The scatology and physiology of push and pull
« previous post | next post »
Having just written about "Drainage issues" (6/25/23), with a graphic depiction of what causes the problem with the drainage system in question, I am emboldened finally to answer a question that one of my graduate students has been asking about for several years. Namely, why do Chinese say "pull poo / shit / excrement" (lāshǐ 拉屎 / lā dàbiàn 拉大便)? What's the logic of that usage? How can one pull excrement when one defecates? Wouldn't it make more sense to say "push" (tuī 推)? Think about it. A bowel movement involves peristalsis,
And what do doctors (and husbands) always say to a woman in labor? "Push", of course. And the baby comes out from the birth canal.
Bear in mind that the person who asked this question is a native speaker of Mandarin with a strong background in linguistics. To tell the truth, though, I had often pondered the reasoning behind "pull poo / shit / excrement" (lāshǐ 拉屎 / lā dàbiàn 拉大便) when I encountered this common expression during my first year learning Mandarin.
lāshǐ 拉屎 21,800,000 ghits
lā dàbiàn 拉大便 2,150,000 ghits
I wonder, though, whether anyone beside myself and my learned graduate student were puzzled by the notion of "pull poo / shit / excrement" (lāshǐ 拉屎 / lā dàbiàn 拉大便).
The closest I can get to it in this long list of 22 definitions for lā 拉 in Wiktionary is #7, "to drag out; to draw out", but that's still a bit of a stretch, if I may put it that way.
Selected readings
- "Greater and lesser conveniences" (6/25/14) (with links to additional posts)
- "Pull!" (11/11/21) — MUST READ (if you're interested in the subject of this post), with a long bibliography
- "Fecal Intensifiers" (3/25/17)
Russell said,
June 26, 2023 @ 7:41 pm
No stranger than English’s “Take a piss.” A non-native speaker once asked me “Why do you say ‘take’? You are GIVING piss.”
Victor Mair said,
June 26, 2023 @ 7:46 pm
Cf. "give a shit".
Gene Anderson said,
June 26, 2023 @ 7:57 pm
Well, why do we say "take a shit" and "take a piss (leak)"? We don't take them anywhere.
Victor Mair said,
June 26, 2023 @ 8:06 pm
@Gene Anderson:
Are you asking essentially the same question as Russell or a different question?
Victor Mair said,
June 26, 2023 @ 8:08 pm
I think it's like "take a shower", i.e., taking a break from what we normally do.
Victor Mair said,
June 26, 2023 @ 8:12 pm
OED calls take "one of the elemental words of the language;" take up alone has 55 varieties of meaning in that dictionary's 2nd print edition. Basic sense is "to lay hold of," which evolved to "accept, receive" (as in take my advice) c. 1200; "absorb" (take a punch) c. 1200; "choose, select" (take the high road) late 13c.; "to make, obtain" (take a shower) late 14c.; "to become affected by" (take sick) c. 1300.
Take five is 1929, from the approximate time it takes to smoke a cigarette. Take it easy is recorded by 1880; take the plunge "act decisively" is from 1876; take the rap "accept (undeserved) punishment" is from 1930. Phrase take it or leave it is recorded from 1897. To take (something) on "begin to do" is from late 12c. To take it out on (someone or something) "vent one's anger on other than what caused it" is by 1840.
etymonline https://www.etymonline.com/search?q=take
Jacob said,
June 26, 2023 @ 9:57 pm
It has always amused me that – tones aside – 饿 and 屙 are homophones
VHM: https://en.wiktionary.org/wiki/%E5%B1%99
Jacob said,
June 26, 2023 @ 9:59 pm
I forgot to add “in Sichuanese”
Jonathan Smith said,
June 26, 2023 @ 10:35 pm
Mand. la1 拉 'pull' and related issues were discussed at length 10 (!) yrs ago here… yikes, round and round we go like Seinfeld :/
pull
Several etyma have been written "拉" over the course of the last 1000+ years — and different items can collapse together due to pressure to "read" a given written form in a single way (a peculiarly East Asian "egg-corning" mechanism.) So the now-marginal Mand. la2 'break off', written the same way, could be relevant (I guess this reflects the MC -p item); cf. thread cited above. Or Cant. laai6 si2 'have the shits', where laai6 ~'rush (of water)' resembles laai1 'pull'.
TL/DR this verb could easily not really reflect 'pull'
Victor Mair said,
June 26, 2023 @ 11:30 pm
"Where did Chinese tones come from and where are they going?" (6/25/13) certainly was a wonderful post, one of our best Language Log discussions on Sinosphere historical phonology ever (a true classic), including the vagaries of tones and pronunciation of 拉 and related words, not only in Sinitic, but also in Korean and Japanese, but it didn't explain the derivation of the meaning of 拉 in "pull poo / shit / excrement" (lāshǐ 拉屎 / lā dàbiàn 拉大便), which is the aim of this post.
Hiroshi Kumamoto said,
June 27, 2023 @ 6:07 am
拉大便 is already found in 児女英雄伝 Chap. 20 according to 近代漢語大詞典.
For the interchange (?) of take / give in Indian English, see an interesting tweet:
https://twitter.com/avzaagzonunaada/status/1671363762648154114
Victor Mair said,
June 27, 2023 @ 6:43 am
Good one, Hiroshi!
Incidentally, "give and take" is enshrined in English as an idiomatic expression meaning "make mutual concessions and compromises".
Victor Mair said,
June 27, 2023 @ 6:45 am
BTW, I knew the meaning and etymology of "peristalsis" from the time I was in middle school and learned a couple dozen synonyms for "vomit", "reverse peristalsis" (also called "retroperistalsis" and "antiperistalsis") being one of them.
Cervantes said,
June 27, 2023 @ 9:13 am
We also say "take the train" or "take the bus," but we aren't actually stealing it, in fact it's taking us. BTW the usage is the same in Spanish, "Tomar el tren."
Frans said,
June 27, 2023 @ 9:34 am
Eating something like asparagus peel without cutting it up may indeed cause a need to pull to get it out, and when everything's normal perhaps there's little need to discuss it.
But of course, in English one pulls a prank. The verbs may be fairly random.
Rodger C said,
June 27, 2023 @ 9:39 am
George Carlin: "You don't take a shit, you *leave* a shit."
KevinM said,
June 27, 2023 @ 11:17 am
@Cervantes: "Take" in relation to urination might perhaps be in the same family as the idiom "take a meeting"? A Spanish speaking friend (Colombian) from college, fluent in English but sometimes unfamiliar with idioms, invariably said "get a piss."
Jonathan Smith said,
June 27, 2023 @ 12:54 pm
If ctext electronic versions are accurate, 兒女英雄傳 Chap. 20 rather has the expression 拉青屎 la1(?) qing1shi3. The application is interesting in its specificity… I don't think qing1shi3 is current in Mandarin for infants' green liquidy shit, but equivalent tshuah tshenn-sái with the specific verb tshuah 'piss/shit incontinently' is regular Taiwanese.
Incidentally I also see 忽然要出大恭 '[he] suddenly needed to take a shit' in Chap. 31 with interesting chu1 da4gong1. :D
… to repeat the tricky point from the previous thread more clearly, it seems possible that the verb which is now Mand. la1 of in la1 shi3 was originally not 'pull' — maybe it related to incontinent expulsion of liquidy doodoo. I see wiktionary also thinks Cant. laai6 si2 'shit incontinently', written "瀨屎" or "攋屎", is relevant here, while Cantodict thinks there is a Mand. equivalent la(i)4 shi3 with 4th tone…
Victor Mair said,
June 27, 2023 @ 4:08 pm
It's still tricky, but slightly less so than before..
