How often are dictionary etymologies wrong?

Question

How often are the etymologies in dictionaries incorect?

Sometimes when reading a dictionary I see a derivation of a word which contradicts my intuition. For example I read that "ball" comes from Proto-Indo-European *bʰel- "to inflate", cognate with Ancient Greek φαλλός • (phallós) "penis" when I had thought it would rather be related to Ancient Greek βᾰ́λλω • (bállō) "I throw".

Or that "zero" comes from Arabic صِفْر • (ṣifr) "zero", believable enough, which supposedly comes from Sanskrit शून्य • (śūnya) "zero", which supposedly comes from "Proto-Indo-European *ḱówH-, related to Latin cavus ('hollow')". In this case the meanings are all clearly related, but I don't understand the phonological relationship between "ṣifr" and "śūnya" and "*ḱówH-".

The problem for me is that when I get confused about things like this, I don't know where to look for more information; there are no citations. Presumably the primary sources are scattered across many documents. But who compiled them, and was the compilation ever questioned or debated? Where should I go for more certainty? And how much error, in general, should I expect to exist in dictionary accounts of word etymology?

Answer 1

I'm going to describe the situation in Modern Greek.

In Modern Greek, you will get good etymologies in the the contemporary dictionaries, Babiniotis' and Triantafyllidis', both of which date from the 90s. The current etymologist of the former, Moysiadis, has written a textbook on etymology; the original etymologist of the latter, Petrounias, has written extensively on the subject as well.

What they have not done is give citations. If you want citations, you go back to Andriotis' etymological dictionary done in the 1950s, which does give citations, and where you'll realise that there was controversy around a lot of the proposals; Andriotis gives proposals that have been rejected, as well as those that have prevailed. If you go to Kriaras' dictionary of Early Modern Greek (started in 1968, now up to sigma), you'll see more up-to-date etymologies for words that attested before 1669, again with details of proposals and counterproposals.

If the etymology is not in Kriaras or Andriotis, well, you're out of luck. Etymological work is being published to this day in Greece, usually in squibs in journals like Hellenika; and if you're not someone like Moysiadis, you're not going to be able to keep up to date with it readily.

The etymological proposals cited in Andriotis and Kriaras are scattered all over the place. Much of the proposals cited in Andriotis' earlier dictionary is the work of only a handful of scholars working between 1880 and 1910 (Hatzidakis, Filindas, Meyer-Lübke), and they argued with each other quite acrimoniously. Luckily, those scholars' collected works have been compiled into single sets of volumes. So you can usually track Andriotis' sources down, at least.

Now, Modern Greek etymology is a much smaller field than the etymology of English, let alone Indo-European; but you can already see that you have to hop between three or four generations of opinion, that the literature is scattered, and that the contemporary dictionaries are at best tertiary sources when it comes to etymology.

... So where do you look for more information about English etymologies? The OED tends to have detailed discussion, but it does not cite sources; you want specialist etymological dictionaries for that (which have Etymological in their title); Skeat and Onions for English, for example. You're asking about Indo-European; you'll want to look at the etymological dictionaries of Proto-Indo-European, and of the language families within Indo-European. It's not for the faint hearted, but it is feasible.

Answer 2

We can't really determine whether an etymology in a dictionary is "correct" or not since we don't know the ground truth to compare. But the editors of etymological dictionaries have taken a great job in preparing them. You can tell good etymological dictionaries by two features

1. Admitting sometimes that the etymology of a certain word is unclear or completely unknown
2. Presenting different theories of the etymology on some words with some evaluation by the dictionary author (e.g, telling which one is preferable or which one is considered outdated or probably wrong).

But your first example is indeed rock-solid, there are lots of correspondences between Greek φ and Germanic b, leading to the reconstruction of *bʰ. The Greek word for "to throw" is just a chance coincidence.

In your second example, the Arabic word is calqued from Sanskrit, i.e., it is a loan translation, extending the meaning of a root formerly only meaning "void". In a calque, there is no phonetic similarity preserved.