Talk:Slavic liquid metathesis and pleophony

→ and >

In linguistics ">" sign is used to denote the "regular unconditional derivation". Some books use "→" or "⇒" to denote "irregular derivation" (by analogy etc.), so some readers might get confused seeng "→" in place of ">". However, I have no problem using either as long as one of them is unambiguously and consistently used.

Also, it would be good to leave (in parenthesis) forms such as "*eRC/*aRC", because they're much easier/faster to mentally parse than "a liquid consonant (either *l or *r) following either *a or *e and preceding another consonant" ^_^

I've bolded back complete and incomplete because some handbooks make a difference on these terms to distinguish whether lengthening occurs or not..--Ivan Štambuk (talk) 06:21, 8 July 2008 (UTC)

That all makes sense, but I don't think that would be a condition for Wikipedia's conventions on making text bold. We could put "quotation marks" around them. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 06:43, 8 July 2008 (UTC)

OK. I'm not all that familiar with the details of WP manual of style (I'd rather write/translate some articles than spend time reading hundreds of KBs of documentation pages..), so feel free too correct anything you please. --Ivan Štambuk (talk) 09:25, 8 July 2008 (UTC)

Out of curiosity, where have you seen usage of "→" being preferred to as opposed to ">" in linguistics-related articles/books/journals etc. ? Is this some local WP convention? --Ivan Štambuk (talk) 19:42, 9 September 2008 (UTC)

I can't say for certain if I have seen → over >, but as I said in my edit summary, the arrows are clearer. So this could be a Wikipedia convention. It could just be me. Maybe other editors can chime in on this. — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 20:03, 9 September 2008 (UTC)

Clearer to whom exactly? ">" is in linguistics is what "+" and "-" are in mathematics, the very basic notational apparatus. We shouldn't be redefining notation that has been in consistent use for 2 centuries. ">" is a well-defined operator meaning "lexeme on the left side, either reconstructed or attested, yields by regular sound changes lexeme on the right side". Arrows can mean all kinds of things, esp. in mathematical logic and some programming languages.. --Ivan Štambuk (talk) 09:40, 10 September 2008 (UTC)

A quick glance at a phonology book at hand (Problems in the Theory of Russian, Lightner, 1972) shows > to be used for derivation and → for phonological rules. Watson in Phonology and Morphology of Arabic (2002) also uses > for derivation. I'm just of the opinion that an arrow is a clearer arrow than a bracket. I suppose if there's a hard set convention that that trumps my opinion. Should we move this discussion to the talk page at WP:PHON? — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 15:39, 10 September 2008 (UTC)

Wherever. I suggest that we stick to the prevalently used symbols ">" and "<" for denoting the relation of "descended from", until contributors are obliged to use much less common symbols by some policy/guideline page. --Ivan Štambuk (talk) 14:48, 14 September 2008 (UTC)

<grumble>All right</grumble> — Ƶ§œš¹ _{[aɪm ˈfɻɛ̃ⁿdˡi]} 19:38, 14 September 2008 (UTC)

What was the effect on short high vowels?

The article currently talks about the mid-vowels *e and *o, but what about the high vowels *ĭ and *ŭ? It is known that Old Church Slavonic has metathesis there too, although the vowels are not lengthened. But the later South Slavic languages including Bulgarian seem to have undone the metathesis, so it seems that the high vowels were probably dropped by Havlík's law and then a new epenthetic vowel was inserted before the liquid. East Slavic seems to have had no changes at all and retained the diphthongs. Can some information about this be added to the article? CodeCat (talk) 17:40, 30 December 2012 (UTC)

Seems to rather be an orthographic tradition than a metathesis, and has to do with South Slavic development of syllabic r and l. See Lunt's OCS grammar 2.63. --Ivan Štambuk (talk) 06:16, 31 October 2013 (UTC)

In East Slavic the results are still curious. Are there any examples for -ĭRTV (where V is not a yer) or similar sequences word-finally? Havlík's law says that the yer must disappear there, but it never does and always appears in Old East Slavic writing and is always vocalised in the later languages. At the same time there's no evidence for pleophony either, only one yer is written in old texts as far as I can tell? CodeCat (talk) 22:27, 3 November 2013 (UTC)

