User:Calthinus/Phonological history of Albanian

This page traces the phonological development of the Albanian language over time, with a special consideration at the end of sound correspondences.

Overview and significance

Earlier developments in Albanian may be especially important for the reconstruction of Proto-Indo-European, although the early period remains understudied and historical phonological methods currently do not have sufficient precision in identifying and dating shifts attributed to this that period. Of especial interest to scholars of Indo-European is the theory that Albanian may have preserved the proto-Indo-European laryngeals, although this is controversial and debated.

Later Albanian sound changes are much easier to identify due to the large amount of Latin words present in Albanian from the Roman Era onward. The split between Gheg and Tosk is dated near the end of the Roman Empire, before the arrival of Slavs in the region.

Typology and Dialectology

Albanian is a satem Indo-European language, but unlike most other satem languages except Armenian and possibly Baltic, it consistently treats all three sets of original velar consonants differently even in its modern form, and in some contexts the "palato-velars" have remained velar. The date for the "satemization" of Albanian is late, as some of the earliest Greek loanwords appear to be affected.

The presence of the close front rounded vowel /y/ in Albanian is due to neither Germanic nor Turkic influence, but instead it developed regularly from Roman era segments of long /u:/ or the diphthongs /iu/ and /ui/, which also occurred in French and some other Gallo-Romance languages, affecting most Latin loanwords in Albanian. However, Cabej and Orel discovered a set of unaffected words that may <CITE OR DELETE>> be loanwords from some extinct Romance dialect in the region <FIND A SOURCE THAT IT ISN'T AROMANIAN?>

Other "French-like" shifts in late Roman-era and medieval Albanian include vowel breaking of long stressed vowels, the closing of /o/, the deletion of unstressed vowels, regressive nasalization and deletion of following nasal stops, and the development of a central vowel phoneme. <IS THIS CONSIDERED SYN? IF SO AND CANT FIND IN OREL, DELETE>. It is unlikely that these actually arose due to actual influence from Gallo-Romance dialects, but rather it represents a case of parallel converging developments <CITE OR DELETE>. Limited intervocalic lenition also happened with voiced stops in Albanian, but this is likely connected to similar developments in Greek and Aromanian. Other developments shared with Greek include the development of nasal + oral stop clusters. <SAME AS BEFORE> Similarly, final devoicing in the Korce dialect, although shared with French, occurred much later, and is likely due to Turkish influences.

Romanian sound shifts during the relevant early medieval period also share much with contemporaneous Albanian shifts, and the two languages are thought to have been in contact.

A recurrent theme in Albanian phonological history is repeated palatalization, which occurred at least three times, often associated with the velarization of certain non-palatalized dental elements. In this sense, Albanian phonological history resembles that of Irish, Russian and French <find that book on palatalization and cite it>. The most recent round of palatalization occurred at a similar time as it occurred (and is still occurring) in Greek and Turkish and may <CITE???> be connected.

Despite Albanian assertions as well as those of some non-Albanian scholars that Albanian sounds are "old", Albanian is actually one of the more innovative languages of the area, being more innovative than Greek or Italian and comparable to Romanian or Bulgarian <CITE THIS??>. Within Albanian, some dialects are much more conservative than others, especially Cham, which preserves many segments that are lost or changed elsewhere. Many Albanians believe that Gheg is "older" than Tosk but this is not actually the case as the level of conservatism varies markedly between Gheg dialects just as it does between Tosk dialects <CITE THIS>. Like French, many unstressed vowels were lost or reduced, but the case system was preserved despite this, as in Irish.

Albanian, Greek and Aromanian are the only languages in the Balkans to retain phonemic interdental spirants, and in Europe the only other major languages that retained them are English and Welsh, not counting the Elfdalian idiom in Sweden. Albanian and Elfdalian are the only two languages in Europe to have both voiceless interdental spirants and front rounded vowels.

The split between Tosk proper and the other Southern dialects -- Cham, Lab, Arvanite, Arbereshe -- happened in the High Middle Ages. The far southern dialects are generally conservative, but do show some innovation perhaps <WHERE WAS THE CITATION FOR THIS> due to Greek or Central Romance influence, in the unrounding of /y/ and the development of velar spirants. On the other hand, central Lab dialects appear to have uniquely retained nasalization in the South <some 1990s Albanian author> as well as extra front rounded vowels.

There are no significant differences between the speech of Christian and Muslim Albanians in terms of the numbers of Turkishism or Greekisms phonologically; both influences instead follow geographic boundaries. On the other hand, in the north, Christian dialects differ phonologically from Muslim dialects to the point that the faith of ones ancestors can be deduced from speech. In general, Christian dialects are more conservative in some ways, but also more innovative and generally divergent, such as the development of triphthongs among Catholics in and around Shkoder, absent from the phonological inventories of every other Albanian dialect <CITE?>. In Shkoder, the speech of Muslims resembles that of Kruja, while Christian speech resembles most of all Malesia and Zadrima. The dichotomy is not true in the South, where Christians speech shows no divergence in development from Muslim speech.

Developments from Proto-Indo-European to Pre-Roman Albanian

Palatalization of proto-Indo-European palato-velars, late enough for Greek loans to also be affected
Most laryngeals are eliminated

Developments to Roman-Era Albanian

fronting of long u
Second round of palatalization, this time affecting what were the proto-Indo-European labiovelars
Proto-Indo-European back velars become plain velars
/s/, depending on its position, becomes either palatal/postalveolar or debuccalized. The mutation of s in this period also occurred in Greek (sex> hex, for example). By the time of Latin contact, /s/ is long gone, and Latin loans with /s/ are rendered as postalveolar, as Albanian sh (Latin summus > Albanian shume)
- Intervocalic s ended up as /h/, and was later deleted
- Initial s ended up as /j/, later closing to a voiced palatal stop, as Albanian gj.

Developments to Late Roman Albanian and early Medieval "Old Albanian"

Long u fronting, merger with ui and iu, ultimately as /y/
Long o is also fronted
Long stressed vowel breaking
Unstressed vowel apocope

The Gheg-Tosk split

Intervocalic dental nasal flapping bleeds nasalization in Tosk, and nasalization is lost (except in some Lab dialects?)
Unstressed vowels before nasals are deleted in Gheg, and flapping never occurs.
Velar fricative in Tosk becomes debucallized, whereas in Gheg it becomes labiodental

Developments during Medieval Albanian

front rounded /ø/ becomes /e/ (ex: okto:ti > tø:ti > tetë)
intervocalic r is geminated to rr, l geminated to ll. (ara > arra )
Palatalized r also becomes rrj , intervocalic or not.
single l before palatal vowels or consonants to j
ll becomes a single dark l, as part of velarization
single trill r becomes a tap
degemination of all consonants except for rr, which becomes a trill, taking the place of the old single trill

Further dialect splits: Lab, Cham, Arvanite, and Arbereshe emerge in the South

Developments to Modern Albanian

Another round of palatalization (doesn't effect some dialects, like Cham and Kruja)

Indo-European Phonological correspondences

<don't actually include all phones, just the ones for which the correspondence is fairly simple and 1 : 1 ish or at least close>

c | all but praeconsonantal | ts | th,s | k (>s,th,c in most Romance languages) | k |

Latin phonological correspondences -- for Latin vocabulary

u: | hiatus | y: > i | u | u | y | y:>i | u | u | u | u | u| u u: | elsewhere | y | u | u | y | y > u-bar | u | u | u | u| u | u o: | everywhere | ø:>e | o | o | ø:, u: if closed by labial, y: if otherwise closed|

Albanian Historical Phonology Compared