Linguistic Miscellany Thread

[Travis B.]

Back when I studied Japanese in middle and high school, I specifically pronounced what is written in romaji as ⟨r⟩ as a lateral flap, but when I see it described in most contexts I see it being referred to just as a flap, without any mention of lateralness or at most mention that it is sometimes lateral. I know some people on this board are familiar with Japanese ─ do any of you pronounce this consistently as a lateral flap?

[Travis B.]

I mentioned in another thread how my English can be analyzed as having nasal vowel phonemes. However, the problem with this analysis is that it is purely an analysis ─ there is no consciousness of the English here having any sort of nasal vowel phonemes.

Take, for instance, couldn't [kʰɵːnː(ʔ)(t)]. A radical analysis would be /kɵːnː(t)/ but I would be damned well surprised if anyone thought of the word this way. A conservative analysis would be /kʊdn(t)/, which is far more in line with how most people likely think of it.

Likewise, take can't [kʰɛ̃̆ə̯̃̆ʔ(t)]. A radical analysis would be /kɛ̃t/, but again I'd be surprised if anyone believed this. A conservative analysis would be /kɛnt/, akin to GA /kænt/, which likewise is how most people probably think of it.

From a purely theoretical perspective the radical analyses make sense. They avoid problems with phonemes that do not surface except in very deliberately overpronounced speech, which can be explained away as conscious influence from a standard variety, and reduce the layers of rules needed to explain the realized surface forms. They also take into account anomalies caused by lexically unpredictable elisions that cannot be explained by regular rules.

Yet at the same time, they are not consciously accessible, as shown by the fact that the radical analyses are not directly created when coining or borrowing words. A good example that I mentioned in the other thread was the borrowing of foreign words containing nasal vowel phonemes ─ the nasal vowel phonemes, as borrowed, are first deconstructed into nasal vowel-nasal consonant pairs, and only afterwards may the resulting nasal consonants be lost, using the same rules as in native words.

Another matter is that morphologically-related words demonstrate alternations that hint at conservative analyses. Take bank [pẽ̝ʔk] versus banking [ˈpẽ̝ŋ.kɘ̃(ː)ŋ] ─ the historical nasal consonant lost in bank resurfaces in banking as the syllable boundaries change. This hints at an analysis of /beŋk/, akin to GA /beɪŋk/, to which a separate, productive /-əŋ/ is added. Furthermore, this alternation is present in all similar words, such as think and thinking, thank and thanking, sink and sinking, plunk and plunking, and so on.

However, both analyses cannot be right at once. If we say that the conservative analysis is the right one, we are faced with edge cases such as why twenty is either [tʷʰw̥ʌ̃j], a careful [tʷʰw̥ʌ̃ːni(ː)], or a markedly careful [tʷʰw̥ʌ̃nti(ː)] and not the expected [tʷʰw̥ʌ̃ɾ̃i(ː)] for people such as myself. The radical analysis implies three simultaneous separate analyses, a colloquial /twʌ̃j/, a careful /ˈtwʌ̃ːni/ likely reflecting dialect borrowing, and an overly careful /ˈtwʌ̃nti/ under the influence of orthography, whereas a conservative analysis of /ˈtwʌnti/ cannot explain the gap where [tʷʰw̥ʌ̃ɾ̃i(ː)] does not reliably surface.

(Contrast this with winter, where there are two common pronunciations [ˈwɨ̃ːʁˤ] and [ˈwɨ̃ɾ̃ʁ̩ˤ(ː)] along with a markedly careful [ˈwɨ̃ntʁ̩ˤ(ː)], reflecting /wɨ̃ːr/ and /ˈwɨ̃ɾ̃ər/ (and you could easily argue that these can be combined as /ˈwɨ̃ɾ̃ər/), along with an overly careful /ˈwɨ̃ntər/ under the influence of orthography.)

Another example of where conservative analyses fail is the word Saturday [ˈs̟ɛːʁˤˌde̞(ː)], with the radical analysis /ˈsɛːrˌde/. The problem is that for individuals like myself, the conservative pronunciation [ˈs̟ɛɾʁ̩ˤːˌde̞(ː)], which is required by the conservative analysis /ˈsɛtərˌde/ (GA /ˈsætərˌdeɪ/), scarcely surfaces at all, something that the conservative analysis simply cannot explain.

