Immediately after the passage quoted from Kivelson in yesterday’s post comes an even more interesting one dealing with the term translated there as “uninhabited field.” In my innocence, I thought that was relatively straightforward, but no:
The vast majority of properties mentioned in seventeenth-century sources were pustoshi, uninhabited but plowed, or formerly plowed and settled, plots. Their evocative name, pustosh’, derived from the linguistic root meaning “empty,” “empty spaces.” They evoke the perpetual shortages of labor encountered by Muscovite landlords as they attempted to eke a living out of their meager estates, and the ever-harsher efforts of the state to bind fleeting taxpayers to predictable locations. “In the village there are old sites of houses, now empty, in the pustosh’ that was the village Savelovskoe, also called Bulgakova.” In other circumstances, on other maps, colonizing powers deliberately presented empty lands, devoid of human settlement. […] In the Russian context, however, emptiness was a problem, not a solution. This dilemma is evident in a map from Uglich that lists “field and unplowed woods of the pustosh’ Nikola, which borders the village of Tolpygino which belongs to the Nikitskii Monastery. The wood and hay field and swamp of Tolpygino border the hay fields and woods of the pustosh’ Nikola, but the woods are held jointly by Nikola and Tolpygino. And between the pustosh’ Nikola and the village Tolpygino there are a spring and a ravine, but no boundary markers at all.” […]
Not all empty land qualified as pustoshi. Pustoshi were recognizable by particular priznaki, signs or indicators known to members of this agrarian society. A proper pustosh’, like one described on a map from Kaluga Province, was formerly a point of settlement: “Pustosh’ Fedosovo of the Trinity-St. Sergius Monastery. This was formerly a village.” Distinguishing carefully between a “pustosh’” and empty land that was not a pustosh’, peasant witnesses in Moscow Province testified that “pustosh’ Shchadra is on the River Shchadra at the end of the field of the village Zhdanovskoe. But in the second spot there are no signs of settlement.” […]
In another case, this distinction becomes more explicit. One of the litigants protested the dishonest and inaccurate map that the local town clerk had sketched of his holding. Having interviewed only biased witnesses, the Vladimir town clerk “drew the map according to their testimony, as if there were a pasture for livestock reaching from that pustosh’ to the River Ialma. But on that so-called pustosh’ up to this day there is no trace of settlement whatever and no hamlets belonging to anyone along the Ialma near that pustosh’, and there is no one at all to drive livestock from the pustosh’ [to the pasture]. And thus the criminality of those witnesses and the clerk became clear.” The claim advanced in the map was patently preposterous. A pustosh’ was a field that had once been inhabited and worked but was no longer. A place with no signs of habitation was something else altogether: a meadow, a field, or swamp. […]
Seventeenth-century vocabulary abounded in terms for emptiness. An astonishingly rich array of terms conveys a sense of the extent of the depletion or echoing emptiness and openness of even central Muscovy. Pustoshi adjoined porrozzhie zemli, unclaimed land. Pomest’e land was carved out of dikie polia, wild fields. Primernye zemli, distant fields assigned to one or another settlement, add to the sense of distance. A pozhnia was a wet, low-lying field, suitable for use only as hay fields or pastures, marginally useful but nonetheless passionately contested.
Another family of terms refers to fields and pastures recently won back from the forest or the open steppe, land reclaimed from the marsh, clearings, and wetlands. This vocabulary communicates powerfully the give-and-take between cultivated and uncultivated, claimed and unclaimed land. A wealth of terms refers to swamps, marshes, bogs, gullies, and ravines, which also were fiercely claimed and contested.
It turns out there’s another word, осёлок [osyolok], presumably derived from село [selo] ‘village,’ that’s synonymous with pustosh’. Perhaps the cliché about Eskimo words for snow could be replaced by Muscovite words for empty land.
Addendum. R Devraj on Forty Names of Clouds, about the language of the Thar (“The act of naming — chhinto for a drizzle of rain or ghuTyo for the asphyxiating stillness of un-raining clouds — is a way of paying homage, recognizing worth, according importance of these events that are vital to their survival”).
I must learn all these words and nuances for empty land, I feel a novel coming.
In German there’s the word Wüstung for a village that was left and not resettled – which apparently only happened during the plague. Wüste today means “desert”.
a village that was left and not resettled
a ghost town?
Remains of ruins of a ghost village.
-criminality
My favorite 17th century Russian word. It’s vorovstvo ‘thievery’ in the original, of course.
Full connotations are still not quite clear to me after reading tons of documents. Yes, it’s more or less ‘criminality’, but there are some nuances.
Vory ‘thieves’ are mostly political criminals, their crime is against interests of the sovereign. Even if they are stealing some land, they are committing crime against the Tsar and this makes them ‘vory’, not the act of stealing.
For fans of SF, Vor in Vorkosigan saga of Lois McMaster Bujold is of same origin. Only in her Barayar, vory became the local aristocracy.
Bujold has said many times that this is sheer accident, and she doesn’t remember the origin of the prefix Vor- (which is not explained in the books either). Indeed, in both the official and the fan translations of her work into Russian, the prefix is given as Фор-, as Вор- would be just too ironic. After Bujold was informed of the coincidence, she picked it up and started playing with it, having characters say things like “It’s true, then, Vor does mean thief!” However, Vorkosigan was consciously derived from the surname Косыгин with the nobiliary prefix.
The Counts, the 60 people who control the regions of Barrayar, are named in English as short for accountant, with supposedly no link to the historic title on Earth. The Russian pun is given by WP.ru as “барраярский титул «граф» восходит к термину «графа», которыми были заполнены налоговые ведомости” — alas, Google Translate is unhelpful. I recognize graf as a borrowing from German.
In general, the fan translations seem to handle names of Russian origin (of which there are lots) better than the official ones. The character known in English as Ludmilla Droushnakovi appears in the fan versions as Людмила Друшнакова, unchanged except for proper gendering, but is mangled by the official versions as Люймилла Друшикко, Ghu knows why. Can’t they read?
Russian pun is indeed ingenious.
They translate count as Graf /noble title of German origin/ and explain that it comes from graph. In Russian, graph means column (eg, in tax declaration), not chart as in English.
Chancellor.
As a teenager I lived a mile or so from Øde-Hastrup, which however wasn’t waste any more. (My sources for etymology hint at cognacy with Latin vastus aso, and that German öde/wüst may therefore be a doublet).
Pustosz must have existed in Polish, since we have a verb derived from it, pustoszyć ‘devastate’. But we still have a large word-family based on the adjective pusty ’empty’ (= пусто́й):
— pustka ‘void, nothingness’;
— puszcza ‘wilderness’ (arch.), now used in the sense ‘primaeval forest’;
— pustać (bookish) ‘wasteland, heath’;
— pustynia ‘desert’;
— pustelnia ‘hermitage’;
etc.
See also Hungarian puszta ‘grassland, steppe’.
