Linguistic Miscellany Thread

[Richard W]

Butː

mēhar qoḏšô (Psalms 3:5)

from-hill holiness-his

from His holy hill

Oh, that's pretty interesting. Why not 'from the hill of his holiness'? Is it because har qoḏš-ô is some sort of very lexified phrase? I guess I could eventually find this if I look for it in grammars, but do you happen to know?

I think it's because a definite noun can only have one possessor, and qōḏeš has already taken that slot. *hārô qōḏeš is not possible, at least, according to Gesenius, who describes tacking the pronoun on the end of the phrase as a standard construction. Also remarkable is that as ben-yᵉmînî 'Benjaminite' is indefinite, it is made definite by making the second element definite ben-hayᵉmînî (Judges 3:15).

[Richard W]

Does Pali descend from Sanskrit? In particular, is it legitimate to say of most Pali words that they are inherited from Sanskrit? This is chiefly a question about what the term 'Sanskrit' includes and excludes in plain English.

From most of my research: no.

Thanks for the treasury of links.

The problem I am wrestling with is that a lot of Pali entries on the English Wiktionary say 'From Sanskrit...' or 'Inherited from Sanskrit'. The word 'Sanskrit' is an automatically generated link to the wikipedia page, and we don't have an easy way of changing that link. Wiktionary has redefined the language code 'sa' to mean the Old Indic dialect continuum including Sanskrit. What is needed, I believe, is something meaning 'As near as damn it, inherited form Sanskrit...'. Would you consider such an etymology as legitimate, and if so, is there a snappy and professional looking way of putting it? I have been toying with 'Effectively inherited from Sanskrit...' as a solution.

[WeepingElf]

Is it correct to say that the relationship of the Middle and Modern Indic languages to Sanskrit is comparable to that of the Romance languages to Classical Latin?

[2+3 Clusivity]

The problem I am wrestling with is that a lot of Pali entries on the English Wiktionary say 'From Sanskrit...' or 'Inherited from Sanskrit'. The word 'Sanskrit' is an automatically generated link to the wikipedia page, and we don't have an easy way of changing that link. Wiktionary has redefined the language code 'sa' to mean the Old Indic dialect continuum including Sanskrit. What is needed, I believe, is something meaning 'As near as damn it, inherited form Sanskrit...' Would you consider such an etymology as legitimate, and, if so, is there a snappy and professional looking way of putting it? I have been toying with 'Effectively inherited from Sanskrit...' as a solution.

As an initial excursus, I'd note that there's a whole tradition of categorizing words in Indo-Aryan Languages that may or may not be inherited from Sanskrit. Wiktionary does not seem to reflect any of this.

(1) tatsamas -- words borrowed from Sanskrit after Middle Indo-Aryan ("MIA") sound changes.
(2) ardha- or semi-tatsamas -- words borrowed from Sanskrit but affected by MIA sound changes.
(3) tadbhava -- ~Old Indo-Aryan words that have been affected by MIA and earlier sound changes but not necessarily the same as those in Sanskrit/Vedic.
(4) deśi -- arising from non-IA sources

A more accurate categorization for MIA and NIA language items would probably account for the category of borrowing or inheritance rather than dumping everything into "inherited from Sanskrit."

What root(s) are you looking at also? For certain liturgical items, I could see "inherited from Sanskrit" as potentially being correct. It doesn't seem all Pali entries are dumped into [sa]. Taking a quick look on Wiktionary, I see PIA root /*ȷ́ʰas-/ ("laugh") giving a reduplicated form with zero grade root /*ȷ́ʰágẓʰati/ I.e., /*ȷ́ʰá~gẓʰ-ati/. Skt. developed /has-/ and /jákʂ-/ and dial. /jajjʰ-/ (the latter per The Sanskrit Language (T.Burrow) showing the diversity of Sanskrit itself). Pali developed seperately into /has-/ and /jággʰ-/. Note that the /-ggʰ-/ is not the expected outcome of either an earlier /-kʂ-/ or /-jjʰ-/. The Pali entry only has the following tags: Pali terms inherited from Proto-Indo-Aryan, Pali terms derived from Proto-Indo-Aryan, Pali lemmas, Pali verbs. Not sure why the first two categories are differentiated.

