Grammaticalization Quickie Thread

[hwhatting]

Dog old-GEN man-ABL young-GEN cat black-GEN chased
The young man's old dog chased the black cat

SOV languages use suffixes all the time, and they often mark both dependent and head. The Turkish translation of your sentence is:
Delikanlının yaşlı köpeği kara kediyi kovaladı
Young_man-GEN old dog-NOM.POSS3SG black cat-ACC hunt-3SG.PRET
So possession is both marked on the head with a genitive suffix and a possessive marker on the dependent. And languages with default SOV order like Latin can have head marking only, with suffixes, plus genitives and adjectives following their dependents:
Canis senex iuvenis felem nigram persecutus est.
dog-NOM old-M.NOM.SG young_man-GEN cat-ACC black-F.SG.ACC hunt-PERF.PARTC.M.NOM.SG be-3SG.PRES

[StrangerCoug]

Let ERGVERB stand for any ergative verb in a language that does not mark grammatical voice. Suppose a speaker of such a language wants to make a distinction between what in English would be "He ERGVERBed" and "He was ERGVERBed." Are there ways to do that besides recasting the latter as the equivalent of "Somebody/something ERGVERBed him"?

[Creyeditor]

You could just drop the agent, e.g. 'ERGVERB him'.

[bradrn]

Also, there’s no reason why a language with ergative alignment can’t have a passive.

[Travis B.]

Let ERGVERB stand for any ergative verb in a language that does not mark grammatical voice. Suppose a speaker of such a language wants to make a distinction between what in English would be "He ERGVERBed" and "He was ERGVERBed." Are there ways to do that besides recasting the latter as the equivalent of "Somebody/something ERGVERBed him"?

The stereotypical way of doing this would be to have "He ERGVERBed" be "He ERGVERBed-ANTIPASSIVE" and "He was ERGVERBed" be just "ERGVERBed him". However, there is nothing saying that you must do things this way. You could have the former be "He ERGVERBed" and the latter be "ERGVERBed him".

[äreo]

Here's one I've been turning over for a while. I think I've got an answer, but it feels a bit too good to be true.

I'm imagining a language that starts out with a relative clause structure like that of Japanese, where the verb of that clause can simply be placed directly before the noun, without any relative pronouns. This language has adjective-like attributive verbs, likes SOV word order, and is left-branching in general, so all of this checks out. We can easily differentiate boar.ACC attack.PAST man and boar.NOM attack.PAST man.

However, this language has a much more fusional noun morphology than Japanese. Sound change results in complete syncretism of the nominative and accusative. As a result, we can no longer tell by boar attack.PAST man who attacked who. Rather than introduce an obligatory pronoun (since most of our pronouns still distinguish nom/acc, allowing boar him attack.PAST man and he boar attack.PAST man), and rather than coming up with some new accusative marker, I've got a more interesting solution:

1. When the modified noun is the subject, the relative clause structure is exactly the same as before. So boar attack.PAST man is taken by default to mean "the man who attacked the boar."
2. When the modified noun is the object, we no longer use the finite form of the verb. We resort to a class of participles derived from the verb that have (or begin to take on) passive meanings. Thus, in order to say "the man who was attacked by the boar," we say boar attack.PASSP man.

Is this tenable? I considered having the verb take a participial form even when the modified noun is a subject (perhaps a "past agentive participle": boar attacked-er man), but I'm not sure this is necessary.

[Travis B.]

1. When the modified noun is the subject, the relative clause structure is exactly the same as before. So boar attack.PAST man is taken by default to mean "the man who attacked the boar."
2. When the modified noun is the object, we no longer use the finite form of the verb. We resort to a class of participles derived from the verb that have (or begin to take on) passive meanings. Thus, in order to say "the man who was attacked by the boar," we say boar attack.PASSP man.

Is this tenable? I considered having the verb take a participial form even when the modified noun is a subject (perhaps a "past agentive participle": boar attacked-er man), but I'm not sure this is necessary.

This sounds perfectly good to me at least - I'd say go with it!

[äreo]

1. When the modified noun is the subject, the relative clause structure is exactly the same as before. So boar attack.PAST man is taken by default to mean "the man who attacked the boar."
2. When the modified noun is the object, we no longer use the finite form of the verb. We resort to a class of participles derived from the verb that have (or begin to take on) passive meanings. Thus, in order to say "the man who was attacked by the boar," we say boar attack.PASSP man.

