

dt responds:

Let me try to summarize where I think we’ve come. As usually happens when one tries to summarize in other words, I realize that it may not look, from your viewpoint, like I’m saying the same things. So please straighten me out if it seems like I’m misrepresenting what we said. Anyway, as I see it:

(1) I asked what you intended the list to suggest. I didn’t say the list was not a meaningful addition to the discussion.

(2) I said that others have used that kind of evidence to argue (fallaciously, I believe) that the very existence of such a list somehow proved that the sound-symbolic sound/meaning correlations were illusory, unreal or at best negligible.

(3) I pointed out that if such an argument is accepted, then I can argue on the same grounds that usually undoubted morphemes are unreal or negligible.

(4) You responded that unless the posited correlation always holds true, i.e.:

If your hypothesis is that a certain sound combination sometimes has specific semantic import, then lists of words are no longer evidential. Are there neurological, psychological, etymological or other studies that might support the hypothesis? If not, then the hypothesis remains untestable at our current level of knowledge. [emphasis mine]

So you (not I) seem to be saying that the list of words is relevant only for demonstrating that the sound combination does not always have the meaning. Once that is established, the list is “no longer evidential”.

(5) I countered again that undoubted morphemes do not always exhibit the sound/meaning correlations you are demanding of sound-symbolic “thingies”, and I asked how you do morphology. I was thinking (though I did not say) that neurological and psychological evidence would be very nice, but as far as I know they are sparse, iffy and difficult to interpret. I know no one who bases their morphology purely or even largely on such evidence.

(6) You said you look to etymology and historical data.

(7) I said that looking to etymology and historical data does not avoid the problem. To identify something as a historical or etymological morpheme, you have to do morphology on the historical data, or accept the authority of people, who (all of them) readily identify certain sounds as having certain meanings even when those sounds do not always have those meanings.

(8) I added a comment to the effect that even when you don’t have an absolute correlation, i.e. when the correlation holds only sometimes, sure, one could take refuge in a statistical percentage-type comparison. I do in fact believe that such counts are relevant, and relate to people’s surprisingly accurate notions of how common some linguistic structure is. I didn’t say, but would strongly maintain, that the same sort of calculation would then be necessary to argue for undoubted morphemes as well, and some of them would not fare any better than the putative sound-symbolic correlations. Which returns to points (3) and (5) above.

(9) You expressed surprise that I was now admitting that the lists of non-correlations had some relevance.
.
Now, in response to all of that!!:
.
I never denied that they had relevance. (a) I deny that their relevance is only for demonstrating that a correlation is non-absolute (and therefore perhaps illusory). Also, (b) I deny that they have relevance only for sound-symbolic things and not for regular morphemes (points (3) and (5)).
.
Whew!
.
I’m really not sure how far apart we are on this. But I do think that if you were to talk about how -s marks ‘plural’, you would not respond acquiescently to someone who would respond: “Except in squints, chintz, prince, plops, copse, Cyclops, ax, sex, sis, miss, purse, hearse, worse, curse, nurse, horse, force, … ”, implying “and therefore the sound/meaning pairing you call a morpheme is illusory.

Yes, those words are not examples of -s ‘plural’. So what? That doesn’t come anywhere near denying that -s ‘plural’ exists. The list of those words is not precisely irrelevant—it is of interest to know how far from absolute the correlation -s/plural is. But even if there is a huge list of them, and a much smaller list where the correlation does hold, it doesn’t mean the correlation is illusory.

Last edited by DavidTuggy (2009-05-04 22:34:12)


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 

#2 2009-05-04 22:46:25

kem
Eggcornista
From: Victoria, BC
Registered: 2007-08-28
Posts: 2147

Re: morphology and the recognition of sound-symbolic quasi-morphemes

Whew indeed!

These analogies you make with “in-“ and “com-“ and “-s” are just that-analogies. Like all analogies, some aspects of the comparison fit, some don’t. The similarities between these analogous cases and an initial “cr-“ or a final “-p” seem small to me. One similarity that all the examples share, of course, is a point I think is incontrovertible: English morphemes are not confined to word units. Both words and word parts can have semantic roles.

It’s the differences in the analogous cases that give me pause. Do you really believe that a statement such as “final -p on an English verb conveys a sense of motion coming to an abrupt stop” is in the same class as the statement “final -s on English nouns indicates more than one exemplar of what the word refers to?” The differences between the two claims is huge. The formation of the English plural involves lexemes and word forms and grammatical changes and long stories about etymology, the history of the language, and comparisons with other languages. Final “-p” has almost none of this. Not only is there a long list of verbs that seem to contradict the hypothesis (“yap,” “help,” “grip,” “seep,” etc.), the verbs that do seem to fit the hypothesis fail to display the morphological behavior we would expect from a semantic unit. To say that the final “-p” in your list of verbs means something is too strong for me. Whatever is happening there is submorphemic.

