Upper Sorbian phonology

The vowel inventory of Upper Sorbian is exactly the same as that of Lower Sorbian.[1] It is also very similar to the vowel inventory of Slovene.

• Word-initial vowels are rare, and are often preceded by a non-phonemic glottal stop [ʔ], or sometimes /ɦ/. /e, o/ never appear in word-initial position, whereas /i, u, ɛ, ɔ/ appear in word-initial position only in recent borrowings.[3]
• /i/ is mid-centralized to [ɪ] after hard consonants.[4]
• /e, o/ are diphthongized to [i̯ɛ, u̯ɔ] in slow speech.[1][5]
• /ɛ/ has three allophones:
  • Open-mid [ɛ] between hard consonants and after a hard consonant;[6]
  • Mid [ɛ̝] between soft consonants and after a soft consonant (excluding /j/ in both cases);[6]
  • Diphthong with a mid onset [ɛ̝i̯] before /j/.[6]
• /ɔ/ has two allophones:
  • Diphthong with a mid onset [ɔ̝u̯] before labial consonants;[7]
  • Open-mid [ɔ] in all other cases.[7]
• The /e–ɛ/ and /o–ɔ/ distinctions are weakened or lost in unstressed syllables.[8]
• /a/ is phonetically central [ä].[1][2] It is somewhat higher [ɐ] after soft consonants.[9]

• /m, mʲ, p, pʲ, b, bʲ, β, ɥ/ are bilabial, whereas /f, v/ are labiodental.[11]
  • /mʲ, pʲ, bʲ/ are strongly palatalized.[12]
  • /β/ is a somewhat velarized bilabial approximant [β̞ˠ], whereas /ɥ/ (the soft counterpart of /β/) is a strongly palatalized bilabial approximant [ɥ].[13]
  • /v/ is very rare. Apart from loanwords, it occurs only in two Slavonic words: zełharny /ˈzɛvaʀni/ 'deceitful' and zełharnosć /ˈzɛvaʀnɔst͡ʃ/ 'deceitfulness', both of which are derivatives of łhać /ˈfat͡ʃ/ 'to lie'. Usage of these words is typically restricted to the Bautzen dialect, as speakers of the Catholic dialect use łžeć /ˈbʒɛt͡ʃ/ and its derivatives.[14][15]
• /n, l/ are alveolar [n͇, l͇], /ɲ/ is alveolo-palatal [ɲ̟], whereas /t, d, t͡s, d͡z, t͡sʲ, s, z, zʲ/ are dental [t̪, d̪, t̪͡s̪, d̪͡z̪, t̪͡s̪ʲ, s̪, z̪, z̪ʲ].[1][16][17]
  • /t, d, l/ before /i/ (in the case of /l/ also before /e, ɛ/) are weakly palatalized [tʲ, dʲ, lʲ]. Šewc-Schuster (1984) also reports palatalized allophones of /f, v, k, ɡ, x, ɦ/, but without specifying the vowels before which they occur.[18] Among these, the palatalized [fʲ, vʲ] are extremely rare.[3]
  • /n, nʲ/ are velar [ŋ, ŋʲ] in front of velar consonants.[19]
  • /d͡z/ is very rare. In many cases, it merges with /z/ into [z].[20][21]
  • /t͡sʲ, zʲ/ are very rare.[20][21] According to Stone (2002), the phonemic status of /t͡sʲ/ is controversial.[3]
• In most dialects, /t͡ʃ, d͡ʒ, ʃ, ʒ/ are palato-alveolar. This is unlike Lower Sorbian, where these consonants are laminal retroflex (flat postalveolar) [t͡ʂ, ʂ, ʐ] (Lower Sorbian /t͡ʂ/ does not have a voiced counterpart).[22][23] Laminal retroflex realizations of /ʃ, ʒ/ also occur in Upper Sorbian dialects spoken in some villages north of Hoyerswerda.[12][24]
• /k, ɡ, x/ are velar, whereas /ʀ, ʀʲ/ are uvular.[25][26]
  • An aspirated [kʰ] is a morpheme-initial allophone of /x/ in some cases, as well as a possible word-initial allophone of /k/.[27]
  • /x/ does not occur word-initially, whereas /ɦ/ does not occur word-finally.[28]
  • The alveolar realization [r͇, r͇ʲ] of /ʀ, ʀʲ/ is archaic.[29]
  • Soft /ʀʲ/ is strongly palatalized.[12]
• /ɦ/ is voiced [ɦ], unlike Lower Sorbian where it is voiceless [h].[30][31]
• An epenthetic /j/ is inserted before a post-vocalic soft consonant, yielding a diphthong. If the soft consonant occurs before /ɛ/ or /e/, it is often realized as hard, and /e/ is lowered to [ɛ].[3]
• In literary language, the contrast between hard and soft consonants is neutralized in word-final position. Word-finally, the letter ⟨ń⟩ represents a post-vocalic sequence /jn/, as in dźeń /ˈd͡ʒɛjn/ 'day'.[3]

• Words consisting of up to three syllables are stressed on the first syllable.[33]
• Foreign words, such as student /stuˈdɛnt/ 'student', preserve their original accent.[34]