Loanwords from Germanic and, more recently, Russian have been feeding *f into Finnic for a good while. Today /f/ has been established as a loanword phoneme in most Finnic varieties (including, I think, all of the literary standards), but for most of the last 2000 years, the consonant has been adapted into native Finnic phonology in various shapes.
Five substitutions are usually recognized:
- *f → /p/
Mostly in oldest loans from Proto-Germanic or Proto-Scandinavian. The oldest examples could feasibly even precede Grimm’s Law, and therefore actually involve *pʰ → *p (the likes of *pëlto ‘field’). Others can be dated as slightly later, e.g. *pasto ~ *paasto ‘fast’ ← Gmc. *fastōn-, showing the probably relatively late *ā > *ō. A few examples are found even in much more recent loanwords such as Fi. porstua ‘porch’ ← Sw. förstuga or förstuva; Fi. upseeri ‘officer’ (perhaps since expected **vo and **hs are or were not phonotactically possible). - /f/ → ∅
In initial consonant clusters, e.g. Fi. läski ‘(pork) fat’, riski ‘strong’ ← Sw. fläsk, frisk. - /f/ → /v/
This is found initially (Fi. vaari ‘grandfather, old man’ ← Sw. far ‘father’) and after a consonant (Fi. konvehti ‘confectionary’). I suspect the switch from the first substitution pattern to this marks the onset of *w > [ʋ]. This may have been completed only after Proto-Finnic, since several dialects of Finnish have been recorded even in the 20th century with [w] adjacent to rounded vowels: kuva [kuwa] ‘picture’, vuosi [wuosi] ‘year’, vyö [wyø] ‘belt’ etc. Dialectal variants such as kasva- ~ kasua- ‘to grow’, kivi ~ [kiw] ~ [kiu] ‘stone’ could also speak in favor of *kaswa-, *kiwi and not **kasva-, **kivi as the starting points. Likewise the Estonian metathesis *Vuh > /Vhv/, more easily rewritten as *wh > *hw. - /f/ → /h/
Found initially preceding a labial vowel (Fi. huotra < *hootra ‘scabbard’ ← Gmc *fōdra-) and word-internally preceding a consonant (Fi. luhti ‘loft’, sahrami ‘saffron’, uhri ‘sacrifice, offer’). - /f/ → /hv/
Found between vowels, e.g. Es. Fi. sohva ‘sofa’, Es. kohv ~ Fi. kahvi ‘coffee’ (contrast though Livonian and dialectal Fi. kaffe, Karelian koffi ~ koofi ~ koufi etc.)
Altogether we have, in the newer layers, /h/-substitutions for preserving voicelessness, /v/-substitutions for preserving labiality and continuancy, and /hv/ for covering both.
There’s however also a sixth that I have usually not seen mentioned: substitution as /uh/ (~ /yh/), unpacking the consonant in the opposite order from the kahvi type. At least two other examples appear to be known. One is the Russian loan ‘kaftan’: Fi. Krl. Izh. Vot. kauhtana, Ludian–Veps kauhtan. (Estonian has rather kahvtan.) The other is Fi. Krl. Izh. tiuhta ‘reed; awn’ ← Gmc. *stifta- (> Sw. stift), with the sound development remarked on in LÄGLOS in the word’s entry (in the 3rd volume), but not in the foreword overview of sound substitutions (in the 1st volume). I think a few additional examples could be adducible too:
- Fi. Izh. vyyhti ‘weft’ (← Gmc. *wifti-), whose vowel length is usually attributed to sporadic lengthening before coda /h/ and labialization to irregular influence from /v/. But Karelian shows viyhti; I think this is likely to be more original. In Finnish and Ingrian, evidently *iü > yy. Finnish and Karelian dialects plus Ludian show also viihti ~ viihť, which could be instead the real example of secondary lengthening (but also maybe a parallel development of *iü).
- *riuhto-, *riuhtat- ‘to rip, tug’ (Fi. Krl. Izh. Lu.), with a variant reuhto- in Finnish. Maybe a derivative from Germanic *rīfan- (> Sw. riva) or *reufan (> Eng. reave) ‘to tear’? For *t-suffixed forms, I only know of the noun rift though.
