Rare/unusual natlang features

[Chuma]

English distinguishes θ/ð from s/z, and has gendered pronouns but not gendered nouns.

[dhok]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

[Linguoboy]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Colloquial French is headed that way, isn't it? In most varieties, the 3P and all singular forms fall together. Then a historically 3S conjugation replaces the 1P, leaving only the 2P/2S.POL as a distinct conjugation. (The details are different for Cajun, but the ultimate result is similar.)

[dhok]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Colloquial French is headed that way, isn't it? In most varieties, the 3P and all singular forms fall together. Then a historically 3S conjugation replaces the 1P, leaving only the 2P/2S.POL as a distinct conjugation. (The details are different for Cajun, but the ultimate result is similar.)

You can't just say that and spare us the details! What happened in Cajun?

But yes, since on replaces nous in colloquial French, and the 3pl and 3sg have merged unexpectedly for non-phonological reasons (not unlike Lithuanian)...

[Travis B.]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Furthermore it only occurs in one mood (indicative).

[dhok]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Furthermore it only occurs in one mood (indicative).

The subjunctive isn't even really real, though, is it? It only operates after a closed stock of expressions and in that- clauses acting as complements of verbs like recommend and suggest, and then mostly in the formal speech of the relatively educated. I think it makes more sense to say that you get the "short infinitive" form there--if I subscribed to generative epicycles I'd go so far as to say there's an invisible modal in the T node that doesn't have any spoken exponence.

[Travis B.]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Furthermore it only occurs in one mood (indicative).

The subjunctive isn't even really real, though, is it? It only operates after a closed stock of expressions and in that- clauses acting as complements of verbs like recommend and suggest, and then mostly in the formal speech of the relatively educated. I think it makes more sense to say that you get the "short infinitive" form there--if I subscribed to generative epicycles I'd go so far as to say there's an invisible modal in the T node that doesn't have any spoken exponence.

The matter is that it is not an infinitive form, since it is clearly finite. E.g. in "I recommend that Elmer Fudd shoot the wascally wabbit" "shoot" is clearly a finite verb. Also, is positing an invisible modal not evident from the surface forms except in the lack of -s a more complex explanation? The only evidence I can see for this in the negation of these forms, e.g. "I recommend that Elmer Fudd not shoot the wascally wabbit", does not follow the normal English negation pattern, i.e. implying "I recommend that Elmer Fudd [modal] not shoot the wascally wabbit". On the other hand, it could be argued that this is consistent with "shoot" having to remain finite, which would not be possible if it had an auxiliary verb.

[Linguoboy]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Colloquial French is headed that way, isn't it? In most varieties, the 3P and all singular forms fall together. Then a historically 3S conjugation replaces the 1P, leaving only the 2P/2S.POL as a distinct conjugation. (The details are different for Cajun, but the ultimate result is similar.)

You can't just say that and spare us the details! What happened in Cajun?

A couple of things. For starters, the 3S completely replaces the 1P, but in a surprising twist, distinct 3P forms survive (usually looking identical to Standard French 1P forms). However, this is only the case when the subject is the 3P pronoun (which takes various forms, including y, eusses, and eux-autres); nominal subjects take 3S agreement and can be replaced with ça. Vous-autres completely replaces vous as a plural familiar pronoun and also takes singular agreement whereas the 2s/p.POL forms fall into desuetude in many areas (since the language is rarely spoken to folks one isn't already on intimate terms with).

So, for instance, in the Vermilion dialect, you end up with:

1s /ʃpɑ(r)l/ "I speak"
2s /tipɑ(r)l/ (2s.POL /vupɑ(r)le/)
3s.m /ipɑ(r)l/
1p /ɔ̃pɑ(r)l/
2p /vuzɔtpɑ(r)l/
3p /ipɑ(r)lɔ̃/ ~ /sapɑ(r)l/

[dhok]

Furthermore it only occurs in one mood (indicative).

The subjunctive isn't even really real, though, is it? It only operates after a closed stock of expressions and in that- clauses acting as complements of verbs like recommend and suggest, and then mostly in the formal speech of the relatively educated. I think it makes more sense to say that you get the "short infinitive" form there--if I subscribed to generative epicycles I'd go so far as to say there's an invisible modal in the T node that doesn't have any spoken exponence.

The matter is that it is not an infinitive form, since it is clearly finite. E.g. in "I recommend that Elmer Fudd shoot the wascally wabbit" "shoot" is clearly a finite verb. Also, is positing an invisible modal not evident from the surface forms except in the lack of -s a more complex explanation? The only evidence I can see for this in the negation of these forms, e.g. "I recommend that Elmer Fudd not shoot the wascally wabbit", does not follow the normal English negation pattern, i.e. implying "I recommend that Elmer Fudd [modal] not shoot the wascally wabbit". On the other hand, it could be argued that this is consistent with "shoot" having to remain finite, which would not be possible if it had an auxiliary verb.