Can you provide their full text definition of [sa]? At this point, it seems so broad that it's not very useful. Also, do you know how these are being auto-populated? I.e., is there a particular source they are pulling from or is someone just linking MIA and NIA language words automatically to Sanskrit?

As for a good quote on the relationship, I found this one in the footnotes on wiki (I've requested the source will report back): "Oberlies, Thomas (2001). Pāli: A Grammar of the Language of the Theravāda Tipiṭaka. Indian Philology and South Asian Studies, v. 3. Berlin: Walter de Gruyter. p. 6. ISBN 3-11-016763-8. "Pāli as a MIA language is different from Sanskrit not so much with regard to the time of its origin than as to its dialectal base, since a number of its morphonological and lexical features betray the fact that it is not a direct continuation of Ṛgvedic Sanskrit; rather it descends from a dialect (or a number of dialects) which was (/were), despite many similarities, different from Ṛgvedic."

The Kobayashi source I gave above also puts it as: "Vedic was probably a specific dialect of Old Indo-Aryan; it was quite close to, but not identical with the language from which Middle Indo-Aryan developed."

Is it correct to say that the relationship of the Middle and Modern Indic languages to Sanskrit is comparable to that of the Romance languages to Classical Latin?

I will admit that I am not familiar enough with the vicissitudes of the early Italic languages to fully opine on the analogy. If the current situation was that modern Romance languages are at their core based on non-Latin Italic language substrates with a heavy multi-layered superstratum of Latin, and that the "direct" Latin language went extinct or is represented only in small understudied minority language pockets, then maybe.

It seems to me, without too much research, that Latin outcompeted its neighboring languages and spread with perhaps only vestiges of the neighboring Italic languages. Drawing on of my example of divergence in the form of the numeral six from before, I do not see, for example, P-Italic forms for numbers four and five surviving but being surrounded by various more sophisticated loan words. I view number 1-9 as relatively core vocabulary and 2-5 as extremely core. Extreme aside, if you imagine lets say predominantly Osco-Umbrian speakers being settled/disbanded/retired into Gaul, I'd think the P-Italic + P-Celtic influence would have been stronger here in trade related terminology.

[WeepingElf]

Of course, no Romance language descends from Sabellic languages, nor from Faliscan or whatever. They certainly do descend from Latin - but not from Classical Latin, rather from spoken Latin dialects in the provinces.

[zompist]

Of course, no Romance language descends from Sabellic languages, nor from Faliscan or whatever. They certainly do descend from Latin - but not from Classical Latin, rather from spoken Latin dialects in the provinces.

I think I know what you mean, but this seems misleading to me. Of course the Romance languages derive from Vulgar Latin, which just means what the people spoke. If modern linguists were in the 1st century Roman Empire, though, we surely wouldn't call Classical and Vulgar Latin different languages, but different registers. It's probably inaccurate to say that VL "came from" Classical Latin, but CL— like most written languages— was closer to the ancestral form, though not an unchanged form of it. And FWIW people were not really conscious that CL and VL were different things until well into the medieval period.

[Richard W]

What root(s) are you looking at also?

The tadbhava words are the ones of interest. The most critical ones are those whose form is compatible with the notion that Pali descends from Sanskrit, e.g. harati.

For certain liturgical items, I could see "inherited from Sanskrit" as potentially being correct.

Do you mean "borrowed"? I believe a fair few grammatical terms are calqued.

The Pali entry only has the following tags: Pali terms inherited from Proto-Indo-Aryan, Pali terms derived from Proto-Indo-Aryan, Pali lemmas, Pali verbs. Not sure why the first two categories are differentiated.

There are three relevant categories: X terms inherited from Y, X terms borrowed from Y, and X terms derived from Y. The first two are subcategories of the third, but to get them into a single list, a term in one of the first two must also be put into the third category. There are also deprecated template usages that output terms directly into the third category.

Can you provide their full text definition of [sa]? At this point, it seems so broad that it's not very useful. Also, do you know how these are being auto-populated? I.e., is there a particular source they are pulling from or is someone just linking MIA and NIA language words automatically to Sanskrit?