Interesting! Russian (unlike Ukrainian and Belorussian) doesn’t appear to have пустка except as a dialect word for ‘deserted/unoccupied house,’ but it has пуща, пустыня, and пустынь; I don’t know what pustać would correspond to.
I don’t know what pustać would correspond to.
That ought to be пустать, which doesn’t seem to exist.
Another good Russian empty-word is пустырь, an empty (urban) lot.
Right!
Slavic *pustъ is an interesting word: it has no obvious external cognates apart from Old Prussian pausta-, attested in compounds meaning ‘wildcat’ and ‘wild horse’, in the derivative paustre ‘wilderness’ and possibly in one or two placenames; no serious etymology has been proposed so far. Quite a challenge.
It’s generally agreed that the verb *pustiti ‘let go, release’ (but also ‘let in, let, allow’, hence the imperative particle пусть ‘let’) is deadjectival from *pustъ (formally, it can hardly be anything else). Why, then, doesn’t it mean ‘to (make) empty’? Perhaps the original meaning of the adjective was ‘wild, untamed’ (of animals as well as land) > ‘uninhabited’, hence finally ’empty’.
I’ve seen the suggestion that it’s ‘to empty your hand in the course of letting something go,’ but that seems a desperate attempt to make sense of it.
pustać (1.1) pustkowie, nieużytki porośnięte niską roślinnością bez drzew[1]
seems to correspond almost exactly to Russian пустошь
Is -ać a common noun suffix in Polish? It doesn’t seem to correspond to anything in Russian.
To me a verb “to let go” seems obviously appropriate to be related to adjectives and nouns about emptiness and wildness–I think controlled, tamed, held things are an obvious dialectical opposite to wild, unoccupied, and un-held things.
as for the -ac ending, isn’t it similar to -ek and -ich?
Is -ać a common noun suffix in Polish?
No, pustać is a 19th-century poetic neologism, probably a hypercorrect distortion of dialectal pustoć. The suffix -oć forms several deadjectival abstract nouns like dobroć ‘goodness’ or wilgoć ‘humidity’. It’s a variant of Slavic *-ota, as in OCS dobrota ‘goodness’, čistota ‘cleanness’, Pol. ślepota, Cz. slepota, Russ. слепота́, etc. ‘blindness’. More distant relatives include English -th, as in wealth, youth, truth, and Lat. -itās.
Talking of *-ota, we also have the noun pustota ‘frivolousneess, flippancy, vanity’ (now old-fashioned, but attested since the 16th c.), whose OCS and Russian cognates mean ’emptiness, void’, i.e. the same as Polish pustka. I suspect the Polish meaning of pustota may be a semantic calque from Latin (vānitās).
Thanks! But why would you hypercorrect something to a unique noun form when -oć was already a pattern-forming suffix?
Because (1) the suffix is a non-productive relic, so the pattern is not very salient; (2) Standard Polish a often corresponds to a vowel symbolised å by dialectologists and pronounced [ɔ ~ ɒ] in most traditional dialects*); (3) Polish also has a high-frequency noun postać ‘figure, form, shape, character’, which may have influenced pustoć (here the suffix is just -ć, while the -sta- part is the verb root ‘stand’).
*) They are both reflexes of Old Polish ā, distinguished from those of short a in nearly all regional varieties of Polish, but not in the standard. Although this /ɒ/ is usually kept distinct from /ɔ/ (which is much higher and prelabialised, [ʷɔ], in the dialects), speakers of the standard variety identify it with their own /ɔ/, and they perceive dialectal ptåk ‘bird’ as “ptok” (standard ptak). This is also the likely reason why the country’s capital, originally Warszewo or Warszowa has ended up as Warszawa; someone must have thought that the [ɔ] of the variant Warszowa was a dialectal realisation of /a/.
So French Varsovie must reflect the actual pronunciation ofl Warszowa, while the English name is a spelling pronunciation based on Warszawa.
German Warschau is inconclusive, BTW, because Slavic -ova was routinely turned into (folk-etymologized as?) the common placename element Au, “vegetated floodplain”.
The usual Latinisation is Varsovia, adj. Varsoviensis, underlying the French and Spanish names of Warsaw, for example. Note the Lithuanian version Varšuva (resembling Lietuva); it’s likely to be old, since the Warsaw area had been well known to the Balts and often raided by them before Poland and Lithuania established an alliance. Before the 15th c. Warsaw was a rather inconspicous place and didn’t even have a stable formal name, hence the attested variants. Then it became the capitaI of the Dukedom of Mazovia and its importance increased. Parliament met in Warsaw and royal elections were held there (first in 1573), but its Polish name became known internationally as the royal court moved from “Cracovia” to “Varsovia” in 1611. The new variant with a was adopted by the neighbours (German, Russian, etc.), and then spread far and wide.
There is also an alternative scenario: in the dialect of Mazovia /a/ tended to become /ɛ/ after palatal consonants in the 15th c., so Warszawa might also be a hypercorrection of the variant with -szew-. Either way it’s hypercorrect. This despite the fact that the toponymic/possessive suffix -ow-/-ew is extremely productive in Poland (just as in Russia).
David Marjanović: German Warschau is inconclusive…
That’s true, see Krakau for Kraków/Cracow/Cracovia.
Old Prussian pausta-
Paustowski?
Konstantin Paustovsky’s father was partly of Polish (or Polonised Ukrainian) descent, and Paustowski ~ Pawstowski can be found among surnames of Polish nobility (of the Niezgoda coat of arms). I don’t know its etymology, but if I were to venture an educated guess, I’d bet on its being derived from a Polish adaptation of the given name Faustus (borne by several martyrs popular in the Middle Ages). Proto-Slavic had no /f/ phoneme, and the same was true of the earliest stages of Polish. In Christian names borrowed at that time, foreign /f/ was usually replaced by /p/: there were names like Ożep, Krzysztopor, Pabian, corresponding to modern Józef, Krzysztof, Fabian. In a few cases we still have coexisting doublets: Szczepan is the older version of Stefan, borrowed at a time when palatalisation before /ɛ/ was obligatory and /f/ was an alien sound. Not unlike French Étienne vs. Stéphane.
Dziękuję bardzo!
Pustoš (f) is a common word in Bosnian/Croatian/Serbian, but it’s not something “empty”, rather something “bare”, esp. devoid of people, useful things or plants. There’s the (impf.) verb pustošiti and the perfective opustošiti, “waste”, “make bare”, used esp. with catastrophic weather events and war, but also when something is abandoned.
Pust is “bare, empty”, and pustinja is “desert”. Pustinjak is “hermit”.
The same happened in the South Slavic, e.g. Croatian Stipan, Stjepan, Montenegrin Šćepan etc.
Paustowski?
Unbegaun lists it under Belorussian surnames and says Паусто́вский < Па́устово < Фа́встово.
