Zompist Bboard Again

Torco wrote: ↑Wed Mar 10, 2021 9:01 pm Quick question. is there a way in linguistics to speak about how commonly a word is known? something like the centrality of a word to its lexicon? for example dog is a word everyone knows, but sparkplug is something a speaker of English could get by without knowing what it is (beyond just a car part), and I don't expect most people to know the words spiralizer, or basinet. I sure as hell didn't before i googled weird cooking implements and helmets respectively.

In paedagogy, the term "core vocabulary" is used a lot. There's a lot of attention devoted to identifying core vocabulary items and concentrating on teaching them (often through stultifying repetition). But I can't think of a linguistic term which captures that dimension. (IME, linguists talk more about frequency, which isn't exactly the same thing.)

Richard W wrote: ↑Tue Mar 09, 2021 8:12 pm

bradrn wrote: ↑Tue Mar 09, 2021 5:14 pm

Richard W wrote: ↑Tue Mar 09, 2021 4:27 pm Likewise, there is nothing accusative about English verbal agreement if one only considers the third person singular.

But I’m not quite sure how you come to this conclusion… English verbs agree with a 3s argument in S and A, but not in O, giving an accusative system.

If you only consider 3s arguments, the simple present indicative ends in -s.

If you only consider 3s arguments, the simple present indicative does indeed end in -s… but only if the 3s argument is in S/A, and not if it is in O, giving, as I have already said, an accusative system.

bradrn wrote: ↑Tue Mar 09, 2021 5:14 pm Well, by that criterion, all linguistic data shows interpretation. I’m not sure how this is a useful criticism.

It's why I wanted some indication of how the prefixes were combined. The initial /ŋ/ of the 1p and 2p object tagging might just be something else.

OK, fair request. Here you go:

g-a-kərə-na
3p.O-1s.A-hit-PRS
I hit them.

n-umbwa-ŋa-kərə-ŋa-na
INV-3p.A-1s.O-hit-INV-PRS
They hit me.

(Incidentally, this also shows the direct-inverse system: the second sentence takes the inverse circumfix n⟩⟨ŋa, and the affix order is consistently ‘less animate-more animate’. But this is irrelevant for our purposes.)

Admittedly, these examples involve pronominal affixes I didn’t introduce in my table above, so here’s an expansion of that table to 1s (which you might notice agrees in a tripartite pattern):

	S	A	O
1s	ma-	a-	ŋa-
1p	e-	e-	ŋe-
2p	o-	o-	ŋo-
3p	g-	bo-/mbwa-	g-

He writes words has ///write/// ///-s/// ///-∅///.
I write words has ///write/// ///-s/// ///SUPPRESS-s///

And this analysis satisfies the 'universal' that it is the 3s which is the person etc. expressed by zero! It makes oddities like "So says you" and "I wants it" less bizarre.

Interesting analysis… it could work, and it certainly does explain phenomena like e.g. why we consider ‘**He see you’ ungrammatical but ‘*I sees you’ merely dialectal. I should note, though, that it doesn’t make the system any less accusative; it just flips it around, so the previously accusative pattern in 3s now becomes direct with both subject and object triggering zero marking, while the previously direct pattern elsewhere becomes accusative with suppression when such NPs are in subject position:

	S	A	O
1s	SUPPRESS	SUPPRESS	-∅
2s	SUPPRESS	SUPPRESS	-∅
3s	-∅	-∅	-∅
1p	SUPPRESS	SUPPRESS	-∅
2p	SUPPRESS	SUPPRESS	-∅
3p	SUPPRESS	SUPPRESS	-∅

(Explanation: something like ‘he tells me’ is now ‘tell-3s.A-1s.O’ //tell// //-s// //-∅// //-∅//, while ‘I tell him’ is ‘tell-1s.A-3s.O’ //tell// //-s// //SUPPRESS// //-∅//. English doesn’t have object agreement, so you have to assign //-∅// to the ‘object agreement affixes’ if you want to include all of S,A,O in your analysis.)