The Wiktionary definition is buried in the banner notice at https://en.wiktionary.org/wiki/Wiktiona ... t_Sanskrit. The entry for jagghati, that you looked at above, seems to have been composed to support the discussion referenced in the banner. There was an attempt to modify the Wikipedia article to include their definition, and there is now disagreement on whether it is good enough to say 'inherited from Sanskrit' without lying, even if one must hold one's nose.

'Just linking' - the categories are populated from the templates inh and bor and some others of lesser importance or deprecated. There have been some very dubious claims of core vocabulary being borrowed from Sanskrit. I was correcting them when I encountered them, but then the formula 'Inherited from pre-Sanskrit' was objected to.

Is it correct to say that the relationship of the Middle and Modern Indic languages to Sanskrit is comparable to that of the Romance languages to Classical Latin?

Thank you, 2+3 Clusivity, for answering the question. I knew I couldn't do it justice.

[Richard W]

Of course, no Romance language descends from Sabellic languages, nor from Faliscan or whatever. They certainly do descend from Latin - but not from Classical Latin, rather from spoken Latin dialects in the provinces.

And where did those provincial dialects come from? The question for the analogy is which part of Latium did the provincial dialects come from? The Romans seem to have spoken a strange form of Latin.

[2+3 Clusivity]

The Wiktionary definition is buried in the banner notice

Ah, I see.

There was an attempt to modify the Wikipedia article to include their definition, and there is now disagreement on whether it is good enough to say 'inherited from Sanskrit' without lying, even if one must hold one's nose.

If one is amendable with defining "Sanskrit" as being broadly synonymous with OIA-at-large, then sure MIA or NIA--even the divergent dialectical stuff or variations over time--descends from "Sanskrit" if perhaps an unattested flavor thereof. It looks like the wiki-folks are at least trying to differentiate "Vedic Sanskrit" from the broader lot which makes sense. Again, as noted below, even putting aside Vedic, the attested Sanskrit we have is by no means homogenous--tons of dialectic variants and one-off forms if you dig into it.

Ultimately from a definitional standpoint, I think this comes down to a bit of a lumping and splitting decision from a number of angles.* It is worth quoting from the intro to "The Prakrit Languages" which can be found at https://archive.org/details/in.ernet.dli.2015.531194

"...Grammarians and Rhetoricians of later days however explain prakrtam [MIA] as derived from the prakrti [OIA], i.e. samskrtam. This explanation is perfectly intelligible even if it be not historically correct. Practically we take Sanskrit forms as the basis and derive Prakrit forms therefrom. Nevertheless modern philology insists on an important reservation ; Sanskrit forms are quoted as the basis in as far as they represent the Old Indo-Aryan forms, but sometimes the particular Old Indo-Aryan form required to explain a Prakrit word is not found in Sanskrit at all, or only in a late work and obviously borrowed from Prakrit.

If in “Sanskrit” we include the Vedic language and all dialects of the Old Indo-Aryan period, then it is true to say that all the Prakrits are derived from Sanskrit. If on the other hand “Sanskrit” is used more strictly of the Panini-Patanjali language or “Classical Sanskrit” then it is untrue...

Is the link below the discussion you referenced or is it somewhere else? Interestingly, the discussion thread immediately following that also seems to show they have thrown out tadbhava categories, etc. It seems that this is an area of active discussion -- ha! Gotta love the internet for allowing people to discuss stuff like this.

https://en.wiktionary.org/wiki/Wiktiona ... Indo-Aryan

From this discussion, I also found another helpful source I had not seen before: https://www.academia.edu/34906743/The_d ... y_of_Indic. See especially pp. 424-425 re Pali which lays out some theories on origins and connections.

*I am mindful also that notions and discussions of the descent and origin of liturgical languages can be understandably ... thorny.

[Ares Land]

And where did those provincial dialects come from? The question for the analogy is which part of Latium did the provincial dialects come from? The Romans seem to have spoken a strange form of Latin.