Is this tenable? I considered having the verb take a participial form even when the modified noun is a subject (perhaps a "past agentive participle": boar attacked-er man), but I'm not sure this is necessary.

This sounds perfectly good to me at least - I'd say go with it!

Thanks! Maybe I'll revive my old scratchpad thread with some examples soon.

[Ahzoh]

so I have a distal demonstrative idd- that triples as a third person personal pronoun and a relative pronoun.

However, I am looking for confirmation that this grammaticalization pathway makes sense/is likely:

idd-im paruḫ-ta "that man [who] spoke" > id-paruḫ-ta > i-pparuḫ-ta "[he] who spoke"
idd-im na-praḫ-ta "that man [who] may speak" > id-na-praḫ-ta > i-nna-praḫ-ta "[he] who may speak"
idd-īya ta-praḫ-tan "those men [who] shall speak" > id-ta-praḫ-tan > i-tta-praḫ-tan "[they] who shall speak"
idd-ūwa tā-ya-praḫ-tan "those women [who] would have spoken" > id-tā-ya-praḫ-tan > i-ttā-ya-praḫ-tan "[they] who would have spoken"

[Travis B.]

so I have a distal demonstrative idd- that triples as a third person personal pronoun and a relative pronoun.

However, I am looking for confirmation that this grammaticalization pathway makes sense/is likely:

idd-im paruḫ-ta "that man [who] spoke" > id-paruḫ-ta > i-pparuḫ-ta "[he] who spoke"
idd-im na-praḫ-ta "that man [who] may speak" > id-na-praḫ-ta > i-nna-praḫ-ta "[he] who may speak"
idd-īya ta-praḫ-tan "those men [who] shall speak" > id-ta-praḫ-tan > i-tta-praḫ-tan "[they] who shall speak"
idd-ūwa tā-ya-praḫ-tan "those women [who] would have spoken" > id-tā-ya-praḫ-tan > i-ttā-ya-praḫ-tan "[they] who would have spoken"

Demonstratives very commonly double as third-person personal pronouns and relative pronouns, so that part is perfectly plausible. And words for "person" becoming pronouns is also perfectly plausible, take StG man for the 3rd person impersonal, or for that matter NAE guys as a 2nd person plural vocative marker (which is derived from guy meaning man, which originally came from Guy Fawkes of all things -- talk about grammaticalization of all things!).

[Oxygenman]

What characteristics of a language influence how agreement develops diachronically (hopefully I am using this term correctly)? For example, Spanish has adjective-noun agreement, where adjectives and nouns in a noun phrase agree in number/gender, but Basque seems to just stick suffixes on the last word of a noun phrase, whereas other languages exhibit "suffixaufnahme" where multiple cases are applied to each word in a phrase?

[bradrn]

What characteristics of a language influence how agreement develops diachronically (hopefully I am using this term correctly)? For example, Spanish has adjective-noun agreement, where adjectives and nouns in a noun phrase agree in number/gender, but Basque seems to just stick suffixes on the last word of a noun phrase, whereas other languages exhibit "suffixaufnahme" where multiple cases are applied to each word in a phrase?

(Side note: Basque, in fact, also has Suffixaufnahme.)

I feel you’re conflating several different phenomena here:

• ‘Adjective-noun agreement’ is mostly something you find in languages with a noun class (or gender) system. Noun class is distinctive in that it is not overt but marked on other parts of speech, and adjectives are just one of the other elements of the sentence where it can be marked.
• Suffixaufnahme is related to case concord, which is when case is marked on multiple elements of the noun phrase. It’s not really true that the noun gets case and everything else agrees with the noun: rather, noun case marks the semantic role of the whole noun phrase. It can be marked on the noun alone, at the beginning or end of the NP, or distributed through the NP (i.e. case concord).
• You haven’t mentioned it, but I see ‘agreement’ most commonly used for agreement of the verb with its arguments, most commonly in person and number. This is a different thing yet again. Most notably, the line between this sort of agreement and subject/object pronouns is very blurry: it often happens that its presence is dependent in some way on other components of the clause being unexpressed.

Since these are all different, their diachronic development will be different too. Thus, you might get case concord by overextending a nominal casemarker to other elements of the NP, and Suffixaufnahme follows straightforwardly from that. (I don’t know if this has actually happened, but it seems the obvious pathway to me.) Whereas verbal agreement will often result from incorporating a pronoun into the verb, which naturally gives it quite different characteristics.

