The merger of /ɛ/ and /i/ is not allowed in formal speech and is regarded as a provincial (East Bulgarian) dialectal feature; instead, unstressed /ɛ/ is both raised and centralized, approaching the schwa ([ə]).
The theory further posits that such neutralization may nevertheless not always happen: vowels tend to be distinguished in emphatic or deliberately distinct pronunciation, while reduction is strongest in colloquial speech.
It is traditionally regarded as a semivowel, but in recent years, it has largely been treated as a "glide" or approximant, thus making it part of the consonant system.
While certain Western Bulgarian dialects (in particular, those around Pernik), have had a long-standing tradition of pronouncing [ɫ] as [w], the use of the glide in the literary language was first noted by a radio operator in 1974.
[27] Another discovery of the study was that in particular positions, certain participants enunciated neither [ɫ] nor [w], but the high back unrounded vowel [ɯ] (or its corresponding glide [ɰ]).
The semivowel /j/ forms a number of diphthongs, which are summarized below:[28][29] The main point of contention between the two schools of thought on Bulgarian consonantism has been whether palatalized consonants should be defined as separate phonemes or simply as allophones of their respective hard counterparts.
[31][13][14] It has proposed alternative notation of palatalized consonants in the form of C-j-V (consonant-glide-vowel) clusters and has made a tentative hypothesis about the decomposition of Bulgarian palatals into consonants + glide using the following arguments:[4][32][2] The second school of thought came to being rather unexpectedly in the late 1940s, as a refinement of Trubetzkoy's rough draft a decade before.
It posits that apart from ⟨й⟩ (/j/), there are 17 separate palatal phonemes that are in minimal pairs with their hard counterparts, including дз' (/d͡zʲ/) and х' (/ç/), which are not found in any native Bulgarian words and were excluded from Trubetzkoy's draft.
They argue that Bulgarian phonemic inventory consists of a total of 45 phonemes, whereof 6 vowels, 1 semivowel and 38 consonants, and present the following arguments:[41] Proto-Slavic underwent three separate rounds of palatalization and one of iotation, but the resulting palatal consonants eventually hardened in Western and South Slavic.
As a result of the contraction and closure of the syllable in the Middle Bulgarian period, unstressed /i/ in many cases turned into the semivowel /j/ or attached to a consonant, palatalising it.
Moreover, according to Stefan Mladenov,[55]"If we disregard individual cases of old, stronger palatalization, which may be found in Eastern and Western dialects alike, Contemporary Standard Bulgarian has developed a very distinctive "semi-palatalization", which is often neglected.
Examples include the complete elimination of end-word palatals in a number of words ending in ⟨р'⟩ (/rʲ/), ⟨н'⟩ (/ɲ/), ⟨л'⟩ (/ʎ/) and ⟨т'⟩ (/tʲ/), e.g., writing and saying кон [kɔn] ('horse') instead of конь [kɔɲ], път [pɤt] ('road') instead of пъть [pɤtʲ]), etc.
; the adoption of the hard suffix -не instead of -нье for verbal nouns, i.e., писане [pisɐˈnɛ] instead of писанье [pisɐˈɲɛ] ('writing'); labelling palatalization before front vowels as dialectal: (поле [poˈlɛ] instead of полье [poˈʎɛ] ('field'), тиква [tikˈvɐ] instead of тьиква [tʲikˈvɐ] ('pumpkin')), etc.
An alternative postulates the following palatalized consonants /pʲ, bʲ, tʲ, dʲ, kʲ, gʲ, tsʲ, dzʲ, mʲ, nʲ, rʲ, fʲ, vʲ, sʲ, zʲ, xʲ, lʲ/.
A large number of other Bulgarian linguists have come out in support of this more minimalistic view of Bulgarian consonantism, e.g., Kiril Mirchev,[69] Petar Pashov,[70] Bozhil Nikolov,[71] Todor Boyadzhiev,[72] Борис Симеонов [bg], who has argued that there was no logic that could explain why a consonant affected by yat mutation (e.g., /b/ in бял-бели [bʲaɫ]-[ˈbɛli]) would be palatal in some of its forms and hard in others, and so on.