Piotr: Szczepan is the older version of Stefan, …. Not unlike French Étienne vs. Stéphane.
In terms of history, yes, but it is not obvious that Étienne and Stéphane are not only related but “doublets” of each other, Stéphane being a modern reformation. Another old version of the same name occurs in Saint-Estève, the name of at least one town.
The inhabitants of Saint-Étienne (a large city near Lyon) are called les Stéphanois: that is because in many cases the (at least official) names for ‘inhabitants of’ are adapted from the Latin forms of the names (often preserved in historical documents dating from before the 16C decree making French the official language).
marie-lucie,
Most Poles probably aren’t aware that Szczepan and Stefan constitute a doublet, though they should be able to figure it out, since St. Stephen (the first martyr) is traditionally called św. Szczepan in Polish (as opposed to King St. Stephen of Humgary = Szent István király, who is usually referred to as św. Stefan). I wonder what percentage of native English speakers know that James and Jacob form a doublet too.
Aigidios/Aegidius has become French Gilles, Spanish Gil, English Giles, but as if he hadn’t suffered enough, he’s also German Ilg (among half a dozen or so variants), Czech Jiljí, Polish Idzi, etc. Some languages have recreated something closer to the original (Ägidius, Egidiusz, Égide, Egidio). I don’t know about other countries, but my compatriots (other than linguists, church people and miscellaneous erudites) are invariably surprised to learn that Idzi and Egidiusz are the same name — and there’s no way they could connect it with Giles. I’ve seen St. Giles’ Cathedral in Edinburgh referred to as katedra św. Gilesa (though Polish Wikipedia correctly calls it katedra św. Idziego).
I know a pair of twins who have Ægidius as their middle name, so I just spent five minutes trying to see if there was a ‘popular’ Danish equivalent like Hans, Jens for Iohannes (or Ib for Iacobus). Doesn’t look like it, though, I suppose the popular transmission of saints’ names is precarious.
(Stephanus became Steffen in Danish, not very exciting. I’m remotely named for Saint Lawrence, of course).
Tolkien’s Farmer Giles of Ham is known in the Book-latin as Ægidius Ahenobarbus Julius Agricola de Hammo.
Idzi sounds like the old imperative of iść ‘go’ (Modern Polish idź). Since in most dithematic personal names with a verbal first member the form of the verb was indistinguishable from the imperative, it was only natural for some Old Polish speakers to treat Idzi as a native name. They must have assumed that Idzi and its diminutives Idzik, Idźko were hypocoristic truncations of something longer, which they attempted to restore (analogically after Bronisław, Stanisław etc.). Hence the curious folk-etymological hybrid Idzisław, attested since the 13th century.
Tolkien’s Farmer Giles…
The Polish translation (1962, by Maria Skibniewska) made him Rudy Dżil (rudy means ‘red-haired’). Skibniewska, the first Polish translator of Tolkien, was on the whole very competent, but this “Dżil” thing is so awful that it borders on linguistic vandalism.
Huh. I wonder what it is in Russian?
Фермер Джайлс из Хэма (1986, translated by Galina S. Usova).
Good enough. Thanks!
Thanks Piotr.
I wouldn’t know about James (old French name, long ago supplanted by Jacques) and Jacob if I had not studied some Romance linguistics. But I had absolutely no idea of the origin of Gilles !
Pronunciation note: Gilles rhymes with villes, not with filles. Gil sometimes occurs also.
Lars: I’m remotely named for Saint Lawrence, of course
I had no idea either!
As for Szczepan, isn’t the initial szcz- an analogical replacement of Czech šť- in Štěpan (Czech dissimilated Common Slavic šč into šť)? Otherwise one would expect *Ściepan.
I had no idea of any of this.
I only know Ilg as a placename. Perhaps similarly, St. Pölten is named after St. Hippolytus, but apparently no people ever were.
As for Szczepan, isn’t the initial szcz- an analogical replacement of Czech šť- in Štěpan (Czech dissimilated Common Slavic šč into šť)? Otherwise one would expect *Ściepan.
There must be a connection — many of the oldest Christianity-related loans reached Poland via Bohemia (note that Doubravka Přemyslovna married Mieszko I of Poland the year before his formal conversion). But the Czech outcome is unexpected for the same reason: in West Slavic, šč reflects the iotated cluster *stj, but not *st palatalised by a front vowel). We have the regular development in Polish kościół, Czech kostel ‘church’, ultimately from Lat. castellum. I’m not sure why we have Štěpán and Szczepan rather than forms with *Ste-. Presumably the loan was mediated by an OHG dialect in which /s/ and its voiced allophone were sufficiently close to Slavic š, ž to inspire such a substitution pattern. Note Czech Ježíš ‘Jesus’, Old Polish Ożep ‘Joseph’, Polish krzyżmo, Czech křižmo ‘chrism’, etc.
Presumably the loan was mediated by an OHG dialect in which /s/ and its voiced allophone were sufficiently close to Slavic š, ž
When did German /st/ > /ʃt/?
It’s complicated: https://www.uni-due.de/~lan300/10_Phonotactically_Conditioned_Alternation_(Hickey).pdf
I’m name-dropping Hans Egede, missionary and ethnographer of Greenland.
I have often wondered whether the development of James had some contamination from the Hebrew name Chaim. Modern English James comes to us mostly through Scottish heritage, where the Scots form is Hamish. The Greek substitution of mu for beta* also suggests (to me, at least) a possible confabulation with חַיִּים.
I admit that I am not a disinterested observer in this, since one of my sons is named “Reuven James,” and the choice of his Hebrew names was a somewhat emotionally fraught issue.
* I really, really wanted to got automatically into LaTeX math mode for entering Greek characters. I had to back up and retype the “mu” and “beta” several times. I kept getting things like “$\mu$”.
Type μ and β and you’ll get μ and β instead.
I will recommend typeit.org for most of your Unicode typing occasions. (No देवनागरी, الأَبْجَدِيَّة , flags or super/subscript numbers, but I’m sure the developer will accept patches).
(Back to being original, WP is not my friend).
I didn’t express myself clearly enough but my thinking was along the lines of: Czech borrows Štěpan from a language with an apical/alveolar s and this becomes OPl Szczepan due to speakers being aware of the OCz šť – OPl szcz correspondence.
Czech borrows Štěpan from a language with an apical/alveolar s and this becomes OPl Szczepan due to speakers being aware of the OCz šť – OPl szcz correspondence.
Just so, and I was wondering if that language might be some kind of German.
On the phonemic level that’s part of the dissolution of the retracted-s phoneme, which conveniently happened in the gap between Middle and Early New High German. On the phonetic level I have no idea.