So if neither the radical analysis nor the conservative analysis is right, then what? Do we chalk the concept of phonemes and phonemic awareness up to the influence of literacy? Do we dismiss any notions of phonemic awareness having any reality and declare supposed evidence for the conservative analysis, such as restrictions on word-coining and borrowing and effects of suffixing, up to a mixture of the influence of literacy and frozen historical sound changes?

[zompist]

So if neither the radical analysis nor the conservative analysis is right, then what? Do we chalk the concept of phonemes and phonemic awareness up to the influence of literacy? Do we dismiss any notions of phonemic awareness having any reality and declare supposed evidence for the conservative analysis, such as restrictions on word-coining and borrowing and effects of suffixing, up to a mixture of the influence of literacy and frozen historical sound changes?

It kind of sounds like you're reinventing phonological theory. I've certainly seen theorists who do think phonemes are more a corollary of alphabetic writing systems than a real thing, and others who think that phonetic realizations are laboriously put together by applying rules that look like the last few centuries of sound change.

Personally I find the standard idea of phones and phonemes pretty useful. A universal about brains is that the brain does a lot of things that are not directly accessible to consciousness. (A cute example: most people can't explain how they change lanes in a car, in a way that would teach correct behavior to a robot. Maybe they knew in driver ed, but now they just do it without thinking.)

Michalel Tomasello has pointed out that children seem to learn language construction by construction, not following adult rules at all. I wonder if this has been applied to phonology. Brains don't have to operate like grammar books, with a systematic, coherent set of rules. Maybe there are separate phonologies for (say) words learned early on, words learned late on from hearing, and words learned from dictionaries.

[linguistcat]

Back when I studied Japanese in middle and high school, I specifically pronounced what is written in romaji as ⟨r⟩ as a lateral flap, but when I see it described in most contexts I see it being referred to just as a flap, without any mention of lateralness or at most mention that it is sometimes lateral. I know some people on this board are familiar with Japanese ─ do any of you pronounce this consistently as a lateral flap?

It definitely has a lateral component for some speakers. But I think while it's an interesting feature, it's not necessary to correctly pronounce the sound. But I practiced saying it that way so now that's just how I pronounce it.

[Travis B.]

So if neither the radical analysis nor the conservative analysis is right, then what? Do we chalk the concept of phonemes and phonemic awareness up to the influence of literacy? Do we dismiss any notions of phonemic awareness having any reality and declare supposed evidence for the conservative analysis, such as restrictions on word-coining and borrowing and effects of suffixing, up to a mixture of the influence of literacy and frozen historical sound changes?

It kind of sounds like you're reinventing phonological theory. I've certainly seen theorists who do think phonemes are more a corollary of alphabetic writing systems than a real thing, and others who think that phonetic realizations are laboriously put together by applying rules that look like the last few centuries of sound change.

The key thing to me is that it seems like neither approach is correct. The very consistent distribution of plosive realizations in a given variety of English, which holds firm not just for native words but also for coinages and loans, convinces me that at least at some level that phonemes are real. Yet at the same time, the idea that speakers are constantly subconsciously playing back the last few centuries of sound change seems fundamentally flawed.

Personally I find the standard idea of phones and phonemes pretty useful. A universal about brains is that the brain does a lot of things that are not directly accessible to consciousness. (A cute example: most people can't explain how they change lanes in a car, in a way that would teach correct behavior to a robot. Maybe they knew in driver ed, but now they just do it without thinking.)

The big problem I have here is do phonemes themselves have to be consciously accessible, or can there be restrictions on the conscious application of phonemes for coining and borrowing that need not apply to internalized lexemes?

Michalel Tomasello has pointed out that children seem to learn language construction by construction, not following adult rules at all. I wonder if this has been applied to phonology. Brains don't have to operate like grammar books, with a systematic, coherent set of rules. Maybe there are separate phonologies for (say) words learned early on, words learned late on from hearing, and words learned from dictionaries.

Definitely. In my own case, I seem to have a strong dichotomy between internalized, lower-register phonology and self-conscious, standard language-influenced higher-register phonology.