Cf. lāxī 拉稀 ("diarrhea"), where xī 稀 means "sparse, diluted; thin; watery"
dàgōng 大恭 ("excrement; feces") < "big / great / large" + "respect / politeness / reverence" -- quite a euphemism! | | | And here is a remarkable map of Sinitic topolectal equivalents for lāshǐ 拉屎 ("defecate”), 52 varieties spread all over the length and breadth of the PRC, and down into Sinophone communities in Southeast Asia.
John Chew said,
June 27, 2023 @ 5:59 pm
FWIW, in Japanese, it's haiben 排便 (expel excrement) and used where one would say defecate in English.
Chris Button said,
June 27, 2023 @ 6:32 pm
I think "turd" and its association with words like "tear" might provide an answer. The meaning of 拉 seems to be "pull" and "break" and perhaps therefore "pull apart".
I suppose that all ultimately then ties back into 屎 via its clear word family association with senses of "separation, split" as I recently noted here:
https://languagelog.ldc.upenn.edu/nll/?p=59405#comment-1606477
David Deden said,
June 27, 2023 @ 11:38 pm
Turd ~ throwed (expelled away) tossed
Taylor, Philip said,
June 28, 2023 @ 2:40 am
David, the OED concurs with Chris' analysis rather than yours — "Etymology: Cognate with Middle Dutch, Dutch tort < a Germanic base of uncertain origin, perhaps a weak past participle of the Germanic base of tear v.1"
Robert C said,
June 28, 2023 @ 6:25 am
Q: Did you take a bath today?
A: Why? Is there one missing?
Benjamin Orsatti said,
June 28, 2023 @ 7:26 am
(1) I love Language Log — we begin by exploring the push-pull of defecation, and end with a collaborative etymology of "turd." (BTW, one of the best words for "turd" in any language has to be "strónzo" (It.) / "shtrúnz'" (It. dial.). French has "étron", but it doesn't have the same… "earthiness".).
(2) Seems like "take" serves more of a grammatical purpose in English than a semantic one in many cases, no? In other words, it doesn't have any meaning in itself, but it serves to "verbify" (linguists — help! What's the word for this?) a substantive which could, itself, also be a verb: take a piss / piss; take a nap / nap; take a bath / bathe, etc.
But it exists because there's a difference between "pissing" and "taking a piss". If I'm "pissing", I could be sitting right here in my chair piddling all over myself. But if I'm "taking a piss", it's presumed that I've gone off somewhere, in privacy, to a socially-acceptable urine receptacle, and micturated. I can "bathe" in a stream, blood, or adulation, but I can't "*take a bath" in a stream or in blood or in adulation. (But see, "I took a bath on that investment" / *"I bathed on that investment").
TL/DR: "Take" means you've gone off and done something with some degree of forethought in an archetypical, stereotyped manner. A layman's guess, but did I "guess" or "take a guess"?
Coby said,
June 28, 2023 @ 11:40 am
English sometimes has "take" where Spanish has dar ("give"), e.g. dar un paseo {"take a walk").
Hiroshi Kumamoto said,
June 28, 2023 @ 10:41 pm
@Jonathan Smith
You're right. I posted it from faulty memory hours after I had looked it up.
Theo said,
June 29, 2023 @ 4:36 pm
Could 拉 la1 be related to 落 la4 (to leave behind)?
https://en.m.wiktionary.org/wiki/%E8%90%BD#Pronunciation_2
loonquawl said,
June 30, 2023 @ 3:12 am
There is the expression "to pull shit" in english vernacular, is there not?
Terry Hunt said,
June 30, 2023 @ 8:02 am
@ Robert C
Akin to the old (and possibly true) joke about an advertising poster for The Times newspaper stating "80% of Clergymen Take The Times", to which some wag added "The other 20% pay for it."
Benjamin E. Orsatti said,
June 30, 2023 @ 8:59 am
loonquawl,
I was about to say, "not in the Western Hemisphere, that I'm aware of," but then I hit the web search, and found:(https://forum.wordreference.com/threads/the-shit-they-pull.3482105/), which reminded me of the expression. The reason it didn't immediately spring to mind, however, must be because, in that expression, the words "shit" and "pull" are never adjacent. That is, you'd never say. "You're pulling shit."; there's always a word or two between them, as in, "Don't pull that shit on me!" or "What kind of shit are you pulling here?".
Jichang Lulu said,
June 30, 2023 @ 12:13 pm
@Chris Button
What ‘split’ words do you have in mind?
Not that a semantic link with ‘split, tear’ is surprising cross-linguistically. Germanic shit from IE ‘split’ has perhaps been noted already in one of these threads. It’s also there in excrement from Latin.
Common (overlapping) semantic domains for shit to be in seem to be, in addition to ‘tear’, ‘expel’, ‘filth’, ‘manure’. Cf. how 去 ‘eliminate’ has been linked to Tibetan རྐྱག rkyag ‘excrement’ (e.g., Schuessler s.v.; cameo in ‘Ancient eggcorns’). Arguably, an etymology not clearly involving any of these (or other common semantic derivations I might be overlooking) would be more remarkable. Commonly adduced cognates of 屎 came up in ‘Dung Times’ (e.g., Tibetan ལྕི་བ lci-ba, Burmese ချေး hkye:, Tangut 2lhiq4, all ‘faeces’); taken to go back to something like *kləy, STEDT #572. But Sagart has recently argued that two separate etyma are involved in this set, given that some languages have two representatives, one traceable to ~*kl-, one to just ~*k-. The Tibetan, Burmese and Tangut examples reflect the former, but Sagart would place the Chinese with the second group based on Min data. ST aside, I wonder what your views are on the objection to the lateral (cluster) initial in 屎.
Chris Button said,
June 30, 2023 @ 10:24 pm
@ Jichang Lulu
I’m not familiar with the languages in question and so would prefer to refrain from speculation. However, very broadly, it looks to me like an attempt to retro-fit a rather odd looking Old Chinese reconstruction *[qʰ]ijʔ to Tibeto-Burman evidence when a lateral is really warranted in the OC form. Ultimately I think it probably connects with words associated with 多, but that’s another story,
Jichang Lulu said,
July 1, 2023 @ 9:29 am
@Chris Button
My comment only casually brought up the alleged cognates of 屎 and the cross-linguistic commonality of SPLIT~EXCREMENT semantic proximity, perhaps in a distracting way.
I was specifically wondering what some other members of the word family you mention in your comment are.
My other question was precisely how warranted the well known (e.g., Starostin, Zhengzhang, Schuessler) reconstructions with a lateral are. Presumably the existence of your word family would offer new support for that reconstruction, adding to known arguments (such as loan-character uses of 尸).
Sagart's objection to the lateral is independent of any discussion of TB etyma, and of the broader B&S system.
