Something I've asked before, but WITH CONTEXT!

[alice]

I've found the source of a curious statement I asked about before, so I can now ask it again with the correct context. It's in Ralph Penny's A History of the Spanish Language, on page 105 at the end of section 3.2.1 about the development of the noun system, in item 3. The statement is:

Leaving aside the universial constraint that plurals may not show a greater degree of morphemic contrast than the corresponding singulars...

and is followed by an explanation why the Vulgar Latin of Spain chose /annos/ over /anni/ as the plural of /anno/.

The question is: is this "universal constraint" an actual thing, a linguistic universal that if a nominal morphemic distinction is made in the plural, it must also be made in the singular? I've never see it referred to anywhere else.

[KathTheDragon]

No, it's not true. Albanian conflates its dative and ablative cases except in the definite plural.

[Rounin Ryuuji]

Yes, the entire idea sounds rather silly. Also, I don't see how changing a single vowel is a "greater degree of morphemic contrast" than attaching a suffix. Italian seems to get on just fine changing o into i, or maybe a, and e also into i, and a into e.

[alice]

Also, I don't see how changing a single vowel is a "greater degree of morphemic contrast" than attaching a suffix. Italian seems to get on just fine changing o into i, or maybe a, and e also into i, and a into e.

That's not actually what he says, but that's not the point.

Of course you can find counterexamples, but the point is how many there are.

[Rounin Ryuuji]

I'm guessing it was speaking specifically of the plurals of nouny things, and not of verby things, but it occurs to me that regular verbs in French would probably qualify: j'écris, tu écris, il écrit, nous écrivons, vous écrivez, ils écrivent, three distinct plural forms, the singular forms having merged in many contexts.

Also, a probably marginal counterexample with English personal pronouns:

he, him, his, his, himself
she, her, her, hers, herself
it, it, its, its, itself
they, them/'em, their, theirs, themself
they, them/'em, their, theirs, themselves

In only one (for me, at least) of the third person pronouns does the singular have as many forms as the plural (and "singular they" is probably the most recently-innovated of them).

[Moose-tache]

Irish (at least formerly) had several noun classes with more distinctions in the plural than the singular. I'm told some of these extra forms are archaic now, though.

[Znex]

I'm guessing it was speaking specifically of the plurals of nouny things, and not of verby things, but it occurs to me that regular verbs in French would probably qualify: j'écris, tu écris, il écrit, nous écrivons, vous écrivez, ils écrivent, three distinct plural forms, the singular forms having merged in many contexts.

Pretty sure the distinction collapses in a lot of colloquial French though. Probably the only one that remains there is vous, otherwise you have (for the same verb for instance):

j'écris /ʒ-ekʁi/
t(u) écris /t(y)-ekʁi/
il écrit /il-ekʁi/
on écrit /ɔ̃n-ekʁi/
vous écrivez /vu-ekʁive/
ils écrivent /il-ekʁi(v)/

Or with another common verb (jeter for instance):

je jette /ʒə-ʒɛt/
tu jettes /ty-ʒɛt/
il jette /il-ʒɛt/
on jette /ɔ̃n-ʒɛt/
vous jetez /vu-ʒəte/
ils jettent /il-ʒɛt/

[Estav]

I'm guessing it was speaking specifically of the plurals of nouny things, and not of verby things, but it occurs to me that regular verbs in French would probably qualify: j'écris, tu écris, il écrit, nous écrivons, vous écrivez, ils écrivent, three distinct plural forms, the singular forms having merged in many contexts.

Pretty sure the distinction collapses in a lot of colloquial French though. Probably the only one that remains there is vous, otherwise you have (for the same verb for instance):

[...]
ils écrivent /il-ekʁi(v)/

Which varieties of French don't show distinct third-person plural forms for verbs like écrire?

[hwhatting]

The question is: is this "universal constraint" an actual thing, a linguistic universal that if a nominal morphemic distinction is made in the plural, it must also be made in the singular? I've never see it referred to anywhere else.

What I remember from a course on grammatical categories about 30 years back at university is that there's a tendency to have less distinctions in marked instances of a category than in unmarked instances. E.g., for number most languages treat plural as the marked instance, and you find that many languages collapse distinctions in the plural that they make in the singular (e.g., Russian or German don't distinguish gender in the plural, but do so in the singular). The French examples quoted aren't really a counter-example - person is distinguished in both singular and plural, French just has shifted the marking from verbal endings to using non-droppable pronouns, and the fact that there are more distinctions left in the plural that are still made by verbal endings is due to the longer endings in the plural taking longer to erode (longer morphemes for marked categories is another tendency). That said, less stacking of distinctions in marked categories is a tendency, not an exception-less universal; also languages may differ in what are they treat as the more and less marked instances of a category,

[vegfarandi]

Many Icelandic words show more case distinctions in the plural. Weak nouns only have a distinction between nominative and non-nominative (or none at all if neuter): masc. api, apa, apa, apa; 'ape' fem. dama, dömu, dömu, dömu; 'lady' neut. hjarta x4 'heart'. But the plural dative always ends in -um and the plural genitive always ends in -a, meaning words minimally distinguish nom./acc., dat. and gen. in the plural: dömur, dömur, dömum, dama; hjörtu, hjörtu, hjörtum, hjarta. The masculine makes a four-way distinction in the plural: apar, apa, öpum, apa.