There is *vьrsta "species" > Russian verstá,with pre-vocalization form attested in Finnish virsta. According to Schenker and some others early *ir/ur, *il/ul became syllabic in soft and hard variants depending on the preceding jer. According to Shevelov no such thing happened however^[1], and individual developments are complex, depending on the region (i.e. language/dialect). In East Slavic these forms never became syllabic and their development was as if jer was always strong, with some mergers occurring in *ьl/ъl which parallel that of *TelT/*TolT merger. There are apparently many theories and exceptions on these developments, and that section will be the toughest to write. The reason for this behavior is that *r and *l were not always a syllable coda in TьRT and TъRT clusters (as opposed to *(T)oRT and (T)*eRT clusters where they always were), where jers would have the same sonority level as liquids, and each could be/become syllable nucleus. --Ivan Štambuk (talk) 23:49, 3 November 2013 (UTC)

I went through this whole issue before and eventually wrote the section on this in History of the Slavic languages#The liquid diphthongs in a very non-committal way. Unfortunately I didn't stick in the references and I don't quite remember which sources I read but it might be enough just to mention that there is a divergence of opinion and cite various sources; it might not be necessary to mention every scholar's particular theory. Benwing (talk) 07:56, 4 November 2013 (UTC)

BTW I agree with Ivan. I don't think there was a South Slavic metathesis and then re-metathesis. It appears that the supposedly metathesized ĭr ĭl etc. sequences were actually syllabic liquids, and behaved differently in OCS from actual rĭ lĭ etc. The two fell together in syllabic liquids after the loss of yers, eventually turning into Vr Vl sequences, and were written rĭ lĭ etc. only as a convention. Benwing (talk) 08:08, 4 November 2013 (UTC)

Ivan -- it seems strange to claim that yers and r/l had the same sonority level. Normally /j/ at least is always more sonorous than any other consonant. Benwing (talk) 08:11, 4 November 2013 (UTC)

@Benwing: Why strange? They are next to each other on the sonority hierarchy.. If the rising sonority/law of open syllables/NoCoda constraint has anything to do with this at all, because it seems to me that these three different types of clusters were three independent developments, with more than a century between them, and the only reason why they are traditionally grouped together in the literature is due to the similarity in shape. oRT is the oldest development and its reflexes are the most uniform, while TъRT/TьRT were resolved very late and their reflexes are the least uniform, and belong to the widest chronological layer, that also happened to coincide with the fall/vocalization of jers and the establishment of new phonotactical rules. --Ivan Štambuk (talk) 11:22, 4 November 2013 (UTC)

@Ivan Štambuk: It seems to me, if you assert that jers and liquids have the same sonority, that would mean that jers would function as consonants in jer-liquid sequences when the liquid became syllabic. But there's no evidence for that, and in general a claim of equal sonority would violate the normal sonority hierarchy and you'd need sources for that. (Even in PIE, where i u r n l m could all be either syllabic or consonantal, there was a hierarchy with i u > r l > n m and possibly even more refined, although with special-casing of initial w-, e.g. *wlk^wos "wolf" and *wrh₂d- "root", cf. German Wurzel.)

I think you're claiming that the reason why TъRT/TьRT clusters were supposedly resolved late (or supposedly not at all in East Slavic) was because the jers and liquids had equal sonority but (a) it's not obvious to me that this late resolution necessarily happened (some authors claim that there was early syllabification followed by desyllabification post-jer-fall); (b) even if there was in fact a late or non-existent syllabification of these clusters, there doesn't necessarily have to be a particular reason for this. It seems reasonable to argue that the NoCoda restraint dispreferred closed syllables, which led to sound changes eliminating these syllables, approximately working their way up the sonority hierarchy (hence nasal codas eliminated before r/l codas), but there's no need for these changes to be deterministic. In general sound change isn't deterministic or teleological. If in fact TeRT/ToRT resolution occurred before TъRT/TьRT, it may have simply been a coincidence. Cf. Brazilian Portuguese, where historically n/m codas were eliminated (nasalization), then l codas (l > w), now r codas (r > h, often dropped entirely), but s codas are still allowed, and stop-consonant codas were eliminated only around the time of l-coda elimination and some are still allowed, esp. the /ks/ cluster: obter "to obtain" [obiˈteh], McDonalds [mɛkiˈdõnawdʒis], Xerox [ˈʃɛroks], Ivan Štambuk [ˈivɐ̃w iʃˈtɐ̃buki] (or similar). Benwing (talk) 19:13, 4 November 2013 (UTC)