Offline

 

#3 2009-05-05 03:27:17

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

Re: morphology and the recognition of sound-symbolic quasi-morphemes

Analogies between in- and com- and -s and cat and galaxy and walrus are also analogies, in which some aspects fit and some don’t. All categories have that nature: some aspects of their members fit and others don’t. When you set up the category you decide which ones you will focus on and which you willl not.
.
So, “do [I] really believe that a […] final -p is in the same class […] final -s?” Yes, of course, when I’m focussing on what they have in common, and no, of course not, when I’m focussing on what is different between them.
.
I have tried to be careful not to call cr- and -p “morphemes” simpliciter , precisely because of the differences that there are. But I think the similarities are important too, and I also feel very comfortable calling them a (peripheral, if you like) kind of morphemes.
.
The main similarity I see is that in both cases a particular meaning is associated in some degree with a particular sound. (And, no one subpart of the meaning is clearly associated with a subpart of the sound; i.e. they are simplex (morphemes) rather than complex (constructions).)
.
The differences show up starkly when paradigm examples are chosen that are maximally opposed. -p is nothing (one might say) like walrus. walrus is a morpheme of the most optimal or prototypical kind; -p is very peripheral or non-optimal. Yet it remains true that both consist of sound/meaning coordinations.
.
An obvious difference is the specificity and extent of the phonological structures involved, and of the meanings. The sound and the meaning of walrus is much more detailed and characteristic in the case of walrus than in the case of -p.
.
Tied functionally in to that is the degree of uniqueness: it is hard for an English speaker to think of a walrus without at least priming the pronunciation walrus, and very hard to pronounce the word without at least priming the meaning. It is relatively easy to pronounce a p, even a word-final p, without particularly priming the notion of a blow of medium force (or however you define it), and easy to think of such a blow without particularly priming a word-final p.
.
This is very closely tied in to a major difference, that of productivity; the likelihood that the sound will be called on for communicating the particular meaning in some new context. In virtually any context where one wants to communicate the notion of a walrus, one is very likely to use the sound walrus. Similarly, when one wants to communicate the idea of a group of instances of a particular kind of thing, one is extremely likely to use the sound -s. When one wants to communicate the idea of a medium-force blow, however, one is considerably less likely to use the sound -p.
.
Another very major difference is in the subtlety of the connection, or as otherwise phrased, in people’s conscious awareness of the connection. People are very aware that the sound walrus has a certain meaning, and that that meaning is typically symbolized by the sound walrus. They are much less aware of the -p/blow connection.
.
Anyway, I could go on and on. Sure these things are different. But there are undoubted morphemes that are not productive, not unique, not highly detailed, etc. These sound-symbolic things combine a bunch of those less-than-optimal characteristics. But (I think, anyway) they still are real, and their non-optimality is precisely what makes them so interesting. And some of them are less non-optimal that others. For me fl- and sl- and str- are much nearer to being full-fledged morphemes than are cr- or -p. But the differences between them, or between them as a class and undoubted morphemes, are matters of degree rather than clear differences in kind.

Last edited by DavidTuggy (2009-05-14 19:34:10)


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 

#4 2009-05-05 05:49:29

kem
Eggcornista
From: Victoria, BC
Registered: 2007-08-28
Posts: 2147

Re: morphology and the recognition of sound-symbolic quasi-morphemes

You have qualified the morphemic status of a final ”-p” to such an extent that it is hard for me to disagree with you about it. Seems like the main point of difference is what labels we choose to apply to the case. In my (admittedly strange) world, differences in kind are just large differences of degree.

Is there any evidence at all for the hypothesis that “final -p on a verb is sometimes an indicator of motion with an abrupt stop?” The list of words you cite is part of the hypothesis, not part of the evidence-one could take any list of words with some common phonemic and semantic aspects and build a morphemic hypothesis around it. What we need is something that makes this hypothesis worth investigating. Throw me a bone here. There must be something that makes you single out this particular connection between sound and semantics.