Commonly reconstructed OC *hl- has been observed to give *tšh- in Norman's Proto-Min. The latter is reflected as e.g. Xiamen sh-. 首, 試 are relevant examples. Now 屎 seems to have no reading with an affricate initial (e.g. Xiamen (the dictionary I have at hand…) has s-, which can come from Proto-Min *š- but not *tšh-). Therefore, if these correspondences are regular and the Xiamen and other readings are inherited, 屎 can't have a lateral initial.
Whatever initial is posited instead of the lateral, one would want any OC reconstruction to account for the Min data when the Min data is there.
This objection may or may not apply to items in the rest of your proposed word family. Likewise, there's the issue of why it does or does not apply to, say, 尸, where Sagart agrees with the traditional reconstruction with *hl-; presumably any Min readings with a fricative initial can be recognised as loans. (B&S see early forms of (resp. ) as pictograms of a person passing stool (resp. water) and so don't see 尸 as a phonophore.)
This is how my casual remarks were arguably relevant to the specific questions about your comment. SPLIT-near-EXCREMENT might argue in favour of your word family, Min 屎 might argue against it. It all depends on what's in that family.
Jichang Lulu said,
July 1, 2023 @ 11:32 am
Typo — apologies:
PMin *tšh- gives e.g. Xiamen tsh-
Chris Button said,
July 1, 2023 @ 12:17 pm
@ Jichang Lulu
Yes, the B&S system makes no distinction there between sl- (ʰl-) and ɬ-, which understandably were very close (perhaps even allophonic for many) in the same way as say sq- (kʰ-) and χ-.
I was thinking of things like 移, 侈, 施. Also 多, although B&S have a t- prefix there when it was surely a velar cluster.
Chris Button said,
July 1, 2023 @ 12:19 pm
To be clear sl- gives ʰl- and then overlaps with ɬ- (I doubt anyone could reasonably distinguish ʰl- and ɬ-) but by then ɬ- is already moving.
Jichang Lulu said,
July 1, 2023 @ 1:16 pm
@Chris Button
Thanks. So if I'm understanding this correctly: you reconstruct (for a certain stage) *sl- vs *ɬ- in some cases where B&S have *hl-. Which of those initials do you use for 屎 and which for e.g. 試? And what about 移, 施, 多?
And is your point that having an extra lateral or lateral cluster allows you to explain the Min reflexes better, or that the Min argument as sketched isn't sound in the first place?
Also, fixing another broken bit of my earlier comment…
> see early forms of 屎 (resp. 尿) as pictograms of a person passing stool (resp. water) and so don't see 尸 as a phonophore
Jonathan Smith said,
July 1, 2023 @ 2:18 pm
@Jichang Lulu
Traditional approaches to the onsets of "Old Chinese" basically try to fold together "Middle Chinese" aka rime book onsets with info from the script. So in light of xiesheng contacts, recent presentations interpret the MC rime book onset shu1 書 (often transcribed ś or ɕ) as earlier "OC" hl-, hn- etc. Thus voiceless lateral in 'feces', etc.
—-
totally separately from this, (esp. Norman's) work in comparative Min suggests proto-Min onset categories written *tš- *tšh- *š- (distinguishable from *ts- *tsh- *s- via "alveopalatal" reflexes in some western varieties.)
—-
It is not at all clear how the hypothetical OC voiceless sonorants relate to the (in practice non-straightforward) comparative proto-Min categories *tš- *tšh- *š-. It is not correct to say that "OC *hl- has been observed to give *tšh- in Norman's Proto-Min": often, etyma in MC "ś-" belong to PMin š-, but the situation is complicated… see esp. W. South Coblin, "Alveopalatal affricate/fricative mixing in common Mǐn cognate sets".
—-
B&S (2014) misrepresents Norman's tš- tšh- š- and constituent etyma. One of many examples is 'water' (ś- in MC). This item shows a mixture of affricates and fricatives in Min daughters and thus was left unassigned in Norman (1974), but B&S (2014) assign it to PMin tš- without comment and further treat the "correspondence" MC ś- : (ersatz) PMin tš- as pointing to OC complex onsets like "*s.t-". Then they (that is Sagart) can compare this to TB in a way they enjoy. I would emphasize that across the board, B&S (2014) is unreliable wrt Min data and Norman's hypotheses, and one must always go back to the sources.
—-
Similar problems affects B&S on OC "uvulars". As this is a family website we leave the matter aside. :D
—-
To say that much work remains to be done in understanding the relationships that pertain between Min and mainstream Chinese inclusive of "MC" would be a massive understatement — actually misleading in suggesting that such work is actively being done.
—-
In short the idea of a voiceless resonant in OC 'shit' remains perfectly plausible. BUT Chris Button's style is categorical unsubstantiated statements about "word families" and the like + laments that no one else will ever get it, so given his support for this idea I am automatically deeply skeptical :P
Chris Button said,
July 1, 2023 @ 2:40 pm
@ Jichang Lulu
So I haven't looked into this at all and would need to look a little closer before posting a detailed argument. But it seems to me that 屎 and 試 would have the same OC ɬ- based on Middle Chinese, while the Min evidence (if I'm understanding you correctly) suggests OC *sl- for 屎 but not for 試.
If that is the case, then there's nothing particularly untoward there in terms of the allophonic variation I mentioned above (please excuse the typo of kʰ- for qʰ- in my earlier comment). Perhaps the most well-known example would be 羌 sk- > kʰ vs 姜 x- > k- (attesting the regular hardening in onset position noted by Pulleyblank) as variants of the same word. The phonetic is 羊 ɣ-, so x- rather than sk- is perhaps more likely as the "original" onset for 羌/姜).
Chris Button said,
July 1, 2023 @ 2:54 pm
One pedantic point to add:
There were no "voiceless laterals" / l̥ / in Old Chinese. The same goes for all supposedly voiceless resonants. The use of the IPA symbol for voiceless (subscript circle) is misleading. Phonetically we are talking about something like [l̥͡l] , which in the case of laterals would most likely be [ɬ] versus a phonemic ʰl- (hence sl- ~ ʰl- ~ ɬ). The source of aspiration in OC was an s- prefix (as in Burmese), although Pulleyblank tried hard to find an alternative explanation (cases like k.k- > kʰ- do hold in loanwords like 車 from Tocharian though)
Chris Button said,
July 1, 2023 @ 2:58 pm
So, yes, there is a voiceless lateral fricative, which then is voiceless but is fricated. Other like n̥- would never have been fully voiceless (much like in modern Burmese phonology, for example, where it is only partially devoiced)
Chris Button said,
July 1, 2023 @ 3:08 pm
@ Jonathan Smith
Not sure what you mean. But a lack of appreciation of phonology and phonetics does sometimes make Old Chinese into a game of algebraic symbols with a forced logic that is devoid of much lingustic reality.
Chris Button said,
July 1, 2023 @ 4:40 pm
@ Jichang Lulu
Another thing that occurs to me (again, not having looked at the matter closely, nor being an expert on Min) is that we're simply dealing with a form of the old ts ~ s vacillation that is very common across xiesheng/phonetic series, albeit not so much across etymologically related words. If it helps, you might compare the pronunciations of "sense" and "cents" in English (granted we're talking about onsets here rather than a coda conditioned by a preceding coronal). Having said that, tsʰ- ~ s- does seem to occur across etymologically related words too, as might be expected from a comparative and phonetic perspective–aspirating a sibilant is a tough distinction to maintain!
