European languages before Indo-European

[Travis B.]

As opposed to baseless speculation, on the very flimsiest of grounds, about supposed Semitic languages in Europe for which no evidence exists whatsoever and which, to exist, would have to predate proto-Semitic itself...

[WeepingElf]

As opposed to baseless speculation, on the very flimsiest of grounds, about supposed Semitic languages in Europe for which no evidence exists whatsoever and which, to exist, would have to predate proto-Semitic itself...

I admit that there is very little known about the prehistoric languages of Europe. Well, that is pretty much by definition: "prehistoric" roughly means, "before the appearance of written records". But I think we can catch faint glimpses of the lost languages, from surviving languages, loanwords and other substratum effects in known languages, old geographical names and all that.

Baseless speculation of the sort our friend Octavià indulges in may be useful as source of inspiration for conlangs, but you always have to be aware that you are speculating, and must not peddle them as the truth. I speculate about prehistoric languages myself in my conlanging, but I am aware that Old Albic and all that is not real, only (hopefully) realistic - it could have been, but there is no way of knowing whether it was.

[Tropylium⁺]

In fact, Proto-Germanic has many words with no likely PIE etymnology.

There's a big and very important difference here between "no likely" and "no known", which is at the root of the reasons why nobody here agrees with your methodology. I'm not going to waste any time explaining it, because it should be obvious.

No, he's right here actually: PGmc has plenty of words that are unlikely to be directly from PIE. The most prominent layers is that of words of the shape *T-T (eg "cut", "trap") and *D-F. PIE didn't allow *D-D or *Dʰ-T, but whatever substrates PGmc went thru, we do not expect them to have had a problem with two voiceless consonants in a root, and probably not with *D-T either.

I'm told some of these words are ascribed to a fairly well-estabilish'd substrate known as the "Northwest Block" which is thought to have been spoken around the North Sea coast SW of Jutland, before the expansion of (West) Germanic there. This substrate is curious in that it appear to have been IE, and thus loans ascribed to it seem to derive from PIE while bypassing Grimm's Law. It is at least plausible that "crab" is one such a word. I've yet to read much on this topic however…

(A large part of the first set of exceptions can actually be explain'd by Kluge's Law: PIE stop + n voiceless geminate, when post-tonic — eg. *glébʰ-no- clip (*kleppan?). However not all cases yield to this.)

[Octavià]

Baseless speculation of the sort our friend Octavià indulges in may be useful as source of inspiration for conlangs, but you always have to be aware that you are speculating, and must not peddle them as the truth. I speculate about prehistoric languages myself in my conlanging, but I am aware that Old Albic and all that is not real, only (hopefully) realistic - it could have been, but there is no way of knowing whether it was.

I don't think my theories are actually much more speculative than yours.

[WeepingElf]

I don't think my theories are actually much more speculative than yours.

I admit that my hypotheses are somewhat speculative. But the key difference between us two is that you think you know what happened and claim the truth, while I know and openly admit that I don't know what was really going on, and consider my model a mere possibility - it may be false, and there are certainly at least some details which I have got wrong.

Indeed, for instance, the interpretation of the Old European Hydronymy is very difficult because we don't know the original meanings of those names, with some rare exceptions where we can at least guess because a name element coincides with a specific geographic feature (such as *hal- which is found specifically in the names of salt production sites and therefore probably meant 'salt'). And where the meaning is unknown, any attempt to say anything about language relationships remains speculative. My assumption that the names come from a sister language family of Indo-European makes things work out nicely, but I cannot claim that it was the truth. I may have utterly misinterpreted the evidence.

And finally, Old Albic and its relatives are nothing but conlangs. I make no claims to truth about those languages - they are my mere invention.

[Count Iblis]

Indeed, for instance, the interpretation of the Old European Hydronymy is very difficult because we don't know the original meanings of those names, with some rare exceptions where we can at least guess because a name element coincides with a specific geographic feature (such as *hal- which is found specifically in the names of salt production sites and therefore probably meant 'salt').