As it happens, this ‘flipped’ system does also explain another oddity in the naïve analysis: the fact that 3s is non-zero while everything else is unmarked, in violation of the animacy hierarchy. I’ll have to think about this.

The past tense doesn't have much bearing on this, though there are elements of deletion in past tense forms like sent and shot.

Not sure what you mean here… could you clarify what ‘deletion’ you’re talking about?

Richard W wrote: ↑Wed Mar 10, 2021 5:12 pm

Zju wrote: ↑Wed Mar 10, 2021 2:54 pm Is it, outside of those dialect areas? I haven't heard anybody talk like that, even as a slip up. Was the discussion about South Wales English all along?

I didn't say which areas had the NSR. Now -s not suppressed by adjacent pronouns is quite common, and "So I says to him" gets 300,000 raw google hits. You've presumably encountered, "You pays your money and you takes your choice". I'm not sure I'd notice the NSR in action - I'd probably dismiss it as a typo.

As I briefly mentioned above, those sound very dialectal and non-standard to me. Synchronically, I certainly wouldn’t count either of these as something I would say, even as a mistake — the only reason they’re not totally ungrammatical for me is that they’ve become fossilised idioms, and because I’ve already encountered such forms in books and such when authors want to represent dialectal speech.

bradrn wrote: ↑Thu Mar 11, 2021 7:44 am Here you go:

g-a-kərə-na
3p.O-1s.A-hit-PRS
I hit them.

n-umbwa-ŋa-kərə-ŋa-na
INV-3p.A-1s.O-hit-INV-PRS
They hit me.

(Incidentally, this also shows the direct-inverse system: the second sentence takes the inverse circumfix n⟩⟨ŋa, and the affix order is consistently ‘less animate-more animate’. But this is irrelevant for our purposes.)

Admittedly, these examples involve pronominal affixes I didn’t introduce in my table above, so here’s an expansion of that table to 1s (which you might notice agrees in a tripartite pattern):

There does seem to be something odd going on there. Is it just odd patterns of syncretism in an old tripartite system?

bradrn wrote: ↑Thu Mar 11, 2021 7:44 am

Richard W wrote: ↑Tue Mar 09, 2021 8:12 pm He writes words has ///write/// ///-s/// ///-∅///.
I write words has ///write/// ///-s/// ///SUPPRESS-s///

And this analysis satisfies the 'universal' that it is the 3s which is the person etc. expressed by zero! It makes oddities like "So says you" and "I wants it" less bizarre.

Interesting analysis… it could work, and it certainly does explain phenomena like e.g. why we consider ‘**He see you’ ungrammatical but ‘*I sees you’ merely dialectal. I should note, though, that it doesn’t make the system any less accusative; it just flips it around, so the previously accusative pattern in 3s now becomes direct with both subject and object triggering zero marking, while the previously direct pattern elsewhere becomes accusative with suppression when such NPs are in subject position:

That's why I think it's unsound to say one or the other is accusative but the other is direct. Additionally, it's reassuring to see similar speech varieties have similar analyses.

I found some good examples of the NSR rule in antebellum white overseers' correspondence from the USA. It was in an article whose main topic was the survival of the NSR in varieties and relatives of AAVE, from Nova Scotia to the Dominican Republic to Liberia. The overseers' English is indubitable English, just slightly odd inflections.

Consigning '-s' on the verb to history is perfectly normal for Norfolk!

bradrn wrote: ↑Thu Mar 11, 2021 7:44 am

Richard W wrote: ↑Tue Mar 09, 2021 8:12 pm The past tense doesn't have much bearing on this, though there are elements of deletion in past tense forms like sent and shot.

Not sure what you mean here… could you clarify what ‘deletion’ you’re talking about?

shoot > shot, send > send.d > sent, sleep > slept, read > read, and light > lit.

Richard W wrote: ↑Thu Mar 11, 2021 3:45 pm That's why I think it's unsound to say one or the other is accusative but the other is direct.

I also find Brad's analysis weird, but only because he seems to insist that non-3s forms follow a different alignment system. It would be more palatable if he said no alignment system.