Jichang Lulu said,
July 1, 2023 @ 5:11 pm
@Jonathan Smith
Many thanks, especially for the Coblin reference. I wasn’t even aware so much material drawing from Norman’s unpublished work was already available. I don’t work in this field and it shows.
In short, you’re saying that one shouldn’t worry too much about Sagart’s objection to the voiceless resonant in 屎 based on Min evidence, because there are too many problems with B&S’s treatment of it (e.g. 水) to accept their claimed correspondences with Proto-Min as correspondences. In that sense, the lateral in 屎 remains as ‘plausible’ as it was in the earlier reconstructions that have one.
Now I had tried, perhaps unsuccessfully, to state the problem with credit to B&S but with as little reliance to their broader claims as logically possible.
is admittedly vague; I didn’t mean it as a ‘meme’ accepting a statement as an axiom because it’s in a popular work. All I meant is that some pre-B&S reconstructions have that onset of their own accord, which in some cases happen to match that Norman initial — I got the examples from my recollection of Norman 1974 and a quick check of Xiamen readings. (My recollection was wrong, I now see, but 深 would have worked just as well — it’s *tšh- in N. 1974, unmixed in N. via Coblin, *hl- in Zhengzhang and others.) This is of course some of the same material B&S use to claim a correspondence here; my point is that we can restrict the discussion to cases where their use of that material seems straightforward.
Indeed, it might be possible to informally discuss comparanda without that turning into assessing the overall merits of one work or ‘school’. The latter kind of discussion is a symptom that a field is sterile and debates are mostly about ‘affiliation’ (mentors, institutional colleagues, gatekeepers).
It seems to me that if someone is going to work on OC at all, somehow they have to try and account for Min data — however messy and poorly studied — and it’s encouraging if anything that more work based on Norman is available than B&S used. Here we have claimed *hl- onsets, of which a couple match (also unmixed) Min *tšh-, while 屎 is unmixed Min š-. The issue is there whatever one thinks of the one published proposed solution discussed so far.
Chris Button said,
July 1, 2023 @ 5:20 pm
I wonder how B&S would make sense of the fact that 矢 (which they reconstruct with a lateral) is used for 屎?
So in the Zuo Zhuan we apparently have 馬矢 for 馬屎 "horse shit". Or is the argument that all trace of the lateral would have disappeared by then in any case?
Chris Button said,
July 1, 2023 @ 5:30 pm
@ Jichang Lulu
I've never fully understood comments like this. The idea that people like Pulleyblank ignored Min data is manifestly false. In particular, I personally thought his way of accounting for the so-called "softened stops" was brilliant.
Chris Button said,
July 1, 2023 @ 5:38 pm
Yes, it looks remarkably like the broad "tsʰ- ~ ts- ~ s-" vacillation I mentioned above.
Jichang Lulu said,
July 1, 2023 @ 5:50 pm
@Chris Button
They are symbols used to discuss a certain level of abstraction. Other than in a phonetic transcription of real-world utterances, it’s the use of dedicated IPA symbols that is potentially misleading. Doesn’t your own work notate a vowel system understood to be quite distant from its phonetic realisation?
Isn’t this, on the other hand, too phonetically precise at this level of abstraction?
This of course happens (e.g. in Mongolic, in that direction).
I don’t think there’s a problem with the phonetic plausibility of the Min reflexes coming from laterals or clusters with laterals. The question rather seems to be how that diversity of reflexes can be accounted together with other sources of evidence, other than as vacillation.
I said nothing about Pulleyblank ignoring Min data. We’re talking about some specific Min data (likely unavailable to many in the past) and its relevance to one specific etymon.
Jonathan Smith said,
July 1, 2023 @ 5:59 pm
Ah, thanks @Jichang Lulu, yes I took your "OC *hl- has been observed to give *tšh- in Norman’s Proto-Min" to mean that "the development OC *hl- > PMin *tšh- is a thing". To reduce the situation to a core fact, we might simply state that "some etyma that belong to MC ś belong to Norman's PMin *tšh-". Yes this "issue is there" and involves e.g. 'deep' and 'rat'… why such canonical examples of Norman's *tšh- corresponding to MC ś should be left out of B&S (2014) is a good thought exercise :D
Objections to B&S (2014) do not concern "schools"; indeed, it is B&S that tend to retreat to such notions (e.g. 2017, where Schuessler's criticisms are held to be due to "divergences within the 'new school'.) The core problem in B&S 2014 is rather, sadly, just data, Min and other. If you have any interest, my paper here presents one specific instance of a more general problem. Wading through details is terribly tiresome… but I guess ultimately necessary as many TBist AAist etc. colleagues assume that at least the Min, etc., data in the book is faithful to the sources listed in the bibliography.
Re: "working on OC", there are no methodological rails (as you know from reading LL) and people with sense (Coblin, etc.) basically quit… there are of course people working more or less seriously on comparative Min, but "OC" makes an appearance in such work only incidentally (a good thing.) Who knows, maybe this work will make a meaningful contribution to our understanding of early Chinese sooner rather than later. But Min has been profoundly impacted by mainstream Chinese and one can only squeeze so much out of extant material… who knows if 'feces' is even really reconstructable to PMin, let alone proto-Chinese…
Chris Button said,
July 1, 2023 @ 6:01 pm
@ Jichang Lulu
I think Matisoff called it “proto-form stuffing” or some such thing.
To your point about the IPA: Yes, but it would help if people did not use misleading symbols. A completely voiceless resonant is a fricative (I think there is a good John Ohala article all about that).
Chris Button said,
July 1, 2023 @ 7:00 pm
Perhaps sonorant would be a better term to avoid any association with vocoids. Or better still, nasal, since I think the Ohala article specifically talks about how devoiced nasals are fricatives. The voiceless lateral fricative has its own symbol of course.
Chris Button said,
July 1, 2023 @ 9:34 pm
Regarding 屍 EMC ɕ- from OC ɬ-, the association is presumably with 死 EMC s- from OC sl-. So, we have proposals that the oracle-bone form for 死 may be interpreted with the meaning of 屍 instead. That works from a phonological and graphic perspective (although whether it is a reasonable interpretation of the meaning of the inscriptional form in the oracle-bones is questioned by Takashima for lack of evidence)
B&S seem correct that 屎 is not connected graphically with 尸/屍 in the earliest inscriptional forms. However, that does not prevent 尸 from serving a phonetic role later on as the script evolved, which B&S seem to take issue with. More significantly, B&S's association of 夷 with 尸 in the earliest inscriptional forms definitely needs refinement. The earliest forms of 夷 do not represent a human figure, and there was just some graphic confusion with 人 (as Takashima has discussed). The graphic confusion is perhaps what later resulted in 夷 and 尸 as variant forms of the same word. The graphic divergence seems larger than it really was. For parallel examples (notably with phonologically unrelated words), consider Takashima's discussions of how 㣇 (for 殺) and 求 came from the same inscriptional form, as did 甾 and 西. Then there is of course the classic case of 音 and 言.
Chris Button said,
July 2, 2023 @ 11:27 am
Straying a little off-topic, but the "additional note" in Pulleyblank (1989) "Ablaut and Initial Voicing in OC Morphology" has a great analysis of where velar nasals correspond with "h" in Min and also Tai loanwords. If you go with a partially aspirated nasal to account for it, then you must assume that the aspiration (or the original s- prefix) simply did not exist in OC originally.
