SAE phonology and grammar tests

[HazelFiver]

When going through WALS for ideas on developing Sparrowgrass further, I felt unsatisfied with the original grammar test, so I designed an alternative that covers more ground. I don't know if it's an improvement, but it's another view of the issue. I don't have any grammars complete enough to get an actual score on this version.

[masako]

Kala gets an 18 for grammar. Feel pretty good about that. Not gonna bother with the phonology as I basically stole Nahuatl phonology and sprinkled in bits from African and east Asian langs.

[akam chinjir]

I get less than 75% for English.

Section 1:

• 1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.
• 1.2. According to WALS the English verb inflects for 2-3 categories, and that seems about right to me, though maybe not more than one at a time.
• 1.3. I don't really know how to count this, but I wouldn't say English has a strong preference for suffixes. (But the linked WALS chapter is only counting inflectional affixes, maybe that was the intent here.)
• 1.4. English does not have grammatical gender.
• 1.9. The "-self" forms are both reflexive and intensifying.
• 1.26. English has neutral alignment when the arguments are full NPs, which is what the linked WALS article cares about (and anyway it's marked nominative, not nom/acc, in its pronouns).
• 1.27. Third person agreement is zero-marked in the past tense. (Maybe this isn't supposed to count?)
• 1.28. "Give" can occur without a recipient.
• 1.29. The double object construction is an applicative construction.
• 1.30. English negation usually uses the -n't suffix.
• 1.31. English negation requires do-support.

Section 2:

• 2.1. Shm- reduplication isn't especially common, but it's productive.
• 2.4. No politeness distinction in pronouns.
• 2.14. First-person subjects can sometimes be pro-dropped. There are no verbal affixes in the first- and second- person.
• 2.15. As WALS defines things, English has object agreement. (Object clitics that must attach to the verb and are present only in the absence of an overt object count, and English has those.)
• 2.16. No reflexive reciprocal construction.

(59/80, 73.75%.)

I hadn't realised that WALS definitions count English as having object agreement, that's fun. (Of course English isn't listed as having object agreement, but that's pretty clearly an error.)

Edit. Wow, it doesn't even count French as having object clitics. (Aside: given the WALS definition, it's hard to believe lack of object markers is really going to turn out to be an SAE feature.)

[bradrn]

I get less than 75% for English.

Are you sure about some of these? Specifically:

[*] 1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.

I wouldn’t call -ing periphrastic.

[*] 1.4. English does not have grammatical gender.

It does in pronouns (though I haven’t checked whether WALS considers that enough).

[*] 1.29. The double object construction is an applicative construction.

But it’s hardly productive — in my speech at least, I can only use it for certain verbs.

[*] 2.1. Shm- reduplication isn't especially common, but it's productive.

But only some dialects use it!

[*] 2.15. As WALS defines things, English has object agreement. (Object clitics that must attach to the verb and are present only in the absence of an overt object count, and English has those.)

Does English has object clitics now? Yes, there are plenty of clitics, but I’m pretty sure there aren’t any for objects.

I hadn't realised that WALS definitions count English as having object agreement, that's fun. (Of course English isn't listed as having object agreement, but that's pretty clearly an error.)

Yes, I’ve long considered that particular definition to be wrong. (Are there any languages without verbal agreement which don’t have reduced pronominal clitics?)

[akam chinjir]

[*] 1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.

I wouldn’t call -ing periphrastic.

be + participle and have + participle are periphrastic.

[*] 1.4. English does not have grammatical gender.

It does in pronouns (though I haven’t checked whether WALS considers that enough).

WALS does classify English as having grammatical gender in its pronouns (and Corbett does that in his book as well), but it's a mistake because there's no agreement, pronoun choice in English is purely semantic.

[*] 1.29. The double object construction is an applicative construction.

But it’s hardly productive — in my speech at least, I can only use it for certain verbs.