- Fi. töyhtö ‘tuft, crest’, Krl. töyhäkkä ‘fluffy’, [1] töyhistyö ‘to puff, bristle up’ (probably ← Fi, per öy). Has immediate resemblance with the English, though Scandinavian only seems to have an s-affixed variant tofs (→ Fi. tupsu ‘tuft, tassel’). Looking at Low German could maybe turn up a suitable loan original?
It can be noted that all of these examples occur in the context *-Vft-. This is probably not an accident: **-fk- does not occur in Germanic (possible enough in Russian though, and giving e.g. colloq. Fi. lafka ‘store’ ← Ru. лaвка), while **-VUhR- does not seem to occur in Finnic. A few rare examples of -VihR- can be found, but usually with simplified variants alongside: in Finnish e.g. (standard form first) kaisla ~ kaihla ~ kahila ‘reed’, laina ~ laihna ‘loan’, raihnas ~ raina ‘decrepit, geriatric’, saiho ~ saihvo ‘corral, pen’.
[1] Krl. töyhäkkä has also a 2nd sense ‘haughty’, which together with töyhteä ‘to fuss about’ are probably better compared with Fi. touhuta ‘id.’, touhu ‘fuss, bustle’ (with typical affective/deminutive fronting).
Freelance reconstruction 
← Gmc. *stifta- (> Sw. stift)
*pʰ → *p / *ā > *ō
When do you use an arrow (←), as opposed to a ?
Arrows for loaning and derivation (transfer relationships), greater-than / less-than for semantic and sound changes (descent relationships).
Interesting, it had never occurred to me that -hv- in Finnish kahvi, Estonian kohvi were a substitution for -f-. In fact, the origin of all European coffee words, Ottomanian Turkish kahve, has the exact same cluster.
While a direct loan is unlikely and I’m not sure about exact transmission paths, wouldn’t re-approximation to the original source be one motivation? Moreover, (Central/Southern) Eastern European languages typically have a -v- sound in their coffees – this includes Polish and Lithuanian (but not Latvian or Russian). So while this alone does not explain the cluster, it may have nudged the development towards “something with v” as opposed to other options.
(Btw., for me the most repelling aspect of linguistics is talking about boring or nasty word referents, like unheard-of species of fish or diseases. I just had to pick this opportunity when it’s different.)
It’s bound to happen, I think, that sometimes we have two foreign sound substitutions “undoing” each other, seeing how /hv/ is entirely general for this purpose.
For some other culinary examples, cf. vohveli ‘waffle’, pilahvi ‘pilaf’, pihvi ‘steak’ (← Sw. biff ← Eng. beefsteak), karahvi ‘carafe’, kahveli ‘fork’ (← Sw. and Low German gaffel); colloquial/dialectal puhvetti ‘buffet’.
There is an interesting “orthographic” parallel in the history of the letter F itself, which involves /w/ instead of /v/. When the Etruscan adopted the Western Greek alphabet, they needed compensated the lack of /f/ in Greek by the digraph FH/HF (both are found in old inscriptions), where F stands for digamma, used for /w/ (In Greek, Etruscan and Latin). This adaptation, transferred to Latin, corresponds to a synchronic orthographic rule, /f/</h/+/w/. In the 6th century, both Etruscan and Latin got rid of the digraph to restore the alphabetic principle, the former with a new 8-shaped letter, the latter by using the vowel V /u/ to also note /w/, so that digamma F without H can note /f/ and not /w/ anymore.
I can always be convinced with enough food …
While cancelling out is not hard to find in descent sound developments, I am not aware of a lot of examples in loan substitutions. I may not have spent as much attention on those; but also loan substitutions tend to be more trivial than sound changes (e.g. identity X -> X), or they are non-reversible (e.g. C1 C2 -> C2).
One sort-of kind-of non-trivial cancellation that comes to my mind is k -> q -> k in
Greek k´alamos -> Arab qalam -> Swahili kalamu, Turkish kalem – “pen, stylus”.
(While q is not available in the latter languages, I am not sure what the motivation for q in the first step is. Is it about Greek a? Is it regular at all?)
Some other cycles I would rather bet to exist than not would include a “voicedness toggle” like d -> t -> d or vice versa (in short d t), the “patch in another lip” change v w, and nasal replacement [nasal vowel] [oral vowel] + N. But I have no idea whether or where to find those. Perhaps it is easier to find full cancelling cycles if we allow a descent development at one of the stages.