@Benwing: Jers were vowels they couldn't act as consonants. But both jers and liquids could turn syllabic as is evidenced in attested forms, which show no predictability in their reflexes, whence it follows that, if the rising sonority rule was motivation at all, both were perceived at the same level in the sonority hierarchy when resolving TьRT and TъRT clusters. But that's just one theory and there are others, within various "frameworks". All sound changes occur for a reason, be it energy efficiency of the vocal tract, entropy increase by phonemicizing new features etc. though their spread often depends on extra-linguistic factors (prestige, transmission errors from parents to children..). If there is no obvious reason, it's because it hasn't been discovered yet, not because it doesn't exist! Or perhaps it cannot be conclusively discovered due to ambiguous evidence. I also find the the syllabic development theory far-fetched - I mean, why would it undo itself, and why develop back into the sequence VR and not the more logical RV sequence, with new vowel (imagine the surprise) also being front or back as was the pre-syllabification jer. Developments in Lechitic are too complex and that theory falls to pieces. The soft/hard syllabic r/l contrast is fantasy not attested in any Slavic language (or any language in the world for that matter). --Ivan Štambuk (talk) 20:23, 4 November 2013 (UTC)

But it's known that front vowels caused a nondistinct palatalisation of a preceding vowel in at least some dialects, and this is preserved in Polish and East Slavic. So it's likely, assuming that syllabic liquids existed in Common Slavic, that older CiRC and CuRC sequences were still distinguished as C'RC versus CRC at some point in certain dialects. The appearance of the later vowels before the liquid would then be a matter of epenthesis. I don't know if this holds any water though, because it implies that CuRC and CRuC should both merge into CRC and give the same outcome, I don't know if that's actually the case in East Slavic or Polish. CodeCat (talk) 20:55, 4 November 2013 (UTC)

I agree with CodeCat here, that the "soft vs. hard" syllabic r/l may simply have been palatalization or lack thereof of the preceding consonant. As for why a syllabic consonant would "undo itself", and why normally VR not RV, neither are strange to me -- same thing happened in the IE languages. Once you allow closed syllables it's probably more preferred to have VR sequences than RV sequences -- e.g. with RV sequences you can get complex onset clusters. As for "all sound changes occur for a reason", I think this is too strong. Some changes are more preferred than others but it's impossible to predict whether a particular change will happen, and changes don't always move in the direction of increased "efficiency" by any definition -- otherwise it would be impossible to have changes like s > ɬ (in Taos) or *ta > tɬa (well-known in Nahuatl, with the more "logical/efficient" changes tɬ > t or tɬ > l occurring in most modern dialects). Sound changes that "undo" earlier sound changes are extremely common -- probably among the most common of all patterns, but frequently not observed historically precisely because of the canceling-out effect. To me a better analogy for sound change is gene mutation -- the changes themselves are constant and fairly random, but most of them decrease the system's overall "fitness" and hence tend to disappear.

In any case this is all speculation ... Benwing (talk) 23:27, 4 November 2013 (UTC)

@Benwing: But the theory goes that it wasn't a palatalization of the preceding consonant, and that it was a palatalization of the sonant itself. The change ьR > 'er in Polish is perfectly normal and expected, while the change r̥' > 'er requires an intermediate stage of magically restoring the supposedly lost jers. It's much more economical to assume that syllabic sonorants were not an archaicism (where they occurred, that is) of an earlier stage, but later developments, with jers being lost in such forms earlier than the general loss of jers. As I said - to my knowledge there are no languages that contrast palatalized and non-palatlized syllabic r/l - these are fantasy segments supported by old theories that placed too much confidence in South Slavic/OCS forms. God forbid that Lechitic and East Slavic were phonologically more archaic in some respects... I've never heard of sound changes undoing themselves consecutively. --Ivan Štambuk (talk) 03:18, 5 November 2013 (UTC)

