Russian Phonology

This article discusses the phonological system of standard Russian based on the Moscow dialect (unless otherwise noted). For an overview of dialects in the Russian language, see Russian dialects. Most descriptions of Russian describe it as having five vowel phonemes, though there is some dispute over whether a sixth vowel, , is separate from . Russian has 34 consonants, which can be divided into two types:
* ''hard'' ( ) or ''plain''
* ''soft'' ( ) or '' palatalized''

Russian also distinguishes hard consonants from soft (palatalized) consonants and from consonants followed by , making four sets in total: , although in native words appears only at morpheme boundaries. Russian also preserves palatalized consonants that are followed by another consonant more often than other Slavic languages do. Like Polish, it has both hard postalveolars () and soft ones ( and marginally or dialectically ).

Russian has vowel reduction in unstressed syllables. This feature also occurs in a minority of other Slavic languages like Belarusian and Bulgarian and is also found in English, but not in most other Slavic languages, such as Czech, Polish, most varieties of Serbo-Croatian, and even the closely-related Ukrainian.

Vowels

Russian has five to six vowels in stressed syllables, and in some analyses , but in most cases these vowels have merged to only two to four vowels when unstressed: (or ) after hard consonants and after soft ones.

A long-standing dispute among linguists is whether Russian has five vowel phonemes or six; that is, scholars disagree as to whether constitutes an allophone of or if there is an independent phoneme . The five-vowel analysis, taken up by the Moscow school, rests on the complementary distribution of and , with the former occurring after hard (non- palatalized) consonants and elsewhere. The allophony of the stressed variant of the open is largely the same, yet no scholar considers and to be separate phonemes (which they are in e.g. Slovak).

The six-vowel view, held by the Saint-Petersburg (Leningrad) phonology school, points to several phenomena to make its case:
* Native Russian speakers' ability to articulate in isolation: for example, in the names of the letters and .
* Rare instances of word-initial , including the minimal pair 'to produce the sound ' and 'to produce the sound ы', as well as borrowed names and toponyms, like , the name of a river and several villages in the Komi Republic.
* Morphological alternations like ('ready' predicate, m.) and ('to get ready' trans.) between palatalized and non-palatalized consonants.

The most popular view among linguists (and the one taken up in this article) is that of the Moscow school, though Russian pedagogy has typically taught that there are six ''vowels'' (the term ''phoneme'' is not used).

Reconstructions of Proto-Slavic show that and (which correspond to and ) were separate phonemes. On the other hand, numerous alternations between the two sounds in Russian indicate clearly that at one point the two sounds were reanalyzed as allophones of each other.

Allophony

Russian vowels are subject to considerable allophony, subject to both stress and the palatalization of neighboring consonants. In most unstressed positions, in fact, only three phonemes are distinguished after hard consonants, and only two after soft consonants. Unstressed and have merged to (a phenomenon known as rus, а́канье, ákan'je); unstressed and have merged to ( rus, и́канье, íkan'je); and all four unstressed vowels have merged after soft consonants, except in the absolute final position in a word. None of these mergers are represented in writing.

Front vowels

When a preceding consonant is ''hard'', is retracted to . Formant studies in demonstrate that is better characterized as slightly diphthongized from the velarization of the preceding consonant, implying that a phonological pattern of using velarization to enhance perceptual distinctiveness between hard and soft consonants is strongest before . When unstressed, becomes near-close; that is, following a hard consonant and in most other environments. Between soft consonants, stressed is raised, as in ('to drink'). When preceded ''and'' followed by coronal or dorsal consonants, is fronted to . After a cluster of a labial and , is retracted, as in ('to float'); it is also slightly diphthongized to .

In native words, only follows unpaired (i.e. the retroflexes and ) and soft consonants. After soft consonants (but not before), it is a mid vowel (hereafter represented without the diacritic for simplicity), while a following soft consonant raises it to close-mid . Another allophone, an open-mid , occurs word-initially and between hard consonants. Preceding hard consonants retract to and so that ('gesture') and ('target') are pronounced and respectively.

In words borrowed from other languages, often follows hard consonants; this foreign pronunciation usually persists in Russian for many years until the word is more fully adopted into Russian. For instance, (from French ''chauffeur'') was pronounced in the early twentieth century, but is now pronounced . On the other hand, the pronunciations of words such as ('hotel') retain the hard consonants despite a long presence in the language.

Back vowels

Between soft consonants, becomes , as in ('five'). When not following a soft consonant, is retracted to before /ɫ/ as in ('stick').

For most speakers, is a mid vowel , but it can be a more open for some speakers. Following a soft consonant, is centralized and raised to as in ('aunt').

As with the other back vowels, is centralized to between soft consonants, as in ('narrowly'). When unstressed, becomes near-close; central between soft consonants, centralized back in other positions.

Unstressed vowels

Russian unstressed vowels have lower intensity and lower energy. They are typically shorter than stressed vowels, and in most unstressed positions tend to undergo mergers for most dialects:
* has merged with : for instance, 'bulwarks' and 'oxen' are both pronounced , phonetically .
* has merged with : for instance, (''lisá'') 'fox' and 'forests' are both pronounced , phonetically .
* and have merged with after soft consonants: for instance, (''mésjats'') 'month' is pronounced , phonetically .