The criterion says nothing about how many verbs need to be involved, and as far as I can tell it's productive across verbs with appropriate semantics. Like, if English borrowed a new verb with a meaning in the neighbourhood of buy, I bet you could use a double object construction just as you can with "buy" or "purchase." What's necessary is change of possession (as is pretty common for applicatives).

[*] 2.1. Shm- reduplication isn't especially common, but it's productive.

But only some dialects use it!

Sure, but a really large number of English speakers will at least understand it.
[/quote]

[*] 2.15. As WALS defines things, English has object agreement. (Object clitics that must attach to the verb and are present only in the absence of an overt object count, and English has those.)

Does English has object clitics now? Yes, there are plenty of clitics, but I’m pretty sure there aren’t any for objects.

Say "I saw him" as you normally would. I bet "him" doesn't get its own syllable, or at least not stress; it's a clitic, =m or maybe =əm.

I hadn't realised that WALS definitions count English as having object agreement, that's fun. (Of course English isn't listed as having object agreement, but that's pretty clearly an error.)

Yes, I’ve long considered that particular definition to be wrong. (Are there any languages without verbal agreement which don’t have reduced pronominal clitics?)
[/quote]

(You just said you thought English doesn't have pronominal clitics, that's confusing.)

I agree the definition is bad, but it's an area where something like WALS can't do a good job, because figuring out what's going on with verb-adjacent pronominal markers takes a lot more than surveying a grammar, it's something specialists regularly disagree about.

[bradrn]

[*] 1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.

I wouldn’t call -ing periphrastic.

be + participle and have + participle are periphrastic.

Ah, right, I hadn’t noticed that the question asks whether all TAM markers are affixes or clitics.

[*] 1.4. English does not have grammatical gender.

It does in pronouns (though I haven’t checked whether WALS considers that enough).

WALS does classify English as having grammatical gender in its pronouns (and Corbett does that in his book as well), but it's a mistake because there's no agreement, pronoun choice in English is purely semantic.

Does that make it any less gendered? I was under the impression that even purely semantic gender is still a grammatical gender system. (And it isn’t even purely semantic: we use her for ships, for example.)

[*] 1.29. The double object construction is an applicative construction.

But it’s hardly productive — in my speech at least, I can only use it for certain verbs.

The criterion says nothing about how many verbs need to be involved

Good point.

and as far as I can tell it's productive across verbs with appropriate semantics. Like, if English borrowed a new verb with a meaning in the neighbourhood of buy, I bet you could use a double object construction just as you can with "buy" or "purchase." What's necessary is change of possession (as is pretty common for applicatives).

But what exactly is the ‘appropriate semantics’? And there’s some pretty odd verbs which can take a double object, without any apparent rules: I read you the book, I sang him a song, but ??I photographed him the view.

[*] 2.1. Shm- reduplication isn't especially common, but it's productive.

But only some dialects use it!

Sure, but a really large number of English speakers will at least understand it.

That is true.

Does English has object clitics now? Yes, there are plenty of clitics, but I’m pretty sure there aren’t any for objects.

Say "I saw him" as you normally would. I bet "him" doesn't get its own syllable, or at least not stress; it's a clitic, =m or maybe =əm.

Yes, you’re right here as well: /ɑ͡iˈsoː.əm/ (I can’t imagine what I was thinking when I wrote that, actually: I’ve just finished reading some stuff on clitics, because I wanted to know more about them, and I know full well by now that English has object clitics!)

Yes, I’ve long considered that particular definition to be wrong. (Are there any languages without verbal agreement which don’t have reduced pronominal clitics?)

(You just said you thought English doesn't have pronominal clitics, that's confusing.)

I agree the definition is bad, but it's an area where something like WALS can't do a good job, because figuring out what's going on with verb-adjacent pronominal markers takes a lot more than surveying a grammar, it's something specialists regularly disagree about.

I agree that this is a tricky case. But you haven’t answered my question: is there any language with nothing that could be called a pronominal clitic? (I’m genuinely interested in the answer to this, by the way: it’s something I’ve been pondering for a while.)

