Palatalization in Old Church Slavonic
[For a discussion of the image, see the post in the LanguageHat blog.]
In our examination of the Old Church Slavonic (OCS) declension system, we have noticed some peculiar consonant alternations, for example: člověkъ ‘man’ (nominative singular), but člověcě (locative singular) and člověče (vocative singular). The root also ends in “k” in accusative, genitive, dative, and instrumental singular, all six forms of the dual, and accusative, genitive, dative, and instrumental plural; the alternant “c” is also found in nominative and locative plural. The same alternation also appears in other inflectional forms of both nouns and verbs, as well as in some derivational (word-formation) patterns. Where do they come from?
These alternations are a result of a process known as palatalization.* In OCS, there was a phonotactic rule (i.e., a type of rule that governs what sounds can or cannot appear next to what other sounds) that prevented velar sounds from occurring before front vowels. Velar sounds are consonants articulated with the back of the tongue against the soft palate (i.e., the back part of the roof of the mouth, known also as the velum). In OCS, there were three velar consonants: k, g, and x. Front vowels are vowels pronounced with the tongue moving forward; in OCS, there were five front vowels: i, e, ě, ь, and ę. Thus, k, g, and x could not appear in front of i, y, e, ě, and ę. What happened if for some reason such a configuration emerged? The answer is palatalization, which changed velars into coronal and palatal sounds. (Coronal sounds, like /ts/, are articulated with the front part of the tongue, “corona” in Latin; palatal sounds, like /č/, involve a movement of the body of the tongue up towards the hard palate.) Palatalization of velars before front vowels is quite common cross-linguistically, as it is an assimilative type of process: velar sounds, which are articulated in a way quite distinct from that of front vowels, are turned into consonants whose articulation occurs more “in the front of the mouth”. The specific alternations that arose in Slavic as a result of two different palatalization processes, as explained below, are summarized in the following table:**
Alternations resulting from First and Second Slavic Palatalization:
original |
/k/ |
/g/ |
/x/ |
First Slavic Palatalization |
/č/ |
/ž/ |
/š/ |
Second Slavic Palatalization |
/ts/ (“c”) |
/dz/ (“ʒ”) |
/s/ |
In historical linguistics, the two distinct processes of palatalization of velars are distinguished, called “First Slavic Palatalization” and “Second Slavic Palatalization” (they are thought to have occurred in that order). Although historically they were triggered by distinct subsets of front vowels, some of these distinctions have become obscured by later changes that affected the vowels. The First and Second Slavic Palatalizations applied in Proto-Slavic, resulting in changes to root consonants. The effects of the First Slavic Palatalization are best illustrated by the following old Germanic loanwords (asterisks indicate forms that have been reconstructed but not attested):
Germanic *helmaz ‘helmet’ > PSl. *xelmu > *šelmu > OCS šlěmъ, Rus. šelóm, šlem
Germanic *kinda ‘child, infant’ > PSl. *kinda > *činda > OCS čędo, Rus. čado
These words were borrowed from Proto-Germanic to Proto-Slavic as *xelmu ‘helmet’ and *kinda ‘child’ and then subjected to the First Slavic Palatalization, which changed /x/ to /š/ and /k/ to /č/. With a few additional sound changes that need not concern us here, the first word was further transformed into OCS šlěmъ, which also shows up in modern Russian as šlem, alongside the “native” Russian šelóm (archaic). The second word morphed into OCS čędo and Russian čado.
The Second Slavic Palatalization produced the OCS word cělъ ‘full’ and its Russian counterpart celyj; a detailed history of this word is provided in my earlier post.
PIE *koylo- > PSl. *kajlu ‘whole, healthy’ > OCS cělъ, Russ. célyj
Within OCS, the same palatalization processes applied if a velar preceded a front vowel across a morpheme boundary, for example, when a nominal stem ending in /k/ was followed by a suffix starting with a front vowel. This is exactly what we see in the declension of člověkъ ‘man’, as mentioned above.