Last edited by kem (2009-05-05 05:50:07)

Offline

 

#5 2009-05-06 23:03:40

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

Re: morphology and the recognition of sound-symbolic quasi-morphemes

I inhabit a similarly strange world. For me it is usually less the size of the difference in degree that makes for a difference in kind than whether one of the compared entities is a limiting case or not.
.
As far as I’m concerned, “any list of words with some common phonemic and semantic aspects” is worth making a morphemic hypothesis around it, and the fact that there are those common aspects makes it worth investigating. I don’t understand what you mean by saying that the list of words is not part of the evidence. That to me is like saying that words like words, aspects, lists, bones , etc. are not evidence for the hypothesis that -s means ‘plural’. Of course they are!
.
The reason I singled out this particular connection between sound and semantics was that you mentioned that

we now have six words in the large cr*p fuzzy spot: “cramp,” “crap,” “cram,” “crimp,” “clamp” and “crump.”

Did you think your list was not worth investigating?
.
In any case, I recognize both the “cr” and “V(m)p” from thinking about these things before. I don’t think they are the clearest examples, but I do think they are examples.
.
I failed to include in my post the other day what actually may be the most important difference between these things and clear cases of English affixal morphemes: they only occur attached to each other and rarely if ever to stand-alone stems. E.g. we recognize en – as a morpheme in enlist, entitle, embalm because we know list, title, balm as stems that can occur by themselves. But fl and sl and ap and op only occur next to things that, like they themselves, do not occur independently.

Last edited by DavidTuggy (2009-05-07 17:37:01)


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 

#6 2009-05-13 18:49:07

kem
Eggcornista
From: Victoria, BC
Registered: 2007-08-28
Posts: 2147

Re: morphology and the recognition of sound-symbolic quasi-morphemes

Did you think your list was not worth investigating?

The words in these “large fuzzy spots” are certainly worth investigating. I’m just not sure how to do it. All I know at this point is that the group of words share both a phonological component and a semantic component, and that the interaction between these two components causes the words to substitute for each other in ways that violate correct usage. I can think of ways to explain this behavior without advancing the hypothesis that certain sound fragments shared by the words in question are serving as discrete morphemic units (in the sense, that is, that I would use the word “morpheme”–seems to me that you have a much more flexible understanding of “morpheme”). So the hypothesis you are offering doesn’t seem all that compelling. Still, it is an interesting idea to think about. If it turned out that phonemes such as “cr-” did have an identifiable and persistent semantic import, then the hypothesis would have wide ramifications for the way we understand the development of the English language.

Offline

 

#7 2009-05-13 20:06:45

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

Re: morphology and the recognition of sound-symbolic quasi-morphemes

kem wrote:

The words in these “large fuzzy spots” are certainly worth investigating. I’m just not sure how to do it. All I know at this point is that the group of words share both a phonological component and a semantic component,

I agree completely

and that the interaction between these two components causes the words to substitute for each other in ways that violate correct usage.

Yes, those two facts are for me as well the crux of their fascination.

I can think of ways to explain this behavior without advancing the hypothesis that certain sound fragments shared by the words in question are serving as discrete morphemic units (in the sense, that is, that I would use the word “morpheme”–seems to me that you have a much more flexible understanding of “morpheme”).

To me it’s the same thing all over again. There is an overlapping fuzzy spot of words beginning with ‘cr’ and another of words ending with ‘V(m)p’. In all three cases (cr, V(m)p, and the combination crV(m)p) you are dealing with a “group of words [that] share both a phonological component and a semantic component”. Yes, to me that amounts to some sort of morphemic unit, though “discrete” is not the first word I’d think of to describe them. (Many morphemes overlap their neighbors more or less, on my view. I don’t know how discreteness fits into that—depends on what one means by the term, I reckon.)

So the hypothesis you are offering doesn’t seem all that compelling. Still, it is an interesting idea to think about. If it turned out that phonemes such as “cr-” did have an identifiable and persistent semantic import, then the hypothesis would have wide ramifications for the way we understand the development of the English language.

How badly would it shake you up if there were words like crop, creep, crepe that fit the crV(m)p pattern only phonologically but not semantically? Would you then deny that that pattern had “an identifiable and persistent semantic import”? If you did, then we would be back to the argument that “-s” and “in-” don’t have an identifiable and persistent semantic import either.
.
Yes, I think these quasi-morphemic “thingies” have a definite influence on the way the English language has developed, and so of course we should want to understand them in order to understand that development.
.
May have mentioned this in some other post, but consider the recently coined verb “to munge”. When I first encountered it (on the Internet), I believe I would have had a reasonable idea of the general semantic area of its meaning with little or no context. It would not easily mean “suddenly perceive” (cf. “grok”) or “bow profoundly” or “explode energetically”, or “argue cogently”, “sigh in exasperation”, “place precisely” “experience a pleasurable sensation” or any of a number of dissimilar concepts. It could easily mean “soil, dirty” “destroy/disorder/mess up” “execute sloppily” or things like that. In fact it means “systematically deform (electronic) data so as to render it irrecoverable”, a meaning which, I maintain, fits very naturally with the overlapping fuzzy semantic areas I associate with “məC(C,#)” and “əndʒ”. A lot of English word stems seem to have been coined in a somewhat similar manner.