[akam chinjir]

Does that make it any less gendered? I was under the impression that even purely semantic gender is still a grammatical gender system. (And it isn’t even purely semantic: we use her for ships, for example.)

WALS does define it in terms of agreement, though (and so does Corbett's book).

If you call ships "her," does this depend on the noun you use? Like you'll do it with "ship" but not "boat"? Even when talking about the same thing?

But what exactly is the ‘appropriate semantics’? And there’s some pretty odd verbs which can take a double object, without any apparent rules: I read you the book, I sang him a song, but ??I photographed him the view.

Ah, nice examples---maybe they're true benefactives, not like the ones I was thinking about, and I'd agree that's not productive. I was thinking more of "threw/kicked/tossed/levitated her the ball," that sort of thing, where the double object construction seems to involve a change of possession meaning.

I agree that this is a tricky case. But you haven’t answered my question: is there any language with nothing that could be called a pronominal clitic? (I’m genuinely interested in the answer to this, by the way: it’s something I’ve been pondering for a while.)

Yeah, it's a good question. Mandarin and Cantonese at least don't let object pronouns reduce, as far as I know (though it wouldn't shock me if there are varieties/registers of Mandarin where object pronouns lose their tones; Ser?). The WALS article also mentions languages in which there are object markers that can't attach to the verb (like, they attach to TAM particles), I don't know how common that is though.

I expect that on the WALS definitions, a really large majority of languages in which it makes sense to talk about clitics at all (maybe not in Cantonese, for example), you have either object agreement or object clitics.

[KathTheDragon]

[*] 2.15. As WALS defines things, English has object agreement. (Object clitics that must attach to the verb and are present only in the absence of an overt object count, and English has those.)

You mean object marking, not object agreement. English does not have object agreement, and the question asks about (and WALS talks about) object marking.

[akam chinjir]

[*] 2.15. As WALS defines things, English has object agreement. (Object clitics that must attach to the verb and are present only in the absence of an overt object count, and English has those.)

You mean object marking, not object agreement. English does not have object agreement, and the question asks about (and WALS talks about) object marking.

Yeah, that was a slip.

[bradrn]

Does that make it any less gendered? I was under the impression that even purely semantic gender is still a grammatical gender system. (And it isn’t even purely semantic: we use her for ships, for example.)

WALS does define it in terms of agreement, though (and so does Corbett's book).

Ah, right, I didn’t realise that.

If you call ships "her," does this depend on the noun you use? Like you'll do it with "ship" but not "boat"? Even when talking about the same thing?

Maybe that was a bad example, because I personally don’t call ships ‘her’! But I know that there are some people who do.

But what exactly is the ‘appropriate semantics’? And there’s some pretty odd verbs which can take a double object, without any apparent rules: I read you the book, I sang him a song, but ??I photographed him the view.

Ah, nice examples---maybe they're true benefactives, not like the ones I was thinking about, and I'd agree that's not productive. I was thinking more of "threw/kicked/tossed/levitated her the ball," that sort of thing, where the double object construction seems to involve a change of possession meaning.

In those cases, I’d agree that it’s productive. But I was thinking more about those benefactive cases.

I agree that this is a tricky case. But you haven’t answered my question: is there any language with nothing that could be called a pronominal clitic? (I’m genuinely interested in the answer to this, by the way: it’s something I’ve been pondering for a while.)

Yeah, it's a good question. Mandarin and Cantonese at least don't let object pronouns reduce, as far as I know (though it wouldn't shock me if there are varieties/registers of Mandarin where object pronouns lose their tones; Ser?).

Interesting; I didn’t know that. (Though I suspect that their rather simple phonotactics has something to do with that.)

The WALS article also mentions languages in which there are object markers that can't attach to the verb (like, they attach to TAM particles), I don't know how common that is though.

Where in WALS do you see that? I can’t find it in the article.