No specific part of Latium. The dialects of Latium were long gone in Caesar's day. (It's been suggested Osco-Umbrian languages influenced Southern Italian but there's little evidence.)

Latin spoken in the Empire was very homogenous. Local dialects really begin to diverge, from each other and from classical Latin around the 4th or 5th century (*) and they remained mutually intelligible until at least the 7th century(**); whereas Classical Latin is more or less what Cicero wrote. An analogy would be modern English vs. the King James Bible.

(*) There were minor differences before. The emperor Hadrian is said to have spoken in an Hispanic accent at the beginning of his career. We don't know how that sounded like. We know that in Africa people had stopped distinguishing vowel length sometime around the third century, I believe?
(**) And probably later in some parts. We known Romance speakers in what would become Northern Gaul had trouble with Latin by Charlemagne's time; but their southern neighbours had no such problem.

[Richard W]

Ultimately from a definitional standpoint, I think this comes down to a bit of a lumping and splitting decision from a number of angles.* It is worth quoting from the intro to "The Prakrit Languages" which can be found at https://archive.org/details/in.ernet.dli.2015.531194

If in “Sanskrit” we include the Vedic language and all dialects of the Old Indo-Aryan period, then it is true to say that all the Prakrits are derived from Sanskrit. If on the other hand “Sanskrit” is used more strictly of the Panini-Patanjali language or “Classical Sanskrit” then it is untrue...

Note the use of the word 'if' rather than 'when'. Another issue has occurred to me. If Vedic as we receive it has undergone some standardisation, and Sanskrit means language varieties that we normally designate as Sanskrit, then the rejected original forms have been excluded from Sanskrit as a source of ancestry.

Is the link below the discussion you referenced or is it somewhere else? Interestingly, the discussion thread immediately following that also seems to show they have thrown out tadbhava categories, etc. It seems that this is an area of active discussion -- ha! Gotta love the internet for allowing people to discuss stuff like this.

https://en.wiktionary.org/wiki/Wiktiona ... Indo-Aryan

No, it's the link in the last two words of the banner's text! In other words, it's https://en.wiktionary.org/wiki/Wiktiona ... Indo-Aryan.

I am mindful also that notions and discussions of the descent and origin of liturgical languages can be understandably ... thorny.

Is that a pun on PIE thorn clusters? I detect an OT allusion to the crown of thorns.

Addendum:
In some ways the situation is analogous to English. Wiktionary gets round the problem with English because it mostly uses just the temporal division into Old, Middle and Modern English, so we don't have the issue of trying to claim that a Modern English word (or a Scots word) is inherited from a West Saxon word. A problem only arises when the Anglian word is different. By contrast, Indic is formally divided up into separate languages, which naturally causes problems when these languages are just key points in a dialect continuum. Just possibly, I'm worrying about which electron went through which slit.

[zompist]

(**) And probably later in some parts. We known Romance speakers in what would become Northern Gaul had trouble with Latin by Charlemagne's time; but their southern neighbours had no such problem.

This needs a caveat, I believe. My understanding is that people read Latin out loud as if it were the vernacular. E.g. if an Italian read sæculum, they'd read it as secolo. This is of course exactly what we do today in French or English— use a centuries-old spelling with modern pronunciation— and what the Chinese do with wényán. If they wrote Latin, it was with modern syntax as well. With these adjustments, they were a long way to the vernacular anyway.

[2+3 Clusivity]

... [tl;dr] forgive me, I am not trying to take over the misc forum, but typed out some research for myself and felt like it was hopefully worth sharing.

No, it's the link in the last two words of the banner's text!

Oh gotcha ... I had read that, but thought I was missing something given that those time stamps appear to be from 2018.

Another issue has occurred to me. If Vedic as we receive it has undergone some standardisation, and Sanskrit means language varieties that we normally designate as Sanskrit, then the rejected original forms have been excluded from Sanskrit as a source of ancestry.

I am a bit unclear on the latter part especially regarding rejection of forms as a source. Like most ancient texts, Vedic Sanskrit (and some of the later periods) should be viewed through the lens of an extended oral period of communication, subsequent writing(s) down, and further recommunications. I.e., its not a perfect snapshot of that "original" time/communication.

