Linguistic Miscellany Thread

[Zju]

There are languages in which different stems or genders or classes have slightly different inflection paradigms. For example, if a noun class only contains inanimate nouns, it often lacks vocative forms, even if the language as a whole has them. In some other languages, animate nouns decline for locative cases differently than inanimate nouns. If a language has perfective/imperfective opposition, some verbs may have only perfective or imperfective forms, depending on their semantics.

How far does this go? Are there categories of parts of speech in some languages that have additional categories, or lack even bigger parts of the inflection paradigms, like the categories for number or tense? IIRC there are languages in which only animate nouns have plural. Can two classes of the same part of speech have vastly different inflection paradigms?

[Richard W]

Can two classes of the same part of speech have vastly different inflection paradigms?

For masculine nouns, the Latin 2nd and consonantal 3rd declension have very different endings. The nominative, accusative and locative singular are quite similar, and the 2nd declension genitive plural morpheme incorporates the 3rd declension genitive plural morpheme, but beyond that, they're rather different. The same case and number combinations exist, with a slight differences in syncretism.

[Richard W]

There are also contrasts such as that between prefix and suffix verbs in Somali, though as the person/number markers that precede/follow the verb are very similar, that might be similar to the contrast between the Basque synthetic and periphrastic conjugations.

The difference in the perfect tenses between normal and semi-deponent verbs in Latin also comes to mind. The perfect active has its own set of endings, while semi-deponent verbs use a past participle inflected for number and gender of the subject plus an auxiliary (the verb 'to be') inflected for person and number. In a language where the present tense of the verb to be was normally omitted (cf. Russian), this system would yield a class inflected for person v. one inflected for gender, with the two systems marking number quite differently. However, I can't think of an actual example.

Gender marking is variable in personal pronouns. For example, Latin marks it in the 3rd person pronouns, but not the first and second persons. By contrast, Semitic (primitive for AA?) marks it in the 2nd and 3rd persons, but not the 1st person.

[Linguoboy]

How far does this go? Are there categories of parts of speech in some languages that have additional categories, or lack even bigger parts of the inflection paradigms, like the categories for number or tense? IIRC there are languages in which only animate nouns have plural. Can two classes of the same part of speech have vastly different inflection paradigms?

Well, a number of languages have members of inflectional classes that take no inflections at all (e.g. -er gentilics in German, invariable adjectives in Hindi, invariable verbs in Cajun French), so I would say pretty damn far.

[Kuchigakatai]

A few of the Spanish personal pronouns have an extra prepositional case form: mí, ti, sí. Some pronouns similarly have an extra comitative form: in the modern language it's the same previous pronouns, /ˈmigo ˈtigo ˈsigo/, spelled without a preceding space as <conmigo/contigo/consigo>, although in older stages the language also had /ˈnusko ˈbusko/, spelled <connusco/convusco>, even though these latter ones didn't have a corresponding prepositional *nus/*vus. The third person pronouns él/ella/ellos/ellas have never had any of these forms.


The most stative verbs in English, such as "to be", "to have", "to love" and "to want", lack the usual contrast between the simple forms and the progressive constructions, restraining the latter to very particular meanings. This means the usual distinction between she spoke vs. she was speaking is not there in she was silly vs. she was being silly, or she had none of it vs. she was having none of it, or she loved him vs. she was loving him. Notice how all three verbs drop the past habitual meaning: compare she was working so much back in the day, the unacceptable *she was loving him so much back in the day, and the acceptable she loved him so much back in the day. "To want" can't be in the progressive constructions at all (*I am wanting some ice cream), unless it's used with the (obsolete) meaning 'to lack' (she was wanting a decent education).

A lot of the trouble English speakers have with the Spanish verbs era/fue, tenía/tuvo or quería/quiso, or their French/Latin equivalents, which they don't have with verbs like hablar 'to speak' or trabajar 'to work', can be traced back to this, as these three languages don't have this phenomenon. Interestingly, English "to like" doesn't have this behaviour even though it means something similar to "to love". "To lie [on a surface]" is also weirdly normal, even though the equivalents of that particular verb in other languages sometimes have unusual grammatical behaviour.

In a similar vein, the most stative transitive verbs in Mandarin can't be modified with the aspect particles.

The difference in the perfect tenses between normal and semi-deponent verbs in Latin also comes to mind. The perfect active has its own set of endings, while semi-deponent verbs use a past participle inflected for number and gender of the subject plus an auxiliary (the verb 'to be') inflected for person and number. In a language where the present tense of the verb to be was normally omitted (cf. Russian), this system would yield a class inflected for person v. one inflected for gender, with the two systems marking number quite differently. However, I can't think of an actual example.

I'd say Latin is already that kind of language when its deponent and semideponent verbs appear in this two-word construction with the copular auxiliary. Latin forces the speaker to distinguish gender in the perfect and other such tenses with those verbs (sum lapsus/lapsa/lapsum 'I tripped, I slipped and fell'), even though it doesn't in regular ones (cecidī 'I fell down').