Chris Button said,
July 2, 2023 @ 7:15 pm
@ Jichang Lulu
On the algebra point. Yes, pulleyblank's stuff is admittedly algebra, and on the surface can look quote bizarre at times. His 1991 "Ganzhi as phonograms" being a case in point. But it is algebra that is grounded very solidly in distinctive features. Look beneath the surface at what he's doing, and it's bewilderingly impressive at times. Unfortunately, that is often missed with a cursory or lay reading, and can even be offputting to some readers. His work is the only stuff that I get more out of each time I read. It's like I'm peeling another layer of the onion each time.
Jichang Lulu said,
July 3, 2023 @ 5:04 pm
Who knows. I’m aware of some of the difficulties, and surely unaware of many others. This is an item Norman does reconstruct. Per Norman via Coblin cited abive, the forms are
which gives a realistic idea of the range of variation.
That reconstruction at the Min level faces considerable challenges at the moment doesn’t necessarily have to mean all work in OC must be interrupted. Baltic and Slavic data are crucial to IE, but work on PIE doesn’t seem to have stopped to wait for fundamental questions on Balto-Slavic to be settled — such as whether Proto-Baltic, or even Proto-Baltic ever existed, or how the Baltic and Slavic accents relate to each other and to Greek, Vedic etc.
— and furthermore, as noted, some of those etyma have been reconstructed with lateral onsets to account for phonetic series data. This is the sense in which this is ‘a thing’.
Thanks for this. Fellner and Hill, which it cites, might also be relevant to this discussion.
Appeals to general semantic shift and derivation have their use, but they can easily be abused to justify pretty much anything. This is why I brought up the Min reflexes in response to conjectured links between 屎 and words with ‘move/split’ semantics.
Chris Button said,
July 3, 2023 @ 6:54 pm
@ Jichang Lulu
Unfortunately, I fear you are wasting your time with this one.
The standard reconstruction in Old Chinese has a lateral. There is sound inscriptional, comparative and phonological evidence for that. If someone want to change that, they need to come in with really strong evidence that is not solely premised on the awkward fact that the standard approach does not fit within the confines of their system for old Chinese reconstruction. The entire premise of this discussion seems quite simply wrong.
As noted above, I only like to comment on languages I know something about so that i can avoid the pitfall of drawing erroneous conclusions based on a lack of sufficient understanding. So, let's take Burmese. I could easily reconstruct the word without a lateral and account for almost all the modern Burmese data. On that basis, I am now going to propose that there was an alternative form of the word without a lateral to support the confines of my system for Old Burmese. But that would be nonsense!
***
As for semantics, the point is to find comparable shifts elsewhere to turn conjectures into something more solid. In my dictionary, I only propose semantic relationships when I can find support elsewhere–ideally (for consistency) and usually from Indo-European. I haven't looked at "excrement" closely, but my hunch is that the overlap with "split" will be there.
Jichang Lulu said,
July 3, 2023 @ 7:24 pm
On voiceless resonants, yes, in addition to Burmese there’s work on Icelandic (and surely more recent work).
Is this the Ohala paper you meant?
However, these are examples of a more general issue — categories traditionally defined by VOT end up being based on phenomena other than VOT. This has come up several times on LL, with examples from German dialects and Korean.
Instead of focusing so much on the use of the voiceless lateral symbol, I would rather take this as a further reminder that any IPA symbol in a context like this discussion is just a representative of a broader class supposed to include the value of those symbols. Somewhat like in the ‘Int’l Phonemic Alphabet’ joke attributed to Beckman. In this sense I find overtly non-phonetic writing like ‘hl’ algebraically healthier than IPA in hypothetical forms.
Chris Button said,
July 3, 2023 @ 9:23 pm
Sure. Pick your language. My personal choice would be one of the Kuki-Chin languages that make such a distinction because I am personally familiar with them.
I don't really think so. What we are discussing here at least is sonorants, like n-, and aspirated sonorants, like ʰn-. They may be compared to obstruents, like t-, and aspirated obstruents, like tʰ-. It's pretty straightforward really. Unsurprisingly both tʰ- and ʰn- seem to have come from *st- and *sn- originally in PST/PTB.
No, I meant "The Phonetics of Nasal Phonology" (1993)
My point is that if someone uses IPA symbol then they should be aware of what that the symbol conventionally entails. I have no problem with the use of a subscript circle for voicelessness if the usage is specified to mean "aspirated" rather than "voiceless" since there is a perfectly good superscript ʰ for aspiration that is normally used for that purpose.
Chris Button said,
July 3, 2023 @ 9:37 pm
Quoting myself here:
However, if you must persist with this feacal investigation, might I recommend you dismiss the Min evidence entirely and turn to Mon-Khmer? Shorto (2006) has *ʔic ~ *ʔiə[c] ~ *ʔ[ə]c, which was incidentally loaned into Proto-Northern Chin (of Tibeto-Burman stock) where we have something like *ʔeˀ. And if you squint a little, you might be able to see something like B&S's Old Chinese *[qʰ]ijʔ in there!
Chris Button said,
July 3, 2023 @ 9:54 pm
And to quote myself again (apologies for the lack of humility)
As for semantics, the point is to find comparable shifts elsewhere to turn conjectures into something more solid. In my dictionary, I only propose semantic relationships when I can find support elsewhere–ideally (for consistency) and usually from Indo-European. I haven't looked at "excrement" closely, but my hunch is that the overlap with "split" will be there.
I just took a quick look for "shit" in Calvert Watkins PIE dictionary.
Turns out "chine" (cut of meat including the backbone) is related to "shit". It's interesting that 多 graphically comes from two 肉 "meat" as 夕 …
Jichang Lulu said,
July 4, 2023 @ 7:15 am
> Turns out "chine" (cut of meat including the backbone) is related to "shit"
So is Slavic ‘clean’ (чистый etc), which illustrates the general value of these semantic analogies.
Chris Button said,
July 4, 2023 @ 7:27 am
@ Jichang Lulu
I think you're missing the point. Or just being provocatively facetious.
Of course there is no value in plucking semantic analogies out of the air on a one by one basis. What you are soing is precisely one of my biggest criticisms of most work in this field (including Schessler' fantastic dictionary). People just make one-to-one comparisons, which are liable to be entirely misleading.
The power comes from series of words connected in meaning and sound. Real "word families". When you get it right, unsurprisingly it all starts hanging together.
When you find an isolate, you're most likely onto a loanword.
Chris Button said,
July 4, 2023 @ 7:34 am
One day I'll take a look at the word family for 屎 and perhaps even report back here. Maybe my hunch is wrong and it will turn out to be a loan from Mon Khmer.
Min Khmer ʔic ~ *ʔiə[c] ~ *ʔ[ə]c and Old Chinese *[qʰ]ijʔ . It's nice, no?
It's almost certainly nonsense too though.
Chris Button said,
July 4, 2023 @ 7:59 am
One more point here.
It’s is absolutely paramount to make the distinction. Aspirated nasals in Old Chinese often unsurprisingly evolved into voiceless fricatives in Middle Chinese. However they did not always become voiceless fricatives and were therefore not voiceless nasals (i.e. fricatives) at the stage we are reconstructing. We also have the evidence that they came from things like *sn-. I appreciate that is controversial in Old Chinese because it doesn’t fit people’s nice “systems” (I accept it), but it is not debatable in Old Burmese for example.