A lot of the dictionaries and sources note two categories of "intrusions" into the older texts: (1) natural MIA forms injected by scribes, etc. (e.g., IMHO, the "dial." /jajjʰ-/ form noted above); (2) what I would call "artificial" forms where a writer is trying to replicate earlier/different forms of Sanskrit***; and (3) gradients between those two poles. It would seem the first at least are a good source for direct forms from OIA. Again, this all depends on how you define on one hand "inherited/natural" versus "artificial" especially depending on how you define Sanskrit (which outside of Panini et al was by no means a standardized language). If you take a broader view, I think in some cases it can be really hard to differentiate inherited and artificial.

For a view on how hard it can be, take for example Whitney's treatment of the root you mentioned above (https://www.sanskrit-lexicon.uni-koeln. ... p?page=207). Note that the forms of the verb root (hr-) in the present section mostly fall into Sanskrit Class 1 (Thematic presents, accent on the root) across time periods but also has a form showing up in class 2 (Athematic root presents) and even stranger single form appearing to be in class 3 (Reduplicated athematic presents). So, should the latter forms be rejected as "artificial" relative to the first larger group? Hard to say.

Given that the perfective and other reduplicated forms are in /jV-/ it's perhaps not surprising that /ji~har-ti/ might crop up by analogy off of the perfect, etc. in /j-/. Or conversely, perhaps the /j-/ in the reduplicated forms is being produced by analogy off of similar roots in /h-/.* On the other hand, semantically related /bhr-, "bear"/ a typically class three present in /bi~bhar-ti/ is noted by Whitney as potentially being a source for a variant in /hr-/ despite the reduplicant in /j{i, a}-/. Is it indeed a variant of /bhr-/? Again, hard to say. This could make sense given that OIA and especially MIA show {bh, dh, gh} > /h/ in certain, sometimes conflicting, circumstances BUT especially word internally following pre-verbs.** Indeed, Whitney notes that /hr-/ is rare until later, so one could imagine this as being a late OIA or MIA variant form added to the text with analogized reduplicated forms not matching the original source.

Drawing back to the discussion of vulgar latin, I imagine there must be similar confusion with various periods of borrowing and reborrowing of forms.

I am mindful also that notions and discussions of the descent and origin of liturgical languages can be understandably ... thorny.

Is that a pun on PIE thorn clusters? I detect an OT allusion to the crown of thorns.

No -- face value. I can't do/communicate humor in text form effectively. Just noting that certain stakeholders have beliefs/interests invested in connecting certain languages in certain ways.

***See for example discussion of the Suparṇākhyāna which is most likely an epic-era poem written down in an artificial attempt at Vedic Sanskrit.
*Getting way over my skis, I am not finding a great etymology of /hr-/ but it is tempting to link it also to a PIE form like /*ǵʰs-ḗr ~ *ǵʰs-r-és; *ǵʰésr̥ ~ *ǵʰs-r-és/ with some sort of cluster reduction. Williams suggests a link to Ionic Gk. <χείρ> "hand" which falls under that form. IIr. at least tended to remove intervocalic /-s-/ between stops and, I believe, also between a stop and a plosive.
**MIA also notably has root (/bhav- ~ ho-/ < /bhu:/).

[Kuchigakatai]

And where did those provincial dialects come from? The question for the analogy is which part of Latium did the provincial dialects come from? The Romans seem to have spoken a strange form of Latin.

Well, there was always some variation within Latin, so presumably this variation was messily carried out outwards. I don't view the Romans as speaking a particularly strange thing.

For an example of the messiness, Latin tends to conserve Old Latin <ai> as <ae>, and same goes for the <ai> of Greek borrowings, except in the core most common words, where it tends to have <ē> at least as a variant, if not as the preferred form even: *dayh₂wḗr > laevir ~ lēvir 'father-in-law'. Varro in the 1st c. BC has a quip about people who say obscaenus/a/um 'obscene' instead of the (as surprising as it sounds, probably etymologically correct) obscēnus, saying that only something that is obscene at the threatre (scaena) should be referred to as obscaenum. (De Lingua Latina 7.95) The quip is doubly amusing from the point of view of Greek too, since Greek had σκηνή scēnḗ, but somehow even learned Romans weren't bothered by the hypercorrection scaena.