Could this possibly come from Greek?

[WeepingElf]

Indeed, for instance, the interpretation of the Old European Hydronymy is very difficult because we don't know the original meanings of those names, with some rare exceptions where we can at least guess because a name element coincides with a specific geographic feature (such as *hal- which is found specifically in the names of salt production sites and therefore probably meant 'salt').

Could this possibly come from Greek?

I have heard of that explanation, but I don't think it is plausible. These names probably date back to the Neolithic, and Greek was never spoken in Central Europe. I have also have heard of a theory that those names were Celtic, but the /s/ > /h/ shift happened only in Britain, not in Gaul or Central Europe.

[Octavià]

I don't think my theories are actually much more speculative than yours.

I admit that my hypotheses are somewhat speculative. But the key difference between us two is that you think you know what happened and claim the truth, while I know and openly admit that I don't know what was really going on, and consider my model a mere possibility - it may be false, and there are certainly at least some details which I have got wrong.

Well, what it really matters to me is the truth, not the one who actually finds it.

Indeed, for instance, the interpretation of the Old European Hydronymy is very difficult because we don't know the original meanings of those names, with some rare exceptions where we can at least guess because a name element coincides with a specific geographic feature (such as *hal- which is found specifically in the names of salt production sites and therefore probably meant 'salt'). And where the meaning is unknown, any attempt to say anything about language relationships remains speculative. My assumption that the names come from a sister language family of Indo-European makes things work out nicely, but I cannot claim that it was the truth. I may have utterly misinterpreted the evidence.

IMHO, the IE word for 'salt', commonly reconstructed as *sa(:)l-, is a Neolithic Wanderwort whose ultimate origin is PNC *q’eɦlV (˜ -ɫ-) 'bitter', a root apparently also reflected in Kartvelian *q’wel- 'cheese'. Apparently, the uvular stop was lenied to a fricative *χ (= PIE *H₂), reflected in hal-, but which shifted to /s/ in an unknown language from it which it was borrowed by IE.

A similar sound shift happened in *sel- 'dwelling, settlement', a regional word found in Balto-Slavic (e.g. Russian seló 'village'), Germanic *sala- 'hall' and Latin solum 'ground, floor' (< *selom), from Eurasiatic *χelA- 'to dwell, to live' (Dolgopolsky's ND 2579), or *sam- 'summer' < Eurasiatic *χEmV (˜ ħ-) ‘warm’ (ND 2586). In this case, *χ was kept in *H₂eHm-ºr- 'heat (of the day)' (Greek hēméra, Armenian awr 'day').

[WeepingElf]

I don't think my theories are actually much more speculative than yours.

I admit that my hypotheses are somewhat speculative. But the key difference between us two is that you think you know what happened and claim the truth, while I know and openly admit that I don't know what was really going on, and consider my model a mere possibility - it may be false, and there are certainly at least some details which I have got wrong.

Well, what it really matters to me is the truth, not the one who actually finds it.

Indeed, for instance, the interpretation of the Old European Hydronymy is very difficult because we don't know the original meanings of those names, with some rare exceptions where we can at least guess because a name element coincides with a specific geographic feature (such as *hal- which is found specifically in the names of salt production sites and therefore probably meant 'salt'). And where the meaning is unknown, any attempt to say anything about language relationships remains speculative. My assumption that the names come from a sister language family of Indo-European makes things work out nicely, but I cannot claim that it was the truth. I may have utterly misinterpreted the evidence.

IMHO, the IE word for 'salt', commonly reconstructed as *sa(:)l-, is a Neolithic Wanderwort whose ultimate origin is PNC *q’eɦlV (˜ -ɫ-) 'bitter', a root apparently also reflected in Kartvelian *q’welV 'cheese'. Apparently, the uvular stop was lenied to a fricative *χ (= PIE *H₂), reflected in hal-, but which shifted to /s/ in an unknown language from it which it was borrowed by IE.