To riff on Scott Alexander: the categories were made for language, not language for the categories.

I'm not convinced that "finite" and "infinite" have any sort of metaphysical meaning--they describe what a verb is doing, morphologically and syntactically; they don't define what it is. In particular, the Latin infinitive cannot act as the sole verb of a clause (which must be the indicative, imperative or [in restricted cases] the subjunctive), has no agreement marking, and has only limited tense/aspect marking. Lots of languages have a verbal form with many or all of these features, so by consensus we call these infinitives. But, of course, there is no requirement that an infinitive even exist: many languages (e.g. Georgian) don't have one, and even many that do are quite restrictive as to what you can do with it (e.g. Persian).

In English, the "short infinitive" already does double duty as an imperative, and generative models already take the imperative with a phonologically null T. If the generative model is correct (there is no guarantee that it is), it is probably not coincidental that the subjunctive survives in structures like I recommend that you not... which are semantically similar to imperatives, i.e. commands, suggestions and exhortations, albeit indirect. Though the difference in negation does suggest that something funny might be going on, of course.

Linguoboy: 3pl /-ɔ̃/ a retention of the expected reflex of -ent, or are we looking at something else? (Somewhat reminiscent of 1pl-3pl syncretism in German.)

[Linguoboy]

Linguoboy: 3pl /-ɔ̃/ a retention of the expected reflex of -ent, or are we looking at something else? (Somewhat reminiscent of 1pl-3pl syncretism in German.)

I don't know the historical development here. It's a feature inherited from Acadian, but I don't know how it came about there. There is a lot of /ɔ̃/ ~ /ɑ̃/ interchange in Cajun French so a retention of -ent isn't out of the question.

[vegfarandi]

Are you thinking about the infamous Hale (2000) paper that analyzed Marshallese with four vowel phonemes //, //, /☎/, /☯/?

TBH makes about as much sense as any other analysis for Marshallese.

[zompist]

The subjunctive isn't even really real, though, is it? It only operates after a closed stock of expressions and in that- clauses acting as complements of verbs like recommend and suggest, and then mostly in the formal speech of the relatively educated. I think it makes more sense to say that you get the "short infinitive" form there--if I subscribed to generative epicycles I'd go so far as to say there's an invisible modal in the T node that doesn't have any spoken exponence.

The matter is that it is not an infinitive form, since it is clearly finite. E.g. in "I recommend that Elmer Fudd shoot the wascally wabbit" "shoot" is clearly a finite verb.

This doesn't clearly show anything of the sort! "Shoot" isn't modified by person or tense, which means you've got a non-finite clause going here. It's also formally identical to the infinitive!

(You're not thinking the infinitive requires "to", are you? It doesn't; compare "I can shoot", "What he did was shoot the rabbit". But if necessary dhok's "short infinitive" can disambiguate here.)

Also, is positing an invisible modal not evident from the surface forms except in the lack of -s a more complex explanation?

FWIW, there are a lot of dubious or ad hoc nodes in generative grammar, but T is one of the best supported ones. It's hard to get the rules for English verbs to work without it, particularly the rules for do-support.

To me, it's far less parsimonious to posit a whole concept (the subjunctive mood) for English, which is almost always identical to the indicative and never introduces a new morphological form. Some syntactic rules should cover it.

[Travis B.]

The subjunctive isn't even really real, though, is it? It only operates after a closed stock of expressions and in that- clauses acting as complements of verbs like recommend and suggest, and then mostly in the formal speech of the relatively educated. I think it makes more sense to say that you get the "short infinitive" form there--if I subscribed to generative epicycles I'd go so far as to say there's an invisible modal in the T node that doesn't have any spoken exponence.

The matter is that it is not an infinitive form, since it is clearly finite. E.g. in "I recommend that Elmer Fudd shoot the wascally wabbit" "shoot" is clearly a finite verb.

This doesn't clearly show anything of the sort! "Shoot" isn't modified by person or tense, which means you've got a non-finite clause going here. It's also formally identical to the infinitive!

(You're not thinking the infinitive requires "to", are you? It doesn't; compare "I can shoot", "What he did was shoot the rabbit". But if necessary dhok's "short infinitive" can disambiguate here.)

"Elmer Fudd" is the subject of "shoot" though. "Shoot" is "infinite" in the sense that it does not inflect for person or number or tense, yes, but I always understood "finite" as meaning "takes a subject".