I've already said why I think he's wrong about the 3p, and he seemed to agree at the time.

On the other hand, I think "no alignment" is correct about the 1p and 2p, and about the past tense— just speaking about morphology, of course.

On the other hand, I find the idea that -s is the basic form and the others are a deletion to be even weirder. The root is found all over in English verbal morphology, and is also the basis for the past tense form and the participles. And changing the analysis because of a supposed universal is highly questionable. "Universals" almost never hold up, and certainly can't be used to browbeat the exceptional languages.

Dialectal forms have to be evaluated within their dialects. I assume we're all talking only about our own dialects.

shoot > shot, send > send.d > sent, sleep > slept, read > read, and light > lit.

Note that this change is very common, just normally hidden by the orthography— e.g. shopped, looked, worked, helped, voiced, passed, laughed, stretched, etc. Final /d/ is common too: turned, seemed, opened, closed, called, starred, followed, loved, used, bagged, stayed, stabbed, nagged, judged, etc.

Richard W wrote: ↑Thu Mar 11, 2021 3:45 pm I found some good examples of the NSR rule in antebellum white overseers' correspondence from the USA. It was in an article whose main topic was the survival of the NSR in varieties and relatives of AAVE, from Nova Scotia to the Dominican Republic to Liberia. The overseers' English is indubitable English, just slightly odd inflections.

Sorry, I've conflated recollections of various articles.

NSR in AAVE: https://historynewsnetwork.org/article/174259
Sinoe, Liberia: https://ling.yale.edu/verbal-s-liberian ... ally-south
Devon and Samanáː httpː//www.sociolinguistics.uottawa.ca/shanapo ... te2004.pdf
(NSR in Liberia is disputed!)
Overseers' English: https://www.google.com/url?sa=t&rct=j&q ... U93clMwLT8

zompist wrote: ↑Thu Mar 11, 2021 5:30 pm On the other hand, I find the idea that -s is the basic form and the others are a deletion to be even weirder. The root is found all over in English verbal morphology, and is also the basis for the past tense form and the participles.

I put it forward as the basic form for the simple present.

zompist wrote: ↑Thu Mar 11, 2021 5:30 pm Dialectal forms have to be evaluated within their dialects. I assume we're all talking only about our own dialects.

I'm primarily talking about dialects for which we'd assume good intelligibility.

zompist wrote: ↑Thu Mar 11, 2021 5:30 pm shoot > shot, send > send.d > sent, sleep > slept, read > read, and light > lit.

Note that this change is very common, just normally hidden by the orthography— e.g. shopped, looked, worked, helped, voiced, passed, laughed, stretched, etc. Final /d/ is common too: turned, seemed, opened, closed, called, starred, followed, loved, used, bagged, stayed, stabbed, nagged, judged, etc.

None of your examples change the stem. In my examples, its vowel shortens, except that its only allophonically shortened in 'send', and in several of them a dental consonant is dropped.

Richard W wrote: ↑Thu Mar 11, 2021 3:45 pm

bradrn wrote: ↑Thu Mar 11, 2021 7:44 am Here you go:

g-a-kərə-na
3p.O-1s.A-hit-PRS
I hit them.

n-umbwa-ŋa-kərə-ŋa-na
INV-3p.A-1s.O-hit-INV-PRS
They hit me.

(Incidentally, this also shows the direct-inverse system: the second sentence takes the inverse circumfix n⟩⟨ŋa, and the affix order is consistently ‘less animate-more animate’. But this is irrelevant for our purposes.)

Admittedly, these examples involve pronominal affixes I didn’t introduce in my table above, so here’s an expansion of that table to 1s (which you might notice agrees in a tripartite pattern):

There does seem to be something odd going on there. Is it just odd patterns of syncretism in an old tripartite system?

Oh, I suppose you could also analyse it that way if you wanted to.

bradrn wrote: ↑Thu Mar 11, 2021 7:44 am

Richard W wrote: ↑Tue Mar 09, 2021 8:12 pm He writes words has ///write/// ///-s/// ///-∅///.
I write words has ///write/// ///-s/// ///SUPPRESS-s///

And this analysis satisfies the 'universal' that it is the 3s which is the person etc. expressed by zero! It makes oddities like "So says you" and "I wants it" less bizarre.