Maybe. I cannot say "No way, you are wrong!" - with my current state of knowledge (or whoseever state of knowledge), such a judgment lies beyond my capabilities. But I feel that my hypothesis (namely that PIE *sal and Hesperic *hal are regular reflexes of a Proto-Europic *sxal) is simpler. A relationship between the Proto-Europic form and the Caucasian words you cite is not impossible, but not necessary for my hypothesis.

A similar sound shift happened in *sel- 'dwelling, settlement', a regional word found in Balto-Slavic (e.g. Russian seló 'village'), Germanic *sala- 'hall' and Latin solum 'ground, floor' (< *selom), from Eurasiatic *χelA- 'to dwell, to live' (Dolgopolsky's ND 2579), or *sam- 'summer' < Eurasiatic *χEmV (˜ ħ-) ‘warm’ (ND 2586). In this case, *χ was kept in *H₂eHm-ºr- 'heat (of the day)' (Greek hēméra, Armenian awr 'day').

The connection between 'to dwell, to live' and 'heat (of the day)' is remote and doubtful, but not impossible, if one considers a dwelling as a place where a hearth spends warmth.

[Octavià]

Maybe. I cannot say "No way, you are wrong!" - with my current state of knowledge (or whoseever state of knowledge), such a judgment lies beyond my capabilities. But I feel that my hypothesis (namely that PIE *sal and Hesperic *hal are regular reflexes of a Proto-Europic *sxal) is simpler. A relationship between the Proto-Europic form and the Caucasian words you cite is not impossible, but not necessary for my hypothesis.

AFAIK, there's no Nostratic/Eurasiatic etymology for this. Here is when Vasco-Caucasian comes to rescue.

A similar sound shift happened in *sel- 'dwelling, settlement', a regional word found in Balto-Slavic (e.g. Russian seló 'village'), Germanic *sala- 'hall' and Latin solum 'ground, floor' (< *selom), from Eurasiatic *χelA 'to dwell, to live' (Dolgopolsky's ND 2579), or *sam- 'summer' < Eurasiatic *χEmV (˜ ħ-) ‘warm’ (ND 2586). In this case, *χ was kept in *H₂eHm-ºr- 'heat (of the day)' (Greek hēméra, Armenian awr 'day').

The connection between 'to dwell, to live' and 'heat (of the day)' is remote and doubtful, but not impossible, if one considers a dwelling as a place where a hearth spends warmth.

No, these are actually two different roots:

- Eurasiatic *χelA 'to dwell, to live' > *sel- 'dwelling, settlement'
- Eurasiatic *χEmV (˜ ħ-) 'warm' > *sem- 'summer' ~ *H₂eHm-ºr- 'heat (of the day)'

Notice that Dolgopolsky didn't include the IE forms in his reconstructions. This is my own contribution.

My point is the forms with initial *s- (thus including 'salt') aren't PIE-native but borrowings from an extinct language (notice that the words 'dwelling' and 'summer' have a limited spread), while the corresponding native PIE forms would have *H- instead (e.g. *hal- 'salt'). So it might turn out that *sa(:)l- 'salt' is the actual "Hesperic" word, much to Jörg's despair.

[Serali]

This is my real name.

So I'm replying to a parasitic wasp? Cool!

[WeepingElf]

Maybe. I cannot say "No way, you are wrong!" - with my current state of knowledge (or whoseever state of knowledge), such a judgment lies beyond my capabilities. But I feel that my hypothesis (namely that PIE *sal and Hesperic *hal are regular reflexes of a Proto-Europic *sxal) is simpler. A relationship between the Proto-Europic form and the Caucasian words you cite is not impossible, but not necessary for my hypothesis.

AFAIK, there's no Nostratic/Eurasiatic etymology for this. Here is when Vasco-Caucasian comes to rescue.

