A question about Proto-Indo-European phonology

[FlatAssembler]

What do you guys think about the controversy over PIE having voiceless aspirates? I think it's actually very plausible that it did have (mainstream linguistics holds that it didn't). Consider the pairs of Latin and Ancient Greek words such as "sapientia" and "sophia". Both of them mean "wisdom", and appear to be obviously related. Yet, if you assume PIE didn't have voiceless aspirates, the Latin word points to the reconstructed PIE root *seh2p, and the Ancient Greek word points to the reconstruction *soHbh. However, if you assume PIE did have voiceless aspirates, you can reconstruct the root *seh2ph, from which both the Latin and the Ancient Greek word can be derived. There are a few more such words. Let's hear your thoughts!

[Daniel]

1. I don't have a strong opinion about this. (Nor is it my research area.)

2. What's the starting point for your comments? Grimm's Law for example indicates that PIE had voiceless plosives, voiced plosives, and "voiced aspirated" plosives (e.g., murmured voicing, or something else). Since this reconstructs contrasts rather than phonetic articulation, there is a single category of voiceless plosives, which might have been phonetically aspirated, or not. You are suggesting there was a phonemic contrast in aspiration for voiceless plosives?
All I would have to say about that is that from a typological perspective the current model for PIE is not unusual. Many languages have no aspiration contrast in unvoiced plosives (English is one, which has phonetic aspiration, or Spanish is one which does not). The only unusual part of (this component of) the PIE reconstruction is the "voiced aspirated" consonants, which in some modern (Indic) languages behave phonemically similarly to true aspirated voiceless plosives, but are not phonetically the same (not really aspirated), so there is no reason to believe that the presence of this (quasi) "aspiration" in the voiced stops would imply true aspiration (and contrastive non-aspiration) in the voiceless series. The "voiced aspirated" series are a puzzle on their own to be solved otherwise, independently.

3. Do not rely on a single pair or handful of pairs to make an argument in reconstruction! As I've said elsewhere, that is a statistical mistake, and you can always find coincidences to make arguments like that. The difficulty is in finding widespread patterns (both similarities and differences, systematically distributed) that also align with the rest of the reconstruction in general. If you did follow this through, then wouldn't the rest of the reconstruction also fall apart? Wouldn't explaining these few pairs of words this way actually do more harm than good by taking away the explanation for many more other pairs of words? Or is there some way around that such as also suggesting that there was a merger at some point so that the aspiration contrast disappeared and what we see in the daughter languages are just some traces of that? The problem is that it would have needed to independently merge in different daughter languages (unless you can somehow show evidence for a large-scale macro-subgrouping as in the proposed but problematic satem-centum divide).

4. Looking at the etymology for those two words in particular:

https://en.wiktionary.org/wiki/sophist#English
https://en.wiktionary.org/wiki/%CF%83%CE%BF%CF%86%CE%B9%CF%83%CF%84%CE%AE%CF%82#Ancient_Greek
https://en.wiktionary.org/wiki/%CF%83%CE%BF%CF%86%CE%AF%CE%B6%CF%89#Ancient_Greek
https://en.wiktionary.org/wiki/%CF%83%CE%BF%CF%86%CF%8C%CF%82#Ancient_Greek
Greek sophos < PIE *seh₁p

https://en.wiktionary.org/wiki/sapient#Etymology
https://en.wiktionary.org/wiki/sapio#Latin
And there we have an explanation:

From Proto-Italic *sapiō, from Proto-Indo-European *sh₁p-i- (“to notice”), from Proto-Indo-European *seh₁p- (“to try, to research”). Cognate with Ancient Greek σοφός (sophós), Old English sefa (“mind, spirit, mood”).

The two words are in fact related, via known/established sound changes in relevant languages/subgroups.
No alternative theory is needed, unless you find a problem with this one.


So in conclusion, there are two ways to approach a question like yours:
1. Argue against the proposal based on the data presented (done).
2. Reject based on having too little data (also reasonable in this case, unless more systematic patterns can be identified).

If you want to maintain the analysis, you would need to so that there is some data not explained by existing theories, or that your explanation is more efficient at explaining the same data (which is not a strong argument, but still a somewhat reasonable one for assuming the simpler theory over the one with more assumptions).

[FlatAssembler]

So, how is the reconstruction *seh1p supposed to explain the aspiration in the Greek word?

