Carpathian historical development

Carpathian is descended from Proto-Indo-European. This language in turn is the parent language of the vast majority of European languages (including English, German, Spanish, French, etc). Proto-Carpathian gradually evolved into various modern Carpathian dialects during the first millennium CE, concurrent with the Slavic, Avar and Hungarian contact. There is no scholarly consensus concerning either the number of stages involved in the development of the language and their periodisation, but for convenience, three stages are usually defined as follows:

• Proto-Carpathian (3500 BCE — 1000 BCE) — a long period of gradual development from Proto-Indo-European. No dialectal distinctions can be reconstructed from this period. Loanwords from an unknown, likely pre-Indo-European substratum entered the language.
• Common Carpathian (1000 BCE — 500 CE) — the stage with the earliest identifiable dialectal distinctions and borrowings from other languages. At this stage Paleo-Balkan influence is prominent.
• Late Carpathian (500 BCE — present) — gradual development of individual Carpathian dialects. The influence of the Slavic languages is prominent at this stage.

Split from Late Proto-Indo-European

Proto-Carpathian exhibits the satem development wherein Proto-Indo-European (PIE) palatovelar consonants became affricates or fricatives, conventionally indicated as *ś and *ź.

• *ḱ → *ś
• *ǵ, *ǵʰ → *ź

These later became retroflex fricatives *š and *ž in the Western dialects, but dental fricatives *s and *z in the Eastern dialects, which is the feature that defines each group (and became the main difference for the subsequent Western and Eastern standard languages).

This sound change was incomplete, in that the Carpathian languages has instances where PIE palatovelars appear as *k and *g, such as kalaušētei “to be listening” from *ḱléwseti, gansis “goose” from *ǵʰh₂éns. Some of these instances are explained by prohibitive dissimilation: *Ḱ_s > K_s — when a palatovelar is next to a sibilant within the same root, it loses its palatalisation (this, however, does not explain words such as hakmū “stone” from *h₂éḱmō. According to Matasović, the depalatalisation of palatovelars also occurred before sonorant followed by a back vowel: *Ḱ_RV_back > K/_RV_back).

Other satem sound changes are the unconditional delabialisation of labiovelar consonants (*Kʷ > K) and the ruki sound law (*{r,u,k,i}s > {r,u,k,i}š), according to which the dental sibilant *s became retroflex *š after *r, *u, *k or *i. In Eastern Carpathian this new š reverted back to s before plosive consonants (in some southern dialects before all consonants), but was retained in other positions, being the only source of the retroflex sibilant in these dialects. In Western Carpathian, on the other hand, *ś and š merged with very few exceptions, such as tor̃sā “bristle” corresponding to Eastern tar̃šā “crumb”.

The final consonants, except for *n and *s, were lost. Final *m changed to n. Final resonants were preserved by addition of a new ending: PIE *péh₂wr̥ “wheat” > *púhr-an > Car. pūrha “bromegrass”. Final plosives were dropped: PIE *tod > Car. ta “it”.

PIE voiced aspirated consonants merged with their plain counterparts:

• *bʰ → *b
• *dʰ → *d
• *gʰ → *g

However, the distinction between the two series manifested as Winter's law, which operated before plain voiced stops, giving the syllable rising acute accent and lengthening the vowel (with some exceptions, such as stògas “stack” from PIE *stógos.

The short vowels *o and *a merge early in Proto-Carpathian resulting in the back likely labialised vowel *a, which was preserved unchanged in Eastern Carpathian. This is supported by later loanwords, such as hamaras “sombre, gloomy” likely from Avar homor (however, Eastern dial. humuras also exists, suggesting that *a was an open vowel with weak labialisation. The borrowing from Proto-Slavic *xmura “cloud” is possible). In most Western dialects new short o arose from *a under certain conditions, such as under circumflex accent in a closed syllable, or near labial consonants: gor̃šatas “earthquake”,wongìs “fire”. Some a>o changes are irregular, likely of an affective nature: dòštei “to resemble” (Eastern: dasētei).

The PIE diphthong *ew became *jau after consonants (*au before labials and h) and *aw after vowels: háugetei “to grow” from PIE *h₂ewgeti, liaũbas “sweet” from *léwbʰos.