I expect that on the WALS definitions, a really large majority of languages in which it makes sense to talk about clitics at all (maybe not in Cantonese, for example), you have either object agreement or object clitics.

Yes, that’s pretty much what I was saying.

[akam chinjir]

Yeah, it's a good question. Mandarin and Cantonese at least don't let object pronouns reduce, as far as I know (though it wouldn't shock me if there are varieties/registers of Mandarin where object pronouns lose their tones; Ser?).

Interesting; I didn’t know that. (Though I suspect that their rather simple phonotactics has something to do with that.)

I think more likely it's lack of stress-related reduction (which is supposed to be absolute in Cantonese, but Mandarin does have unstressed/toneless syllables, including some that are often classed as affixes). So a place to look for more examples is languages that aren't supposed to have stress.

The WALS article also mentions languages in which there are object markers that can't attach to the verb (like, they attach to TAM particles), I don't know how common that is though.

Where in WALS do you see that? I can’t find it in the article.
[/quote]

I was basing it on this:

By contrast, person markers which cannot be bound to the verb, i.e. independent person forms such as the subject markers in Woleaian (Oceanic; Micronesia) in (8), or free-standing combinations of person forms fused with tense as in Anejom (Central-Eastern Oceanic; Vanuatu) in (9), have been excluded.

(I guess that's describing TAM particles with object markers, but maybe there's another analysis.)

[bradrn]

The WALS article also mentions languages in which there are object markers that can't attach to the verb (like, they attach to TAM particles), I don't know how common that is though.

Where in WALS do you see that? I can’t find it in the article.

I was basing it on this:

By contrast, person markers which cannot be bound to the verb, i.e. independent person forms such as the subject markers in Woleaian (Oceanic; Micronesia) in (8), or free-standing combinations of person forms fused with tense as in Anejom (Central-Eastern Oceanic; Vanuatu) in (9), have been excluded.

(I guess that's describing TAM particles with object markers, but maybe there's another analysis.)

Yes, I agree that a fused TAM+person marker probably doesn’t count in the same way as a person marker. But that Woleaian example is interesting: the grammar (luckily open-access) doesn’t give enough information to tell whether it’s a clitic or a standalone word, although a grammar of the related Satawalese language suggests that it’s a clitic.

[KathTheDragon]

But what exactly is the ‘appropriate semantics’? And there’s some pretty odd verbs which can take a double object, without any apparent rules: I read you the book, I sang him a song, but ??I photographed him the view.

Ah, nice examples---maybe they're true benefactives, not like the ones I was thinking about, and I'd agree that's not productive. I was thinking more of "threw/kicked/tossed/levitated her the ball," that sort of thing, where the double object construction seems to involve a change of possession meaning.

Aren't all of these cases of dative alternation? "I read the book to you", "I sang a song to him", "I threw the ball to him", etc. The reason there's no "I photographed him the view" is that there's no "I photographed the view to him", you have to say "for him".

[akam chinjir]

Aren't all of these cases of dative alternation? "I read the book to you", "I sang a song to him", "I threw the ball to him", etc. The reason there's no "I photographed him the view" is that there's no "I photographed the view to him", you have to say "for him".

Sure, that's another name for it. But it promotes something to object, which is what applicatives do, so presumably it's also an applicative construction, at least superficially.

One reason for saying the argument becomes an object is that you can then apply passive: "Mary was thrown the ball." (Compare: "*The wall was kicked the ball," "*I kicked the wall the ball", "I kicked the ball to the wall": there's an animacy restriction on the promoted object in this construction that you don't get with an indirect object, and you do get in the passive, reason to think the passive derives from the double object construction, not the one with an indirect object. ---So those bad examples might be fine in a context where you were personifying the wall, just like "Paris was sent the package" becomes fine if you're using "Paris" to refer to a particular group of people in Paris.)

[Kuchigakatai]

1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.

Suppletive case marking? In personal pronouns, you mean?