This mess is reflected in Romance outcomes too. In Classical times, there was variation on whether words had (the more urban and prestigious) au or (the more rustic and less prestigious) ō. You've probably heard of the mid-1st c. BC aristocrat siblings who changed their clan name from Claudius to Clōdius as part of a larger populist mocking of the upper class. A millennium later, Old French shows words that go back to au and others that go back to ō: OFr coue [ˈkowə] > late OFr queue [ˈkøə] reflects a Classical rustic cōda (urban cauda would've produced late OFr *coe [ˈkɔə] then modern *coe [ko], cf. laudō > OFr lo [lɔ]), but OFr torel [ˈtɔɾɛɫ] 'bull' (modern taureau [tɔʁo toʁo]) and oreille [ɔˈɾɛʎə] 'ear' (modern [ɔʁɛi]) reflect Classical *taurellus and auricula (cf. Old Occitan aurelha and taur [tawɾ]; *tōrellus and *ōricla would've produced *[ˈtoɾɛɫ] > modern *toureau [tuʁo] and modern *oureille [uʁɛi]).

There's a few individual words in a given Romance language where Oscan and Umbrian influence have been claimed, in Spanish notably the [​u] of octubre 'October' and nudo 'knot' (Latin octōber, nōdus) and the use of fui 'I was' for 'I went' (something found mostly just in Western Romance), but it's very little stuff.

(*) There were minor differences before. The emperor Hadrian is said to have spoken in an Hispanic accent at the beginning of his career. We don't know how that sounded like. We know that in Africa people had stopped distinguishing vowel length sometime around the third century, I believe?
(**) And probably later in some parts. We known Romance speakers in what would become Northern Gaul had trouble with Latin by Charlemagne's time; but their southern neighbours had no such problem.

Augustine in the 4th c. says he couldn't understand when the natives of Rome said <i> or <e> when they spoke. This has often been interpreted as showing that Augustine's African Latin (at least in today's Algeria) had the same merge as Sardinian, ī ĭ > /i/, ē ĕ > /e/, rather than the Roman ī > /i/, ĭ ē > /e/, ĕ > /ɛ/. I think there's some other evidence from Libyan inscriptions? In terms of the misspellings present or absent.

The Council of Tours is now often thought to say what it says because Charlemagne's reform about 20 years earlier had standardized a new way of reading Latin with one letter = one sound. So what was previously <saeculum> [ˈsjɛglə] (or [ˈsjeglu] for pre-Astur-Leonese speakers, etc., what zompist said) was now [ˈsɛkulum], and for people who only knew [ˈsjɛglə] this was incomprehensible. (EDIT: I corrected the attestation of lingua romana rustica...)

[Richard W]

And where did those provincial dialects come from? The question for the analogy is which part of Latium did the provincial dialects come from? The Romans seem to have spoken a strange form of Latin.

Well, there was always some variation within Latin, so presumably this variation was messily carried out outwards. I don't view the Romans as speaking a particularly strange thing.

My memory of the specific oddities may be wrong, but I think the hardening of voiced fricatives wasn't shared with Faliscan and Praenestine. There may be other features. Praeneste was destroyed about the time Julius Caesar was on the run.

[Richard W]

... [tl;dr] forgive me, I am not trying to take over the misc forum, but typed out some research for myself and felt like it was hopefully worth sharing.

I wouldn't object if this subthread were hived off - perhaps as 'Dodgy Descent from Classical Languages', though I'm not actually urging such a hiving off.

Another issue has occurred to me. If Vedic as we receive it has undergone some standardisation, and Sanskrit means language varieties that we normally designate as Sanskrit, then the rejected original forms have been excluded from Sanskrit as a source of ancestry.

I am a bit unclear on the latter part especially regarding rejection of forms as a source. Like most ancient texts, Vedic Sanskrit (and some of the later periods) should be viewed through the lens of an extended oral period of communication, subsequent writing(s) down, and further recommunications. I.e., its not a perfect snapshot of that "original" time/communication.