[Ahzoh]

How would I go about creating this sort of consonant mutation?

w <-> b/p
w <-> m

y <-> dz/ts or g/k
y <-> n or n̮

Proto-lang: *ˀuś- / *ˀawś- "fire"
> ūś-am/wāś-am "fire" (concrete)
> apś-as "heat" (abstract)
> ˀ-b-ś "burn" (e.g. abaśni "I burn")
> ˀ-m-ś "cook" (durative, e.g. amaśni "I cook")

I only want it to be limited to a small set of words that were originally "biconsonantal roots" (e.g. monosyllabic roots). So it's not really intended to be productive.

[bradrn]

How would I go about creating this sort of consonant mutation?

w <-> b/p
w <-> m

y <-> dz/ts or g/k
y <-> n or n̮

This is more of a question for the Sound Change thread than here. You just need some sound changes which create these alternations and then delete whichever sound originally triggered the alternation. Being more concrete than this is difficult, because I don’t know the precise origin of your triconsonantal system, but I’d suggest making *b/*p/*dz/*ts/*g/*k the protoforms, then leniting or nasalising those as appropriate. (You could start with *w/*y and then fortite, but then you’d need to explain why there’s multiple outputs, and it’s less obvious to me what could trigger fortition.)

[Darren]

maybe
*m → w / V_V, *sm → *zb,sp → b,p
*ŋ → *ɰ → y / V_V, *sŋ → *zg,sk → g,k, *ŋ → n

Proto-lang: *ˀuś- / *ˀamś- "fire"
*ˀamāś-am → wāś-am
*ˀa<s>mś-as → apśas
*ˀa<s>maś-ni → abaśni
*ˀa<m
*ˀəmaś-ni → *ˀammaśni → amaśni

[StrangerCoug]

All other parameters being the same, is it attested to unambiguously contrast the perfective and imperfective aspects and have a three-way tense contrast among past, present, and future in the imperfective aspect but only a two-way contrast between one of past/nonpast or future/nonfuture in the perfective aspect?

[Travis B.]

All other parameters being the same, is it attested to unambiguously contrast the perfective and imperfective aspects and have a three-way tense contrast among past, present, and future in the imperfective aspect but only a two-way contrast between one of past/nonpast or future/nonfuture in the perfective aspect?

To my knowledge yes. It is very common to lack a perfective present altogether or to only have a performative perfective present, as, after all the present is what is now as it happens rather than being a defined moment in time.

[Richard W]

All other parameters being the same, is it attested to unambiguously contrast the perfective and imperfective aspects and have a three-way tense contrast among past, present, and future in the imperfective aspect but only a two-way contrast between one of past/nonpast or future/nonfuture in the perfective aspect?

Yes, Russian is an example. The imperfective future is periphrastic, whereas the the perfective future is expressed using what looks like a plain present tense.

[hwhatting]

All other parameters being the same, is it attested to unambiguously contrast the perfective and imperfective aspects and have a three-way tense contrast among past, present, and future in the imperfective aspect but only a two-way contrast between one of past/nonpast or future/nonfuture in the perfective aspect?

Yes, Russian is an example. The imperfective future is periphrastic, whereas the the perfective future is expressed using what looks like a plain present tense.

Polish is the same, and the aspect systems of other Slavic languages are usually being described as being similar. This is a case of convergent development, as older stages of the languages, like e.g. Old Church Slavic, still show non-future uses of the perfective present.

[Ahzoh]

How do I diachronically justify this ablaut class system?

ablaut.png (9.39 KiB) Viewed 254 times

In the irrealis, there is a stem theme vowel where:
a = transitive, stative or dynamic
u = dynamic intransitive (e.g. run, walk, fall)
i = stative intransitive (e.g. be old, stand, sleep)

And given the table there seems to be two axes:
1) stative versus dynamic
2) transitive versus dynamic

But I don't know how to justify why and where.

My current speculation is a realis versus irrealis system that involves 4 morphemes:
a dynamic modal suffix that a-colour
a dynamic modal suffix that u-colours
an intransitive modal suffix that a-colours
a stative modal suffix that doesn't colour vowels (or /i/ somehow resist u-colouring)