I have a lot to say about the Hickley paper, but I had to get up early today and am now too tired to write that much. In short, I have a very heterogeneous opinion of it. I’m completely on board with the Voyles bashing, but when Hickley, living in Bonn, shows he’s aware that alveolar consonants can be laminal or apical and that this is used as a phonemic distinction in some languages, and at the same time shows he hasn’t noticed that the alveolar consonants except /s/ are apical in English but laminal in German*, its gets cringeworthy. The subject of the paper is one where phonetic pedantry really pays off, as I hope to show tomorrow – though actually you can already look it up in scattered places on Wikipedia, where I wasn’t involved.
* Except close to the coast, far north of Bonn.
Where to begin…
Perhaps with the abstract.
That’s misleading. It wasn’t distinguished from /ts/ most of the time, both being represented as a letter that has the shape z in the Wikipedia photos of various manuscripts. The shape ʒ existed but was a merely graphical variant. The practice of consistently using z for the affricate and ʒ for the fricative is, AFAIK, limited to modern editions.
That’s where the promised phonetic pedantry begins to pay off.
First, there are two quite different voiceless apico-alveolar fricatives in the world. The IEng. one sounds a lot like [θ]. The other one occurs in Castilian Spanish, sounds similar to [s] and [ʃ] but not to [θ] (which is after all a different phoneme there) and, importantly, is written as s.
Second, the Greek /s/ sounds a lot like the Castilian one, though Wikipedia says it’s a retracted laminal sound, not an apical one. The Finnish /s/ sounds like both, so it’s probably one or the other. The Dutch one also seems to belong here.
It seems to me that:
– If there’s only one sibilant in a language, it will likely be a retracted [s] (whether laminal as in Greek or apical as in Castilian). A possible reason is that a non-retracted laminal [s] is an extremely high-pitched sound that may be harder to hear than a retracted one for some people or under some circumstances. Anyway, a retracted [s] should be the default assumption for Latin, Proto-West Germanic, Proto-Germanic and PIE. (Also Proto-Semitic for that matter, which had three sibilant affricates but only one sibilant fricative.) Northern Andalusian and American Spanish is the big exception, possibly to blame on relative frequencies of occurrence before a merger.
– If there are two sibilants in a language, they’re most likely [ʃ] and a non-retracted laminal [s]. This is found in modern English, modern German, all those Slavic languages that don’t have a third one, and Classical Nahuatl. That last one is clear from the fact that the letter s was almost never used to write it, and then apparently only for /ʃ/, not for /s/.
– If there are three, one common pattern is the Polish one (non-retracted [s], [ɕ], [ʃ~ʂ]); the other common pattern is the Basque one, which was by all evidence shared by 15th- and 16th-century Spanish.
The comparison of the High German consonant shift to the origin of the IEng. retracted [θ] is a compelling insight into the mechanism that turns aspirated plosives into fricatives. But if you start from different plosives, you’ll probably get different fricatives. The English /t/ is apical, and so is its new fricative allophone; there’s no reason to think that the pre-OHG /t/ was apical, so it was probably laminal, just like the [ts] and [s] that have come from it.
A beautiful hypothesis. Alas, I have an ugly fact to offer: in those German accents that retain consonant length, like both of mine, the affricates are long in all positions where length can be distinguished, meaning a long stop released into a short fricative. I think that’s the true reason the spelling tz exists. Moreover, in Central Bavarian dialects like mine, a lenition process has paired these long affricates with short versions that consist of a short (and lenis) stop which is released into a short (and curiously fortis) fricative.
So, instead of one step ([tʰ], [tːʰ] > [s], [ts]), I have to postulate two: [tʰ], [tːʰ] > [ts, tːs] > [sː], [tːs]. First the aspiration became affrication, then the short affricates became long fricatives in “weak” positions (as defined by Hickey).
Exactly: unlike in English, /t/ was laminal, and so was ʒ.
Bad move. The “circumflex accent below” is the IPA sign for voicing (rarely needed because usually a separate symbol is available). As far as I can tell, [ṱ] would be the exact same thing as [d] (all my efforts to pronounce a voiced fortis plosive have failed).
Oh dear. No. That’s not a retention! If it were, the [ks] would have been spelled x the whole time. It’s an innovation; [xs] is retained in Icelandic and reportedly (as [χs] of course) in Swiss German.
(Endnote 8 is just a waffling repetition.) The reason that t remained in st, ft and the ignored ht isn’t that fricative clusters weren’t allowed, it’s that *t wasn’t aspirated in this position and that only aspirates underwent the High German Sound Shift.
For the determination of the phonetic value of ʒ, that of s is of only indirect relevance. It is not so when comparing the treatment of loanwords in OHG nor when viewing the further development of /s/ in late MHG. As is often remarked, Slovene /ʃ/ was rendered by OHG s and not ʒ in a set of manuscripts known as the Freising documents (Sonderegger 1974: 44f.). Now the /ʃ/ sound was a palatal fricative which was formed with extended constriction behind the alveolar ridge. As OHG had only an apical fricative and a laminal one, it is understandable that the latter with its more distributed [endnote 9] constriction would be chosen to represent Slovene /ʃ/.
Here Hickey plainly underestimates his ability to tell which sounds will sound most similar to other people. It’s simply a fact that retracted [s] sounds a lot more like [ʃ] than non-retracted [s] does.
It’s not just the Freising manuscripts. Hungarian uses s for /ʃ/ to this day, and all the early German loans into Slavic have their s represented as /ʃ/ or /ʒ/, as do the Italian ones in coastal Croatian.
Sure, but why would groove width change like that?
/ʃ/ does no such thing in the accents I’m used to, but that may be different in Bonn. /l/ is not at all palatalized in German; it sounds more front than in English simply because it’s laminal. Except when it’s not, which is the case farther north than Bonn; in such accents, you can hear the tongue bending back during /iːl/.
I’ll take this as further evidence that Hickey hasn’t been listening very closely. The realization of ei, au, eu/äu as the “tense” bizarro diphthongs [ae̯ ao̯ oø̯] is a deliberately artificial feature of German stage pronunciation, where it was codified to improve perception under the acoustic conditions of a theater. Otherwise, nobody talks like that, not even newsreaders while doing their job.
It is. It’s also completely off-topic, having nothing to do with places of articulation.
No, just of /s/, interestingly. For /f/ it was undone, presumably by pressure from the [w] > [ʋ] shift; for /x/ it may never have happened, but given that short /x/ is lost in almost all of the accents that have [z] and all of those that have [ɣ], it’s hard to tell.
Both are true in Bonn and many other places. The second is not true almost anywhere south of the White Sausage Equator, and the first only trivially so because [z] isn’t in the inventory there in the first place; contrary to the next sentence, that includes Austrian and Swiss Standard German.
That may be the case in Dutch and Polish, where it really is word-final. But in the German varieties that have it, it’s syllable-final. It’s striking how Hickey spends the rest of the page not noticing this, even though it easily explains all the cases he adduces there.
I’m too tired to continue, more tomorrow.
Oh dear. No. That’s not a retention!
Man, even I know that’s wrong, and I know hardly anything about this stuff.