Interesting analysis… it could work, and it certainly does explain phenomena like e.g. why we consider ‘**He see you’ ungrammatical but ‘*I sees you’ merely dialectal. I should note, though, that it doesn’t make the system any less accusative; it just flips it around, so the previously accusative pattern in 3s now becomes direct with both subject and object triggering zero marking, while the previously direct pattern elsewhere becomes accusative with suppression when such NPs are in subject position:

That's why I think it's unsound to say one or the other is accusative but the other is direct. Additionally, it's reassuring to see similar speech varieties have similar analyses.

Now I understand your point of view a bit more: the alignment of the verbal agreement for each person will depend on the analysis. Though either way, you still have a direct/accusative split.

zompist wrote: ↑Thu Mar 11, 2021 5:30 pm

Richard W wrote: ↑Thu Mar 11, 2021 3:45 pm That's why I think it's unsound to say one or the other is accusative but the other is direct.

I also find Brad's analysis weird, but only because he seems to insist that non-3s forms follow a different alignment system. It would be more palatable if he said no alignment system.

But ‘no alignment system’ is itself an alignment system — so-called ‘direct’ or ‘neutral alignment’. And there’s several reasons to consider it as such:

• An alignment with all three possibilities separate (‘tripartite’) is widely accepted, so it makes sense that the opposite possibility should also be considered to be an alignment system.
• It greatly simplifies analysis of systems with optional case-marking, making their analysis consistent with more traditional split systems — e.g. with this analysis, Hebrew can be analysed as having a definiteness split between accusative and direct case-marking (with et as the accusative marker), while English can be analysed as having the same split between pronouns and nouns. Note that both of these are consistent with the animacy hierarchy.
• 
  

  I've already said why I think he's wrong about the 3p, and he seemed to agree at the time.

  I don’t recall this… I do remember that you did give an analysis, but I believe it was consistent with my own.

bradrn wrote: ↑Thu Mar 11, 2021 8:22 pm

zompist wrote: I also find Brad's analysis weird, but only because he seems to insist that non-3s forms follow a different alignment system. It would be more palatable if he said no alignment system.

But ‘no alignment system’ is itself an alignment system — so-called ‘direct’ or ‘neutral alignment’.

Sure, and it's appropriate for (say) Swedish verbs. E.g. the present tense of 'be' is är, for all persons and numbers-- no problem, morphologically this is no alignment, or neutral/direct alignment.

English does have person/number inflection-- it's just highly restricted. I don't accept that any non-inflected form within a paradigm means that that particular form, for some reason, has no inflection. It may be a null marking.

It's tricky to say when we have null marking, and when a type of inflection is simply missing... that's what people have been arguing about, with various edge cases.

I've already said why I think he's wrong about the 3p, and he seemed to agree at the time.

I don’t recall this… I do remember that you did give an analysis, but I believe it was consistent with my own.

Here's the post where you agreed with me; my analysis is just above it.

viewtopic.php?p=41644#p41644

I think I have some more evidence that the distinctively* phonetically zero ending of the English present tense actively agrees with the subject. Some of it comes from areas at the edges of grammatical competence.

• When native standard English speakers who do not naturally use the pronoun 'thou' attempt to use it with the -st ending, it tends to inflect with -s. It is as though the agreement condition were 'singular except first singular', suggesting that 'I' as the effective subjective actively affects the choice of ending.
  
  An annoying complication is that for many native speakers who include 'thou' (or equivalent) in their basic active vocabulary, it traditionally does have the 3s ending for most verbs.
• One of the rules that has been taught in school is that only one of the subject and verb can have the '-s' ending. This is a fairly widespread type of rule, so it is liable to be learnt by induction if it is not taught. Thus, plural noun subjects may actively condition phonetic zero.
• An agreement condition such as "singular except for 'I'" is supported by the relative subject pronoun inducing 1st person agreement if the antecedent is 'I' but 3rd person agreement if the antecedent is 'me'. YMMV.