If there are no cognates in other Eurasiatic languages, it may of course be possible that Proto-Europic borrowed the word from some other language, perhaps in the Caucasian region ... wait ... Eurasiatic??? I thought you consider Indo-European and Uralic unrelated? Eurasiatic, as it is usually understood, comprises IE, Uralic, Altaic and, depending on the author, various other languages of northeastern Eurasia. Or do you use the term in some nonstandard way?

A similar sound shift happened in *sel- 'dwelling, settlement', a regional word found in Balto-Slavic (e.g. Russian seló 'village'), Germanic *sala- 'hall' and Latin solum 'ground, floor' (< *selom), from Eurasiatic *χelA 'to dwell, to live' (Dolgopolsky's ND 2579), or *sam- 'summer' < Eurasiatic *χEmV (˜ ħ-) ‘warm’ (ND 2586). In this case, *χ was kept in *H₂eHm-ºr- 'heat (of the day)' (Greek hēméra, Armenian awr 'day').

The connection between 'to dwell, to live' and 'heat (of the day)' is remote and doubtful, but not impossible, if one considers a dwelling as a place where a hearth spends warmth.

No, these are actually two different roots:

- Eurasiatic *χelA 'to dwell, to live' > *sel- 'dwelling, settlement'
- Eurasiatic *χEmV (˜ ħ-) 'warm' > *sem- 'summer' ~ *H₂eHm-ºr- 'heat (of the day)'

Oops, my fault.

Notice that Dolgopolsky didn't include the IE forms in his reconstructions. This is my own contribution.

My point is the forms with initial *s- (thus including 'salt') aren't PIE-native but borrowings from an extinct language (notice that the words 'dwelling' and 'summer' have a limited spread), while the corresponding native PIE forms would have *H- instead (e.g. *hal- 'salt'). So it might turn out that *sa(:)l- 'salt' is the actual "Hesperic" word, much to Jörg's despair.

The Hesperic word, i.e. the one found in the Neolithic names, is *hal-. The Proto-Europic form which I reconstruct as *sxal- I do not reconstruct that way just because IE has /s/ and Hesperic has /h/. There is evidence for the /x/ from IE alone: the /a/-vocalism of the IE word hints at the presence of a *h2 (= Proto-Europic *x) before the vowel, which was as usual lost in IE. A cluster *sx is comparable to clusters of the *tk kind which are found in some IE words (see for instance PIE *dhghom 'Earth' (should have some twiddle on the <g>)).

[Octavià]

AFAIK, there's no Nostratic/Eurasiatic etymology for this. Here is when Vasco-Caucasian comes to rescue.

If there are no cognates in other Eurasiatic languages, it may of course be possible that Proto-Europic borrowed the word from some other language, perhaps in the Caucasian region ... wait ... Eurasiatic??? I thought you consider Indo-European and Uralic unrelated? Eurasiatic, as it is usually understood, comprises IE, Uralic, Altaic and, depending on the author, various other languages of northeastern Eurasia. Or do you use the term in some nonstandard way?

I personally don't think Uralic is the closest relative of IE, but Altaic instead. I also use the term Eurasiatic as a synonym of "Nostratic" in other theories, thus explicitly excluding Afrasian. The actual perimeter of Eurasiatic might vary, but I tend to include Kartvelian and Dravidian.

The Proto-Europic form which I reconstruct as *sxal- I do not reconstruct that way just because IE has /s/ and Hesperic has /h/. There is evidence for the /x/ from IE alone: the /a/-vocalism of the IE word hints at the presence of a *h2 (= Proto-Europic *x) before the vowel, which was as usual lost in IE. A cluster *sx is comparable to clusters of the *tk kind which are found in some IE words (see for instance PIE *dhghom 'Earth' (should have some twiddle on the <g>)).

Unfortunately your theory is unable to explain the other words I mentioned, and also it doesn't gives any clue about the external cognates of 'salt'. This is why I can't accept it.