Thanks for doing all that work and reporting back!
Piotr Gąsiorowski says: Slavic *pustъ is an interesting word…
Apart from the “desert”, “devastation” and related meanings, there is an interesting Croatian compound:
“pustolovina” (adventure) and “pustolov” (adventurer), made from “pust” and “lov” (hunt).
Also, what is this nonsense about Wagner being pronounced /vaknər/? No doubt there’s plenty of dialects where it’s devoiced, but surely none where it is fortis.
David: If there’s only one sibilant in a language, it will likely be a retracted [s] (whether laminal as in Greek or apical as in Castilian).
I would put is slightly differently. If there’s only one sibilant, it will almost certainly be [CORONAL], but the exact tongue configuration will not be essential, and, as a consequence, a wide range of allophones can be expected, because there will be no strong pressure to constrain the place of articulation and the tongue shape more narrowly. Naturally, the “centre of gravity” of all those allophones will be halfway between extreme realisations — the kind of sound you describe rather than something dental, palatal, emphatically subapical or emphatically “bunched”. In effect, it will be a retracted [s] (as a central tendency) +/- a good deal of dispersion.
/l/ is not at all palatalized in German; it sounds more front than in English simply because it’s laminal.
Depends whose English you talk about. Ray Hickey is Irish an I can clearly remember his own /l/, which is strikingly “light” and sounds palatalised (even to my Polish ear) at least in comparison with most other Englishes.
Agreed.
Yeah, I haven’t heard much Irish English. However, I wouldn’t describe the Polish l as palatalized either: it occupies the front half of the German allophone range, but doesn’t seem to go beyond that (except maybe before /nʲ/), noticeably unlike a Russian or Serbian /lʲ/.
Also, what is this nonsense about Wagner being pronounced /vaknər/? No doubt there’s plenty of dialects where it’s devoiced, but surely none where it is fortis.
I can only say that in my idiolect, the syllabification for this and similar words (e.g. regnen “to rain”) is -V-gnV-, so no devoicing at all.
If there’s only one sibilant in a language, it will likely be a retracted [s] — but not in Danish.
Danish has two alveolar fricative/approximant phonemes. One is /s/, the voiceless apical alveolar non-retracted sibilant [s̺]; the other is /ð/, a voiced velarized laminal alveolar approximant [ð̞ˠ] which has a rare non-sibilant fricative variant. I assume the two have been pushing each other apart, keeping /s/ non-retracted and moving /ð/ into a space that would be heavily contested by shibilants if they existed.
I wouldn’t describe the Polish l as palatalized either…
In my accent, it’s always apical (even before /i/, /j/ and /ɲ/), the body of the tongue is generally quite low and the root of the tongue is in a neutral position, as far as I’m aware (no pharyngealisation). There is some coarticulatory influence of the following vowel, but not enough to be perceived as velarisation before /u/ or palatalisation before /i/. In the working-class accent of Warsaw, /l/ is always dark [ɫ] (pharyngealised and slightly velarised), and rather than assimilate to a following /i/, it retracts it to [ɨ] (saying ly for li is a shibboleth distinguishing “low” and “high” varieties of Warsaw Polish). Along the eastern border, /l/ may be palatalised and contrasts with an apicodental /ɫ/ corresponding to mainstream Polish /w/ (spelt ł) — roughly as in East Slavic.
@Hans:
What about words/names like “Adler”? I remember finding it interesting to learn how this differs in textbook German from words like “endlich” (which has Auslautverhärtung). (For people who actually speak a variety of German natively, Google search seems to indicate that there can be a bit of uncertainty about how to describe the consonant in “endlich”, maybe because it’s not usually aspirated.) If syllabification is the explanation, this would seem to indicate that despite not occuring at the start of words, /dl/ is a valid onset cluster in textbook German (/ˈaː.dlər~ˈaː.dlɐ/).
It makes it easier for me to remember if I think of these clusters as resulting from compression of syllabic /əl/ and /ən/ to non-syllabic /l/ and /n/, at some level (are there any German words with a robust surface distinction between these sounds in this context?). E.g. “Adler” < "Adeler", "regnen" < "regenen"). Well, it at least seems to be true in some cases historically, if nothing else. I guess if this mnemonic were to be taken as a serious analysis, it would imply that "d" and "l" are both in an onset, but not part of the "same" underlying onset. Then again, there must be some kind of phonotactic interaction going on that explicitly “permits” /dl/ at some level because as far as I know you don’t get things like /-tʁn-/ from /-tʁən-/.
Oddly enough, I found a book "1001 Pitfalls in German" (Henry Strutz) that does prescribe /k/, the same sound as at the end of "genug", in "Lügner" and "Wagner". It would be ironic if Strutz himself fell prey to a pronunciation pitfall in this context!
I’ll be posting links in a separate post so I won’t have to wait for approval.
Links related to my previous post.
A Wordreference forums thread where a German says he pronounces “endlich” with /dl/ and “”Liebling” with /bl/
“1001 Pitfalls in German” (Henry Strutz), p 232 on Google Books
Thanks for approving that so soon, Hat! Oops, I just realized I had one more thing I wanted to mention. There’s a way that I find a little bit too “neat” or convenient to avoid a somewhat similar problem in certain analyses of English. From “Syllabification and allophony” (John Wells):
(I actually find the general syllabification analysis Wells posted there quite convincing and intuitive, but I also have a vague feeling that it’s wrong (at some fundamental level, I mean, not just in specific parts), and it goes against a lot of other literature and theory about maximizing onset clusters.)
Hm… A good way to master the pronunciation of ch is to practise panting like a dog.
In Poland’s Bieszczady Mountains there’s lots of unpopulated places where villages used to be (their inhabitants were removed in the operation “Vistula” in late 40’s). However, their names are put on touristic maps and still enjoy some degree of use; people talk about them when they describe their hiking routes. Sometimes there are visible leftovers like ruins of buildings (especially churches), tombstones, fruit trees. When something is established or rebuilt in such a place it tends to get its name after the former village. So in a way this is comparable to the pustoshi in Russia, however no special word exists to refer to the desolate settlements.
@Eli Nelson: for me, it’s A-dler, but end-lich and Lieb-ling, so the /d/ in Adler is voiced, but the /d/ in endlich and the /b/ in Liebling are devoiced.
Brett: I have often wondered whether the development of James had some contamination from the Hebrew name Chaim. Modern English James comes to us mostly through Scottish heritage, where the Scots form is Hamish. The Greek substitution of mu for beta* also suggests (to me, at least) a possible confabulation with חַיִּים.
The Hebrew name transmitted through Latin as /iakobus/ (with stress on the first vowel) and preserved in Romance languages (among others) has gone through phonological changes along with words of similar shape and sound elements, especially consonants. Different paths of evolution were followed according to different languages and dialects.