For example, if Vedic had at one stage contained a stem jaggh- but then systematically purged it in favour of jakṣ-, it would not be right to claim that the form in jaggh- could have derived from Vedic Sanskrit, for it was removed by the time the attested language took shape.

For a view on how hard it can be, take for example Whitney's treatment of the root you mentioned above (https://www.sanskrit-lexicon.uni-koeln. ... p?page=207). Note that the forms of the verb root (hr-) in the present section mostly fall into Sanskrit Class 1 (Thematic presents, accent on the root) across time periods but also has a form showing up in class 2 (Athematic root presents) and even stranger single form appearing to be in class 3 (Reduplicated athematic presents). So, should the latter forms be rejected as "artificial" relative to the first larger group? Hard to say.

There seems to be a fair bit of verb reshaping between Sanskrit and Pali, but I'm not too bothered about that. A derivation of Pali Class 1 (all grammarians) harati from Sanskrit Class 1 stem of root hṛ. by inheritance would be straightforward if such vertical transmission could occur. Confusingly, horizontal transmission is significant. For starters, recall Jespersen's remark that the local dialect is beaten into children in school playgrounds, and note that in patrilocal societies mothers tend to primarily teach children their fathers' language.

Given that the perfective and other reduplicated forms are in /jV-/ it's perhaps not surprising that /ji~har-ti/ might crop up by analogy off of the perfect, etc. in /j-/. Or conversely, perhaps the /j-/ in the reduplicated forms is being produced by analogy off of similar roots in /h-/.* On the other hand, semantically related /bhr-, "bear"/ a typically class three present in /bi~bhar-ti/ is noted by Whitney as potentially being a source for a variant in /hr-/ despite the reduplicant in /j{i, a}-/. Is it indeed a variant of /bhr-/? Again, hard to say. This could make sense given that OIA and especially MIA show {bh, dh, gh} > /h/ in certain, sometimes conflicting, circumstances BUT especially word internally following pre-verbs.** Indeed, Whitney notes that /hr-/ is rare until later, so one could imagine this as being a late OIA or MIA variant form added to the text with analogized reduplicated forms not matching the original source.

I'm tempted to add a note on that particular merger, i.e. that Pali harati may also derive from an Old Indic bharati.

[Kuchigakatai]

Oh, that's pretty interesting. Why not 'from the hill of his holiness'? Is it because har qoḏš-ô is some sort of very lexified phrase? I guess I could eventually find this if I look for it in grammars, but do you happen to know?

I think it's because a definite noun can only have one possessor, and qōḏeš has already taken that slot. *hārô qōḏeš is not possible, at least, according to Gesenius, who describes tacking the pronoun on the end of the phrase as a standard construction.

Well, that is also true of Arabic. So how come that phrase is not interpreted as "from the hill of his holiness", but rather "his hill of holiness"? Unless I'm misinterpreting what you meant by (the idiomatic English) "his holy hill"...

Question: with these things you've been mentioning, have you been trying to show that the possessor is the head in some sense? I guess we might need to wonder about what the working definition of "head" we want... I don't know what to use myself, but I'm starting to think this topic in Semitic might not be so easy.

Also remarkable is that as ben-yᵉmînî 'Benjaminite' is indefinite, it is made definite by making the second element definite ben-hayᵉmînî (Judges 3:15).

That is so weird. The -î derives the (nominalized) adjective from ben yəmîn, but then the article still attaches to the etymological possessor!? It's a wtf.

Since you brought up that weird exceptional case, here's an exceptional case in Arabic you might like:

Adjectives in Arabic typically take a noun complement via a preposition (مختلق عن شيء muxtaliqun ʕan ʃaiʔin, "different.NOM away.from thing.GEN" = 'different from sth'), but one possible construction meaning 'of [adj] [noun]', often equivalent to [adj]-[noun]-ed in English (long-legged, of long legs, many-sided, of many sides, multilateral) involves putting the adjective in the construct state with the noun complement as its (definite) genitive possessor, all while the adjective still agrees in gender+number (+definiteness, below) with the head noun it modifies.