*So this argument would not apply to Norfolk.

Richard W wrote: ↑Mon Mar 08, 2021 11:18 pm

bradrn wrote: ↑Mon Mar 08, 2021 6:07 pm Perhaps a non-English example may assist. The Lower Sepik language Murik has the following person-agreement affixes in the plural:

	S	A	O
1p	e-	e-	ŋe-
2p	o-	o-	ŋo-
3p	g-	bo-/mbwa-	g-

(The singular has a similar but less consistent pattern; there’s also a direct-inverse system on top of this, but that’s irrelevant here.)

This verbal agreement system clearly has accusative alignment in first and second persons, but ergative alignment in the third person. My claim is that English has a similar system, but with direct alignment in non-3s and accusative alignment in 3s.

I think that table should have the columns ordered A, S, O. It also cries out for a football team diagram showing the order of the prefixes - the regularity of the subject/object correspondences make me suspect dishonesty in your source.

From Bill Foley's article in the Languages and Linguistics of the New Guinea Area: A Comprehensive Guide, it seems that the order of the prefixes is:

g-	bo-	ŋe-	e-
n-	mbwa-	ŋo-	o-

n- is the initial part of the inverse circumfix. My table is speculative for multiple first- and second-person prefixes.

Can you confirm the reading of the table for Murik, please. Would 'They hit us' have the two prefixes bo-/mbwa- and ŋe-? What determines the selection between bo- and mbwa-?

From the scanty examples, I suspect the form mbwa- is triggered by the presence of a first or second person object.

Richard W wrote: ↑Fri Mar 12, 2021 11:59 am When native standard English speakers who do not naturally use the pronoun 'thou' attempt to use it with the -st ending, it tends to inflect with -s. It is as though the agreement condition were 'singular except first singular', suggesting that 'I' as the effective subjective actively affects the choice of ending.

what about the Quakers who generalized "thee" to the nominative? Ben Franklin's autobiography - not a very reliable source, but the only one I know of without looking - has two quotes from Quakers:
- "At any other time, Friend Hopkinson, I would lend to thee freely; but not now, for thee seems to be out of thy right senses."
- "I being thy servant, why did thee not order me to come down? But thee was willing enough that I should stay and help to fight the ship when thee thought there was danger."
which makes it look like 2sg = 3sg in verbal agreement

elsewhere:
- "Young man, I am concern'd for thee, as thou hast no friend with thee, and seems not to know much of the world, or of the snares youth is expos'd to; depend upon it, those are very bad women; I can see it in all their actions; and if thee art not upon thy guard, they will draw thee into some danger; they are strangers to thee, and I advise thee, in a friendly concern for thy welfare, to have no acquaintance with them."
which looks grammatically confused - IMO the most likely explanation is Ben Franklin mangled it in the telling. for one thing, it's the only instance of a Quaker using the form "thou"; for another, it's hard to harmonize "thee was" with "thee art"

Nortaneous wrote: ↑Fri Mar 12, 2021 2:13 pm

Richard W wrote: ↑Fri Mar 12, 2021 11:59 am When native standard English speakers who do not naturally use the pronoun 'thou' attempt to use it with the -st ending, it tends to inflect with -s. It is as though the agreement condition were 'singular except first singular', suggesting that 'I' as the effective subjective actively affects the choice of ending.

what about the Quakers who generalized "thee" to the nominative?

I'm not sure of your point. For one interpretation, for 'thou' include 'thee'.

As to '-s' as a 2s ending, that goes way back. But I believe general awareness of it is quite rare.

Richard W wrote: ↑Fri Mar 12, 2021 3:08 pm

Nortaneous wrote: ↑Fri Mar 12, 2021 2:13 pm

Richard W wrote: ↑Fri Mar 12, 2021 11:59 am When native standard English speakers who do not naturally use the pronoun 'thou' attempt to use it with the -st ending, it tends to inflect with -s. It is as though the agreement condition were 'singular except first singular', suggesting that 'I' as the effective subjective actively affects the choice of ending.

what about the Quakers who generalized "thee" to the nominative?