[zompist]

Yet at the same time, the idea that speakers are constantly subconsciously playing back the last few centuries of sound change seems fundamentally flawed.

Agreed. I'm sure you know, but in case others don't, that's precisely the position of Chomsky & Halle in Sound Patterns of English.

This is a pitfall in many areas of linguistics... do we have a rule for such-and-such a behavior, or are we just reproducing historical changes? If, say, the speaker applies a rule to novel situations, that's evidence it's a real rule.

The big problem I have here is do phonemes themselves have to be consciously accessible, or can there be restrictions on the conscious application of phonemes for coining and borrowing that need not apply to internalized lexemes?

I don't think phonemes have to be conscious, but I think they are discoverable by the usual tests (minimal pairs etc.).

An analogy might be the tests optical scientists use to find an individual's color response curves. Your unique green might not be the same frequency as mine. But we both reproducibly have a unique green, even if (before the experiments) we couldn't point it out on the spectrum.

Some bits of phonology are less supported than others. E.g. I think it could be argued that we have separate phonological inventories for initial, medial, and final position. Phonologists tell us that top, battle, pot all have /t/, but does the language really care? All we actually need to do is distinguish each sound from those that can occur in the same position.

(And yes, morphology can change the position of a phoneme. But this is also precisely where we get irregularities and phonological puzzles.)

[Travis B.]

Some bits of phonology are less supported than others. E.g. I think it could be argued that we have separate phonological inventories for initial, medial, and final position. Phonologists tell us that top, battle, pot all have /t/, but does the language really care? All we actually need to do is distinguish each sound from those that can occur in the same position.

(And yes, morphology can change the position of a phoneme. But this is also precisely where we get irregularities and phonological puzzles.)

Productive morphology can often be quite enlightening. Take the neologisms swat [sʷwɑʔ(t)] and swatting [sʷwɑɾɘ̃(ː)ŋ]. We can tell by morphology that [ʔ(t)] and [ɾ] are probably the same phoneme here, as these are neologisms and thus frozen historical sound changes should not apply.

[Travis B.]

I would say that unless a sound change can be demonstrated to apply to new coinages, loans, or innovative morphology or cliticized syntax there is no reason to assume that it is currently applicable.

Take the example we had been discussing not too long ago on here of /kn-/ and /gn-/ > [n] in English. It is clearly not a productive sound change in some English varieties such as my own as shown by its inapplicability to names like Knuth or the GNU project, yet at the same time it is still productive in other English varieties as shown by some of our British posters finding it quite difficult to pronounce initial /kn-/ (and when forced to attempt to pronounce it compromising by turning loaned /kn-/ into [kʰən]).

As a result, we can test whether a sound change should be considered part of someone's internal phonology or whether it should be considered a historical sound change, and unless we have evidence of the former we should presume the latter so as to assume the simplest synchronic phonological processes.

About affixation, a good test as to whether a sound change that it may or may not invoke is synchronic is whether one automatically applies it to newly-learnt words where one has not actually heard the word's full paradigm. If someone correctly derives morphologically-related forms without having heard them or other forms to which the same change has applied, then it should be assumed that the underlying phonological processes are synchronic rather than only diachronic.

[zompist]

(And yes, morphology can change the position of a phoneme. But this is also precisely where we get irregularities and phonological puzzles.)

Productive morphology can often be quite enlightening. Take the neologisms swat [sʷwɑʔ(t)] and swatting [sʷwɑɾɘ̃(ː)ŋ]. We can tell by morphology that [ʔ(t)] and [ɾ] are probably the same phoneme here, as these are neologisms and thus frozen historical sound changes should not apply.

Sure, I agree that these are evidence. It's just not entirely conclusive, as there's also analogy.

[Richard W]

Take the example we had been discussing not too long ago on here of /kn-/ and /gn-/ > [n] in English. It is clearly not a productive sound change in some English varieties such as my own as shown by its inapplicability to names like Knuth or the GNU project, yet at the same time it is still productive in other English varieties as shown by some of our British posters finding it quite difficult to pronounce initial /kn-/ (and when forced to attempt to pronounce it compromising by turning loaned /kn-/ into [kʰən]).