If this bleak picture is about right, the change hv -> f -> hv is already a gem – almost too curious to be true!
I am not sure what the motivation for q in the first step is. Is it about Greek a?
I would guess another relevant factor to be how, to my knowledge, Semitic /k/ is usually aspirated, while /q/ is not.
Ah, that makes sense.
Still can’t get my mind off the cancellation thing.
Meanwhile, I’ve come across another class of only slightly boring examples:
Appearance and subsequent disaappearance of prothetic or epenthetic vowels.
E.g.:
Old High German bank -> Italian banco, banca -> Middle French banque (with final vowel) -> Middle English banc “bank”.
Gothic *stampjan / *stap^on -> Spanish estampar “stamp”, Am. Spanish estampida “crash” -> Am. English stampede
(The “bank” example also involves the oral -> nasal -> oral vowel cycle, at a stage of French when N after nasal vowels were still pronounced.)
More to the point of the article:
Does that mean that you assume *pʰ as an intermediate stage in Pre-Germanic between PIE *p and PGmc *f?
If so, what do you expect for PIE *bʰ at the same period?
Yes. If this was phonemic at the time is not clear though: the *B series could have still had some other distinctive feature to it at this stage. I’d have to look over “Kümmel’s Laws” again if that establishes anything on the matter.
PIE *bʰ I expect to have always been plain voiced [b]. Aspiration, with or without devoicing, is IMO an innovation within the Greek-II areal (quite possibly, but not absolutely necessarily, further including Italic and Armenian).
Assuming voiced aspirates in Pre-Germanic allows collapsing two of Grimm’s three laws as “all aspirates become fricatives” – except in a few environments where this was blocked for the voiced ones, and the aspiration was just lost later (or, next to nasals, perhaps earlier).
The same holds for Italic, except there the voiceless aspirates came from word-initial devoicing of the voiced ones, not from the PIE voiceless series.
Assuming voiced aspirates in Pre-Balto-Slavic explains Winter’s law.
It will look slightly neater on the paper, but I don’t think the “pairing” of aspirate fricativization is much progress typologically. Voiced fricativization seems like simple lenition, voiceless fricativization in cases like these (where initial consonants are affected) I believe usually runs through affrication. I also don’t think general voiced fricatives are a good assumption for any stage of PG; at most voiced fricatives would seem to have been medial allophones in a begadkefat fashion. If they had been general, I’d expect e.g. to see some examples of *β- → Finnic/Samic *v-, or *ð- → Samic *ð-, but no such thing is attested.
For Proto-Italic I believe there are a number of signs that the voiced aspirate series went through general devoicing and spirantization and only secondary medial re-voicing, but I’d have to collate some notes to give a good argument. One point at least is the development of final *ns, *nts > /f/ in Sabellic, likely running as something like *nts > *ntʰ > *(n)θ > f. Note that preceding *n and final position are required, hence I would prefer to assume *s to have been nondistinctively [+spread glottis] (as evidenced also in Greek, Armenian and Indo-Iranian) and cheshirifying this feature when lost finally, over a more typical dentalization along the lines of *ts > *t̪s̪ > *s̪ > *θ.
I do think word-initial position was (in Germanic) one of the blocking environments where at least *bʰ and *dʰ did not fricativize, and simply lost their aspiration at a later date (conveniently after *d > *t, most likely).
On the absence of general devoicing in Italic, there’s Stuart-Smith’s book “Phonetics and Philology” – “Sound Change in Italic”, which is downloadable as a pdf somewhere; it’s quite compelling.
I found it, it’s here, and I have to put enough filler text around the link to avoid the spam filter.
Looks like I have Stuart-Smith already in my files. The principal result there seems to be to cement that the common Italic starting point was voiceless fricatives initially, voiced medially, which I do not contest.
Some problems remain:
• I am not convinced that Umbrian ‹-rs-› < *-rs- but ‹-rf-› < *-rVs- would be counterevidence for a lack of general voicing also adjacent to liquids. If the assumption is that *rz > *rð > [rv], then should not the same dentalization (feeding into the common Italic interdental > labial shift) have also applied to voiceless *rs? POA shifts are generally orthogonal to phonation. I’d suggest instead that in Umbrian, primary *rs was early on retracted to something like *rs̱ [rẕ], and this is what protected it from dentalization — not voicelessness.