The form closest to the Latin one is dialectal Italian Iacopo. The change of /b/ to /p/ is a reinforcement between the unstressed middle vowel and the somewhat stronger final one. Such a reinforcement is also attested in other Italian forms. On the other hand, the original /b/ has completely disappeared in French Jacques, while the medial consonant /k/ has been kept intact.
In most other Romance languages the Latin /b/ of the name has been changed to /m/, a rarer but not unknown change, perhaps as an alternative to /v/ which is a more common result of the evolution (as in Latin habere, French avoir ‘to have’).
An example of this is French omelette where one would expect “ovelette from the word meaning ‘egg’.This /m/ shows up in several versions of the name: Italian Giacomo, Spanish Jaime, Catalan Jaume, and the now older French derivatives Jacquemin and Jacquemine (current Jacqueline is probably a more recent reformation on the model of Micheline or Jocelyne, where the sequence line was later taken for a suffix and used to form other feminine names, such as Roselyne).As for the medial consonant /k/, it has disappeared from Spanish, originally leaving the palatal approximant /y/ in its place after the loss of the (unstressed) medial vowel. The Catalan equivalent of the Spanish diphthong /ay/, namely /au/, is in keeping with the velarization occurring in some Catalan cases. In French this medial consonant still shows up in the Haitian toponym Jacmel, where /k/ was preserved after the loss of the medial vowel (with a suffix added). It has completely disappeared in (Old) French James, a name adopted in English but which has lost to Jacques in Modern French. A history of Scotland written in French has the kings called James in English referred to as Jacques.
I have no idea how Hamish ended up with its initial. The Irish version is Seamus, where the initial fricative is perhaps more compatible with those of the Romance versions.
I have no idea how Hamish ended up with its initial
From the vocative “a Sheumais” with initial mutation.
per incuriam,
Thank you! That must be it. I remember now that there is this process of “initial mutation” in Celtic but don’t know enough to think of it offhand in a case like this.
@per incuriam: That’s very interesting. I figured it was just wishful thinking that it might have something to do with “Chaim.”
An example of this is French omelette where one would expect “ovelette from the word meaning ‘egg’.
The OED and the TLFI tell a completely different and much more surprising story:
French omelette (1561; also as aumelete (1603), aumelette (1611 in Cotgrave)), alteration of Middle French, French amelette (1480; now regional), apparently a variant (with metathesis) of an unattested Middle French form *alemette (compare alumette (c1400), alumecte (first half of the 15th cent.)), itself in turn a variant with suffix substitution of alemelle, alumelle ‘thin plate, blade of a sword or knife’ (second half of the 12th cent. in Old French as alemele, alumele; late 14th cent. as alumelle in sense ‘sweet fritter’, perhaps ‘omelette’), ultimately a variant (with metanalysis of the definite article) of lemelle ‘blade’ (second half of the 12th cent. in Old French as lemele; French lamelle (early 15th cent. in Middle French)) < classical Latin lamella.
1548 homelaicte (RABELAIS, Quart Livre, éd. R. Marichal, chap. 9, p.67, 71); 1561 omelette (J. DU FOUILLOUX, Receptes pour guarir les chiens de plusieurs maladies, éd. G. kTilander, chap. 73, p.159). Altération, prob. d’orig. méridionale (cf. la localisation des formes en au- ds FEW t.5, p.136a) sous l’infl. de mots issus du lat. ovum «oeuf», de amelette «omelette» (19 févr. 1480, Reg. des compt. de l’hôtel de ville de Tours, A. Tours ds GDF. Compl.) par métathèse de *alemette, var. de alumecte «id.» (1re moitié XVe s. [date ms.], Ménagier de Paris, éd. G. E. Brereton et J. M. Ferrier, p.244, 33), par substitution de suff. de alumelle «id.» (ibid., p.245, 5), issu de lemelle forme anc. de lamelle*, avec agglutination du -a de l’article, v. aussi al(l)umelle, la forme aplatie de l’omelette ayant été comparée à celle d’une lame.
The OED adds:
The change in the initial vowel from a to probably occurred in southern French under the influence of forms of œuf ‘egg’; compare Middle French œufmollette (1576), French œufmelete (1607), œufmeslete (1615).
So ovum is only a minor influence on this word; its phonemes after the first come from lamella ‘small thin metal plate’.
JC, Thanks for the correction. I am surprised, as I basically repeated what I had read in histories of the French language. That’s why it did not occur to me to verify the history of omelette with the TLFI, my usual trusted source.
So, I withdraw my sentence about the word.
LH, Can I (or you) cross it out?
Done! (If you catch something like that within the edit window, you can cross it out with <s></s>, short for “strikethrough”; otherwise just ask me and I’ll be happy to do it.)
Thanks LH! In this case I could not go back to my posted comment.
That’s interesting about Polish… I’ve encountered the “always dark” accent, but not the “apical” one so far.
Danish has a [ɕ], though, right? Apparently there’s a debate on whether it’s a phoneme or should still be analyzed as /sj/, which is where it comes from; either way, [ɕ] would be a strong push factor on [s].
*lightbulb moment* Regional variation. Gerhard Schröder pronounced Gegner “(political) opponent” with [ç] in the middle – this is only explicable as syllable-final fortition of the ghost of Low German /ɣ/. In keeping with this, there’s a paper on German phonology where Theo Vennemann (yes, that one), very much a northerner, reports pronouncing adlig* “noble” with /t/.
* I would only say adelig, but usually die adligen […].
In English?
Vienna shows up twice in Roman sources, first as Vindobona, later as Vindomina.
BTW, Wells’s “pre-fortis clipping” should be looked at from the other side and called “Winter’s law”.
Writing about maximizing onset clusters in English is slightly perverse, given how frequently English tramples on the Maximal Onset Principle. The metalanguage used should be Georgian or something ….
Winter’s law
I have seen this popping up now and again, and it seems strange to me. “X’s law” is normally a hypothesis about a sound change that happened at a specific time to a specific language. If we found that in 13C Lower Slobbovian, voiced stops became voiceless stops and voiceless stops became voiceless fricatives, it would seem utterly weird to call that “Grimm’s law”.
Ghrassman’s law is an exception, because it took time to realize that it operated independently in Greek and Sanskrit, so that means there are two Ghrassman’s laws.
@David Marjanović:
I was thinking of both phonological theory in general, and one particular argument related to English that I will link to.
and @John Cowan:
I found the following account, whereby English consistently maximizes onsets (to the extent that it is phonotactically possible), even to the extent of resyllabifying word-final consonants to the onset of a following vowel-initial syllable, surprising but well argued. (Things that Wells explains using syllabification are explained by Ricardo Bermúdez-Otero using phrase boundaries and feet-based lenition rules.) I’d appreciate hearing what other commenters here think of it.
“An amphichronic approach to English syllabification“, Ricardo Bermúdez-Otero
Danish has a [ɕ], though, right? Apparently there’s a debate on whether it’s a phoneme or should still be analyzed as /sj/, which is where it comes from; either way, [ɕ] would be a strong push factor on [s].