I have seen an American unmaliciously claim that initial [kn] was not a humanly possible combination. (I successfully counterclaimed that Germans were human.) Note also that the realisation of standard Khmer initial /kn/ is [kʰən], with the aspiration helping to mark it as phonetically a cluster. (The aspiration is inconsistently noted in the spelling, as seen in the borrowing Khmer and the transliteration Phnom Penh.) The vowel length may also be significant in identifying it as a consonant cluster. There may be similar things going on with Thai impure clusters, or they may be an artefact of literacy. I wouldn't seriously claim that English tmesis demonstrates /tm/ > [m], nor that we have initial /ps/ > [s] just because we are taught that the 'p' is silent in that combination.

About affixation, a good test as to whether a sound change that it may or may not invoke is synchronic is whether one automatically applies it to newly-learnt words where one has not actually heard the word's full paradigm. If someone correctly derives morphologically-related forms without having heard them or other forms to which the same change has applied, then it should be assumed that the underlying phonological processes are synchronic rather than only diachronic.

NEWSFLASH! Most English speakers are literate. Thus the result of adding a morpheme may depend on a word's spelling. Thus the transformation of an obscure vowel before the suffix -ian generally depends on its spelling.

[Raphael]

I have seen an American unmaliciously claim that initial [kn] was not a humanly possible combination. (I successfully counterclaimed that Germans were human.)

Thank you, that's a very generous thing to do for a British person.

[Travis B.]

Take the example we had been discussing not too long ago on here of /kn-/ and /gn-/ > [n] in English. It is clearly not a productive sound change in some English varieties such as my own as shown by its inapplicability to names like Knuth or the GNU project, yet at the same time it is still productive in other English varieties as shown by some of our British posters finding it quite difficult to pronounce initial /kn-/ (and when forced to attempt to pronounce it compromising by turning loaned /kn-/ into [kʰən]).

I have seen an American unmaliciously claim that initial [kn] was not a humanly possible combination. (I successfully counterclaimed that Germans were human.)

I should note that what I said was not universal for Americans, as there are Americans who will change initial /kn/ in names to [kʰən], per the pronunciation Wikipedia gives for Knuth.

Note also that the realisation of standard Khmer initial /kn/ is [kʰən], with the aspiration helping to mark it as phonetically a cluster. (The aspiration is inconsistently noted in the spelling, as seen in the borrowing Khmer and the transliteration Phnom Penh.)

One thing this discussion reminds me of is how a Scottish person may argue that final [lm] or [ɫm] or like are impossible, based on the evidence that Scottish English, Scots, and Scottish Gaelic all lack this cluster (as demonstrated by cases of SSE [ˈfɪɫəm] for film).

The vowel length may also be significant in identifying it as a consonant cluster. There may be similar things going on with Thai impure clusters, or they may be an artefact of literacy. I wouldn't seriously claim that English tmesis demonstrates /tm/ > [m], nor that we have initial /ps/ > [s] just because we are taught that the 'p' is silent in that combination.

A good sign is whether new coinages or loans allow a particular cluster. For instance, in the case of tmesis I have to remind myself that you are not supposed to pronounce the ⟨t⟩, implying that /tm/ > [m] is not a living sound change for me, whereas for /ps/ > [s] I would not ever even think of pronouncing the ⟨p⟩, implying that I have a current rule forbidding initial /p/ before /s/.

About affixation, a good test as to whether a sound change that it may or may not invoke is synchronic is whether one automatically applies it to newly-learnt words where one has not actually heard the word's full paradigm. If someone correctly derives morphologically-related forms without having heard them or other forms to which the same change has applied, then it should be assumed that the underlying phonological processes are synchronic rather than only diachronic.

NEWSFLASH! Most English speakers are literate. Thus the result of adding a morpheme may depend on a word's spelling. Thus the transformation of an obscure vowel before the suffix -ian generally depends on its spelling.

I do give you that, so I should amend the above to apply to words learnt through hearing them rather than reading them.

Just because someone is fully literate in a standard language does not mean that one is literate at all in one's native speech variety. Take Scottish English and Scots for instance ─ many people who speak both are fully literate in Standard English but have no literacy in Scots.

