Apocope, meaning change and lexeme

[vox]

In French
-manifestation means either ‘show, display’ as in People’s manifestation of support or ‘protest’ as in A protest in the street
-manif means only ‘protest’, the first meaning is absolutely excluded
The polysemy is reduced with the apocope.

I wonder how to analyze this case : two forms for the same lexeme or two lexemes partially synonymous ?

Thank you.

[Daniel]

That's an interesting question.

In the first place you would need to decide if the first word is actually polysemous or homonymous. Maybe it's just two lexemes that happen to have the same form? And even if it's "polysemous" how would you analyze that as a single lexeme? Maybe even then it's two linked lexemes. Only when you have two closely related meanings would that analysis definitely not work. In the end, the 'right' analysis depends on your assumptions and the particular theoretical approach you're taking.

Additionally, since truncations are not generally rule-based, that must be memorized anyway, so it doesn't seem that odd to also memorize along with that some meaning changes.

So while I can't definitively answer the question about analysis (but maybe those ideas help!), I do want to add something:

This doesn't seem unusual to me at all. In fact, in many cases I think truncation results in highlighting one meaning over others:

• Referee can be anyone who makes judgments (for example, a referee at a journal or the usage referring to writing letters of recommendation). But the truncated form ref only refers to athletic referees.
• Professional can be used widely to refer to anyone in a specialized (maybe even non-specialized?) job. But pro has some narrower and conventional meanings: a golf pro (a trainer/coach for golfing), a professional athlete, a prostitute (not sure if this is the same abbreviation or not!), and also someone who is "a real pro" meaning very good at their job, but not just anyone who has such a job. Doctors, lawyers, etc., would not typically be called "pros" even though you would consider them "professionals".
• Cellular can mean many things, but cell abbreviates only 'cellular phone' (among other I think mostly unrelated meanings of 'cell').
• Carriage has several meanings, but car is more specific (though this may be a later change, and also in parallel to usage for 'train car' etc.).

And so on.

My suspicion for an analysis (from a historical perspective at least) would be that words actually have specific usage/meanings we don't notice until something else also changes. I would imagine that the specific usage that the truncated form takes on follows from conventionalized usage of the original lexeme, which arguably has already split even though there is little evidence for that immediately.

This reminds me of something I was working on and thinking about a while ago with the meaning of words that change when applied to new contexts. The best example is "husband" or "wife" in reference to gay marriage. This has nothing to do with politics! Whoever you ask, two men who married each other are husbands, not wives. And two women who married each other are wives, not husbands. The first edition of the OED (and many other dictionaries) defined husband as roughly "a man married to a woman". The recent revision in the last couple years changed that to "a married man", I assume because of changes in politics/laws. But what is important is that this was not a change that happened because of legalizing gay marriage. It was already what the word meant. It did not refer to a man who married a woman, but to any man who was married. It just also happened to be the case that previously men only legally married women, so the distinction in the definition was undetectable and irrelevant. But once gay marriage became a topic of discourse, it was very quickly discovered (not changed!) that "husband" actually referred to any married man-- thus husband means "male spouse". A logically equivalent possibility would have been that as the spouses of men, gay married men would be called "wives"-- "spouse of a man", while gay married women would be called "husbands"-- "spouse of a woman". But that's not what happened. And it doesn't sound right to my ears. Because we know that "husbands" are men, and that "wives" are women. It isn't due to who they marry, but to who they are. So the word originally meant that (or at least highlighted that) even before the "change" became apparent in usage.

So in short I would suggest that these truncations allow us to view evidence of a pre-existing split in the lexeme, just like words might mean something a little different from what we think based on usage-based definitions. At least that's one hypothesis to consider.

[vox]

Intuitively I don’t agree with this analysis :
lexeme 1 manifestation ‘show, display’
lexeme 2 manifestation, manif  ‘protest’

So to me manifestation is polysemous. But you suggest that truncation reveals an unnoticed homonymy or an interim step between homonymy and polysemy. It’s true that we can’t say *Une manifestation d’étudiants et de colère without producing a zeugma (‘A protest/display of students and anger’). But the fact remains that the two meanings are obviously related : the second one (‘protest') is just a usage-based specialization of the first one. It would seem very exagerated for any French native to consider that we have two homonymous words manifestation. I think the best equivalent word in english is demonstration.