I'm not sure of your point. For one interpretation, for 'thou' include 'thee'.

As to '-s' as a 2s ending, that goes way back. But I believe general awareness of it is quite rare.

they naturally use it, but don't preserve the EME system

for people who don't naturally use it, I've seen -th in addition to -s, which is probably just "what's an archaic person marking suffix to stick on verbs that need to agree with the archaic pronoun" but could I guess also be explained as a generalization of the "singular except 1sg" pattern

Nortaneous wrote: ↑Fri Mar 12, 2021 2:13 pm elsewhere:
- "Young man, I am concern'd for thee, as thou hast no friend with thee, and seems not to know much of the world, or of the snares youth is expos'd to; depend upon it, those are very bad women; I can see it in all their actions; and if thee art not upon thy guard, they will draw thee into some danger; they are strangers to thee, and I advise thee, in a friendly concern for thy welfare, to have no acquaintance with them."
which looks grammatically confused - IMO the most likely explanation is Ben Franklin mangled it in the telling. for one thing, it's the only instance of a Quaker using the form "thou"; for another, it's hard to harmonize "thee was" with "thee art"

Thou survived long amongst US Quakers according to the Language Log discussion George Fox, Prescriptivist, at least in writing.

Richard W wrote: ↑Fri Mar 12, 2021 1:47 pm From the scanty examples, I suspect the form mbwa- is triggered by the presence of a first or second person object.

My suspicion is confirmed by Bill Foley's Direct versus Inverse in Murik-Kopar. The inverse forms at least also have suffixes countering various neutralisations amongst the prefixes.

First and second person subject and object together are handled by converting the subject to third person, so the 3rd and 4th columns of the football team diagram can be collapsed.

Thus, with all participants in the plural, the present tense 'We/they hit you' winds up in Murik as
nu-mbu-ŋe-kɨrɨ-ŋa-na-ro
INV-3PL.A-2PL.O-hit-INV-PRES-2PL

(There seems to be inconsistency over the citation of the inverse 3rd person ergative bound pronoun - this article cites it as mbu; I think I've also seen mbw. Allomorphy can be a pain.)

I'm getting the message that splitting the agreement by person does not yield totally independent schemes, which is not what Bradrn intended to convey.

zompist wrote: ↑Thu Mar 11, 2021 9:38 pm English does have person/number inflection-- it's just highly restricted. I don't accept that any non-inflected form within a paradigm means that that particular form, for some reason, has no inflection. It may be a null marking.

It's tricky to say when we have null marking, and when a type of inflection is simply missing... that's what people have been arguing about, with various edge cases.

But is there even a difference? The two cases are indistinguishable. I’m totally willing, for instance, to declare that all languages have agreement with both subject and object, with languages like Swedish and Mandarin having all agreement affixes null. This is admittedly a very weird (and completely useless) analysis at the descriptive level, but it does makes some sense typologically — e.g. it explains why polypersonal agreement is so much more common than agreement with one or zero arguments, which is basically restricted to a couple of language families/Sprachbunds.

Richard W wrote: ↑Fri Mar 12, 2021 11:59 am I think I have some more evidence that the distinctively* phonetically zero ending of the English present tense actively agrees with the subject. Some of it comes from areas at the edges of grammatical competence.

• When native standard English speakers who do not naturally use the pronoun 'thou' attempt to use it with the -st ending, it tends to inflect with -s. It is as though the agreement condition were 'singular except first singular', suggesting that 'I' as the effective subjective actively affects the choice of ending.
  
  An annoying complication is that for many native speakers who include 'thou' (or equivalent) in their basic active vocabulary, it traditionally does have the 3s ending for most verbs.
• One of the rules that has been taught in school is that only one of the subject and verb can have the '-s' ending. This is a fairly widespread type of rule, so it is liable to be learnt by induction if it is not taught. Thus, plural noun subjects may actively condition phonetic zero.
• An agreement condition such as "singular except for 'I'" is supported by the relative subject pronoun inducing 1st person agreement if the antecedent is 'I' but 3rd person agreement if the antecedent is 'me'. YMMV.