I'd say genitive -'s -s' still counts as a case marker...

I hadn't realised that WALS definitions count English as having object agreement, that's fun. (Of course English isn't listed as having object agreement, but that's pretty clearly an error.)

Edit. Wow, it doesn't even count French as having object clitics. (Aside: given the WALS definition, it's hard to believe lack of object markers is really going to turn out to be an SAE feature.)

You mean object marking, not object agreement. English does not have object agreement, and the question asks about (and WALS talks about) object marking.

Fun thing: a few months ago I had a small argument/discussion with zompist because he also holds this idea that a language can be said to have object agreement if object pronouns (nearly) always appear to replace an absent NP. And of course, we were discussing French. Defining it that way, it also had the interesting effect that Arabic and Akkadian (and Spanish and English...) would also count as having object agreement.

WALS does classify English as having grammatical gender in its pronouns (and Corbett does that in his book as well), but it's a mistake because there's no agreement, pronoun choice in English is purely semantic.

The WALS as usual does not follow its own definition here (which is based on agreement), but I'd say English could probably still count if we were trying to decide how to split the categories, because "masculine" and "feminine" are nevertheless important categories of the morphology of pronouns. Gender by suppletion essentially. Much in the same way you'd talk about the Spanish demonstratives forming a 3-way "paradigm" with esto, eso, aquello.

Yeah, it's a good question. Mandarin and Cantonese at least don't let object pronouns reduce, as far as I know (though it wouldn't shock me if there are varieties/registers of Mandarin where object pronouns lose their tones; Ser?).

Interesting; I didn’t know that. (Though I suspect that their rather simple phonotactics has something to do with that.)

I think more likely it's lack of stress-related reduction (which is supposed to be absolute in Cantonese, but Mandarin does have unstressed/toneless syllables, including some that are often classed as affixes). So a place to look for more examples is languages that aren't supposed to have stress.

While I agree that in Mandarin and Cantonese it's more likely due to pronouncing most syllables with basically equal stress (aside from a few grammatical function words/affixes and the noun-marking suffixes of Mandarin), I wonder if there isn't some truth to what bradrn said. Japanese has simple phonotactics and yet it has heavily-stressed personal pronouns (part of what makes them noun-like, regarding the other thread). Does anyone know what typically happens in Polynesian and Bantu languages? It wouldn't surprise me if in both families pronouns tend to be fully stressed too (not counting the polypersonal agreement prefixes of Bantu verbs).

I'd say reduction of object personal pronouns to the neuter tone is pretty common in Mandarin, in dialects that are most fond of toneless syllables, like those of the east coast in the traditional Wu areas... Even in other dialects like those of Beijing and Taipei, the object pronouns seem to be a bit less stressed than most content syllables, don't you think? Like, if we divided Mandarin stress in three levels, the pronouns would appear in the middle level while still retaining tone. As Yip and Rimmington point out in their Comprehensive Grammar, the usual prosodial restriction against the disyllabic verb + monosyllabic object construction is disregarded if the object is a monosyllabic pronoun, since the monosyllabic pronoun "often becomes unaccented" (i.e. little stressed). I suppose this means it is treated as a phonological suffix of sorts onto the disyllabic verb.

Chinese verbs are either monosyllabic or disyllabic. A monosyllabic verb may govern a direct object of any number of syllables, but a disyllabic verb must govern a direct object of at least two syllables, and, if the direct object is monosyllabic, a closely-associated numeral or the indefinite-referenced 一 yī ‘a/an’ followed by a measure word must be included:

写书 xiěshū ‘to write (a letter)’
*改写书 gǎixiě shū
改写一本书 gǎixiě yī běn shū ‘to rewrite a book’
*改写本书⁴ gǎixiě běn shū

⁴ In this invalid sentence 本 běn is used as a measure word for books, not as the rather literary adjective meaning ‘this’. Otherwise 改写本书 could mean ‘to rewrite this book.’