Danish does, and there is. [ɕ] and [j] are in complementary distribution except when alone in syllable onsets, where sjappe/jappe is one of many minimal pairs (but this one has sappe as well if you want a pair for /s/ vs /sj/~/ɕ/). So it’s silly to have a debate, both analyses are perfectly cromulent.
It is a good point, however, that regardless of phonemic analysis [ɕ] will indeed push hard on [s].
I’ve finally read and understood both articles. Wells’s analysis does seem to work for RP, but fails at American latex, which Wells doesn’t mention and which Bermúdez-Otero explains on p. 9; so, if we want to posit a single set of syllable-based rules for all of the Standard Englishes (which doesn’t work in German), the latter wins.
Bermúdez-Otero’s article also shows nicely that Old English obligatorily inserted [ʔ] before every stressed syllable that would otherwise have begun with a vowel, just like northern German today. The absence of this rule in southern German and in Middle and later English is therefore most likely an innovation. Should we blame a Celtic substrate for Middle English/Anglian and a Romance one for southern German?
Old English obligatorily inserted [ʔ] before every stressed syllable that would otherwise have begun with a vowel
This is why its alliterative poetry still works when the vowel-initial words in a line all start with different vowels: the alliteration is with the initial [ʔ] not the vowel sounds. I guess German philologists had no problem recognizing this, but those unfamiliar with German were puzzled. But I don’t think Bermúdez-Otero’s observation is original, I learned about the alliteration ‘exception’ (actually non-exception with the glottal stop) years ago.
Should we blame a Celtic substrate for Middle English/Anglian
The French influence at the time should be enough of an explanation.
There is another rule of Germanic alliteration that is a bit surprising, especially in English: /st/, /sp/, /sk/ alliterate only with themselves. I wonder if there is a phonological explanation for this.
@JohnCowan:
I remember reading some sources that analyze word-initial [st] [sp] [sk] as phonemes in a number of European languages, due to their special behavior, but this doesn’t seem to be an explanation so much as a generalization.
I don’t know. Alliteration fell completely out of fashion between Old and Middle High German. The Hildebrandslied alliterates; the Nibelungenlied rhymes, and while it contains much older material which did alliterate, this alliteration isn’t used at all; perhaps it’s even actively avoiced. (The first example are the three kings of the Burgunds, Gunther, Gernot & Giselher; instead of occurring in the same line, they’re spread over two stanzas.)
Correct. I should have said this is the first time I see a demonstration with examples instead of just the claim.
Here’s a discussion which adds PIE. More recent sources I’ve seen avoid this hypothesis by sophisticated theories about syllable onsets, but Germanic is definitely the best candidate for phonemic “suffricates” or “presigmatized stops”.
Compare (p. 138 onwards) the prelateralized stops of a Tibetan language spoken in Ladakh; an ancestor of Central Tibetan likely had the same, says the same paper.
Do you suffer from suffricates? Offset them with offricates! Special spoffer, today only!
Don’t mind me, I’m coming to the end of a long book edit and am getting a little batty.
🙂
“suffricates” or “presigmatized stops”
Out of context, these words certainly suggest suffering, whether physical or psychological. Sadistic linguistics?
Eli N: I remember reading some sources that analyze word-initial [st] [sp] [sk] as phonemes in a number of European languages, due to their special behavior
A few years ago I read a review of a recent book (thesis? from Germany I think) about the “S mobile” which shows up in PIE sCVC roots with approximately the same meaning as otherwise identical CVC ones. I am sorry I did not have the opportunity to read the work, but it has occurred to me a number of times that this extra S could have had a function or functions similar to those of glottalization and/or aspiration in some North American languages. For instance, in Wintu (a language of California) there is a large number of pairs or even triples of verb stems with very similar meanings, as in pEt- in ‘to squeeze (also knead)’, pe:t ‘to pound’, phit- in ‘to hammer, smash’, p’it- in ‘to squeeze through a small opening, to roll out (bread dough)’. This seems to be a type of consonant gradation, although apparently without a function (at least in the modern forms), and it and other types of gradation also occur in some other languages of the West Coast to indicate diminution vs augmentation in the case of nouns, as in Chinook -tc’iau ‘snake’, dim. -ts’iau, aug. -djiau (Boas).
So I wonder if the addition of s to initial stops in PIE (or even earlier) could have had a similar purpose, but so long ago that a lot of the potential sC-/C- pairs would have been preserved in just one of the possible forms, either C- or sC-.
Does that sound plausible? Anyone else has had the same idea?
Out of context, these words certainly suggest suffering, whether physical or psychological. Sadistic linguistics?
Particularly if you read the second one as “prestigmatized stops”, as I did.
Actually, that’s the way I must have read it!
I’ve heard that pre-tonic diaresis can lead to severe post-trematic stress.
marie-lucie: Anyone else has had the same idea?
I’ve thought that it may have had an intensifying function, originating as either a reflexive or a deictic pronoun. Not that it’s original to me.
I could see “intensifying”, but not from an extra morpheme, rather from a phonological reinforcement.
Interesting idea, m-l.
Thanks David!
I looked up S-mobile on Wikipedia. Sure enough there was a page about it. It looks like everyone thinks that basically the extra s- comes from the transfer or repetition of an -s suffix occurring before the initial C of a root. But I am not convinced.
The page also has a chart with examples of sC- words and corresponding C- words in various IE languages. One word that struck me among the (s)p- words was Ancient Greek psar ‘starling’, listed along with sparrow and a few others sp- forms. Why ps- rather than sp- ? Actually, sC- from *Cs- had occurred to me as a possibility (although in Greek we find both Cs- and sC-, at least with stop C’s). Such a change could be explained if *Cs had been a possible variant of *C on the same order as *Ch- or *C’ (aspiration or glottalization).
I think misanalysis of -s as s- could only happen after sC- onsets had been established by some other process, quite possibly a morphological s- prefix (or several which merged, cf. How many -s suffixes in Old Chinese?).
Out of context, these words certainly suggest suffering
I remember as a child thinking that suffrage was a high-register equivalent of suffering.
I remember as a child thinking that suffrage was a high-register equivalent of suffering.
Not just you, and not just as a child, apparently. This is why the infamous “End Women’s Suffrage” petition got so popular.
This is in fact regular in Germanic, judging from the very small number of available examples:
shove < *skeuβan- < *kseubʰ-
steal < *stelan- < *tsel-
German stieben < *steuβan- < *tseubʰ-
as described in this paper, if the link works.
But of course this metathesis against the sonority hierarchy was only possible if word-initial sP clusters (P for plosive) were already much more common than Ps clusters.
metathesis against the sonority hierarchy was only possible if…
It’s an approximate guideline, not a law. *ts- > st- is a perfectly wholesome and natural metathesis.