*So this argument would not apply to Norfolk.

Hmm, I find this evidence actually quite convincing. At this point, I see your analysis as being very plausible, though I’m not 100% convinced yet (mostly because it requires disfixes).

Richard W wrote: ↑Sat Mar 13, 2021 7:21 am

Richard W wrote: ↑Fri Mar 12, 2021 1:47 pm From the scanty examples, I suspect the form mbwa- is triggered by the presence of a first or second person object.

My suspicion is confirmed by Bill Foley's Direct versus Inverse in Murik-Kopar. The inverse forms at least also have suffixes countering various neutralisations amongst the prefixes.

First and second person subject and object together are handled by converting the subject to third person, so the 3rd and 4th columns of the football team diagram can be collapsed.

Thus, with all participants in the plural, the present tense 'We/they hit you' winds up in Murik as
nu-mbu-ŋe-kɨrɨ-ŋa-na-ro
INV-3PL.A-2PL.O-hit-INV-PRES-2PL

(There seems to be inconsistency over the citation of the inverse 3rd person ergative bound pronoun - this article cites it as mbu; I think I've also seen mbw. Allomorphy can be a pain.)

I'm getting the message that splitting the agreement by person does not yield totally independent schemes, which is not what Bradn intended to convey.

Interesting article, thanks for finding it! This will be of use to me in my continued attempt to understand the intricacies of direct-inverse systems.

_________________

Anyway, though I’d still like to resolve this dispute, I have an (unrelated) question which I’d like to ask here: are there any attested cases of tone spreading outside the phonological word? (Including clitics, though I realise that many languages place clitics outside the ‘phonological word’ sensu stricto.)

Re phrasal tone spreading: This is definitely attested. The first language that comes to my mind is Bemba. Nancy Kula has worked on it. You should be able to find some handouts if you google it. The spreading is across words though. Of the top of my head I can't think of any clitic data, but I am sure they are out there.

bradrn wrote: ↑Sat Mar 13, 2021 7:41 am Anyway, though I’d still like to resolve this dispute, I have an (unrelated) question which I’d like to ask here: are there any attested cases of tone spreading outside the phonological word? (Including clitics, though I realise that many languages place clitics outside the ‘phonological word’ sensu stricto.)

Yes! There are some examples in Larry Hyman's Presidential Address from the 2018 LSA annual meeting.

Particularly relevant is Giryama, where the rightmost H shifts to the penultimate mora in a phonological phrase. Take the root gul ‘buy’, which is underlyingly toneless, and bánd ‘break’, which has a high tone. When you construct a verb phrase ‘buy/break wooden bowls’, with ‘buy’ you get low tone throughout the verb phrase, but with ‘break’ you get a high tone on the penultimate mora of ‘wooden bowls’...

1. ku-gul-a mu-vuure ‘to buy wooden bowls’
2. ku-band-a mu-vuúre ‘to break wooden bowls’

(He doesn't gloss these examples in the slides, but if Giryama is like every other Bantu language then ku- should be an infinitival marker, -a a "final vowel", and mu- a plural noun class prefix.)

In addition to spreading/shifting, there are also other kinds of long-distance tonal processes that Hyman mentions in that talk. Tone is really cool!

Thanks for those links! I found them very useful, particularly Hyman’s presentation (and then his chapter in The Handbook of Phonological Theory which that led me to).

Oh, and I found another example of the sort of thing I was looking for: lexically specified floating high tones often end up on the first syllable of the next word.

Is there anyone who can shed some light on the -ong final of Mandarin? Written sources tend to say that it's [ʊŋ], and when I took a course in Mandarin, our teacher said to pronounce it [ʊŋ], not [oŋ]. But I've seen claims here on ZBB that people actually do say [oŋ], and I think I too have heard that pronunciation from Chinese people.