It can be seen that the ‘end weight’ principle in Chinese prosody indicates that whilst a monosyllabic verb may take a direct object of any syllabic length, a disyllabic verb must take at least a disyllabic object. If not, the object in question must be compensated in syllable length by other syntactic means which come into direct association with it in the sentence.

However this principle applies only to direct objects that carry new information and are normally stressed to render weight to the utterance. Pronominal items, which usually convey repeated or existing information, can often become unaccented, thus giving free rein to the disyllabic verbs that govern them to take on ‘end weight’ themselves.
[...]

找他 zhǎo tā ‘to look for him’
去找他 qù zhǎo tā ‘to go to look for him’
拜访他 bàifǎng tā ‘to visit him’
去拜访他 qù bàifǎng tā ‘to go to visit him’
你找谁？ nǐ zhǎo shéi ‘who are you looking for?’
你邀请了谁？nǐ yāoqǐng le shéi ‘who did you invite?’
我没邀请她 wǒ méi yāoqǐng tā ‘I did not invite her’
[...]

Before we leave this section, we must hasten to add that a disyllabic verb, particularly with an internal VC [verb + resultative complement] structure [...] may take on a bare (i.e. unmodified) monosyllabic direct object without invalidating the whole construction, but the sentence is nevertheless incomplete and must have a follow-up clause to complete it. For example,

他喝醉了酒，... tā hē zuì le jiǔ ...

[KathTheDragon]

I hadn't realised that WALS definitions count English as having object agreement, that's fun. (Of course English isn't listed as having object agreement, but that's pretty clearly an error.)

Edit. Wow, it doesn't even count French as having object clitics. (Aside: given the WALS definition, it's hard to believe lack of object markers is really going to turn out to be an SAE feature.)

You mean object marking, not object agreement. English does not have object agreement, and the question asks about (and WALS talks about) object marking.

Fun thing: a few months ago I had a small argument/discussion with zompist because he also holds this idea that a language can be said to have object agreement if object pronouns (nearly) always appear to replace an absent NP. And of course, we were discussing French. Defining it that way, it also had the interesting effect that Arabic and Akkadian (and Spanish and English...) would also count as having object agreement.

This is, of course, bullshit, and I'd point him at Haspelmath's paper on... not exactly this subject, but he dispenses with the idea while defining the alternative description of bound person markers.

[akam chinjir]

Fun thing: a few months ago I had a small argument/discussion with zompist because he also holds this idea that a language can be said to have object agreement if object pronouns (nearly) always appear to replace an absent NP. And of course, we were discussing French. Defining it that way, it also had the interesting effect that Arabic and Akkadian (and Spanish and English...) would also count as having object agreement.

French seems at least a step in the direction of agreement, because the clitics clearly aren't just reduced pronouns, and they occur in a different position in the clause from NP objects. (The English ones can with phrasal verbs, but mostly they're in the same spot as NPs.) And isn't Spanish another step closer, since object clitics can occur in the presence of some objects? (The WALS chapter actually counts Spanish but not French or English as having object markers, which seems to me a reasonable place to draw a line, though not the only reasonable place.)

Does anyone know what typically happens in Polynesian and Bantu languages? It wouldn't surprise me if in both families pronouns tend to be fully stressed too (not counting the polypersonal agreement prefixes of Bantu verbs).

Bantu object markers are actually the subject of some disagreement, e.g. Baker, On the status of object markers in Bantu languages.

Still, given the object markers on the verb, you might expect that any additional pronouns would be there for focus or something, and would be phonologically independent.

I'd say reduction of object personal pronouns to the neuter tone is pretty common in Mandarin, in dialects that are most fond of toneless syllables, like those of the east coast in the traditional Wu areas... Even in other dialects like those of Beijing and Taipei, the object pronouns seem to be a bit less stressed than most content syllables, don't you think? Like, if we divided Mandarin stress in three levels, the pronouns would appear in the middle level while still retaining tone. As Yip and Rimmington point out in their Comprehensive Grammar, the usual prosodial restriction against the disyllabic verb + monosyllabic object construction is disregarded if the object is a monosyllabic pronoun, since the monosyllabic pronoun "often becomes unaccented" (i.e. little stressed). I suppose this means it is treated as a phonological suffix of sorts onto the disyllabic verb.