A fun thing is that they don't have passive constructions for those tenses even though they may be transitive, e.g. the deponent sequor 'to follow sb/sth' doesn't have a way to become passive and mean 'to be followed', even though it's transitive with an accusative direct object and all.

Another thing is that the meaning of the past participle is different. Deponent and semideponent verbs naturally treat it with the active-like deponent meaning, so e.g. ūsus/a/um means 'having used sth'. This means that these verbs, unlike regular verbs, usually have a complete set of semantically active participles: ūsus/a/um 'having used sth', ūtēns 'currently/then using sth', ūsūrus/a/um 'about to use sth, using sth in the future'. (And to relate to the above, sequor has no semantically passive participles either, as an after-effect.) Regular verbs don't have a past active, a present passive or a future passive participle.

[Kuchigakatai]

English and probably most other languages in the world often have a category of "labile" verbs with weird behaviour in terms of voice and arguments. These are usually actions that inanimate objects can conceivably carry out on their own without or with minimal conscious, voluntary human/divine/animal intervention: to break, to burn, to freeze, to bend, to split, to harden, to fry...

In English they show up with an alternation between 1) intransitive with inanimate subject, and 2) transitive with animate subject and inanimate object. The bonsai tree bent, He bent the bonsai tree.


In Spanish, the alternation involves 1) reflexive with inanimate subject (essentially a pseudo-passive that expresses an accident), and 2) transitive with animate subject and inanimate object. This verbal category is immensely large in this language, including all sorts of accidents and involuntary events that animate entities can undergo too. Se cocina 'it is being cooked' (literally "it cooks itself"), Lo cocino 'I cook it, I am cooking it'; Se lastimó 'she got hurt (as an accident, e.g. by tripping)' (literally "she hurt herself"), La lastimó 'he/she/someone hurt her (with words)'.

Since these verbs constrain the passive-like reflexive this way, this means that passives where the action is not an accident but has a very voluntary agent must be expressed with the passive construction using the auxiliary ser 'to be' in higher registers (fue lastimada 'she was hurt by someone/some people (with words)'), or with the pseudo-impersonal 3PL form in lower registers (la lastimaron 'she was hurt by someone/some people'). (Some speakers, including me, also accept using a true impersonal while demoting the experiencer to an indirect object: se le lastimó 'she was hurt', literally "there was hurting to her", but this is not that common and is also non-standard... And yes, this "impersonal" does imply a very voluntary agent too.)

Regular non-labile verbs use the reflexive as an actual reflexive, e.g. se vio 'she looked at herself', and don't have a "voluntary agent" vs. "accidental event" distinction in the passive constructions: fue vista, la vieron (se le vio) 'she was seen' (maybe it was an accident that people happened to see her as the people were not trying to see her, they were just hanging out nearby, or maybe they were paparazzis and were trying to find her and see her).

[Vijay]

I don't think Malayalam has verbs like that...

[Kuchigakatai]

I don't think Malayalam has verbs like that...

Neither does Latin, but I'm under the impression most languages do? Maybe I'm wrong. (I'd love it if anyone could show me I'm wrong.)

By the way, what does Malayalam do with regard to "the water boiled" vs. "I boiled the water"?

(Latin tends to use separate lexical words for these two types, like intransitive ferveō vs. derived (semantically causative) transitive fervefaciō, so there's no grammatical phenomenon like the ones found in English and Spanish. Aqua ferbuit "water.NOM boil[vi].PRF.3SG" 'The water boiled', Aquam fervefēcī "water.ACC boil[vtr].PRF.1SG" 'I boiled the water'. Another example: Librī ardent "books.NOM burn[vi].3PL" 'The books are burning', Librōs ūrō "books.ACC burn[vtr].1SG" 'I'm burning the books'.)

[Vijay]

I'm under the impression most languages do?

What gives you that impression? My impression is more the opposite...I can't even think of any languages off the top of my head outside of Europe that do this.

Latin tends to use separate lexical words for these two types

That's what Malayalam does, too.

[Richard W]

The most stative verbs in English, such as "to be", "to have", "to love" and "to want", lack the usual contrast between the simple forms and the progressive constructions, restraining the latter to very particular meanings. This means the usual distinction between she spoke vs. she was speaking is not there in she was silly vs. she was being silly, or she had none of it vs. she was having none of it, or she loved him vs. she was loving him. Notice how all three verbs drop the past habitual meaning: compare she was working so much back in the day, the unacceptable *she was loving him so much back in the day, and the acceptable she loved him so much back in the day. "To want" can't be in the progressive constructions at all (*I am wanting some ice cream), unless it's used with the (obsolete) meaning 'to lack' (she was wanting a decent education).

The past habitual meaning is conveyed by the simple past (or by 'used to').

The progressive He was wanting some ice cream can occur - it would apply to a distinctly transient desire.

I dismissed this set as an inflectional example because there are constructions that just about require the use of the progressive.

[Richard W]

The difference in the perfect tenses between normal and semi-deponent verbs in Latin also comes to mind. The perfect active has its own set of endings, while semi-deponent verbs use a past participle inflected for number and gender of the subject plus an auxiliary (the verb 'to be') inflected for person and number. In a language where the present tense of the verb to be was normally omitted (cf. Russian), this system would yield a class inflected for person v. one inflected for gender, with the two systems marking number quite differently. However, I can't think of an actual example.