Actually, sC- from *Cs- had occurred to me as a possibility
This brings to mind also A. Dolgopolsky, who thinks PIE *sk- and *st- come from earlier *c- and *č- respectively.
(Yes, that’s dental affricate to s+velar stop, versus postalveolar affricate to s+dental stop, which should probably already be a sign to not put too much stock into this idea.)
Not in word-initial position. Outside IE and even inside, not many languages allow sP clusters in syllable-initial position in the first place.
Yeah, that seems… backward and crossed-over!
Outside IE and even inside, not many languages allow sP clusters in syllable-initial position in the first place.
Well, initial ts-, not as an affricate but as a cluster, is probably rarer.
In Hebrew there’s a regular process of t+sibilant metathesis in the reflexive binyan hitpa’el, e.g. √spr ‘cut hair’ > *hitsaper ‘he got a haircut’ > histaper.
That’s probably true.
That looks like it’s an attempt to avoid confusion with the phoneme /t͡s/, and thus inapplicable to PIE, Proto-Germanic or (in initial position) anywhere in between.
The same thing happens with roots beginning with /ʃ/ and /z/ (in the latter case, assimilating to /zd/), but there are no corresponding affricate phonemes in the language.
Moreover the change goes back to when tsade was /sˠ/, and /t͡s/ didn’t exist either.
I don’t think there ever was such a time, and apparently I’m in the Semitist mainstream with this nowadays. Not only does the old idea that the pronunciation as [t͡s] should be blamed on European languages make no sense, but the traditional pronunciation of tsade used by Persian Jews is reportedly [t͡s], which the local languages don’t have. The affricate pronunciation is nowadays generally reconstructed all the way to Proto-Semitic, as an ejective [t͡sʼ].
That the same thing happens with /ʃ/ and /z/ is interesting, and could be analogical, but how old is the metathesis? If you go back far enough, samekh was itself affricate ([t͡sʰ] most likely), as is clear from Akkadian (where e.g. t-š clusters across morpheme boundaries were written with s signs), and could have been the motivation for metathesis in |t-ʃ| < *|t-s̠|.
Classical Arabic is to Proto-Semitic as Sanskrit is to PIE and modern Finnish is to Proto-Uralic: important, but historically overrated at the expense of evidence from elsewhere. …Oh, and on Classical Arabic (and metathesis…!), check this out.
You’re right about the /sˠ/ being probably an Arabic innovation. [s’] is a likelier realization of the older tsade.
Al Jallad’s paper you link to starts by saying that “Since Steiner’s (1982) monumental study of the Ṣade, it has been widely accepted that this sound was originally affricated in Proto-Semitic and in many early Semitic languages.” This is odd, since Steiner himself (Affricated Ṣade in the Semitic Languages, p. 89, online) concludes that even though the affricate is and was much commoner than had been assumes, it coexisted alongside the fricative in the oldest strata where the pronunciations can be discerned.
You can add to the metathesis-inducing consonants also ś, with examples from Isaiah 28:20 and Haggai 1:6. In Isaiah’s time and perhaps Haggai’s too the consonant was probably pronounced [ɬ].
Simpson (The Origin and Development of Nonconcatenative Morphology, pp.168–169, online) gives similar examples in Arabic, Ge’ez, and Akkadian (in another binyan, what he calls the Št- stem). So the phenomenon is old. Why this metathesis applies in some binyanim and not others, and why it doesn’t apply in other environments, that I don’t know.
Ejective fricatives are very rare on a global scale, almost as rare as aspirated ones. There don’t seem to be any in the entire Caucasus, for instance. I know some Semitic languages have them today, but their rarity probably means they aren’t generally a long-lived kind of sound.
I’ll wave my hands in the general direction of paradigmatic analogy generalizing it in some and eliminating it from others.
That’s not a very good argument. Ejective fricatives are rare because they either a) are rare to form or b) are unstable (or both). Since they are manifested in most of Modern South Arabian and Ethiosemitic, stability is not the issue.
That said, I’ve been looking at Kogan’s chapter in Weninger’s recent survey of Semitic. To be sure, there are good arguments for affricates, better than I’d realized. All the same, there is no decisive explanation tat I have seen for all the observations.
Incidentally, I learned from ibid. p. 61 that Cyrillic Ц and Ч come from the medial and final forms of the ṣade, צ and ץ.
I’d suggest that, if ejective sibilants are unstable, they’re then probably unstable towards developing into affricates, and could just keep wandering back and forth.
It is true that we do not see much similar vacillation as in Semitic anywhere else (there could be some occasional exceptions in the Americas). But Semitic does seem to fit into a sweet spot where there are several ⁽*⁾affricate series with full phonation contrasts, versus a much more poor “true” fricative system. This kind of a setup often goes with widespread deaffrication developments, as seen widely in IE languages.
By contrast, languages in the Caucasus, or similarly e.g. the Na-Dene family, or several other examples from Lakota to Xhosa, tend to also have /z ž/ as distinct from /dz dž/, etc. This contrast could protect not just them but also their ejective counterparts from deaffrication.
(And, I think changes like *tsʰ > sʰ likewise mostly occur in languages like Burmese that have a well-developed three-way T Tʰ D system in stops/affricates but no original phonation contrast in fricatives.)
Proto-Siouan is reconstructed to have had *s’ and *š’, retained only in Mississippi Valley Siouan. I don’t know what the reasoning is to interpret them as a retention in MVS rather than an innovation. The online Siouan Comparative Dictionary writes this sound as <sʔ>.
Some languages in the family reflect the *s’ as an affricate, like /ts’/ or /tš’/. Others lose the glottalization, and others move it to after the following vowel. I haven’t looked for any patterns.
Greek is so big on /s/+plosive that it won’t tolerate /s/+liquid initials without inserting a plosive, so “Slav” comes into the language as sklavos ‘slave’, and similar things happen with “*sn” vs. “skn”, though I can’t think of an example offhand. The Latin form sclavus materializes in the Romance languages with the usual epenthetic i/e, but the Germanic equivalents lose the /k/, except for Modern German Sklave. Since MHG had both sklafe and slave, I wonder if the /k/ form didn’t get a boost from Ancient Greek during the revival of learning.
The only skn- word in my Greek dictionary is σκνίπα ‘gnat.’
Recently I read an explanation of the s mobile that immediately made sense. Of course I promptly forgot the source. The idea is that long consonants were strictly avoided even across word boundaries, so that *-s C- and *-s sC- were not distinguishable and got mixed up at random. (It is textbook wisdom that *-s-s- always became *-s-, so that part is unproblematic.)
This requires, of course, that *sC- word onsets already existed. I wonder if they got started by verbs that were prefixed with a noun root that happened to end in *-s; there is the idea, somewhere out there in the literature, that the super-rare onset *tp- comes from *ped p-, so the whole thing would mean “kick with the foot” or something.