The IE root * weid- seems to have meant “to see” and, in its perfective stem * woid-, “to know”.

The “know”-semantics of this root are well attested in all the main branches of IE (English wot, Greek oîda, Armenian gitem, etc.), with two exceptions that I'm aware of:

  • Albanian, which was not attested in writing until fairly recently, and is known for its relative paucity of reconstructible IE vocabulary (as far as I know, it has no reflexes of * weid- in any meaning)

  • Italic, in which the * weid- root is well attested, but is more-or-less restricted to the meaning “see” (Latin vidēre “to see”, Umbrian virseto “seen”, etc.).

First of all: is it correct that the “know” meaning of * weid- is fully absent from attested Italic?

I.e., it doesn't even have any derivatives of this root (such as English wit, wise, etc.) in which the “know”/”knowledge” meaning is retained?

And if it is correct, are there any likely explanations for this gap in Italic?

(I realize that we may not be able to definitively explain this development, but I'm wondering if there are any known factors that seem especially likely to have caused it.)

  • 1
    What do you consider a "main branch" of IE?
    – Draconis
    Commented Jun 22, 2021 at 3:28
  • In this case, all branches that are attested beyond scattered fragments.
    – user8017
    Commented Jun 22, 2021 at 13:21
  • Even if there are other IE branches within the aforementioned scope (besides Albanian and Italic) that lack the "know" semantics of *woid-, the absence thereof in Italic is still remarkable, considering that Italic is a relatively early-attested branch that is not known for the extensive loss of IE vocabulary/morphology.
    – user8017
    Commented Jun 22, 2021 at 13:34
  • In that case, are there reflexes of *woyd- with this meaning in Anatolian, Baltic, Celtic, etc? I only know of some in Germanic, Indo-Iranian, Slavic, and Hellenic (and you mention Armenian). I understand not wanting to count Tocharian because of the scarcity of evidence there, but e.g. Celtic seems very "main" to me.
    – Draconis
    Commented Jun 22, 2021 at 16:37
  • Celtic does have it (Old Irish rofitir, Welsh gwyr "knows"), as do Baltic (Old Prussian waissei ("you (sg.) know"), and Tocharian (Toch. B ūwe "learned").
    – user8017
    Commented Jun 22, 2021 at 17:24

3 Answers 3


One theory that has occurred to me:

In Latin, and perhaps all Italic, the morphological category that * woid- originally represented became the simple preterite, and this change may have leveled out any independent, lexicalized meanings of the perfective stems. Thus, the perfect of videre (vīdī/vīdistī/etc.) simply means “saw / have seen” in the same way that cēpī means “I took / have taken”, ēdī means “I ate / have eaten”, etc.

By contrast, most of the other IE branches had (or developed) some type of protection against this leveling:

  • In Greek, the perfective was not the sole or primary option for expressing the past tense, and in addition to oîda / eidénai, there were many other lexicalized perfectives with a non-past meaning (pepoithénai, eoikénai, etc.).

  • In Germanic, * woid- became part of a robust group of verbs (including can, may, shall, etc.) in which the originally past/perfective stem had present-tense semantics, and on whose base a new past tense was formed (Icelandic vissi, German wusste “knew”, etc.).

  • In most of the IE branches apart from Greek/Germanic, the reflex of * woid- changed from a perfective stem to a normal present-tense stem (Armenian gites, Slovene veš, Old Prussian waissei “you (sg.) know”, etc.), which then existed as a fully separate verb alongside the other variants of the original root (Slovene videti “to see”, etc.), and would thus have been unaffected by the type of leveling that * woid- was vulnerable to in Italic.

What this doesn't quite explain, though, is why Italic would lack even isolated derivatives of * weid- that retain the “know” semantics (English wise, etc.).

  • Small correction: I double-checked, and it looks as though the Old Prussian word has only one "s" (waisei). There is also an alternate form with two "s"s but no final "i" (waisse).
    – user8017
    Commented Jun 23, 2021 at 2:46

You seem to have provided a beginning of solution in your own answer to your question.
From the point of view of Proto-Indo-European, the meaning *woid- "to know" is derived from the Perfect *woid- "to have seen", hence "to know". The original meaning of the root *weid- is "to see".
Latin vid- "to see" is just conservative, it's not a "gap", it's just that Latin did not develop the Perfect *woid- into a separate verb with the innovative meaning "to know", as Greek or Germanic did.

  • And Celtic, and Baltic, and Slavic, and Indic, and Armenian (at least).
    – user8017
    Commented Jun 22, 2021 at 13:24
  • 5
    If @user8017 is right that all other branches of Indo-European except Albanian and Italic have this correspondence, it sounds more like an innovation on the part of Italic, not the exact same innovation made in every other family independently
    – b a
    Commented Jun 22, 2021 at 15:50

First of all: is it correct that the “know” meaning of *weid- is fully absent from attested Italic?

As far as I know, yes. Its Latin cognate, vīdī, means "I have seen" without the lexicalized metaphor of Greek οὶδα, and I've never seen it used differently in other Italic languages. Latin does have a similar perfective usage—the perfect nōvī < noscō < *ǵneh₃-, literally "I have learned", is commonly used to mean "I know"—but to the best of my knowledge vīdī isn't used for that meaning.

And if it is correct, are there any likely explanations for this gap in Italic?

Accident, most likely.

Consider the widespread Indo-European terms *ǵher- "desire" and *welh₁- "want". Both have cognates across many branches of Indo-European, including English "yearn" and "will".

We know they both survived into Italic. But in Latin, vōlō < *welh₁- won out, and in Oscan, herest < *ǵher- won out, with the other vanishing completely. (Latin does retain a derivative of *ǵher-, but its semantics are completely separate from "desiring".)

Why did this happen in Latin and Oscan and not in other branches of IE? Almost certainly just a historical accident. The two words' semantics had drifted together until they became almost identical, then one became more popular, and the other died out. The same thing very well could have happened in English (or Germanic or any other stage of an IE language); it just so happens that it didn't.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.