Ah, interesting. (I remember the first time I looked at Yip ad Rimmington and seeing they had a section on object agreement. Then of course I saw it was about prosody, so happy.)

Would you say that plural pronouns are similarly less-stressed? Maybe (being bisyllabic) they can get counted as a whole phrase (or whatever) in a way that a (monosyllabic) singular pronoun can't.

[akam chinjir]

1.1. English has periphrastic aspect, and what case-marking it has is by suppletion.

Suppletive case marking? In personal pronouns, you mean?

I'd say genitive -'s -s' still counts as a case marker...

Oops, and yeah, it was dumb not to mention that.

[Kuchigakatai]

This is, of course, bullshit, and I'd point him at Haspelmath's paper on... not exactly this subject, but he dispenses with the idea while defining the alternative description of bound person markers.

Okay, I just read the paper, and I feel it does a much better job at raising questions than providing answers... There are a number of problems which it unfortunately does not address.

Many languages (e.g. Latin, Russian) can use bare adjectives to stand for the NPs they'd otherwise be in modifying a noun —how is this different from verbs with indexes, and no NPs next to them, that are supposedly not examples of agreement? Imagine a conlang where verbs index their arguments with only gender + number (as in Bantu), and adjectives that show concordance for their nouns in gender + number too. If such languages can also use bare verbs with indexes only ("bloom-PLANT.PL" = 'they're blooming'), and bare adjectives with concordants only ("beautiful-PLANT.PL" = 'the beautiful ones'), why should these be distinguished as (cross-)indexes vs. concordants?

Also, many languages (e.g. Mandarin, Indonesian, Persian, Luo in Nilo-Saharan) conflate the adjective-noun and possessor-possessed constructions. Are there really no languages that would put a noun modified by an adjective in a Semitic-like construct state? Maybe Persian counts with its noun-adjective ezafe: اتاق /otɑːɢ/ 'room', اتاق کوچک /otɑːɢ-e kuːtʃæk/ "room-EZAFE small". If so, wouldn't that Persian ezafe count as an index of nouns for their adjectives against Haspelmath's classification?

His argument showing Spanish -o -es -e -emos are distinct indexes rather than concordants of agreement largely relies on verbs being able to take the 1PL form with a plural subject NP beside it: las mujeres queremos justicia 'we women want justice', since he's also in the camp against the notion of "pro-drop" (saying it's an over-analogy with English/German). But the pro-drop analysis is also advantageous here, as you can also say (though Haspelmath doesn't mention it) nosotras las mujeres queremos justicia —wouldn't this suggest las mujeres is not the subject in the former sentence?

I think the view akam chinjir posted is better: there are a number of reasonable places at which to draw the line between what counts as object agreement and what is worth placing apart as "indexes". My disagreement with zompist was mainly whether co-occurrence with NPs was a necessary line (Spanish le dije eso a mi madre 'I told that to my mother'), but even then zompist's argument for French also relied on the absence of possible pauses between the subject prefix and the verb (can't say *je, si je peux, t'aiderai, lit. "I if I can will-help you"), and the reduction of some prefixes into segments that can't be words (like je = [ʃ ʒ], e.g. je t'aiderai Parisian [ʃtɛˈdʁɛ], Montreal [ʃtɛˈdʀe]).

[Richard W]

I'd say genitive -'s -s' still counts as a case marker...

Why not as a postposition, though? It can even appear on a verb, typically one at the end of a relative clause. A noun phrase composed of a pronoun is the only clear exception, though coordinate noun phrases (e.g. 'you and me') are near the boundaries of native speakers' competences.