[Tropylium⁺]

The Proto-Europic form which I reconstruct as *sxal- I do not reconstruct that way just because IE has /s/ and Hesperic has /h/. There is evidence for the /x/ from IE alone: the /a/-vocalism of the IE word hints at the presence of a *h2 (= Proto-Europic *x) before the vowel, which was as usual lost in IE. A cluster *sx is comparable to clusters of the *tk kind which are found in some IE words (see for instance PIE *dhghom 'Earth' (should have some twiddle on the <g>)).

Unfortunately your theory is unable to explain the other words I mentioned, and also it doesn't gives any clue about the external cognates of 'salt'. This is why I can't accept it.

*sam- "summer" has that same *sa- pattern so *sxem most certainly would work there.

Further examples BTW: http://tech.groups.yahoo.com/group/Nost ... ssage/5933

[Octavià]

There's another thing it doesn't work in Jörg's "Hesperic" theory, namely that the Old European Hydronymy (OEH) has actually no /h/ but zero, which is the expected result from PIE *H₂- (/χ/). It's actually the non-IE substrate language which has *H₂- > *s-.

An actual example would be the OEH root al- < *H₂el- 'to well up, to flow', which I link to sal- < PIE *sal- 'to jump'.

*sam- "summer" has that same *sa- pattern so *sxem most certainly would work there

Where's the comparative evidence (i.e. external cognates) for such a cluster?

Further examples BTW: http://tech.groups.yahoo.com/group/Nost ... ssage/5933

These are from Arnaud Fournet, a former colleague of mine. This is why I call the sound shift *H₂- > *s- as "Fournet's Law".

[WeepingElf]

AFAIK, there's no Nostratic/Eurasiatic etymology for this. Here is when Vasco-Caucasian comes to rescue.

If there are no cognates in other Eurasiatic languages, it may of course be possible that Proto-Europic borrowed the word from some other language, perhaps in the Caucasian region ... wait ... Eurasiatic??? I thought you consider Indo-European and Uralic unrelated? Eurasiatic, as it is usually understood, comprises IE, Uralic, Altaic and, depending on the author, various other languages of northeastern Eurasia. Or do you use the term in some nonstandard way?

I personally don't think Uralic is the closest relative of IE, but Altaic instead. I also use the term Eurasiatic as a synonym of "Nostratic" in other theories, thus explicitly excluding Afrasian. The actual perimeter of Eurasiatic might vary, but I tend to include Kartvelian and Dravidian.

I think the evidence for an IE-Uralic relationship is stronger than that for an IE-Altaic relationship (Altaic is of course doubted by many scholars, but I think the parallel alternations in the pronoun system makes a relationship of Turkic, Mongolic and Tungusic very plausible). That doesn't mean that Altaic cannot be a more distant relative of Indo-European.

The Proto-Europic form which I reconstruct as *sxal- I do not reconstruct that way just because IE has /s/ and Hesperic has /h/. There is evidence for the /x/ from IE alone: the /a/-vocalism of the IE word hints at the presence of a *h2 (= Proto-Europic *x) before the vowel, which was as usual lost in IE. A cluster *sx is comparable to clusters of the *tk kind which are found in some IE words (see for instance PIE *dhghom 'Earth' (should have some twiddle on the <g>)).

Unfortunately your theory is unable to explain the other words I mentioned, and also it doesn't gives any clue about the external cognates of 'salt'. This is why I can't accept it.

I admit that there are things which my hypothesis does not cover. But so far, you haven't given me solid counter-evidence. If you reject it because you feel that things were different, you are free to do so.

[WeepingElf]

There's another thing it doesn't work in Jörg's "Hesperic" theory, namely that the Old European Hydronymy (OEH) has actually no /h/ but zero, which is the expected result from PIE *H₂- (/χ/). It's actually the non-IE substrate language which has *H₂- > *s-.

An actual example would be the OEH root al- < *H₂el- 'to well up, to flow', which I link to sal- < PIE *sal- 'to jump'.