[zompist]

"Elmer Fudd" is the subject of "shoot" though. "Shoot" is "infinite" in the sense that it does not inflect for person or number or tense, yes, but I always understood "finite" as meaning "takes a subject".

'What I want is for him to shoot the rabbit."
"We're waiting for you to go."

Then there's the fact that in "I want to shoot the rabbit", "shoot" does have a subject, namely "I". There's no confusion over who's shooting.

FWIW Crystal (first glossary I have at hand) defines finite as "a form of a verb that can occur on its own in a main clause, and permits variations in tense, number, and mood."

[Nortaneous]

Are you thinking about the infamous Hale (2000) paper that analyzed Marshallese with four vowel phonemes //, //, /☎/, /☯/?

TBH makes about as much sense as any other analysis for Marshallese.

Given the number of invisible semivowels that analysis posits, many of which emerge diachronically from nowhere, and the fact that most of the consonantal secondary articulation contrasts don't come from feature transfer from vowels (for example, *s *t > tˠ tʲ), I don't know why Marshallese isn't analyzed as having vowel neutralization in closed syllables... but there's probably a reason that I just don't know.

(The only secondary articulation-contrasting pairs that come from feature transfer from vowels are /lʲ lˠ lʷ/, /nʲ nˠ nʷ/, /rˠ rʷ/, /kˠ kʷ/, and /ŋˠ ŋʷ/. Everything else is apparently from a consonant: for example, rʲ < *c. Unless you count the development of the Pʷ series in the Proto-Oceanic stage, which was at least partially from feature transfer.)

Another thing is that Marshallese has initial [j] contrasting with "ɦʲ", but the Marshallese Dictionary's VVS analysis just calls this ɦʲɨ + a prime symbol.

[Travis B.]

"Elmer Fudd" is the subject of "shoot" though. "Shoot" is "infinite" in the sense that it does not inflect for person or number or tense, yes, but I always understood "finite" as meaning "takes a subject".

'What I want is for him to shoot the rabbit."
"We're waiting for you to go."

Then there's the fact that in "I want to shoot the rabbit", "shoot" does have a subject, namely "I". There's no confusion over who's shooting.

FWIW Crystal (first glossary I have at hand) defines finite as "a form of a verb that can occur on its own in a main clause, and permits variations in tense, number, and mood."

In the case of "I recommend that Elmer Fudd shoot the wascally wabbit", I always interpreted "that" as being a relativizer, which leaves open only the possibilities that "shoot" is a finite verb in a non-indicative mood or that there is an invisible auxiliary present.

[akam chinjir]

These looks like cases where it might be helpful to distinguish semantic from syntactic subjects.

In English, infinitives often have semantic subjects. Like "I" is the semantic subject of "shoot" in "I want to shoot the rabbit": to understand that, you know that it's about me shooting the rabbit, not about someone else doing that. And the verb determines the particular thematic role of that subject, whether it's an agent or experiencer or whatever. (Contrast "I want to hate the rabbit".)

But in some sense infinitives don't seem to be able to support a subject, syntactically speaking---their subjects, when they're explicit, need the support of a preposition (the "for" in zompist's examples) or the matrix verb (in raising and control constructions) or something. (There are also cases with accusative subjects: "Me shoot the rabbit? You've got to be kidding." I'm not sure though whether the verb there gets classified as an infinitive.)

The basic distinction here derives from relational grammar, I think. Chomskyans often describe the syntactic deficiency I mentioned in terms of licensing, often more specifically in terms of case; like, you might say that English infinitives can't license nominative case, so the subject of an infinitive has to get case from somewhere else (like "for", or a matrix verb).

I have the impression that one of the factors that influences whether linguists refer to various verb forms as infinitives in whether they get used in purpose clauses.

I remember thinking that the Noonan article on complementation in Language Typology and Syntactic Description (vol 2) is really helpful on this sort of stuff, fwiw.

[Xwtek]

English verbal conjugation is far more bizarre: I don't know of any other language where verbal agreement is limited to one person/number combination out of six and appears only in a single tense (everywhere but the copula, that is).

Trumai, for the limitation to one person/number combination. Although I don't know about the tense.

[Xwtek]

Are you thinking about the infamous Hale (2000) paper that analyzed Marshallese with four vowel phonemes //, //, /☎/, /☯/?

Is that even a phoneme?

[Xwtek]

For Indonesian, after I read the paper, is probably the only language with Indonesian-type voice system, the rest of the language in western Indonesia uses Sundic-type voice system. (I think Indonesian-type voice system is actually saner than Sundic-type, so this might be a good idea for if natlang were conlang thread)