Likewise, even within English, just because one is literate in Standard English does not mean that one is literate in one's native dialect of English, as one's native dialect of English is liable to be purely spoken. This certainly applies to myself, I should say.

[jal]

Regarding ps-, I'd argue that it is meaningless for this discussion (sound changes), as no-one illeterate would ever suspect a "p" before the "s", so "ps" > /s/ is 100% a spelling rule and nothing else.

With regards to loans, I think they are unreliable as test for allophonic rules, as they quite often (at least in Dutch, and I'd say in English too) contain sounds that do not exist in the language's phonemic inventory outside loans.

I would say that unless a sound change can be demonstrated to apply to new coinages, loans, or innovative morphology or cliticized syntax there is no reason to assume that it is currently applicable.

Don't you mean "active" instead of "applicable"?


JAL

[Travis B.]

Regarding ps-, I'd argue that it is meaningless for this discussion (sound changes), as no-one illeterate would ever suspect a "p" before the "s", so "ps" > /s/ is 100% a spelling rule and nothing else.

If someone would naively borrow initial <ps> as /ps/ that implies that someone doesn't have a rule forbidding initial /ps/ -- and the fact that we don't see this implies that there is still a current rule forbidding initial /ps/.

With regards to loans, I think they are unreliable as test for allophonic rules, as they quite often (at least in Dutch, and I'd say in English too) contain sounds that do not exist in the language's phonemic inventory outside loans.

They are less reliable than coinages and productive morphology, of course, but they can still provide significant information (e.g. that initial and final /ʒ/ are disfavored by many native English speakers, as shown by their often being shifted to /dʒ/, and that /ʒ/ is often perceived as a foreign phoneme as shown by its frequent hyperforeign use in Beijing and parmesan).

I would say that unless a sound change can be demonstrated to apply to new coinages, loans, or innovative morphology or cliticized syntax there is no reason to assume that it is currently applicable.

Don't you mean "active" instead of "applicable"?

Both work here.

[zompist]

People were talking about shortening number names in the Conlang Random Thread, so naturally I got to wondering what the longest natlang names for numbers are.

I'm just eyeballing this, but it's close enough:

1 (N Nambilkuara) – naĭúkənadaiş̌
2 (Kuwama) – parkatangkenʸ
3 (Bororo) – ure póbe ma jéw metúya bokwáre
4 (Chaap Wuurong) – pūlīt whummin bā pūlīt whummin
5 (Wathawurung) – bulatj-ba-bulatj-ba-kuimatj
6 (Hamtai) – feca hatuanga akapu hatuanga fati
7 (Moro) – ðέέnaŋ náŋáƚḿέñ́έ gáɽijan
8 (Enga) – mange-lao-waketau
9 (Bunabun) – ambentavdji’rukənem
10 (Mesem) – bainimbegere baje delang

[Travis B.]

4 (Chaap Wuurong) – pūlīt whummin bā pūlīt whummin

Let me guess, two in Chaap Wuurong is pūlīt whummin.

[bradrn]

Oh, somehow I missed that the first word translates the number one, the second two, and so on. I think I read the digits as mere list entries.

[zompist]

4 (Chaap Wuurong) – pūlīt whummin bā pūlīt whummin

Let me guess, two in Chaap Wuurong is pūlīt whummin.

Yep!

Curiously, from other languages in Kulinic, pūlīt is etymologically '2'. I don't know what whummin means. But the language is similarly extravagant with '1', kæp yang gnūrak, where the family comparisons suggests kæp itself is '1'.

[Raphael]

Perhaps, just speculating, something like "The honored and sacred number one [or two]"?

[zompist]

Perhaps, just speculating, something like "The honored and sacred number one [or two]"?

It's a dead language, so it may well have been recorded very early, by someone not trained as a linguist. (That set was sent to me, so I don't have the original source.) So someone trying to elicit '2' maybe got something like 'the number 2' or 'second finger' or whatever.

But equally possible, it's something grammatically required in the language, for whatever reason. As an analogy, Dungan Chinese has yigɨ lyongɨ sangɨ for 1/2/3. The gɨ is not mysterious; it's a measure word that's got affixed to the numbers.