Not really a problem. In my view, PEur *x is normally lost in Hesperic (as in the root *al- < *xal-; the loss may actually have been in the IE languages which adopted the Hesperic names), but the cluster *sx gave *h.

*sam- "summer" has that same *sa- pattern so *sxem most certainly would work there

Where's the comparative evidence (i.e. external cognates) for such a cluster?

Can't you accept a reconstruction for Proto-Europic without further cognates in other, putatively related families? Of course, if a, for instance, Uralic cognate turns up, the better!

Further examples BTW: http://tech.groups.yahoo.com/group/Nost ... ssage/5933

These are from Arnaud Fournet, a former colleague of mine. This is why I call the sound shift *H₂- > *s- as "Fournet's Law".

Which IE languages have which forms? One possibility is that we are dealing with borrowings from an IE dialect with the shift */s/ > */h/, which has happened independently in Iranian, Greek and Brythonic. The matter requires further research.

[Octavià]

These are from Arnaud Fournet, a former colleague of mine. This is why I call the sound shift *H₂- > *s- as "Fournet's Law".

Which IE languages have which forms? One possibility is that we are dealing with borrowings from an IE dialect with the shift */s/ > */h/, which has happened independently in Iranian, Greek and Brythonic. The matter requires further research.

I'm not sure if you actually understood my point. What I said is that in addition to the regular shift *H₂- > zero, some words (usually with a limited spread, thus probably not PIE) have initial s-. This clearly has nothing to do with */s/ > */h/, as it's rather the opposite phenomenon.

IMHO this is a substrate phenomenon, that is, these words with initial *s- were borrowed from some extinct substrate language. Our respective views also differ because I include here the word 'salt', according to the etymology I gave before.

[WeepingElf]

These are from Arnaud Fournet, a former colleague of mine. This is why I call the sound shift *H₂- > *s- as "Fournet's Law".

Which IE languages have which forms? One possibility is that we are dealing with borrowings from an IE dialect with the shift */s/ > */h/, which has happened independently in Iranian, Greek and Brythonic. The matter requires further research.

I'm not sure if you actually understood my point. What I said is that in addition to the regular shift *H₂- > zero, some words (usually with a limited spread, thus probably not PIE) have initial s-. This clearly has nothing to do with */s/ > */h/, as it's rather the opposite phenomenon.

IMHO this is a substrate phenomenon, that is, these words with initial *s- were borrowed from some extinct substrate language. Our respective views also differ because I include here the word 'salt', according to the etymology I gave before.

At least with the 'salt'-word, you have it backwards: PIE had */s/, Hesperic had */h/ (or rather, something that became /h/ in the IE languages through which the names came unto us). But I have to admit that I don't know yet how to consider those other words with "irregular" */s/. I need to know which IE languages have which forms with which meanings.

[Octavià]

Perhaps these links would help (please ignore high-level links):
*sal- 'to jump'
*sam- 'summer'
*sel- 'dwelling, settlement' (+ Latin solum)

[WeepingElf]

Perhaps these links would help (please ignore high-level links):
*sal- 'to jump'
*sam- 'summer'
*sel- 'dwelling, settlement' (+ Latin solum)

They are all regular, showing /h/ instead of /s/ in exactly those branches which have shifted initial /s/ to /h/. The first two may actually have been *sh2al- and *sh2am-, because of their /a/-vocalism. They may have been borrowed into PIE (or an ancestor of PIE) from some other language, and it is of course not impossible that "Fournet's Law" applies to that source language. I cannot say that you are wrong, but I don't see how this contradicts my hypothesis either.

[Åge Kruger]

The thing is many people prefer to explain things from what's already known instead of searching into the unknown. I'll give you a folk story as an example: One night a man who had lost his keys in a street poorly illuminated was walking in circles around the only street lamp working, looking in vain for them. Another man who was walking up the street saw this and asked him:
- What are you doing?
- I'm looking for my keys.
- Are you sure it's here where you lost them?
- No, I lost them down there.
- Then why are you looking for them here?
- Because there's light.