The question you should be asking then is: “Do we ever get actual voiceless nasals in onset position where a subscript circle is justified?” For that, I’d refer you to a nice discussion in Ladefoged and Maddieson’s classic tome.
Jichang Lulu said,
July 4, 2023 @ 8:03 am
It was me who noted in these comments that proximity to ‘split’ is perfectly plausible, and unremarkable. All I'm saying now is that the PIE root you're looking at is so richly attested that it can give you arguments to relate ‘excrement’ to all manner of concepts, such as ‘clean’. So the semantic plausibility test via comparison to a family as vast as IE sets the bar too low.
I'm not engaged in any faecal investigations. You referred to 屎 as having a ‘clear word family association with senses of "separation, split"’. I found that interesting and asked for details, and also how the item matches reflexes in daughter languages.
If all you meant is to wonder if various things that might have had lateral initials could be somehow related, inspired by the general semantic proximity of SPLIT and EXCREMENT, then that's fine — surely a cool conversation topic for an informal forum like this. It's also perfectly fine if you haven't actually reconstructed this item or its alleged ‘word family’ so it can't be put to any more stringent test than general semantic hand-waving.
So maybe it was just premature to assert there's a ‘clear word family association’? Would it be fairer to say ‘it might be worth considering if…’?
Chris Button said,
July 4, 2023 @ 8:17 am
Yes, that’s a very fair point! I was clear overstating. Perhaps just prematurely. I’ll try to get to this (possible) word family soon. I’ve already spent far too much time posting here :)
By the way, to my last comment above, things like s- giving th- is a rare but attested sound change. Not dissimilar from hl- giving th- in Old Chinese. But then there are manifold cases across languages in the area where the aspiration has sometimes (regularly across entire languages or sporadically on individual lexemes) been dropped, so hn- gives n- for example. And I don’t think that would be possible if the nasal were fully voiceless.
Thanks for the, as always, interesting discussion
Jonathan Smith said,
July 4, 2023 @ 11:53 pm
@Jichang Lulu or whoever still cares :P
yes our combined remarks "some etyma that belong to MC ś belong to Norman’s PMin *tšh [and] some of those etyma have been reconstructed with lateral onsets to account for phonetic series data" is what I was trying to convey in my 1st comment. So perhaps we are in total agreement that this is a very interesting open problem? And the proposals of B&S may (must) IMO be safely ignored. I mean, as it happens, I personally suspect that all of Norman's PMin "alveopalatal" onsets go back to liquid clusters of some kind at the "pre-PMin" stage… but whether it will be be possible to assemble even an LL-worthy argument to this effect, let alone one which peers deem publication-worthy, is another matter.
Re: Min 'feces', note in particular the "proto-variation" "*ʃi3 ~ *ʃəi3" needed to account for the Inland/Coastal separation. And FWIW it is only Western points, which seem in part or whole to reflect late intrusions, that support the alveopalatal onset to begin with.
Re: Baltic/Slavic analogy — true to a degree but, of extant material that might plausibly be leveraged to triangulate pre-MC Chinese / "proto-Chinese", Min comprises I suppose 95%+ of it. So it feels like a big miss when Min evidence is lacking. I do think though that more, better work can still be done via the traditional study of the script etc… we shall see.
Re: Chris's arguments, his suggestions for early Chinese generally involve metaphors to IE cognate sets and ex cathedra statements like "clear word family association with senses of 'separation, split'." Behind the curtain turn out to be "things like 移, 侈, 施. Also 多"… Chris, again, please at least have the courtesy to frame your speculations as such.
Chris Button said,
July 6, 2023 @ 9:48 pm
@ Jichang Lulu
I've been racking my brains about where I got the idea for the "split" concept and now I recall it.
He's not relating "excrement" (that's my addition), but this is what Schuessler's Minimal Old Chinese (p.213) has to say at the top of the 多 series:
@ Jonathan Smith
On four-way stop distinction for proto-Min (p, ph, b, bh) versus three-way in OC (p, ph, b):
See Pulleyblank "Word families" (1973). Discussion starts top of p.115. What I particular enjoy is then how in his "Typology of IE" (1993:101-104), he compares the effect of his voicing prefix to the constraint allowing PIE *tek- or *dhegh- but never *tegh- or *dhek-. His notes on Wu dialects are interesting too–particularly "Historical & Prehistorical Relationships" (1995:166-167) with his brief note on the spreading of "voiced aspiration" across the syllable.
Jonathan Smith said,
July 7, 2023 @ 4:42 pm
@Chris Button
Haha, as I suspected :D. PLEASE use this as a teaching moment to try to resist your mournful, laconic allusions to (what you perceive to be) the genius of Pulleyblank, Takashima, etc. Simply provide some LL-tailored synopsis of the idea in question along with a reference of some kind — then others will take you seriously and can help you. Or even learn from you, who knows.
To review, upthread (and previously on LL) you remarked: "I personally thought [Pulleyblank's] way of accounting for the so-called 'softened stops' was brilliant." — no specifics, no source, thus non-serious; readers are sad and have no choice but to ignore you.
Following private encouragement, you just above: "On four-way stop distinction for proto-Min [etc.]" — specifics, source , thus serious; readers are happy and can engage you.
Engagement:
(1) You appear not to know what the "softened stops" of Min are. This is a problem entirely separate from the one you refer to above, viz. aspirates in the historical lower register a.k.a. Norman's PMin *bh-. You can read about it on Wikipedia. Better would be Zev Handel's 2003 paper (2003) "Northern Min tone values and the reconstruction of 'softened initials'".
(2) As it happens, Pulleyblank *did* mention Min "softening" offhand at least a couple times — see "Morphology in Old Chinese" (2000), pp. 39-40, where the context is an arguably parallel phenomenon in Vietnamese (this concerns much of the same material B&S have now tried to incorporate), as well as a mention at "Some notes in Chinese Historical Phonology" (1981), p. 285. Pulleyblank did not AFAIK make a careful study of this issue or propose much in the way of new ideas; instead, he references Norman. One cannot do everything.
(3) As it happens, Pulleyblank's ideas about *Bh- in e.g. 1973 are not lost to history, as your lament suggests (like always), but remain of interest and are referenced in current literature including, if I may, at p. 14 note 26 of my paper linked above. Of course there is now much more and better data, so it is possible to tell more precisely when PMin *Bh- or some equivalent is called for and when it isn't and thus to offer more tailored guesses about the morphology involved. You can learn about it, and develop opinions about it, or not.
(4) Of course I (you, others) can be wrong or not know something or not know a lot. For instance: if Pulleyblank actually DID make an argument about Min softening, guess what? I want to know about it. You could help me. Do you get it?
Chris Button said,
July 7, 2023 @ 5:41 pm
I look to have mistakenly used the wrong term. My bad. Yes they are two different things. Thanks for pointing out my momentary terminological slip so succinctly and politely.