[hwhatting]

Many Icelandic words show more case distinctions in the plural. Weak nouns only have a distinction between nominative and non-nominative (or none at all if neuter): masc. api, apa, apa, apa; 'ape' fem. dama, dömu, dömu, dömu; 'lady' neut. hjarta x4 'heart'. But the plural dative always ends in -um and the plural genitive always ends in -a, meaning words minimally distinguish nom./acc., dat. and gen. in the plural: dömur, dömur, dömum, dama; hjörtu, hjörtu, hjörtum, hjarta. The masculine makes a four-way distinction in the plural: apar, apa, öpum, apa.

What about pronouns? Articles? Do they distinguish these cases in the singular?
I'm asking because what I said above is basically about language as a system - Russsian and German don't distinguish gender in the plural in no part of the language, while they do in the singular of pronouns, adjectives, and partially (Russian) in verbs. The situation you describe for Icelandic nouns looks like another case where longer plural endings have eroded less than the short singular endings. In German, there is a similar situation with the dative plural, which is distinguished in several noun classes that don't distinguish the dative in the singular (at least in outside of poetic / archaising speech and some frozen expressions). But in German, the article has taken up the slack of distinguishing cases in the singular:
Sg. nom. der Mann, gen. des Mann(e)s, dat. dem Mann, acc. den Mann
Pl. nom/acc die Männer, gen. der Männer, dat. den Männern

[vegfarandi]

Many Icelandic words show more case distinctions in the plural. Weak nouns only have a distinction between nominative and non-nominative (or none at all if neuter): masc. api, apa, apa, apa; 'ape' fem. dama, dömu, dömu, dömu; 'lady' neut. hjarta x4 'heart'. But the plural dative always ends in -um and the plural genitive always ends in -a, meaning words minimally distinguish nom./acc., dat. and gen. in the plural: dömur, dömur, dömum, dama; hjörtu, hjörtu, hjörtum, hjarta. The masculine makes a four-way distinction in the plural: apar, apa, öpum, apa.

What about pronouns? Articles? Do they distinguish these cases in the singular?
I'm asking because what I said above is basically about language as a system - Russsian and German don't distinguish gender in the plural in no part of the language, while they do in the singular of pronouns, adjectives, and partially (Russian) in verbs. The situation you describe for Icelandic nouns looks like another case where longer plural endings have eroded less than the short singular endings. In German, there is a similar situation with the dative plural, which is distinguished in several noun classes that don't distinguish the dative in the singular (at least in outside of poetic / archaising speech and some frozen expressions). But in German, the article has taken up the slack of distinguishing cases in the singular:
Sg. nom. der Mann, gen. des Mann(e)s, dat. dem Mann, acc. den Mann
Pl. nom/acc die Männer, gen. der Männer, dat. den Männern

I wasn't responding directly to you but yes, that makes sense. Icelandic pronouns/articles etc. tend to distinguish all the cases/numbers/genders quite fully, but nouns, in particular weak nouns show less distinctions. Interestingly, the marked categories of plural and definiteness always distinguish all cases (except nom/acc in the neuter which is a cross-IE feature).
But yes, what you said about OP's thing being a tendency, that is probably true. But it is definitely not a universal.

[alice]

But yes, what you said about OP's thing being a tendency, that is probably true. But it is definitely not a universal.

Possibly a universal tendency?

[vegfarandi]

But yes, what you said about OP's thing being a tendency, that is probably true. But it is definitely not a universal.

Possibly a universal tendency?

Hm… personally I don't believe in the Chomskyan idea of a Universal Grammar and therefore I'm not sure what "universal" would imply other than being part of UG. I personally subscribe to the idea that cross-linguistic tendencies occur as a byproduct of languages attempting to operate in the most efficient, clear manner.

Take for example the idea of differential object marking (DOM). DOM is the term used to describe when objects of verbs are differentially marked based on some semantic property. A well-known instance of this is Spanish, which requires the preposition a with sentient direct objects:

• Pedro tocó la foto 'Pedro touched the photo'
• Pedro tocó a María 'Pedro touched María'

The reason this kind of a feature has developed in multiple languages around the world is most likely because sentient things are more likely to be subjects as sentient things are able to intentionally do things and the lower you go on the animacy scale, the less like the thing is to be able to do something. So the sentient argument being an object is out of the ordinary and you may want to bring that to the attention of the listener. Spaish also allows the movement of the subject to the end of a clause:

• Tocó la foto Pedro

So a second reason exists within Spanish for the development of this feature: it clarifies that a sentient object is not a displaced subject.

If there is in fact a tendency for more marked categories to distinguish fewer additional categories than less marked categories (having not done or seen comparative research on this I don't want to definitively say there is), I can see three explanations, none of which involve UG:

1. A marked category might occur less commonly than the unmarked one and due to it already being "remarkable" due to its expecit marker there's less pragmatic need for further marking as the listener is more likely to pay attention to it given it is marked (cognitive hypothesis)
2. Multiple morphological markers could be more subject to phonological erosion over time than unmarked roots (diachronic hypothesis)
3. A language may be limited in the available number of marker slots with one category superseding others (morphosyntactic hypothesis)

Any or all of these could be factors leading to a general tendency for what you described.

Personally, I'm averse to call anything in linguistics "universal" unless you're using it in the Greenbergian sense of a "language universal" which is what you can call a "tendency" of correlated features A+B that it occurs in 90%+ plus languages that have A be true. I think the above is far from rising to that level.