Just as in Proto-Balto-Slvaic, Hirt's law operated in Proto-Carpathian, causing the accent to retract to the previous syllable, if the vowel in the preceding syllable was immediately followed by a laryngeal:

• PIE *dʰuh₂mós > Proto-Carpathian dū́mas “smoke”, cognate with Lithuanian dū́mas.
• PIE *gih₃wós "alive" > Proto-Carpathian *gī́hwas > gī́was, cognate with Lithuanian gývas. But in Balto-Slavic Hirt's law was not in effect in this word — Lituanian gyvà “alive, feminine” for Carpathian gī́wā. Carpathian gihwùs “healed” (with preserved laryngeal) was likely formed by analogy and represents later development.

It took place before the addition of epenthetic vowels before syllabic sonorants, so at the time of the change, syllabic sonorants still acted as a vocalic nucleus like the true vowels and could attract the accent as well.

Syllabic sonorants

Syllabic sonorants become liquid diphthongs with *i (sometimes *u) inserted before them. *u instead of *i is expected before labiovelars, which means that the change happened before the delabialisation of velars. This change occurred after Hirt's law, which operated on original syllabic sonorants but not on sonorant diphthongs. Another change that happened during the same period was *wl, *wr > *l, *r word-initially (but *wrōdnas > W. wōrdanas, E. wōrdaras “dawn” (cognate with Sl. *rànъ “morning”), which preserves the initial *w through vowel metathesis, however the reconstruction *wōrdnas is also possible, likely related to war̃dinas “early”). It occurred after the prothetic vowels had appeared: *wĺ̥kʷos > wul̃kas “wolf”.

Laryngeals

Because Carpathian preserves consonantal reflexes of PIE laryngeals at least in some positions, it is often compared to the Anatolian languages and Armenian, although in other regards it shares more in common with Balto-Slavic, than with the former two.

Between consonants PIE laryngeals became *a in initial syllables and were lost in non-initial syllables:

• PIE *sph₁rós > Proto-Carpathian spàras “abundant”;
• PIE **dʰugh₂tḗr > Proto-Carpathian duktī̃ “daughter”.

However, there is some evidence pointing towards preservation of h₃ in initial syllables as *u: PIE *h₂éh₃mōl~*h₂h₃m̥lés > Carpathian hùmōlis “yellow sorrel”.

PIE *h₁ was lost in all positions, but it made its preceding vowel (or a syllabic sonorant) long and gave it a rising acute accent: PIE *spéh₁ti > Proto-Carpathian *spḗtei “to hurry”. A semivowel (*j or *w) was inserted to break a vowel hiatus: *dʰeh₁eti > dējetei “to be doing”.

The laryngeals *h₂ and h₃ merged in Proto-Carpathian, just as in Proto-Anatolian, and was preserved as a glottal fricative /h/. When preceding a consonant this new *h was lost, acting in the same way as *h₁: Proto-Carpathian *stáhtei > stā́tei “to become” (from PIE *stéh₂ti). In other positions the reflex of PIE laryngeals was retained: PIE *stoh₂éyeti > Carpathian stàhītei “to determine”; PIE *kruh₂és (genitive) > Carpathian kurùhis “blood” (dialectal karaũhis, from nominative *kréwh₂s); PIE *welh₃is > Carpathian wel̃his “ghost”. Word-finally, however, the laryngeals were also lost, probably at the same time as all final consonants, except for *s and *n: PIE *dʰoHnéh₂ > Carpathian dṓnā “grain”.

Consonant assimilation

Many heterorganic consonant clusters were assimilated in Proto-Carpathian, giving rise to geminate consonants: PIE *h₁n̥h₃mén > Carpathian immin “name”; PIE *déḱsinos > Common Carpathian deśśinas “right”. The combinations with *w and *j as a second element survived, although the clusters of dental consonants and *j underwent iotation: *weĩśśinjāh > weisiniā “cherry” (Eastern [weɪ̀.si.ɲɑ], Western [weɪ̀.ʂi.ɲɑ], later degemination of *ś after a diphthong); labial consonants were not affected, while velar consonants show variation among various dialects: liaugā ~ liaudiā “puddle” ("diV" represents the [ɟ] sound). Some dialects, particularly in the eastern highlands, also undergo labialisation, though this process is not regular or widespread: Standard swestī “sister” — [sʷɛ̀s.tiː] or [sỳs.tiː], sometimes spelt süstī in those dialects. The consonant *w was elided in clusters with labial consonants, followed by it: PIE *bʰuHeh₁t → Pre-Proto-Carpathian bwēt — Carpathian bē “was”. Clusters with *h simplified by eliding the glottal fricative, except for clusters with sonorants, where h was the second element, which were preserved in most dialects (sometimes as plain voiceless sonorants).

