§ 1. Proto-Indo-European language.
The Proto-Indo-European phonetics was not stable at all: ablauts (vowel interchanges), assimilations, many different consonant processes at the end of the word. Many of these features were developed in branch languages, some of which had very complicated phonetics, e.g. Latin. But still we do not know too much about the Proto-phonetics and its development. These things can only be seen in comparison of different branches and groups with their own phonetic peculiarities. But the question we are interested now is the problem of initial mutations in Proto-Indo-European.
Some interchanges of consonant in in initial position are found in the Proto-Indo-European structure. The most common example of them is the case with the initial k- in some words. It sometimes disappears but sometimes remains as it was, and still linguists cannot say with sure how the word for "a bone" sounded in the Proto-Indo-European language: kost- like in Balto-Slavic, or ost- like in Celto-Italic. This is really a case for Word-A-Week: it is hast- in Hittite, odb in Old Irish, os, ossis in Latin, though it had also costa (a rib), kost' in Russian etc. It seems probable that even in the Proto-language this mutation took place, so different dialects of the common tongue took different variant of the same stem.
This initial k-, as well as other fricative sounds, is also seen disappearing in some other words. The same process was common for the initial s-. It can disappear rarely, but still, we think, regularly. There are two versions of this phenomenon's origin. Some believe it occurred due to the prefix s- with unclear meaning; some state this is a "sandhi", the influence of the previous word ending in -s. The last sound causes a change of the initial one in the next word, and that is called "sandhi" from Sanskrit, where it is very common.
But the above mentioned cases are the only ones in Proto-Indo-European,
where we can speak for sure about the initial mutations. There existed
also internal word mutations, e.g. tw - t, sw - s, but they
are not initial and will not be considered in the present article.
§ 2. Anatolian languages.
Hittite, Luwian and Palaic languages are interesting for they drifted apart from the Proto-Indo-European community shortly before the Proto-language took its latest forms. Anatolian languages bear features that are too archaic even for Proto-Indo-European. It can be noticed in flexion system, in noun genders and, certainly, in phonetics.
But even here, in Hittite and Luwian, initial mutations are visible. Hittite shows two of them that seem to occur regularly in the language: the first one is k - h that appeared possibly due to the non-Indo-European influence on Anatolians. We should bear in mind that Anatolian languages suffered really great changes while contacting with Hatti, Aramaeans and other aboriginal nations of Asia. The dictionary of Hittite has got only about 22% of purely Indo-European words, other ones are of Hatti and other origins. The religious system and terms is almost completely borrowed from aborigins. Many changes took place in phonetics as well. That is, we believe, the case with k - h mutation in initial position. Hittite keššar (a hand) was also pronounced as hiššari. That initial h disappeared completely in Luwian.
The other mutation is even more interesting, for it has strange and mysterious analogues in Baltic, Slavic and Italic languages. We speak about the change n - l - t and we offer you a word lamen (to name) derived from Indo-European *nomen- (a name). A derivative from that Hittite verb was a-timan (a name), which makes us think about the unstability of the initial sound in that verb in Anatolian languages. It seems that in some cases - exactly still unknown - Indo-European *n- was so unstable and weak that easily could turn into l- (obviously, a palatal sound) and t- (palatal, too). Another example, this time from Luwian, will be tapaša (sky) which sounded nepiš in Hittite and *nebho- in Proto-Indo-European. The same law n - t acts here.
And the third mutation that was also irregular in the language: d
- t - š which we will analyze in Hittite, Luwian and Palaic
with only one example. The Proto-Indo-European *deiwos (god
of daylight) was the supreme deity of Indo-European pagan pantheon.
He was worshipped among Anatolians as well, and was called tiwat
(sun god) in Luwian, and tijaz, tiuna (a god) in Palaic.
That happens according to common Anatolian phonetic laws. But in Hittite,
nevertheless, this god's name was like šiu, šiun, and the
word for "a day" cognate to it was šiwat. We don't know for
sure if this initial consonant really sounded as English [sh], that is
why the hypothesis exists that that was a sibilant close to palatal [d],
seen in Ancient Umbrian and Modern Czech
r'. Umbrian
also replaced d with this kind of sound, and what were the
reasons of that, we don't know.
§ 3. Slavic and Baltic
languages.
The n - d problem exists in both Baltic and Slavic branches,
long ago being united in Balto-Slavic community. We shall note that neither
Baltic, nor Slavic languages have now or had ever had initial mutations
depending on the previous words or sounds. No regular signs of it are noticed,
and nevertheless some examples exist, being quite irregular and too scarce
in those languages.
Lithuanian has several words showing this kind of mutation. The Lithuanian namas (a house) is obviously derived from Proto-Indo-European *dom- (a house), also seen in Russian dom, Latin domus, Greek domos, Germanic *tum-. This Lithuanian word has a n-, and the reasons for it are not clear. Some linguists guess this word derives from another stem *nom- (region, district), also in Greek nomos, but most scientific works deny this point of view.
The next example is the same as in Luwian mentioned above. The Proto-Indo-European word for "sky", *nebho-, turned in Lithuanian into debesis (a cloud), though it is nebo, nebesa in Russian, nabhas (fog) in Sanskrit, etc. As we can see here, the mutation is vise versa: n - d. It is evident that these initial n and d were too weak to keep their place and had to change or interchange, replaced by each other. What was the cause of this mutation, unknown.
And the third case we would like to point here is concerned both Slavic and Baltic languages: the word devyni (nine) in Lithuanian and *devet in Common Slavic originated from Proto-Indo-European *newn (last n is syllable). The only explanation of this particular phenomenon is the influence of the next numeral "ten" which sounds dešimt or in Slavic *deset. But still, if this really happened, it proves the weakness of the initial *n- in this word, so it was easily converted into d-.