What determines whether the “k” of the stem changes to “č” or to “c”? It depends on the exact front vowel that triggers the palatalization, but here things get a little bit tricky. The only two vowels that trigger the second palatalization (i.e. k-to-c change) are /i/ and /ě/, but not any odd /i/ and /ě/ trigger the second palatalization—some other instances of /i/ and /ě/ actually trigger the first palatalization! The reason for this complications is this: although by the OCS period, the /i/ that triggered the second palatalization and the one that triggered the first palatalization sounded exactly the same, and the /ě/ that triggered the second palatalization and the one that triggered the first palatalization sounded exactly the same, yet historically the two different i’s and the two different ě’s came from different sounds. (In historical phonology, this is called a “merger”: two distinct earlier sounds end up being pronounced exactly the same.) But their different past would still catch up with them in OCS and cause them to trigger two different palatalization processes. (A similar example of two vowels that sound exactly the same but behave differently with respect to some phonological rule because of their distinct origins was discussed in my earlier post: the /e/ in lev ‘lion’ disappears in the genitive l’va, but the exact-same-sounding /e/ in les ‘forest’ does not disappear in the genitive lesa.)
So how can we tell the two different ě’s or the two different i’s? Unless you speak fluent Proto-Slavic (or Proto-Indo-European), the only way to know is to learn which morphemes contain them. A helpful hint is that the ě that triggers the second palatalization (i.e. k-to-c) correlates with i as the “soft stem” alternant. In the twofold nominal declension (see Lunt 2001: 54), this ě appears in four paradigm slots: masculine/neuter locative singular, feminine locative/dative singular, neuter/feminine dative dual, and in the beginning of the ending for masculine/neutral locative plural (-ěxъ/-ixъ). In other words, in masculine nouns whose stems end in a velar, like člověkъ ‘man’, the c triggered by the following ě appears twice: in locative singular (člověcě) and in locative plural (člověcěxъ). In feminine nouns whose stems end in a velar, like rǫk-a ‘hand’, the c triggered by the following ě appears also twice: in locative and dative singular (rǫcě) and in nominative and accusative dual (also rǫcě). The second-palatalization-triggering i occurs in the two fold nominal declension only in the masculine nominative plural: hence, člověci.
Another place where the second palatalization, triggered by the “right kind” of ě/i shows up is in the imperatives. The imperative is formed by a suffix -ě/-i, following by a person/number ending (see Lunt 2001: 98). As noted in Lunt, the suffix –ě appears in the dual and plural of stems that end in -ǫ or j, while -i appears in all the singular forms and in the dual and plural forms of the stems that end in a consonant other than j or a vowel other than ǫ. Crucially for our purposes, both imperative suffixes -ě and -i are of the kind that triggers the second palatalization, so all velar-final stems undergo the k-to-c change. Consider: mog-ǫtъ ‘be able’ ~ moʒi / moʒěte; strěg-ǫtъ ‘protect’ ~ strěʒi / strěʒěte. As an additional complication, stems in -ek / -eg change to –ьc / –ьʒ; hence, rek-ǫtъ ‘say’ ~ rьci / rьcěte.
With the exception of the “special” i and ě, all other front vowels, including the other instances of i and ě, as well as ь, e, and ę, trigger the first palatalization, i.e. the k-to-č change, as in člověče ‘man’ (vocative singular).
Finally, it should be noted that, as discussed in great detail in an earlier mini-series of posts (see part 1, part 2, part 3, and part 4), the Second Slavic Palatalization did not apply in the dialect of Old Novgorod. Russian linguist Andrei Zalizniak has shown that the Old Novgorod dialect, alongside the Old Russian dialects of the Kiev and Moscow, played a part in the formation of the standard modern Russian. As a result, modern Russian combines some forms from the southern zone that exhibit the palatalization, such as celyj ‘whole’, with forms without the palatalization that derive from the dialect of Old Novgorod, such as na ruke ‘on the arm/hand’.
_______
*Unfortunately, “palatalization” is a confusing term because it can mean one of two different things: addition of a palatal co-articulation to the usual articulation of the consonant or changing the usual articulation of the consonant completely. The first case can be illustrated with the change from /k/ to /k’/ (the contrast between the two sounds can be heard in Russian in k Ire ‘to Ira’ vs. Kire ‘to Kira’), or the difference between the “hard” /t/ and “soft” /t’/ (see image on the left). In this post, we are interested in the second type of palatalization, the one that changes, for example, /k/ to /č/.
**Besides plain velars, consonant clusters consisting of /s/ followed by a velar were subject to these palatalization processes, but for the presentation’s sake, here I focus on the plain velars only.
Sources:
Lunt, Horace G. (2001) Old Church Slavonic Grammar. Mouton de Gruyter.