• The pleading for a transient intermediate stage with voiced fricatives as allophones of voiceless aspirates seems unsubstantial and I’d expect a system like this to be rapidly rephonologized.
Still, on the other hand, maybe her “Early Split” theory can be pushed a bit further back still. If the *B series was something else such as implosive or ejective, and the *Bʰ series were the real plain voiced stops, then no [Bʱ] stage is required medially at all! And while I’d expect the resulting /v ð ɣ/ to end up phonologically detached from initial /pʰ tʰ kʰ/, nothing really prevents them from drifting back into allophones of each other again, once the latter had also fricativized.
The problem you identify is real, but shouldn’t we expect /s/ to have been retracted all along because there was no /ʃ/ or suchlike to contrast with?
I completely agree. However, this is very easily repaired by changing the timing on the assumed shifts a little. Stuart-Smith seemed to assume that “getting rid of the voiced aspirates” was one step, leading to voiceless aspirates initially and voiced fricatives medially. Even regardless of the outcome, that – getting rid of a class of sounds all at once, but in drastically different ways in different conditions – would be a strange sound shift absent massive substrate influence for which I don’t know any evidence. Instead, I think that the devoicing came first, so that the initial voiceless aspirates were the same phoneme as medial voiced aspirates, and then “all aspirates become fricatives” (through short-lived affricates) happened as more or less one step.
This is then easily extended to Germanic, where I don’t think all voiced aspirates were gotten rid of in one step either.
But Stuart-Smith spends a lot of time on arguing that the *Bʱ series really did consist of voiced aspirates in all positions in, well, crown-group PIE.
I do agree, for the usual reasons, that the *B series was something else than plain voiced plosives at some stage earlier than that, possibly as late as Proto-Indo-Tocharian, and that the *Bʱ series represents the original voiced plosives. But for crown-group PIE it does seem to me that we have to assume the pre-glottalist tradition has it right.
The situation reminds me of the Rgyalrong languages. They have a plosive inventory with voiceless aspirated, plain voiceless, plain voiced and voiced prenasalized members, which looks beautifully symmetric – but the plain voiced plosives are markedly rare, quite unlike the prenasalized ones; except for /d/, they’re even more or less limited to loanwords and ideophones if I remember everything right. It is thought that the prenasalized series was plain voiced in earlier times and got out of the way as the new plain voiced series formed (no idea if that’s a pull or a push shift).
Either *s retracted further in *rs, or *s fronted in neutral positions such as intervocalic, makes no difference. (I know I’ve seen a paper arguing for retracted *s in PIE, as seen, besides RUKI, also in e.g. Albanian *s > gj, sh; or in the Germanic → Finnic loans with *s → *š > *h.)
I actually think that not reconstructing voiced aspirates is what glottalic theory is most likely to be right on. Much of the supposed non-direct evidence for aspiration strikes me as simply circular, along the lines of “Germanic shows fricatives, therefore fricatives are likely to come from voiced aspirates, therefore fricatives in Germanic suggests voiced aspirates” where I think every inference is wrong: voiced aspirates are not particularly likely to fricativize, and therefore should not be assumed to have ever existed in pre-Germanic (likewise for deaspiration in pre-Celtic, pre-Albanian, or pre-Balto-Slavic). Stuart-Smith only attempts to defuse this by pointing to some conditional or isolated parallels for these developments from Indic. But this is nothing more than cherrypicking: once we are discussing Indic varieties on the level of individual Hindi dialects, there would be hundreds of others to consider as well, and the vast majority of them show neither general fricativization nor general deaspiration. And, as far as I know, none has anything resembling Grimm’s Law, Winter’s Law etc. — e.g. Indic varieties that medially lenite voiced aspirates always also lenite plain voiced stops.
If that’s true, it has massive implications for the Italic situation. (Unless the pre-Italic plain voiced stops devoiced and then revoiced somehow, which I’m not prepared to claim, even though it miiiiight help with Lachmann’s law.)
Speaking of Lachmann’s law, here is the latest word on it and several other issues from a glottalic perspective. I must say I’m impressed. The paper posits Pacific-Northwest-style preglottalized nasals as the origin of the mediae.