I'd say Latin is already that kind of language when its deponent and semideponent verbs appear in this two-word construction with the copular auxiliary. Latin forces the speaker to distinguish gender in the perfect and other such tenses with those verbs (sum lapsus/lapsa/lapsum 'I tripped, I slipped and fell'), even though it doesn't in regular ones (cecidī 'I fell down').

The problem with the Latin example is that the personal endings are mostly similar. e.g. lapsī sumus v. cecidimus for 1p.

[chris_notts]

I'm under the impression most languages do?

What gives you that impression? My impression is more the opposite...I can't even think of any languages off the top of my head outside of Europe that do this.

It's not just European, but it's not universal. My understanding is that languages basically fall into three or four categories:

1. Few or no ambitransitives
2. Mainly ambitransitives of the S=A type (e.g. I ate, I ate chicken, subject role doesn't change)
3. Mainly ambitransitives of the S=P type (e.g. it bent, I bent it, intransitive subject = transitive object)
4. A mix of (2) and (3)

And then, if you look at morphology, the preferred direction for overt valency change differs. Some language have a lot of transitive verbs and prefer to overly detransitivise, as Spanish does via reflexives, and some languages mostly have intransitives and add an overt transitiviser. Of course, there are some verbs which will always be transitive (it's hard to conceive of something being eaten without an eater around), but the class of verbs which are S=P in English can either be lexicalised as root transitives or root intransitives (or both, as an alternative to voice morphology) in a languages that avoids ambitransitive roots.

[Vijay]

Oh OK, we do have "I ate" vs. "I ate chicken" and direct object dropping in Malayalam.

[hwhatting]

On the pro-drop questions in German:
1) Dropping 1st person pronouns used to be very typical of a kind of impersonal style associated with Prussia and traditional militarism; you'll find a lot of examples for this in texts from the late 19th century up to the 1950s. It also was used in telegrams and similar contexts where brevity was valued; I assume it made it into text speak from there.
2) Like linguoboy, I would see things like "kommst heute?" not so much as pronoun dropping but as extreme reduction of the fused form "kommste" that has become indistinguishable from the pronoun-less form.
3) The beer ad is a reference to the Häschenwitze (bunny jokes) that had their heyday in the late 70s and featured a bunny with speech peculiarities like hattu "hast du", muttu "musst du". The jokes sometimes were absurdist and sometimes just bad puns, here is a selection.
4) On dropping 3rd person pronouns - they can be dropped from the sentence-initial position like any other information that can be understood from the context. But I wouldn't call that pronoun pro-drop, it's more like "topic drop". If I ask "Was macht Peter?", I can drop the pronoun and answer "wäscht sein Auto", but if the verb is already supplied in the question ("Was wäscht Peter?"), the natural response is "sein Auto" or a full sentence, but not * "wäscht sein Auto".

[Pabappa]

Okay thanks ... that explains a lot, such as why it's an animal dressed up as another animal. and if the jokes were popular in the 1970s Im guessing it is not an imitation of Japanese.

[Frislander]

So I've been teaching myself Scottish Gaelic for the last couple of months and I just want to share something I've come across that's super wild.

So as everyone knows, Celtic languages inflect their prepositions. However, I think Scottish Gaelic is unique in not only having the standard set for pronominal objects of prepositions, but also has an entirely separate set for marking the pronominal possessor of a following noun leading to sets like the following:

anns an taigh
in the house

annam
in me

mo thaigh
my house

nam thaigh
in my house

This becomes especially wild when you remember that the direct objects of constructions using verbal nouns is marked by the genitive, leading to examples like the following:

Tha i a' pòsadh a bhráthar
She's marrying his brother

*Tha i a' pòsadh e
*(intended: she's marrying him)

Tha i ga phòsadh
She's marrying him

[Darren]

I've heard somewhere that English is the only Indo-European language in Europe which has lost gender completely. This seems a bit unlikely, but I don't know of any counter-examples. Is it true?

[Pabappa]

I've heard somewhere that English is the only Indo-European language in Europe which has lost gender completely. This seems a bit unlikely, but I don't know of any counter-examples. Is it true?

Armenian has also lost gender, but it's on the very fringe of Europe, so people might just be excluding it. http://enwp.org/Grammatical_gender#Indo-European lists a bunch more, but of those, only Ossetian is spoken in Europe.

[Darren]

I've heard somewhere that English is the only Indo-European language in Europe which has lost gender completely. This seems a bit unlikely, but I don't know of any counter-examples. Is it true?

Armenian has also lost gender, but it's on the very fringe of Europe, so people might just be excluding it. http://enwp.org/Grammatical_gender#Indo-European lists a bunch more, but of those, only Ossetian is spoken in Europe.

Thanks! I probably should have found that myself if I'd looked properly.

[dhok]

Aren't there some dialects of Jutland Danish that no longer have the common-neuter distinction?