مفاوضة متعدّدة الأطراف
mufaawadˤ-at-un mutaʕaddid-at-u (a)l-ʔatˤraafi
negotiation-FEM-INDEF.NOM numerous-FEM-CONST.NOM the-sides.DEF.GEN
'a many-sided negotiation'
(literally: a negotiation numerous of the sides)

Oh, but to form the definite form of the adjectival phrase, the article al- is attached to the construct adjective, something that otherwise never happens in Standard Arabic. Nouns in the construct state don't take al-.

المفاوضة المتعدّدة الأطراف
al-mufaawadˤ-at-u (a)l-mutaʕaddid-at-u (a)l-ʔatˤraafi
the-negotiation-FEM-DEF.NOM the-numerous-FEM-CONST.NOM the-sides.DEF.GEN
'the many-sided negotiation'
(literally: the negotiation [the] numerous of the sides)

And so you also have (to take examples from Karin Ryding's grammar, §10.5 / page 254) واسع النفوذ waasiʕu (a)n-nafuuði "wide of the influence" = 'widely-influential', عميق التفكير ʕamiiqu (a)t-tafkiiri "deep of the thought" = 'deep-thinking', متوسط الحجم mutawassitˤu (a)l-ħadʒmi "middle of the size" = 'medium-sized'.

This is the so-called (in English) "false ’iḍāfa" (Arabic إضافوة غير حقيقية ʔidˤaafa(tun) ɣairu ħaqiiqiija(tin) "truthless addition", i.e. false ’iḍāfa), which is always a nice headscratcher for people learning Arabic, who by the time they learn about it are used to the normal ’iḍāfa that's actually about possession...

[Richard W]

Oh, that's pretty interesting. Why not 'from the hill of his holiness'? Is it because har qoḏš-ô is some sort of very lexified phrase? I guess I could eventually find this if I look for it in grammars, but do you happen to know?

I think it's because a definite noun can only have one possessor, and qōḏeš has already taken that slot. *hārô qōḏeš is not possible, at least, according to Gesenius, who describes tacking the pronoun on the end of the phrase as a standard construction.

Well, that is also true of Arabic. So how come that phrase is not interpreted as "from the hill of his holiness", but rather "his hill of holiness"? Unless I'm misinterpreting what you meant by (the idiomatic English) "his holy hill"...

As these phrases use the possessor slot for an attribute or material, e.g. silver, it is indeed arguable that a possessor of the thing in the possessee slot also possesses the thing in the possessor slot. Thus, 'his vessels of silver' are normally indeed '(the) vessels of his silver'. (Of course, it can get confusing, as when one owns a building on someone else's land.)

Question: with these things you've been mentioning, have you been trying to show that the possessor is the head in some sense? I guess we might need to wonder about what the working definition of "head" we want... I don't know what to use myself, but I'm starting to think this topic in Semitic might not be so easy.

This construction approaches that arrangement, but not wholeheartedly - the Semitic cases I know of are in the wrong place. I've read of a language where the case markings that go on the noun phrase go on the possessor rather than the possessee, while the possessee is marked as such, but I misremember the name of the language. Alternatively, I've misremembered something about Bilin, where the two constructions for a case applied to a compound noun phrase is either possessor-GEN possessee-CASE or possessee possessor-GENADJ-CASE. The case markings differ from non-repeating postpositions by having lexically determined forms - there are special forms for the feminine singular and some 'pseudo-feminine' masculine nouns. Using the second form to say case is marked on the possessor rather then the possessee strikes me as grossly misleading, and not just because of the existence of the first form. The case goes on the second noun when nouns are in apposition. 'GENADJ' forms an adjective from a noun, so its form depends on the gender/number of the possessee.

Source: Bilin Morphology by David Appleyard.

[Space60]

There is no "12 AM" and "12 PM". Noon can't be before or after itself. There's 12 noon and 12 midnight. "12 AM" and "12 PM" are meaningless.

[KathTheDragon]

The people who successfully use 12 am and 12 pm to mean 12 midnight and 12 noon respectively would kindly disagree.