Kurama, It’s true that LSJ included the construction under Final clauses (it’s jeidsath who’s misreading), and they certainly should not have. But ως in this construction is exceptionally hard to classify, and I guess they dumped it here, at the end of the section, faute de mieux. Probably they’d have done better to isolate the usage, or to link it with C. I and II. The more I look at the ως entry the more of a mess it appears. Βut make no mistake, for all its unsatisfactory features LSJ remains a truly excellent dictionary, put together by scholars who really knew Greek (better than anyone does today), and were pretty good amateur lexicographers into the bargain.
What I don’t understand is why you “still don’t understand why on Earth LSJ tell us that both ὡς μὲν ἐμοὶ δοκέειν and <ως επος ειπειν?> have the same syntactical structure.” They are both absolute infinitives (as I explained) preceded by ὡς. They even have the same qualifying function.
You “insist” that “there is something both of these constructions are expressing at a literal level of meaning.” Well, yes, but it’s hardly different from their actual meaning. The closest English can come to expressing the literal meaning of ως επος ειπειν is “so to speak” (I’m not saying that ως means “so”). What it means, as Hylander said, is “more or less” or “just about.” I attempted to account for the επος ειπειν combo by reference to its Homeric antecedents; the ὡς and the infinitive each fall into line with post-homeric usage. Perhaps you’re thinking of the constructions as if they were something like metaphors, where of course there is a literal level of meaning, however much faded, different from the metaphorical meaning. With these quasi-parenthetical qualifying phrases, however, what they say is pretty well what they mean, what they always meant.
If you are wondering how ως επος ειπειν came to mean what it does, I doubt that there is an demonstrable explanation if LSJ doesn’t address the question. But I suspect that a word indicating approximation simply dropped from the expression, perhaps ως σχεδον επος ειπειν.
Just to make sure that I understand the entry, I’d like to ask a question. I’ve noticed that there are two ways, within the entry for ὡς itself, of starting a list that is headed by an uppercase Roman numeral. One way is like in Ab.II. Here it goes like this:
Ab.II. with Adverbial clauses:
parenthetically, in qualifying clauses,
in elliptical phrases, so far as
etc.
Another way is as in B. II. There it goes like this:
B.II with Final Clauses, that, in order that;
2. ὡς is also used with past tenses of the indic. to express a purpose which has not been or cannot be
fulfilled,
3. (text). ὡς c. inf., to limit an assertion,
I.e., in one case the first gloss (to clarify what I mean by gloss, there are three glosses of ὡς in section B.II) gets a separate entry from the text that immediately follows the Roman numeral, in the other case it does not.
What does this mean? If the number system is to have any consistency, I would assume that when they list the 1. separately, it means that 2, 3, etc. are not considered subcases of 1, but rather all of them are considered subcases of the Roman numeral, with 1 being no more special than the others. And when they don’t list the 1, that it means that 2, 3, etc. are subcases of the gloss that is immediately given after the Roman numeral (the one that doesn’t start with a 1). In other words, since B.II doesn’t have a separate 1, I would assume that what it is telling us is that the gloss “that, in order that” also applies to the use of ὡς with infinitive, which is what B.II.3 defines. If this is correct, then indeed jeidsath misread. But if he did not, it seems that ‘Final Clauses’ would only apply to the first gloss (what would have been B.II.1) but not to B.II.3.
If it helps, I went to the library today and looked at the Brill dictionary (in English, they don’t have it in Italian here). There, all 3 uses of ὡς that LSJ place under B.II get unequivocally classified as final clauses. Interestingly, though, even if they class it as a final clause, they also clearly separate the use of ὡς with infinitive from the use that means “that, in order to”, unlike LSJ, where the format makes the entry a bit fuzzy. I also found something interesting in Bailly’s Dictionnaire grec-francais. It tells us that ὡς plus infinitive means “de telle sorte que”. That is, they treat it almost like ὡς as a relative pronoun indicating manner.
What I find so puzzling about this construction with ὡς is that if instead of the infinitive we had a finite form, then we could basically treat ὡς as a relative and translate ὡς ἔπος εἰπεῖν as ‘such as I say’. That would make it more comprehensible how exactly the ὡσ is working with the verb. But alas, we do not have a finite form…
I agree with you that ὡσ works to limit the assertion it governs, but what I am so frustrated about is that I haven’t found an explanation of how exactly it uses the infinitive to limit the assertion. I am working under the assumption that language is compositional, so there must be some one thing that ὡσ is doing to the infinitive of ὡς ἔπος εἰπεῖν and of ὡς ἐμοὶ δοκέειν. Yes, it is clear that both phrases have the same syntax. But what really is going on with the semantics of ὡς here? To say that it somehow compounds with the infinitives to limit the assertion it governs just doesn’t give me more than a very vague account of what is going on here. Quite simply, if it weren’t for the examples and the translations I have consulted, I would be left with no idea at all of what the phrases mean just from reading the dictionary entry.
By the way, all this is extremely interesting, but it is overloading my head. I’m just going to focus for now on finishing Mastronarde (only 5 more units to go) and come back to this issue when I’ve had a few months of reading, so I can see more examples of the construction.
Something to bear in mind is that ως is not governing the construction. ως can be put in front of all sorts of things to more or less subtle effect but without affecting the construction itself (see e.g. LSJ C I & II, as I mentioned), and in such circumstances its “meaning” can be hard to pin down. So here: absolute infinitives don’t really need the ως to be grammatical, as you can see from the examples in Smyth.
all this is extremely interesting, but it is overloading my head. I’m just going to focus for now on finishing Mastronarde (only 5 more units to go) and come back to this issue when I’ve had a few months of reading, so I can see more examples of the construction.
That seems much the best plan. All this is overloading my head too.