What about considering that truncation is a morpho-phonological operation to create synonymous lexemes ? It seems more plausible to me.

[Daniel]

Reasonable points. Let's start here though:

What about considering that truncation is a morpho-phonological operation to create synonymous lexemes ? It seems more plausible to me.

That doesn't work! A synonymous lexeme would have the same meaning(s), rather than narrowing them. And that's the problem I pointed out with the argument to begin with.

So to me manifestation is polysemous.

Yes, that may be reasonable. A polysemous word is one that has multiple related meanings. So maybe we could instead say that truncation is a process that potentially splits polysemous meanings and focuses on just one (or several).

Some technical difference between homonymy and polysemy should be introduced. Maybe the comparison here really should be between polysemy and vagueness. That is, as you suggest with the zeugma test, we should be able to relate vague meanings. "The coordination of the project and the club both went smoothly." Those are slightly different senses of "coordination" but we can consider them vague (and roughly "the same"). But that doesn't work so well with: "The coordination of the project and of my yoga teacher are both great." -- that is a zeugma, and although the meanings are related they seem to be polysemous rather than vague. As for homonymy, we could take something even less transparently related, such as "The coordination of the project and of the sentences was acceptable." -- while that is still arguably homonymy because at some point those derived from the same etymology, it seems reasonable to think that linguistic 'coordination' (and, or, etc.) is lexically distinct from the 'organization' sense of 'coordination'.

So it makes sense that truncation would affect only one homophonous lexeme, and that it would also more or less apply to a whole, even broad, vague meaning. But the tricky case is polysemy.

In summary, I think the answer is that truncation seems to be able to select a subset of polysemous meanings and split a lexeme in that way. I would argue that potentially the lexeme is already split in some sense-- again, we would need a technical sense of "polysemy".

Dictionaries often have three classifications (at least implicitly) by having different entries for homonymy (check1, check2, etc.), different sections/paragraphs (a./1., b./2.) for polysemy, and different listed meanings for vague meanings (just listed in the same paragraph, sometimes numbered for clarity, 1....., 2....., 3....., etc.).

The arguments above seem to suggest that polysemous words are actually distinct lexemes, perhaps somehow 'related' ('networked'?) in the lexicon. They do have related meanings etymologically but their usage is distinct, and we start to develop, for example, lexical usage statistics for each sense independently.

So I think I may have misleadingly focused on homonymy in my first answer when I should have focused on polysemy! I think that's a much better answer.

[vox]

The more we talk the more I think that truncation is a morpho-phonological operation associated with a semantic one consisting in narrowing the meaning. If I follow your hypothesis about the lexeme split (which I agree with) it would be to say : truncation cannot be what that creates this narrowing since there’s a preexisting split between the full meaning and the narrowed one. In other words, truncation doesn’t involve a new meaning just like a suffixation does (for example). That leads to consider truncation as an operation of subset selection, as you say :

I think the answer is that truncation seems to be able to select a subset of polysemous meanings and split a lexeme in that way.

I tend to think that the initial form and the truncated one represent two different lexemes. I have the impression it’s your opinion too. The issue remains how to describe the semantic relation between the two lexemes. I think the hypothesis of synonymy is defendable if we consider that a full semantic equivalence is not necessary. A partial equivalence is sufficient for two lexemes to be synonymous. I admit that leads to other issues.

What's the definition of vagueness that you use ? Because to me a lexeme is vague when its extension can’t be discrete by nature, like the extension of young. I may be wrong, semantic ambiguity is not really my specialism.

[Daniel]

truncation cannot be what that creates this narrowing since there’s a preexisting split between the full meaning and the narrowed one.

Right. Basically, truncation just reveals an existing split. (Whether that is within the same lexeme, or across two, or whatever, I don't know, and the technical details would depend on your theory. This is very interesting evidence to consider though!!)