Chris Button said,
July 7, 2023 @ 8:22 pm
And since you asked for help so nicely, here's your bone:
Pulleyblank "Dialect Studies" (1991:440):
Now this is not a topic I had planned to discuss. My terminological typo should have been obvious to an erudite Min scholar (I was talking about S. Min and Tai h- corresponding to unaspirated ŋ-), but let's see if I can figure out why I muddled up the terminology in my head …
Maybe preglottalization is correct, particularly if "rhinoglottophilia" is involved–hadn't Norman suggested prenasalization after all? And doesn't Handel suggest that there might be an assocation with voiced aspiration there too? Let's take a look at Pulleyblank's "Morphology in OC" (2000), which you cited earlier along with its internal reference to Pulleyblank 1981 (note there are other dotted references here and there too–1995 "Comments" and, specifically just on Vietnamese, "Review of Honey" (2002):
A lost presyllable with a preceding unaccented schwa vowel, eh? Maybe having an effect not too dissimilar to a non-syllabic glide perhaps? Note how Pulleyblank takes the phonetically reasonable assumption that his non-syllabic glide, which caused (breathy) voicing of obstruents (yes, that same one), was also associated with pre-nasalization. (Incidentally, speculation around a nasal N- prefix is utterly unsupportable in my opinion–the nasalization was surely secondary).
Pre-nasalization of a voiced obstruent can often be used as an articulatory mechanism to retain voicing (accounting for otherwise superficially random shifts like g- to ŋ- without an apparent conditioning environment). You may then get allophonic variation between the plain and pre-nasalized variants. A nice comparison comes from the prenasalized stops of Middle Chinese that gave the stops in Sino-Japanese corresponding to nasals in Middle Chinese (note the actual details are inverted because the nasals hardened to obstruents via fortition in onset position–as you incidentally see in some Min dialects–that offers a separate phonetic account from the one that causes obstruents to nasalize in other environments/dialects/languages)
I exploit the allophonic variation between pre-nasalized and nasalized obstruents as part of my OC reconstruction in its evolution to Middle Chinese. You might like to ponder if Min did the same …
So there's your bone… and here's the disclaimer tied to it: This is merely idle speculation on an internal language forum that should be a fun place to share ideas–any comments should be taken in that spirit.
Chris Button said,
July 7, 2023 @ 8:41 pm
Of course Lolo-Burmese teaches us that another source of pre-glottalization may be the same s- prefix that gives aspiration in Old Burmese (and by extension Old Chinese). Go figure …
Jonathan Smith said,
July 8, 2023 @ 12:18 am
Thanks for the mea culpa, however strained — it is hard what with our bloated egos; some of us never get there. And thanks for this *specific* info including a nice find in Pulleyblank — long live the new Chris Button :D
In short, preglottalization is not promising for Min "softening" due to tonal reflexes. The details might be presented most elegantly in Shen Ruiqing (2018) 早期邵将方言响音声母的声调演变, which includes discussion of Tai. The latter analogy IMO suggests that preglottalized obstruents *are* plausible for Norman's "*Dh-" — I have presented this but not sure it is good enough to publish.
Yes "lost presyllable" is now everywhere to account for "softening," esp. B&S 2014. It must be right in some cases.
Re: "fun place to share ideas" — um, exactly? Replace "Sigh. Pulleyblank had a great idea about X. Cant get no respekt. Sigh" with fuller information, and the fun returns. No?
Chris Button said,
July 8, 2023 @ 8:27 am
While the loss of presyllable could well have triggered it, the occurrences seem too frequent unless we want to make some oretty far-reaching assumptions about the nature of the language. My suspicion, as I try to point out above (and never having looked at the issue closely) is that the loss of a presyllable (and most specifically what remains–as Pulleyblank notes) has a similar effect to something more general, such as the role of Pulleyblank's non-syllabic pharyngeal online and its secondary association with prenasalization.
Chris Button said,
July 8, 2023 @ 8:42 am
*pharyngeal onglide
Chris Button said,
July 8, 2023 @ 10:23 am
The issue I have with all of it, including Pulleyblank’s proposal for the four-way Min stop distinction (hence I conflate it in my head with softened stops as one concept), is that it starts to prefix already prefixed forms (excluding cases where we do have a lost presyllable)—e.g. aspiration comes from a prefix and now that aspirated form gets a prefix. That is by no means unheard of typologically in terms of waves of prefixation, but it does complicate things. My money is on allophonic variation where the same voicing (non-syllabic pharyngeal) prefix is either voicing or it is creating pre-nasalized allophones that may become pure nasals or “soften” depending in the dialect. Interdialectal mixing then just confuses it all. But as I say, I’ve never looked at it closely.
Chris Button said,
July 8, 2023 @ 10:40 am
Which would be problematic for Pulleyblank’s 4-way stop proposal. Meanwhile Norman’s prenasalized “softened” stops seems to affect already voiced stops. Unless what is distinctive is the prenasalization and the actual voicing of the stop varied allophonically or dialectally too? Incidentally, that is why I’ve never bought the idea that the voicing prefix could be an unspecified nasal N- on the basis that loans into Miao Yao attest voiced and voiceless stop variants depending on the time of the loan. Clearly there, what mattered was the nasalization.
Jichang Lulu said,
July 8, 2023 @ 12:14 pm
As the discussion has drifted somewhat, I thought I should ‘bump’ the hypothesis on the main subject of this post that Smith (in turn partly summarising ideas from cited threads) mentioned above. It could be recapitulated like this:
Mandarin 拉 lā ‘excrete’ could reflect a conjectural sense of earlier 拉 (Mandarin Chinese *lap, Mand. lá) ~‘break off’. The first-tone reading would have arisen from a reinterpretation influenced by Mandarin lā ‘pull’ (one of the unusual Mand. sonorant-with-first-tone ‘physical action’ verbs alluded to in, e.g., ‘Comrades, "hike up your skirts for a hard shag"’), whose connection to *lap is unclear. In that scenario, lāshǐ does mean ‘pull poo’ today, but originally didn't.
This remains rather speculative unless one can find clearly pre-Mandarin (essentially, pre-loss of -p) attestations of 拉 meaning ‘excrete’. The Canto item is vaguely relevant but it's unclear how it would support this idea.
If this is true, the semantic association might only involve the scatology of ‘pull’ as an artifact of the writing system, with the etymological connection rather reinforcing the scatology of ‘split’ familiar from ‘excrement’, Germanic ‘shit’ or the earlier ‘divide’ post.
Jichang Lulu said,
July 8, 2023 @ 1:05 pm
Another upthread issue of core relevance to the post (since it manifests itself in B&S's temptative reconstruction of ‘faeces’) is ‘signalling’ in notational devices in reconstructions.
B&S notate the onset of 屎 ‘[q^h]-’. In their system, this means ‘aspirated uvular stop, or something else equivalent under the action of some of the theory's sound changes’; i.e., q^h is a representative of an equivalence class of reconstructed phonemes. To establish what exactly ‘something else’ might be, the reader needs to rebuild the equivalence class by running the relevant postulated sound laws backwards, which the presentation doesn't make trivial.
This thread has much harsh criticism of B&S's proposals but, from a purely notational perspective, the use of brackets B&S are known for seems to be a good idea. It leads to a more honest reflection of what the authors are actually prepared to assert than would result from the arbitrary choice of one member of that equivalence class.
A good idea, but still less good than a more explicit algebraic notation — a capital letter, say, that is somewhere explicitly defined as standing for a closed list of phonemes. That would make the reconstruction easier on the reader, also preventing misunderstandings caused by a literal reading that ignores the brackets.