These have included Austrian researcher Merlingen (1957),[74] Americans Carleton Hodge (1957)[75] and Joseph van Campen and Jacob Ornstein (1959),[76] Romanian linguist Alexandru Rosetti, who qualified the degree of palatalization of Bulgarian consonants as "a softening" (1967),[77] Swiss Max Mangold (1988),[78] Korean Slavist Gwon-Jin Choi, who has argued about the decomposition of Bulgarian palatalism (into C + j) (1994),[4][36] as well as phoneticians Ternes and Vladimirova-Buhtz, who have most recently suggested C-j-V notation of palatals, as their limited distribution proved they were allophones rather than phonemes (1999).
[91] In his magnum opus, Principles of Phonology, published posthumously in 1939, he referenced extensively Eastern Bulgarian, even offering a model phonemic inventory it.
[93] The other major postwar Bulgarian linguist, Lyubomir Andreychin, then quickly suggested another two, /d͡zʲ/ and /ç/, arguing that even though they only existed in foreign proper names like Хюстън /xʲustɤn/ ('Houston') and Ядзя [jad͡zʲa] ('Jadzia') and had no contrastive function, they could have one, if need be.
[94] Tilkov designated /d͡zʲ/ and /ç/ as "potential phonemes", adding ф' (/fʲ/) to them in 1982, as it existed in only a handful of words, all of them borrowings (e.g., фюрер [ˈfʲurɛr] ('Führer')).
While the consonant model was lauded in the Soviet Union by linguists such as, e.g., Yuriy Maslov, acceptance in the West, except for Klagstad, has been lukewarm.
The most prolific Bulgarian phonologist and grammarian in the English-speaking world, Ernest Scatton, notes (1993):[99]Alveo-palatal obstruents are weakly palatalized.
[102] Only the articulation of bilabial and labiodental consonants (/pʲ/, /bʲ/, /mʲ/, /fʲ/, /vʲ/) is accompanied by a noticeable raising of the body of the tongue towards the palate, but only to a moderate extent.
[113] All this has raised the question whether Bulgarian palatals have indeed lost their secondary articulatory gesture and have decomposed into CjV sequences, as claimed by Danchev, Ignatova-Tzoneva, Choi, etc.
[114] Nevertheless, based on the phonological distribution of Bulgarian palatals, which was similar to that in English and completely different from that in Russian, the author argued in favour of CjV notation.
[116] Though the ⟨ʃt⟩/⟨ʒd⟩ speaking area currently covers only the territory of the Republic of Bulgaria and the eastern half of the wider region of geographical Macedonia, toponomy containing ⟨ʃt⟩ and ⟨ʒd⟩ that goes back to the Early Middle Ages is widely preserved across Northern and Central Greece, Southern Albania, the Republic of Macedonia, Kosovo and the Torlak-speaking regions in Serbia.
[116][117] For example, in the Struga municipality, the names of 13 out of 43 villages contain either ⟨ʃt⟩ (Kališta, Korošišta, Labuništa, Moroišta, Piskupština, Radolišta, Tašmaruništa, Velešta and Vraništa) or ⟨ʒd⟩ (Delogoždi, Mislodežda, Radožda and Zbaždi).
[116] The same applies to Kosovo, where Russian Slavist Afanasiy Selishchev found а number of place names around the city of Prizren featuring the Bulgarian clusters ⟨ʃt⟩/⟨ʒd⟩ in a Serbian official document from the 1300s (Nebrěgošta, Dobroušta, Sěnožeštani, Graždenikī, Obražda, Ljubižda, etc.).
If the existence of separate palatalised consonant phonemes (39-consonant model) is accepted, 6 more contrastive obstruent pairs are added: /pʲ/↔/bʲ/, /fʲ/↔/vʲ/, /tʲ/↔/dʲ/, /sʲ/↔/zʲ/, /tsʲ/↔/dzʲ/, /ɟ/↔/c/, for a total of 14.
Like all other Slavic languages apart from Serbo-Croatian and Ukrainian, Bulgarian features word-final devoicing of obstruents, unless the following word begins with a voiced consonant.
[125] Assimilation also occurs across word boundaries (in the form of sandhi), for example, от гората is pronounced [odgoˈratɐ] ('from the forest'), while над полето becomes [natpoˈlɛto] ('above the field').