It doesn't have to be restoring of yers if you see it as epenthesis. I don't know if that's compatible with other instances of epenthesis, which occur in word-final -CRŭ and -CRĭ sequences after loss of the yer. If this idea is correct then you'd expect -CRŭ and -CŭRC- to give the same outcome at least as far as the R is concerned. The palatalisation is then the deciding factor in whether a front or back vowel is inserted as epenthetic vowel. To say it another way: if you assume at first a generic "schwa" is inserted before the syllabic liquid (-CəRC-), then this schwa is then vocalised in one way or another depending on the palatality of the preceding consonant. In Russian for example -CəRC- > -CoRC- while -C'əRC- > -CeRC-. CodeCat (talk) 03:31, 5 November 2013 (UTC)

@CodeCat: So this supposed front/back schwa had the exact same reflexes as jers, with the alleged palatalization disappearing without a trace..too much conspiracy. Why not assume they were jers in the first place? The theory has nothing to do with -CRŭ and -CŭRC- developments were no syllabic sonorants are assumed. In Russian those always acted as strong jers but elsewhere (Belarusian, Ukrainian, Polish..) as normal jers in weak/strong variants. --Ivan Štambuk (talk) 09:39, 5 November 2013 (UTC)

Such assumptions would be OR in this case though. I only tried to reason that the presence of syllabic liquids doesn't exclude the possibility that different vowels could later appear before them. I don't think that's what actually happened but I do think that if the loss of yers is assumed, then epenthesis combined with palatalisation can plausibly give the appearance that they never disappeared at all. Epenthesis did occur in Slovene in particular, but Slovene has no palatalisation and the two yers merged. If epenthesis did occur in some languages, it could have happened in others. CodeCat (talk) 20:20, 5 November 2013 (UTC)

@CodeCat: Not OR but a one of the theories proposed by others (read Shevelov above it has an extensive discussion on merits of both approaches). Your theory that it palatalized the preceding consonant and not the liquid, subsequently causing the front or back schwa to manifest itself and then vocalize in the same manner as jers is OR. Of course, all of the theories (there are others as well) require mentioning, good sides plus the criticism, per NPOV. --Ivan Štambuk (talk) 22:17, 5 November 2013 (UTC)

Well, weren't you saying that the soft-or-hard syllabic liquid was fantasy? I was responding to that statement in particular by arguing that under a different interpretation, the attested forms may be compatible with syllabic liquids. That's OR, sure, but your statement was also unsourced so I assumed it was your own. CodeCat (talk) 23:11, 5 November 2013 (UTC)

Examples!

I could criticise the English in parts of the article, but that would be a bit petty, although I'm tempted to change "parallelly" to the correct "in parallel" (no native speaker would dream of trying to say "parallelly"!. All very well, very learned, but what about some examples, so that we all know what we're talking about - I came here to learn, first of all, what an "open" and a "closed" syllable are, but - no help, I'm expected to know that from the start. Some (preferably lots of) examples, please! Maelli (talk) 14:28, 15 April 2013 (UTC)

Right, ruwiki has in the meantime developed a nice article with many examples that we can shamelessly steal from. --Ivan Štambuk (talk) 06:18, 31 October 2013 (UTC)

This page shouldn't explain open syllable and closed syllable but rather link to them. Benwing (talk) 19:37, 4 November 2013 (UTC)

Use tables

I think it would be better, e.g. when describing the outcomes of CeRC/CaRC or CiRC/CuRC, to use tables rather than descriptive text for language outcomes. Languages down the side, environments along the top. For CeLC/CaLC we may need different environments depending on the particular C's (e.g. hard dental, soft dental, etc.). Benwing (talk) 19:35, 4 November 2013 (UTC)

what is T?

I guess most readers will recognize that C means ‘any consonant’ but I forget what T means: any stop? any voiceless stop? any dental? —Tamfang (talk) 21:04, 24 February 2023 (UTC)