This illustrates what I meant above when I criticised the use of IPA symbols in (meta-)phonemic scripts used to notate a reconstructed language. The laryngeals of PIE are a good example of how common this (‘algebraic’) practice in fact is, and of how it doesn't avoid or prevent discussion of phonetic realisation. Arguably, Ancient Greek ζ (as in the echidna post) is another, not so good, example.
An upfront ‘algebraic’ notation signals to the reader that the symbols stands for classes of sounds whose definition is theory-dependent and must be made explicit.
IPA in a reconstructed context makes the reader fill featural slots that the theory typically can't argue for. That choice of features isn't even ‘natural’; it's an accident of the history of the IPA, as I meant to suggest with the comment on VOT above.
Chris Button said,
July 8, 2023 @ 9:14 pm
@ Jonathan Smith
I will take this in the tongue in cheek manner that it is surely intended and move on.
@ Jichang Lulu
Personally I don’t like it. It seems non-committal and fundamentally non-linguistic. If you are going to reconstruct a form, then you need to think about the phonological and phonetic implications. I recall reviews of Pulleyblank’s Lexicon complaining about his use of complex IPA symbols. The irony is that when explaining the details of how one form begets another, even more complex IPA symbols with diacritics and tie-bars should really be employed as necessary. The more specific the better.
A good example would be Li Fang-kuei’s final “phonemic” -g coda in Old Chinese, which (if memory serves correctly) was not actually a stop per se, and more akin to Pulleyblank’s glide/fricative approach. The problem though is that the -g is easily misinterpreted as a stop.
For PIE, I would personally specify a “laryngeal” reconstruction and then endeavor that what I say supports that position. For me, that would mean ʔ, χ, ʁ in that order, while admitting a possible x ~ χ allophonic variation and noting that extending the allophonic variation to ʁ could then pose problems regarding the rounding that is dependent on the conditioning by the uvular articulation (the explanation for the reflexes that result could then need a more formidable array of IPA symbols to adequately account for what is going on).
David Marjanović said,
July 12, 2023 @ 1:09 pm
For approximants, [l] in particular, that seems both obviously wrong (approximants can be whispered, and Tibetan doesn't seem to sound like Welsh or Navajo) and almost certainly irrelevant: I'm not aware of any language that distinguishes voiceless approximants from fricatives, and like you I doubt there are any.
That reminds me of Basque where intervocalic [n] has become [ɦ̃] (and then largely but not entirely disappeared; there's even one dialect that still distinguishes it from the inherited [h] as a separate phoneme).
[ʔ] actually makes a bit less sense than [h], both for Bozzone's law (*h₁j > h and *h₁i > hi in Greek) and for the distinction of the Hieroglyphic Luwian signs a and á, plus for the fact that *h₁ behaves like a fricative in terms of where it can occur in a cluster & stuff.
I would not expect a velar to cause vowel coloring and see no other reason to assume one either.
That assumes that *o was in fact rounded at the time…
Chris Button said,
July 12, 2023 @ 9:52 pm
I noted in the comment beneath that the Ohala paper was all about how voiceless nasals were fricatives. The voiceless lateral has its own lateral fricative symbol since that is how it tends to surface.
I agree regarding x ~ χ, and by extension ɣ ~ ʁ. I suppose it's really a question of what the phonemes represent. Positing them as velars fills a gap in the phonemic inventory; posting them as uvulars makes more sense from a phonological/phonetic perspective. Hence the need for precision when talking about such things. In a sense, I would suggest phonemic /x/ and /ɣ/ surfacing as [χ] (or [ħ] — the pharyngeal to really produce the "a" coloring, as has been proposed) and [ʁ] as a compromise. In a reconstructed form, I suppose would then just go with the uvulars χ and ʁ, and just go with the asterisk in front without the use of any slashes or square brackets.
I like the glottal stop and also subscribe to the glottalic theory. I also struggle with *h for h₁ since wasn't h₂ supposed to have played the major role in forming voiceless aspirates?
David Marjanović said,
July 17, 2023 @ 5:02 pm
It's out of fashion again because it causes more problems than it solves and doesn't solve most of the ones it was supposed to solve.
Yes, but only in Indo-Iranian – where, it turns out, *h₁ caused voiceless aspirates just the same. Perhaps start here.
It's possible that *h₁ also caused voiceless aspirates in Greek at least under some currently unknown conditions. *h₂ definitely did not do that in Greek.
Then there's Bozzone's law which strongly argues that *h₁ was [h] immediately before its loss.
David Marjanović said,
July 17, 2023 @ 5:07 pm
…in Greek. Bozzone's law is specific to Greek.
Finally, the Hieroglyphic Luwian distinction of a < *V- and á < *h₁V- makes a lot more sense for [h] than for [ʔ]; I'm not going to assume a word-initial distinction of /ʔ/ vs. zero if I don't really have to.
Chris Button said,
July 17, 2023 @ 9:51 pm
It looks like Kortlandt has written some interesting stuff here. I particularly like the association of pre-glottalization with gemination. He also mentions (Lolo)Burmese in one place, where he is correct to note some confusion around pre-glottalization and aspiration. He seems unaware that an s- prefix lies behind both (as Thurgood 1978 – "Lisu and PLB" notes), but the connection is nonetheless interesting. It also makes me think of Pulleyblank's proposal (based original on a proposal by Shafer) that consonant reduplication could be a source of aspiration in some cases in OC.
There's an interesting comparison with Lolo-Burmese here. Original a series of pre-glottalized stops and aspirated stops (from s-) were reconstructed, but Graham Thurgood (1977) noted that actually only s- should be reconstructed. He struggles to account for the pre-glottalization The shift of sp > to ph
Actually *h₁ as h does sound quite tempting…
Regarding the glottalic theory: all the issues aside, the lack of b- as opposed to g- is a pretty fatal problem for the standard analysis.
Chris Button said,
July 17, 2023 @ 9:56 pm
Wow, that came out wrong! Here's a second try:
It looks like Kortlandt has written some interesting stuff here. I particularly like the association of pre-glottalization with gemination. He also mentions (Lolo)Burmese in one place, where he is correct to note some confusion around pre-glottalization and aspiration. He seems unaware that an s- prefix lies behind both (as Thurgood 1978 – "Lisu and PLB" notes), but the connection is nonetheless interesting. It also makes me think of Pulleyblank's proposal (based original on a proposal by Shafer) that consonant reduplication could be a source of aspiration in some cases in OC.
Having said that, *h₁ as h does seem quite tempting, and it looks like Kummel is wavering between h and glottal in what you cited above.
I'm no expert on PIE by any means. However, all secondary issues with glottalic theory aside, the lack of b- as opposed to g- is a pretty fatal problem for the standard PIE analysis simply on the basis of basic phonetic and phonological plausability.
David Marjanović said,
July 21, 2023 @ 6:40 pm
Glottalic theory has no answer to it. The supposed rarity of [p'] only applies to most of Semitic; in western North America, the inventory of labial consonants is highly restricted or altogether empty, so the almost complete absence of [p'] has to be expected; outside these areas, in the Caucasus for instance, [p'] isn't rare when any ejectives are present.
The lack of phonemic [b] in the standard reconstruction certainly is a problem, but what I don't know is whether this problem goes away if the two voiced series weren't modally and breathy-voiced but stiff- and slack-voiced. (And I forgot where I got that idea from. It wasn't presented at any length, though.)
I think Kortlandt confused two quite different things there, but I can't remember where I read the details.