In other words, truncation doesn’t involve a new meaning just like a suffixation does (for example).

Not in these cases. There are a few instances of so-called "subtractive morphology" in languages, where for example removing the final consonant from a noun might create the plural form. So truncation can be productive with a consistent meaning, but that's not the case here, and in fact that is very, very rare cross-linguistically. (I'm not sure that more than a handful of uncontroversial examples have been identified, since what I've seen in textbooks/handbooks/overviews is just a few of the same examples cited over and over again. Maybe it's not quite so rare, but it's certainly unusual, and not known in Europe I don't think.)

That leads to consider truncation as an operation of subset selection

Well, it depends on whether that set was properly formed in the first place. It does result in a subset of the meanings, but I just phrased it that way for explanatory purposes. Instead, my hypothesis would be that there are two things in parallel and that truncation simply highlights that they are distinct. Imagine two identical twins who you cannot easily tell apart. But then one gets a very short haircut, and suddenly you can see that they were clearly distinct individuals the whole time even though it was hard to see that.

An alternative possibility would be that truncation actually changes the meaning (by narrowing it), but I find that less likely. I imagine the narrowing has already occurred at least through frequency in usage, and that the truncation instead highlights that, rather than creating it.

You might argue that the act of truncation is what triggers the split, sort of mixing the two ideas. There was a split ready to be triggered (because speakers are aware of distinct uses/senses), but it wasn't really "split" yet, not until something (truncation) came along to trigger it.

I'm not entirely sure how to determine which theory is correct. But the reason I'm leaning toward the "already split" interpretation is that theories often underestimate the amount of memorization we have done as speakers. Syntactic chunking ("How are you?" is not productively generated each time from a tree structure), or frequency statistics are examples of things that fall outside of generative theories but are part of our 'competence' (knowledge) of the language. Even if we COULD still say it's a single lexeme from a minimal theoretical account, I don't know that we SHOULD, because we already have memorized a lot (including different senses of the 'same lexeme' in this case). So why not just let that be the explanation? Theoreticians don't seem to like relying on performance to explain competence, but it's actually non-minimal to not refer to givens like memory-- they come for free, rather than positing something else additional to explain the data in a theory!

I tend to think that the initial form and the truncated one represent two different lexemes. I have the impression it’s your opinion too.

Not quite. I would say that first there are two "twin" lexemes (or a "complex lexeme" if you prefer). Then one of those becomes optionally truncated, leaving the other behind. Again, refer to the haircut metaphor above.

I'll give a concrete example:
Stage 0: Referee refers to anyone who makes decisions, references, etc. "One who refers."
Stage 1: Referee has several meanings. Speakers know that, and can probably list the meanings without much thought.
Stage 2: Referee in the sense of "sports official" is a particularly common usage/meaning, and for many individuals this is the main meaning with the others considered secondary.
Stage 3: "Ref" enters as a truncated form of the sense in Stage 2.

Now I would say that around Stage 2 (not Stage 3!) the lexeme has split (either internally or externally-- that's a question that must be embedded within a theory).

The issue remains how to describe the semantic relation between the two lexemes.

That's an issue however we analyze this data. Clearly there are similar (but distinct) lexemes at some point. And that's a theoretical question that can have very different answers. I don't have one in particular to suggest here. Whether "lexemes" are even proper constructs on a cognitive level is a good question-- are they really distinct nodes in a network or some sort, or are they just related information dispersed on that network? etc.

I think the hypothesis of synonymy is defendable if we consider that a full semantic equivalence is not necessary. A partial equivalence is sufficient for two lexemes to be synonymous. I admit that leads to other issues.

Many other issues. In fact, I don't know that then "synonymy" would have any value as a theoretical concept, since you're just saying "some lexemes are similar", but not that they are actually somehow tied together.

What's the definition of vagueness that you use ? Because to me a lexeme is vague when its extension can’t be discrete by nature, like the extension of young. I may be wrong, semantic ambiguity is not really my specialism.

There is a distinction between two concepts:
Vagueness/specificity: how narrowly defined something is, and how easily it can apply to different usage.
Ambiguity: when there are multiple but distinct senses for words, which could be listed.