Last edited by DavidTuggy (2009-05-13 23:31:24)


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 

#8 2009-05-14 07:37:11

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

Re: morphology and the recognition of sound-symbolic quasi-morphemes

Let me make clear that I am talking about subtle conventional sound-meaning linkages in English, without taking a position regarding any putative universality of such linkages. I.e. I am not saying that cr, V(m)p, mə, əndʒ, etc. would be likely (or unlikely) to have in other languages the vague/fuzzy meanings they tend to have in English.
.
That topic would be more relevant to the question of language origins, the bow-wow theory, and all that stuff.
.
I do suspect there are even more subtle effects of that sort in some cases, but such effects are not enough to fully account for the English (or any other language) data—the role of convention is crucial. I would say that any such tendencies are one factor (of many) that exert a subtle pressure on the conventionalization process, but it is conventionalization that establishes what is or is not normal for a given language.


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 

#9 2009-05-16 02:51:26

kem
Eggcornista
From: Victoria, BC
Registered: 2007-08-28
Posts: 2147

Re: morphology and the recognition of sound-symbolic quasi-morphemes

Would you then deny that that pattern had “an identifiable and persistent semantic import”? If you did, then we would be back to the argument that “-s” and “in-” don’t have an identifiable and persistent semantic import either.

Something like the final “-s” plural only comes into the picture if there is a reasonable analogy between the “morpheme” of initial “cr-” and the morpheme “-s.” I still don’t think there is a useful analogy here. The only thing that the “-s” in the huge list of English plurals that end in “-s” has in common with the “cr-” in the small list of words in the “fuzzy spot” is that both are consonant sound sequences and both are adfixes. The morphemic status of final “-s” rests on many lines of evidence: word inflections (removing the “-s” results in a real word that is singular in number), comparison with other languages that also have suffix markers for the plural, a reconstructed history of the development of English that explains how a final “-s” became a plural marker, etc. We don’t have to go out and collect lists of words to construct the hypothesis about morphemic “-s”. But we could make such lists, and if we did they would, as you say, provide some statistically-interpreted evidence in favor of the semantic content of ”-s” plural: in a random English text an interesting percentage of nouns with terminal “-s” would be plurals and an even larger percentage of nouns without “-s” would be singular. In the case of “cr-“ in the small list of words in the “fuzzy spot,” however, the only evidence we have (so far) is the list of “cr-“ words with the target semantics and the counterlist of “cr-“ words that don’t have the target semantics. And since we built the initial list of “cr-“ words using the target semantics, the list can’t be used to prove what it assumes-that would be a textbook example of a vicious circle. [When I wrote this last sentence I wrote “viscous circle!” I suppose a vicious circle could be viscous mind trap. I can see the day coming when I won’t have to comb the web for eggcorns. I’ll just quote myself.]

I’m not sure where you are going with the “munge” example. Are you really suggesting that if I heard an unknown word that I would start the process of meaning discovery by making some sort of semantic retrieval based on quasi-morphemic sound fragments? When I hear an unknown word I begin to discover its meaning by (1) computing its part of speech, (2) looking for semantic clues in the word context, and (3) extracting recognizable morphemic roots. Somewhere down the line I might entertain a hypothesis about (1) an onomatopoetic source, (2) a foreign language source, or (3) a mishearing/misreading. Only at the very end of the discovery process would I put myself in the place of the word’s inventor and consider how sound/semantic associations might have given rise to a novel word, and even at this late, desperate stage I would probably not make much use of quasi-morphemic sound fragments. My intuition tells me that most word coinings are not a product of blended sound fragments. They are usually portmanteux, malaprops and jabberwockies (ear-pleasing, non-semantic permutations of complete word units.) These are the avenues I would explore first.

If you gave me the word “munge” stripped of all clues except what I might guess about it from the last step in the discovery process, the step where I put myself in the place of a word coiner, my guess would have been that “munge” was a jabberwock of “plunge” or “grunge” or a malaprop for “mange” or “munch” (The latter might not be a bad guess. Given its meaning, “munge” might in fact be some permutation of “munch.”). I don’t know when, or whether, I would turn to “overlapping fuzzy semantic areas” that are associated “with ‘məC(C,#)’ and ‘əndʒ’” in order to guess at a meaning for “munge.”