Of course, the lost keys represent words with no known etymology and the street lamp represents PIE. The moral is some people prefer to invent an IE etymology, no matter how implausible it might be, instead of looking into the "unknown" (dark street).

Yesterday, upon the stair,
I met a man who wasn’t there
He wasn’t there again today
I wish, I wish he’d go away...

When I came home last night at three
The man was waiting there for me
But when I looked around the hall
I couldn’t see him there at all!
Go away, go away, don’t you come back any more!
Go away, go away, and please don’t slam the door... (slam!)

Last night I saw upon the stair
A little man who wasn’t there
He wasn’t there again today
Oh, how I wish he’d go away

[Tropylium⁺]

Perhaps these links would help (please ignore high-level links):
*sal- 'to jump'
*sam- 'summer'
*sel- 'dwelling, settlement' (+ Latin solum)

They are all regular, showing /h/ instead of /s/ in exactly those branches which have shifted initial /s/ to /h/. The first two may actually have been *sh2al- and *sh2am-, because of their /a/-vocalism.

I think the point here is that because *a before *s correlates to *s-versus-*h2 doublets, it is this *s~*h2 itself that is causing the coloring? *sh2 does not seem like a parsimonious reconstruction. If Old European lost *h2 prevocally, that would be expected to happen regardless of a preceding *s. And if it's the *s that's being reflected as *h, then the added *h2 is ad hoc.

At this point we should remember that laryngeal coloring is not a comparativ reconstruction, but an internal reconstruction, and that not all late PIE *a need to come from that. How about: *sal, *sam- are loans from an older stage of Old European to late PIE; *hal-, *heHm- are newer loans from a newer stage into the individual branches.

Let's see if this keeps up (looking thru the rest of AF's wordlist)…
"Cereal" *aw- ~ *sab-
The no-initial form is found in Latin (avēna) and Balto-Slavic. The s-initial form is found in Latin (sabaia meaning "beer") and Germanic derivativs. I'm not sure what to do with the Kartvelian word.
"Pure" *sak-
This comparision is limited to Latin (Italic) sacer ~ Semitic *zaq-. Loaning from there seems better here.
"Hedge" *saj-
Found in Latin (saepēs < *sai-p-), Greek (*sai-m-), Baltic and IA (*sai-dʰ-). The Greek is /haimos/, not ˣ/aimos/, so we actually have no variants here at all to account for.
"Angry" *aj- ~ *saj-
S-initial form in Latin (saevus), Tocharian, Baltic (*saj-w-), Greek (hai-m-), Germanic (*saj-r-), Celtic (*saj-t-). No-initial form in Latin (aemulus), and a h-initial form in Hittite! However, these diverge semantically once the cognates outside Latin are examined (the former seems to be "sick", the latter "similar"). The latter would seem to be better compared to *sVm- "together". Greek zelos doesn't quite fit in any way.
"Thorn" *ab- ~ *sap-
S-initial form in Latin sappīnus (which is usually connected to the "sap" root however). No-initial form in Latin abiēs. Could work, tho the distribution isn't very good.
"To sew" *or- ~ *sar-
S- initial form in Latin (sarciō), Hittite, Greek (with /e/, so *sH1rk-?) No-initial form in Latin (ōrdiōr), Greek (with short /o/). I see no reason to strip the /k, d/ off the roots and the o-vocalism means this cannot be *h2. Arabic zar- doesn't seem to even have a Semitic etymology so I'll dismiss it as a coincidence.

Doesn't seem too good ("cereal" is the best candidate for continuing the pattern). I'll do the rest sometime else.

[TomHChappell]

Yesterday, upon the stair,
I met a man who wasn’t there
He wasn’t there again today
I wish, I wish he’d go away...

When I came home last night at three
The man was waiting there for me
But when I looked around the hall
I couldn’t see him there at all!
Go away, go away, don’t you come back any more!
Go away, go away, and please don’t slam the door... (slam!)