Pokorny supposes the PIE root to be *sap~*sab, if that helps you to explain.

[Daniel]

I'm not sure. You'd have to ask someone who specializes in (or at least knows specifically about) PIE>Greek changes. But I would assume there is some reason for that change, given that wiktionary is usually pretty good about etymologies and you cite Pokorny as well.

[FlatAssembler]

I don't know. I mean, it's definitely not explicable by the basics explained on Wikipedia.
https://en.wikipedia.org/wiki/Proto-Greek_language#Proto-Greek_changes
Nor do the etymological dictionaries appear to agree on reconstruction (now this is not just a spelling difference).

[Daniel]

You'll have to dig deeper than Wikipedia to find the details of the theory. It just seems unlikely that something this obvious was missed by other researchers for 200 years, for well-known roots that are thought to be related.

Note that it falls at the end of a root, which might make some unusual things happen.

[ForumExplorer]

Daniel, you've said I can ask honest questions in the Historical Linguistics subforum. So, can you explain to a layman what this thread is about? And why do you think that rhetoric "How come nobody noticed that?" applies here? FlatAssembler is not asserting a conspiracy. If somebody claims rockets can't exist because of XYZ, then that rhetoric is appropriate, but I don't see how is this analogous to that.

[Daniel]

Daniel, you've said I can ask honest questions in the Historical Linguistics subforum

Well, yes, but as a new thread, rather than in another one that is about a technical topic.

So, can you explain to a layman what this thread is about?

OK, but briefly so it doesn't drag things too off topic. Again, for other or general questions please do start a new thread.

And why do you think that rhetoric "How come nobody noticed that?" applies here?

I assume someone did notice it, but that it isn't a detail that is easy to find online, as I have said.

FlatAssembler is not asserting a conspiracy.

Correct. Just that this detail has been overlooked. But I still find that unlikely given the extensive, detail-oriented research for the past two centuries and that the particular etymology in question is well known. I don't personally know the answer (and I'd rather suggest spending some time in the library to look it up than guessing myself, because the solution is not apparent from the limited data available here or that I can infer at the moment).

If somebody claims rockets can't exist because of XYZ, then that rhetoric is appropriate, but I don't see how is this analogous to that.

I don't follow your comparison either.


So what is this thread about?

1. Proto-Indo-European (PIE) is the accepted ancestor to English, Latin, Greek, Sanskrit, Russian, and so forth. (Look it up on Wikipedia.) In short, there was some shared parent language. The details about it are up for debate, but some such language existed.
2. PIE has been reconstructed by comparing the modern languages. All languages have sound systems, of course, and FlatAssembler is asking whether the reconstruction of the PIE sound system is compatible with some of the sound changes required to derive the daughter languages from it. Specifically about the voiceless stop series [p t k] and whether there might also be an aspirated series [pʰ tʰ kʰ]. Since PIE is the results of about two centuries of tinkering of linguists, there are some competing theories, but most of the possibilities (especially within the range of the most popular proposals) have been explored. You can look up some proposals for PIE sound inventories, such as here: https://en.wikipedia.org/wiki/Proto-Indo-European_phonology#Stop_series
Specifically this section is relevant:

Former reconstructions involved series of four stops: voiceless unaspirated and aspirated, and voiced unaspirated and aspirated: *t, *tʰ, *d, *dʰ. The voiceless aspirated stops, however, came to be reinterpreted as sequences of stop and laryngeal and so the standard reconstruction now includes series of only three, with the traditional phonetic descriptions of voiceless, voiced and voiced aspirated.

So FlatAssembler's idea is not too crazy or anything-- in fact it was a standard assumption in earlier research on PIE. But an apparently better hypothesis was selected instead. Obviously this is some of the research to read about. It will not be 'light reading' in any sense. And it's going to be filled with technical details, and sometimes conflicting accounts. That's why I would be very surprised if this issue hasn't yet been dealt with. But I cannot at this time provide a direct link to a simple explanation on Wikipedia. It's a good question, but one I would assume has been solved.
One problem with Wikipedia, of course, is that while it gives a good overview of the ideas, it doesn't clearly cite all of the sources, including, for example, why the summary there indicates that there are reasons for rejecting FlatAssembler's alternative theory (as in the earlier proposals), but without actually explaining what those reasons are. The next step? Figuring that out.