Those two terms are often used interchangeably, but they should not be confused. They are also not related to other non-technical senses of the same words (such as "vague" meaning "intangible").

So a word is vague when it can refer to many different things. Lots of things can be "good" for example, because there is a wide range of usage for "good"-- a "good person", a "good book", a "good idea", and so forth. Note that we cannot very easily identify clearly different meanings, and they all seem to be on a continuum (or various continuums).

On the other hand, a word is ambiguous when it can refer to distinct different things, and when those different things can be listed out, with relatively clear boundaries. For example, that's how I would interpret your original examples of French manifestation. They are still related ideas, but they are distinct-- it is possible to identify which one.

One test for vagueness vs. ambiguity is whether you can ask "Which one?"-- if so, it's ambiguous. If not, it's vague.

[vox]

Thank you Daniel.

[Audiendus]

An alternative possibility would be that truncation actually changes the meaning (by narrowing it), but I find that less likely.

If that were the explanation in any particular case, I wonder exactly how it would work. What would cause the truncation in the first place? If it were just a convenient shortening of a long word, why would that have any bearing on the meaning?

One possibility might be where a manufacturer uses a truncation of an existing word in order to name a new product or component. One of the meanings of the original word would have some association with the product/component, but the word would never have been used in relation to it (because the product/component is new). I cannot think of any examples at present, but there may be some.

[Daniel]

I agree with you that truncation should not in general change the meaning.

Your example is interesting though. Similarly you could imagine a particular group (for example teenagers) who introduce the shorter form and highlight a particular meaning. For example, sports fans might have introduced "ref" in the context of sports. So that's possible. But I still feel that it highlights an existing de facto split rather than necessarily creating a new one.

[vox]

In French, it seems that truncation always selects a concrete meaning and eliminate the abstract one. It’s obvious when the original form is a noun of property :
actualité ‘currentness;the news’ > actu ’the news’
publicité ‘publicness;advertising’ > pub ‘advertising’
faculté ‘faculty/aptitude;college’ > fac ‘college’
promotion ‘fosterage/support;professional elevation;special offer (in a shop);advertising for the release of a new movie/book/etc.;school group/year group’ > promo all but ‘fosterage/support’
combinaison ‘combination;sort of suits often made in one piece combining the high and low part (wetsuit, ski suit, tracksuit, coveralls, space suit, etc.)’ > combi ‘sort of suits’ (I think the meaning ‘combination’ is not completely impossible in a community of mathematicians or engineers)
cinéma ‘movie art;movie theater’ > ciné ‘movie theater’
chorégraphie ‘art of composing dances;particular sequence of dance movements’ > choré ‘particular sequence of dance movements’


When the original meaning is concrete (animate, inanimate and material, locations, etc.), nothing changes except from the language register :
microphone ‘microphone’ > micro ‘microphone’ (either the one to speak with or the one to spy with)
métropolitain ‘subway’ > métro ‘subway’ (informal)
homosexuel (noun) ‘homosexual’ > homo ‘homosexual’ (informal)
restaurant ‘restaurant’ > resto ‘restaurant’ (informal)
adolescent (noun) ’teenager’ > ado ’teenager’ (informal)
mythomane (noun) ‘mythomaniac’ > mytho ‘mythomaniac’ (very informal)
appartement ‘appartment’ > appart ‘appartment’ (informal)

May be an exception with fanatique (noun) ‘religious zealot;very intolerant person’, the meaning changes in fan ‘admirer, nut’ (borrowed from English, not a French truncation) and fana ’nut’  (as in I’m a moto nut). Neither fan nor fana can be used to express the meaning ‘religious zealot;very intolerant person’, even in an informal way.

[Daniel]

That's an interesting (and very long) list. You might almost be able to argue for a semantically-based derivational process with that data. Well done!

There is still the question of why, though. My guess would be something like this, sort of from an economic perspective: people say shorter things for products and goods that are tangible, so they truncate longer forms to refer to concrete things that are of immediate relevance. So the resulting semantics might just be coincidental (but perhaps almost conventional now, since it is so common).