Last night I saw upon the stair
A little man who wasn’t there
He wasn’t there again today
Oh, how I wish he’d go away

And furthermore:

One bright day, late at night
Two dead boys rose up to fight.
Back to back they faced each other,
Drew their swords, and ... SHOT each other!

A deaf policeman heard the noise
And came and killed the two dead boys.

(If you don't believe this lie is true
Just ask the blind man -- He saw it, too.)

[WeepingElf]

Perhaps these links would help (please ignore high-level links):
*sal- 'to jump'
*sam- 'summer'
*sel- 'dwelling, settlement' (+ Latin solum)

They are all regular, showing /h/ instead of /s/ in exactly those branches which have shifted initial /s/ to /h/. The first two may actually have been *sh2al- and *sh2am-, because of their /a/-vocalism.

I think the point here is that because *a before *s correlates to *s-versus-*h2 doublets, it is this *s~*h2 itself that is causing the coloring? *sh2 does not seem like a parsimonious reconstruction. If Old European lost *h2 prevocally, that would be expected to happen regardless of a preceding *s. And if it's the *s that's being reflected as *h, then the added *h2 is ad hoc.

At this point we should remember that laryngeal coloring is not a comparativ reconstruction, but an internal reconstruction, and that not all late PIE *a need to come from that. How about: *sal, *sam- are loans from an older stage of Old European to late PIE; *hal-, *heHm- are newer loans from a newer stage into the individual branches.

It may very well be that most PIE words with an *a not accounted for by a securely reconstructible adjacent *h2 are rather late loans, entering PIE at a time when the vowel-colouring effects of the laryngeals were already in place. Especially when they are limited to a few neighbouring branches. I don't see much of a problem with a cluster *sh2; if *h2 was /x/, the cluster /sx/ doesn't look weirder than /dhgh/ which is attested. But that cluster certainly wasn't very common. The 'salt'-word may have been the only PIE word with that cluster.

Let's see if this keeps up (looking thru the rest of AF's wordlist)…
"Cereal" *aw- ~ *sab-
The no-initial form is found in Latin (avēna) and Balto-Slavic. The s-initial form is found in Latin (sabaia meaning "beer") and Germanic derivativs. I'm not sure what to do with the Kartvelian word.
"Pure" *sak-
This comparision is limited to Latin (Italic) sacer ~ Semitic *zaq-. Loaning from there seems better here.
"Hedge" *saj-
Found in Latin (saepēs < *sai-p-), Greek (*sai-m-), Baltic and IA (*sai-dʰ-). The Greek is /haimos/, not ˣ/aimos/, so we actually have no variants here at all to account for.
"Angry" *aj- ~ *saj-
S-initial form in Latin (saevus), Tocharian, Baltic (*saj-w-), Greek (hai-m-), Germanic (*saj-r-), Celtic (*saj-t-). No-initial form in Latin (aemulus), and a h-initial form in Hittite! However, these diverge semantically once the cognates outside Latin are examined (the former seems to be "sick", the latter "similar"). The latter would seem to be better compared to *sVm- "together". Greek zelos doesn't quite fit in any way.
"Thorn" *ab- ~ *sap-
S-initial form in Latin sappīnus (which is usually connected to the "sap" root however). No-initial form in Latin abiēs. Could work, tho the distribution isn't very good.
"To sew" *or- ~ *sar-
S- initial form in Latin (sarciō), Hittite, Greek (with /e/, so *sH1rk-?) No-initial form in Latin (ōrdiōr), Greek (with short /o/). I see no reason to strip the /k, d/ off the roots and the o-vocalism means this cannot be *h2. Arabic zar- doesn't seem to even have a Semitic etymology so I'll dismiss it as a coincidence.

Doesn't seem too good ("cereal" is the best candidate for continuing the pattern). I'll do the rest sometime else.

It does indeed look quite shoddy; I feel that none of these words have anything to do with the pattern we see in PIE *sal ~ Hesp. *hal 'salt'.