Offline

 

#10 2009-05-16 16:09:53

DavidTuggy
Eggcornista
From: Mexico
Registered: 2007-10-12
Posts: 1777
Website

Re: morphology and the recognition of sound-symbolic quasi-morphemes

kem wrote:

The only thing that the “-s” in the huge list of English plurals that end in “-s” has in common with the “cr-” in the small list of words in the “fuzzy spot” is that both are consonant sound sequences and both are adfixes.

Plus that both are associated in a less than absolute degree with a meaning.

The morphemic status of final “-s” rests on many lines of evidence: word inflections (removing the “-s” results in a real word that is singular in number),

This is what I was referring to when I said (6 May) that cr- and such “only occur attached to each other and rarely if ever to stand-alone stems.”

comparison with other languages that also have suffix markers for the plural,

This is a relatively weak kind of evidence or argument. It is a good argument for the possibility that the human mind can conceive of a particular meaning, and (if the facts bear it out, as they do in the case of plural) that many groups of people have found that particular meaning generally useful, and that does make it plausible or unsurprising when it is posited for a new case. But in the end it is finding a lot of cases where the particular sound associates with the particular meaning that enables us to conclude that that sound means that meaning.

In the case of “cr-“ in the small list of words in the “fuzzy spot,” however, the only evidence we have (so far) is the list of “cr-“ words with the target semantics and the counterlist of “cr-“ words that don’t have the target semantics. And since we built the initial list of “cr-“ words using the target semantics, the list can’t be used to prove what it assumes-that would be a textbook example of a vicious circle.

So then finding a lot of words ending in s with the meaning “plural” is also a vicious circle. I really don’t get this argument. Why does adopting a practical discovery procedure invalidate the results of discovery? If I look for yellow rocks to help me find gold, does that mean that any gold I find can’t be real? If I build a list of in-initial words using the notion ‘negative’ as target semantics, does that mean the list can’t be used to show there are a number of such words?

I’m not sure where you are going with the “munge” example. Are you really suggesting that if I heard an unknown word that I would start the process of meaning discovery by making some sort of semantic retrieval based on quasi-morphemic sound fragments?

No, I am just saying I could and I think you and others could as well. I also of course look for how it is used in context, and if that does not fit what the quasi-morphemic sound fragments would have suggested, the contextual use overrides the other. If I were to read “John munged gratefully that Sally finally understood”, I would be pretty sure the writer meant munge as a verb of mental apprehension or something of the sort. But that would be surprising to me. When I read “the use of non-replyable / munged addresses is strongly discouraged” I am likely to guess that munge is a verb of messing something up or rendering it useless. That will be much less surprising to me.

My intuition tells me that most word coinings are not a product of blended sound fragments. They are usually portmanteux, malaprops and jabberwockies (ear-pleasing, non-semantic permutations of complete word units.) These are the avenues I would explore first.

What are portmanteaux, malaprops, jabberwockies, etc., if not blended sound fragments? (Blended sound fragments with associated bits of meaning more or less attached, of course.) Things like cr-, məC, sl-, fl-, -ap, -omp, etc. are cases where a bit of systematicity seems to emerge from the mess. Not as much systematicity, I agree, as in well-established central members of the category “morpheme”, but a degree of it anyway.

My guess would have been that “munge” was a jabberwock of “plunge” or “grunge” or a malaprop for “mange” or “munch” (The latter might not be a bad guess. Given its meaning, “munge” might in fact be some permutation of “munch.”).

Yes. I can send you if you like a section (in Spanish) from a book I’m working on where I make exactly that suggestion. “munge” as a combination of məC- and -əndʒ need not prohibit you seeing it (and seeing it correctly) as a deformation or permutation of “munch” (and expunge, smudge, mush, and other words). (Grunge, by the way, is another excellent example of a coinage that uses associations of the sound-symbolic type: gr- is very often used for repellent notions: did you ever sing “great green gobs of greasy grimy gopher guts”? Similarly, -unge [= -əndʒ].)

Last edited by DavidTuggy (2009-05-16 16:23:44)


*If the human mind were simple enough for us to understand,
we would be too simple-minded to understand it* .

(Possible Corollary: it is, and we are .)

Offline

 
  • Forum index
  •  » Soapbox
  •  » morphology and the recognition of sound-symbolic quasi-morphemes

Board footer

Powered by PunBB
PunBB is © 2002–2005 Rickard Andersson
Individual posters retain the copyright to their posts.

RSS feeds: active topicsall new posts