Now let's remember Hittite n - l mutation, because the
similar one can be met in Baltic tongues. Indo-European *nisdo-
(a nest) in all branches has a n- in the beginning of it.
All but Baltic which show lizdas ("a nest" in Lithuanian).
Again the similarity which makes us think there was something in Proto-Indo-European
that in certain cases mutated the initial consonants.
§ 4. Italic languages.
So we see that every irregular phonetic mutation mentioned above is
rotating around d, n, t, l sounds. This is natural, because
exactly these four sounds are dentals and that's why have much in common
in articulation while being pronounced. We summed up cases met in Hittite,
Luwian, Palaic, Lithuanian, Common Slavic. Now the turn of Latin.
In fact, we have two examples to fit the matter being discussed. The first one is concerned the word lacrima (a tear) that obviously comes from Italic *dacerma. So d > l, and this feature cannot at all be called regular, as all other known words in l- with Indo-European origin have their ancestor stem beginning with the same letter. Let's remember lacus (a lake), liber (free, free people), lupus (a wolf), longus (long). They all have cognates in other Indo-European languages with this very l- at the beginning.
Another example is the personal name Ulyssus derived from
Greek Odysseos. It is not exactly initial, but the nature
of this lenition may be the same. But this case has an acceptable explanation.
The word could have come into Latin through Etruscan who had had cultural
ties with Greeks and their mythology long before Romans established such
ties themselves. And Etruscan might have borrowed the legend and the name
even from Mycenaean Greek, with changing d into l
or some other native Etruscan sound according to its own phonetic laws.
So here the source can be non-Indo-European influence. We know that Italic
languages suffered significant phonetic changes due to pre-Indo-European
population of Italy, especially this influenced Umbrian, Picene and Volscian.
It looks like Latin also had not only lexical influence of Etruscan, but
phonetic as well.
§ 5. Celtic initial mutations.
And now about the most famous initial consonant mutations ever occurred in Indo-European languages, regular and stable. Insular Celtic languages have really plenty of them, and these mutations became one of their characteristic features. Let us have some typical examples from all those Insular Celtic tongues:
Irish (2
consonant mutations):
doras - a
door
mo dhoras
- my door (pronounced [goras], with a sort of
Greek "gamma")
bhur ndoras
- your door (pronounced [noras])
Welsh (3
consonant mutations):
Cymru - Wales
i Gymru -
to Wales
yng Nghymru
- in Wales
a Chymru -
and Wales (pronounced [xymru])
Cornish
(5 consonant mutations):
benen - a
woman
an venen -
the woman
Breton
(2 consonant mutations):
kador
ar gador
ar c'hadorioú
The same case as in Lithuanian described above: d - n is regular in most Insular Celtic tongues (Breton is also called Insular though it is spoken on the continent) and has a name of "nasalisation" in Welsh or "eclipsis" in Irish. It is caused by the final sound in the previous word. In Irish and Scottish Gaelic, e.g., it must be final -n or its rudiments which may cause eclipsis. Lenition, provection, lanition, prefixed consonants, other Insular mutations are also caused by previous sounds.
And here we are approaching the most interesting question of mutations: when, why and where from the Celtic mutations appear? The fact is that Gaulish, Celtiberian and Lepontic, three ancient Continental Celtic languages, did not have a sign of those initial changes and used the same phonetic signs as Proto-Indo-European did. And though Late Gaulish sometimes has some varieties of weakening intervocal consonants (between vowels), similar to the same process of Old Irish internal lenition, initial consonants were always strong and stable, and no tendency to drop or be changed. Only in the Insular Celtic branch, which began to develop separately in the last centuries BC, such tendency appears, although in Oghamic inscriptions (the earliest stage of Irish language) again no lenition is noticeable.
Researchers agree in the opinion that initial mutations must have appeared in Insular Celtic dialects between the 1st and the 6th century AD. Some, however, suppose they existed in some forms already in Common Insular Celtic, used everywhere on the British Isles before different separate languages started to grow. Anyway, we can consider proven the fact that this feature of initial consonant mutations is a phenomenon that took place only on the Isles, but not among Gauls, Belges, Celtiberians and Lepontic Celts who lived in continental Europe. Such a strange thing might have appeared, as many linguists believe, due to aboriginal British tongues assimilated by Celtic nation. Maybe, the Stonehenge people, or Picts who carried an unknown family language, or Iberians who are supposed to be the earliest humans in Britain were the source of phonetic changes in Insular Celtic. Otherwise it will be very hard to call all peculiarities of modern Celtic languages an internal process. We mean here not only phonetics, but the strange syntax items, the word order (in Gaelic) abnormal for all Indo-European languages, the prepositional pronouns and many other traits that distinguish Celtic speech from other Indo-European groups and make it so interesting.
We met different points of view concerning this question. There is an opinion that Gaulish and Lepontic languages became extinct before their process of initial mutation, natural for all Celtic tongues. Continental Celtic just did not have time to reach the appropriate stage of the language development. Some linguists disagree that only Celtic tongues were subject to lenitions and nasalisations: I heard an example of some Alpian Italian dialects using similar features, like lenitions of the initial consonant after a vowel. We would not say it is impossible or that the theory described in the present article is true for good. But the problem still remains quite interesting, and there is vast place for research here. We will be waiting for your remarks, additional information on the subject, and we welcome every opinion about the issue here.
It seems more that Proto-Indo-European was not fond of initial mutations. But the development of its branches, growing in various surroundings and sometimes suffering significant influence from other languages, was going each time another way. And that is why some groups chose a way of initial lenitions: a non-Indo-European feature caused by external or internal processes of the language.