These are the relevant sounds ("phones") - you may check on the following pages for which languages possess them:
It is noteworthy that /x/ is the most common across the Slavic languages, and that the phoneme realised as /x/ in modern Slavic languages can be traced all the way back to Common Slavic, of the first millennium CE, and that is also reconstructed as /x/. Most Slavic languages just have /x/, and /ɣ~ɦ~h/ does not exist as a separate phoneme - Russian, Polish, Slovene, Serbo-Croatian-Bosnian-Montenegrin, Bulgarian and Macedonian, as well as Kashubian.
On the other hand, the /ɣ/ of Belarusian, /ɦ/ of Ukrainian, Czech, Slovak, Rusyn and Upper Sorbian are a development of Proto-Slavic /g/; compare Russian голова
with Ukrainian голова
, and Polish głowa
with Czech hlava
. This link of voiced velar or glottal fricatives with the historic voiced velar plosive /g/ is evident in the Cyrillic orthography in Ukrainian and Belarusian.
Hence we see that there are languages that distinguish /x/ from /ɦ~ɣ/. In Belarusian, this "ход-год distinction" is mainly a voicing distinction. We see that this was the case diachronically with Ukrainian; but the additional shift in the place of articulation (velar /x/ vs glottal /ɦ/) may have helped reinforce the phonemic difference. Note Ukrainian reimported the /g/ phoneme, and Standard Ukrainian orthography maintains this with its own letter: Ґ. Belarusian did this to a much lower extent.
Czech also has the "chod-hod distinction" between /x/ and /ɦ/, with the /ɦ/ coming from Proto-Slavic /g/. Like in Ukrainian, /g/ in Modern Czech goes back to loans from Latin or from other languages, such as graf
and regulace
(regulation) - it is a reimported phoneme.
The case of Silesian is borderline, as with many aspects of the Silesian language varieties - the opposition of /x/ and /ɦ~h/ is found in some varieties, e.g. Cieszyn (Teschen), but not in others e.g. Sulkovian, where /x/ contrasts strongly with /g/ like in Russian and Polish. This is highly dependent on how much the /g/ > /ɦ/ change has happened, which is usually attributed to varying amounts of Czech vs Polish influence.
If we were to have a distinction between voiceless /h/ and voiceless /x/, that would strictly be a place distinction. Polish is supposed to have had one previously, more common in Eastern dialects of Polish, as orthographically there is a distinction between the spellings < h > and < ch > - much to the chagrin of many Polish schoolchildren to this day! However, the spelling < h > was reserved for learned Latinate words like herbata
, or other loanwords like handel
(from Low German Handel
) and hałas
(from Ukrainian галас
), and the interjection hej
. It is possible that this was actually a voiced /ɦ/ anyway for those who made the distinction; the majority of Polish speakers never made the distinction and have had /x/ for all of these.
The one Slavic language I can see that has a distinction between /x/ and /h/ is Lower Sorbian. The /h/ phoneme does not diachronically derive from Common Slavic /g/ - that phoneme is kept as /g/ (hence głowa
in Lower Sorbian, hłowa
in Upper Sorbian). Rather it comes from German loanwords, e.g. hengist
from hengest
, in Modern Standard German Hengst
stallion. Orthographically, there is an interesting way that Lower Sorbian adapted the German initial glottal stop in front of words that begin with vowels as < h > in the spelling, but it is unclear whether this was ever pronounced as /h/.
However, synchronic variability between velar and glottal realisations of this phoneme is super common across the world. Romanian is no exception - as a 2016 study stated:
The plain dorsal fricative was realized as velar [x] 86.7% of the time and as glottal [h] 13.3%, based on the analysis of 368 items. There was a slight tendency for males to produce more glottal realizations (males 15.9%, females 11.8%), but this difference did not reach statistical significance. As for the palatalized forms (365 items), all but one (99.7%) were realized as the palatal fricative [ç]; the exception was realized as a glottal by a female subject.
This variation is also found in e.g. Ukrainian, with a set of allophones for /x/ and /ɦ/, but the details are slightly different.
What can we conclude about the borrowing of /h/ into Romanian then? With the long timespan that this could have occurred, we have many candidates:
- Proto-Slavic
- Common Slavic
- Old Church Slavonic
- any of the mediaeval Slavic languages
But also the following non-Slavic languages that use /h/:
- Albanian
- Turkish
- Hungarian
Which was it?
By the time of the scrisoarea lui Neacșu in the 1520s, where Old Romanian is exemplified, the speakers of Slavic have been writing for half a millennium, whereas the first documentation of Albanian that has survived predates it by less than a century. Although we see no evidence of native Romanian /h/ in that letter (although it is found only in the names), Slavic loanwords like duh
spirit and hrean
horseradish point to consistent Slavic influence. Without documentary evidence though, it is difficult to pin down where and how - whether this is from Old Church Slavonic or from the Common Slavic of earlier is very hard to say. From the range of loanwords, especially the more common domestic ones, we can be fairly sure that the Slavic influence was what brought the /x~h/ phoneme to Romanian.
The fact that this was not a preservation from Latin can be seen from the many Romance roots that do not retain their intial h-, e.g. ierba
grass, avea
to have. Many of the h- words in Romanian were reimported in the Re-latinisation of the 18th and 19th centuries, many of them orthographically borrowed, and especially through French (which did not and still does not have them pronounced as /h/, but keeps them silent). Hence the doublet hiberna
hibernate vs ierna
pass the winter.
On the other hand, Romanian did retain a few examples of /h/ coming from Latin /f/, e.g. holba
, something that Castilian Spanish also did. Parts of southwestern Spain and Latin America retain this in e.g. hierro
; compare this with Aromanian her(u)
.