18 Oct 2008

The so-called imitative status of PIE *pneu- "to sneeze"

I don't have a lot of time to talk, folks, so I'll be brief about something bugging me lately. Recently, Bradshaw of the Future wrote in the article Sneeze and pneumatic about the origin of "sneeze" in PIE *pneu- calling it an imitative root. This is in fact a common description of this Indo-European root. However the question I'm posing here is: How do we really know that this is an imititative root?

The cold truth is that we don't. We only assume this to be true because of the semantic nature of the verb. So we should be cautious to distinguish solid facts from these sorts of unverified or unverifiable assumptions. I mean, I don't know about any of you, but personally my sneezes never sound anything close to "pneu".

I've been exploring an alternative origin of this word, not from an echoic origin, but rather as a possible Semitic loan. I've mentioned before in Pre-IE Syncope and possibly expanding the Metathesis rule that there may be some loans from Proto-Semitic that exhibit word-initial metathesis of consonants in PIE after experiencing the event of Syncope. Metathesis is one common tactic for renormalizing syllable structures after awkward clustering is caused by the disappearance of unstressed vowels. Such metathesis is guided by a universal rule in world languages known as sonorancy hierarchy. Certain clusters are universally avoided such as *rp- or *bft- for example. So if it's generally accepted that Pre-IE must have undergone Syncope, then it's naturally implied that awkward clusters like this should have occasionally arose and that there must have been a mechanism in place to restructure these roots.

So in that light, here's yet another hypothetical Semitic loan path to discuss amongst ourselves:

Proto-Semitic *napāḥu "to blow" → Mid IE *napéwa-

This would then become late MIE *nᵊpéwᵊ- (via Reduction) and then *pneu- by early Late IE via phonotactically motivated Metathesis in order to avoid the less desirable outcome of **npeu-.

Enjoy that thought. I know I do.


  1. Definitely like this explanation. Arabic has naffa/yafifu 'to blow one's nose; to snuff' A reduplicated second radical root. I wonder where the ḥ ran off to.

  2. PhoeniX: "I wonder where the ḥ ran off to."

    That I can explain. I reconstruct uvulars only for the Late IE period. I think that the uvular *h₂ was once a velar /x/ in Mid IE.

    So all in all, I think that Mid IE certainly did not have pharyngeals and without uvulars, there would be little to compare to a Proto-Semitic pharyngeal *ḥ in their sound inventory. So it may have been simpler for an MIE speaker to simply omit that throaty sound.

    If we surmise that the Semitic infinitive of this root (*napāḥu) was pronounced as [nə'pʰæ:ħʊ] or even [nə'pʰæ:ħʷʊ], then the pharyngeal neighbouring *u would be a plausible source of [w] in MIE and subsequently in PIE.

  3. Maybe to add further, even without concommitant labialization, I don't think it's impossible that a pharyngeal to MIE ears would appear to have a [+back] acoustic quality, the same quality as their *w which may be one of the closest approximations they could make. Semitic *ḥ- appears to be approximated also as MIE *x in word-initial position (c.f. *ḥāniṭu vs. PIE *h₂endʰ-). So the way in which this sound surfaces in PIE may depend on syllabic position. It would be nice to find more examples to see such a pattern.

  4. Haha, it appears that you misunderstood me. I was referring to the missing *ḥ in Arabic :) Nothing wrong with a root n-f-ḥ, and still it ended up being n-f-f.

  5. PhoeniX: "Haha, it appears that you misunderstood me. I was referring to the missing *ḥ in Arabic :)"

    Ah, damn. You should be more specific. The topic here afterall wasn't Arabic. Oh well. My mind is obsessed by how different languages map sounds from foreign languages. So when you were mentioning "disappearing pharyngeals" I was thinking of other things like PIE *septm̥ from PSem *sabʕatum.

  6. By the way, note Arabic nafaẖa "to breathe" (see link).

  7. I wondered about that as well. There's nothing imitative sounding about *pneu- to me. *baba- sure, but not *pneu-.