Clusters of sonorants or sibilants, followed by plosives were permissible and thus remained unchanged. The reverse clusters (with plosives as the first element of a cluster), as well as clusters with more than two consonants and some heterorganic clusters, were resolved by vowel anaptyxis, also called pleophony after a similar process in the East Slavic languages: PIE *bʰrodʰos > Carpathian baràdas (Western boràdas) “ford”. This also affected borrowings from the neighbouring languages: Slavic *korl′ь > Carpathian koròlias/karàlias “king”, the "rl"-cluster being a heterorganic sonorant cluster, difficult to pronounce. However, some complex consonant clusters were simply reduced instead: *dilbtilas > diltilas “woodpecker”. Clusters with a dental or velar plosive followed by a sibilant were usually resolved by assimilation to that sibilant: PIE *pleth₂-som — Modern Carpathian pelessa “glacial lake”; *muHdʰ-stlom — mūslis “thought” (via *mūssəsla → *mūsslis → mūslis). When a sibilant is in a cluster between two consonants, it disappears: Proto-Carpathian *lōkštawis — modern lōktawis “swallow” (from PIE *lek-ti “to jump”). A special case is initial clusters with a plosive, followed by a sibilant, in which case instead of pleophony, metathesis took place: PIE *k⁽ʷ⁾sowdós — Carpathian skaudùs “small, dispersed”.

The clusters of two plosives typically resulted in a geminate second consonant: PIE *septḿ̥ > Proto-Carpathian septəmas > Eastern Carpathian settimas “seven” (Western settemas). However, early geminate dental plosives were broken into homorganic clusters with fricatives as the first element: PIE *h₁éd-ti > Early Proto-Carpathian ēttei > Carpathian ēstei “to eat”. The kt-cluster was preserved unchanged: naktis “night”, paktas “baked goods” (but *təranktas → tarankatas “crowd, throng”; *penktas → penkitas “fifth” because of a complex "nkt"-cluster).

The clusters with nasal as a second element typically survive as geminate "nn" (PIE *h₂egʷnents > Carpathian hànnēn “lamb”), unless they comprise a cluster with more than two consonants: PIE *ml̥dʰnis — Western Carpathian mil̃dinis “lightning” (Eastern Carpathian lùkkawis from *lukskawis). Labial consonants, however, become [m] before nasals: PIE *swépnos — Standard Carpathian swamnas [ˈswɑ̀m.nɑs] “dream”; *dʰubʰnóm — dumna “bottom”.

The clusters *sr, *śr and *źr were rare in Carpathian, but had a unique development of prothetic dental stops between the two elements, which resulted in the only allowed clusters consisting of more than two consonants: PIE *h₂ewsrom — hauštra “morning, east”. This process likely happened early in the Carpathian development, since the initial clusters were also affected: PIE *sroumḗn → Pre-Proto-Carpathian straumēn — Carpathian staraumēn “stream”. The development is still productive, affecting later loanwords: Izdrajelis “Israel”.

Dialectal differentiation

Approximately by the beginning of the Common Era Proto-Carpathian began splitting into two main branches: Western, spoken nowadays primarily in Slovakia and some districts of the Transcarpathian region, and Eastern, mostly spoken in Southern Poland, Western Ukraine and Northern Romania (region, called Marišas).

Alternations between *e and *a reflexes

Initial *a and *e (as well as their long counterparts) tend to alternate between two Carpathian dialect groups. In Eastern Carpathian the reflex is usually a', while in Western Carpathian it is e instead. Long "ē" and "ā" are more stable, but can appear in affective terminology: Western dēdis — dādis “uncle”; dēla ― dāla “deed, action”; rēnā — rānā “wound”. A special case in the er-ar-vacillation, because the process is regular in the two branches: Western erellas — Eastern arallas “plough”, ermēn — armēn “shoulder”. Rozwadowski's rule, shared with Balto-Slavic, may be a part of this general trend. According to this rule the original *e and *a were mixed, and both became "e" in Western Carpathian and "a" in Eastern Carpathian with few rare exceptions: Proto-Carpathian *éźera “lake” resulted in Western ežera and Eastern azera; *eske “still” → eske — aske; *àlawa “lead” → elowa — alawa; *àpsā “aspen” → espā — aspā (dial. àpasis). The rule only affected the absolute initial position, and is thus different from other e-a-vacillations.