The merger of unstressed and in particular is less universal in the pretonic (pre-accented) position than that of unstressed and . For example, speakers of some rural dialects as well as the "Old Petersburgian" pronunciation may have the latter but not the former merger, distinguishing between and , but not between and (both ). The distinction in some ''loanwords'' between unstressed and , or and is codified in some pronunciation dictionaries (, ), for example, and .

Unstressed vowels (except ) are preserved word-finally, for example in second-person plural or formal verb forms with the ending , such as ("you do") (phonetically ). The same applies for vowels starting a word.

As a result, in most unstressed positions, only three vowel phonemes are distinguished after hard consonants (, , and ), and only two after soft consonants ( and ). For the most part, Russian orthography (as opposed to that of the closely related Belarusian) does ''not'' reflect vowel reduction. This can be seen in Russian (''nébo'') as opposed to Belarusian (''néba'') "sky", both of which can be phonemically analyzed as and morphophonemically as , as the nominative singular ending of neuter nouns is when stressed: compare Russian , Belarusian "village".

= Vowel mergers

=
In terms of actual pronunciation, there are at least two different levels of vowel reduction: vowels are less reduced when a syllable immediately precedes the stressed one, and more reduced in other positions. This is particularly visible in the realization of unstressed and , where a less-reduced allophone appears alongside a more-reduced allophone .

The pronunciation of unstressed is as follows:

# (sometimes transcribed as ; the latter is phonetically correct for the standard Moscow pronunciation, whereas the former is phonetically correct for the standard Saint Petersburg pronunciation; this article uses only the symbol ) appears in the following positions:
#* In the syllable immediately before the stress, when a hard consonant precedes: ('ferry'), ('grass').
#* In absolute word-initial position.
#* In hiatus, when the vowel occurs twice without a consonant between; this is written , , , or : ('to use common sense, to reason').
# appears elsewhere, when a hard consonant precedes: ('cloud').
#* In absolute word-final position, may occur instead, especially at the end of a syntagma.
# When a soft consonant or precedes, both and merge with and are pronounced as . Example: 'tongue'). is written as in these positions.
#* This merger also tends to occur after formerly soft consonants now pronounced hard (, , ), where the pronunciation occurs. This always occurs when the spelling uses the soft vowel variants, e.g. ('wife'), with underlying . However, it also occurs in a few word roots where the spelling writes a hard . Examples:
#** 'regret': e.g. ('to regret'), к сожале́нию ('unfortunately').
#** 'horse', e.g. , (pl. gen. and acc.).
#** in numbers: e.g. ('twenty en., dat., prep.), ('thirty nstr.).
#** ('rye dj. m. nom.).
#** ('jasmine').
#These processes occur even across word boundaries as in ('under the sea').

The pronunciation of unstressed is after soft consonants and , and word-initially ( ('stage')), but after hard consonants ( ('to breathe')).

There are a number of exceptions to the above vowel-reduction rules:
* Vowels may not merge in foreign borrowings, particularly with unusual or recently borrowed words such as , 'radio'. In such words, unstressed may be pronounced as , regardless of context; unstressed does not merge with in initial position or after vowels, so word pairs like and , or and , differ in pronunciation.
*Across certain word-final inflections, the reductions do not completely apply. For example, after soft or unpaired consonants, unstressed , and of a final syllable may be distinguished from each other. For example, ('residents') contrasts with both ('bouta resident') and ('(of) a resident'). Also, ('he goes') and ('they go').
*If the vowel belongs to the conjunctions ('but') or ('then'), it is not reduced, even when unstressed.

= Other changes

=
Unstressed is generally pronounced as a lax (or near-close) , e.g. ('man'). Between soft consonants, it becomes centralized to , as in ('to huddle').

Note a spelling irregularity in of the reflexive suffix : with a preceding in third-person present and a in infinitive, it is pronounced as , i.e. hard instead of with its soft counterpart, since , normally spelled with , is traditionally always hard. In other forms both pronunciations and alternate for a speaker with some usual form-dependent preferences: in the outdated dialects, reflexive imperative verbs (such as , lit. "be afraid yourself") may be pronounced with instead of modern (and phonetically consistent) .

In weakly stressed positions, vowels may become voiceless between two voiceless consonants: ('exhibition'), ('because'). This may also happen in cases where only the following consonant is voiceless: ('skull').

= Phonemic analysis

=
Because of mergers of different phonemes in unstressed position, the assignment of a particular phone to a phoneme requires phonological analysis. There have been different approaches to this problem:

*The Saint Petersburg phonology school assigns allophones to particular phonemes. For example, any is considered as a realization of .
*The Moscow phonology school uses an analysis with morphophonemes (, singular ). It treats a given unstressed allophone as belonging to a particular morphophoneme depending on morphological alternations, or on etymology (which is often reflected in the spelling). For example, is analyzed as either or . To make a determination, one must seek out instances where an unstressed morpheme containing in one word is stressed in another word. Thus, because the word ('shafts') shows an alternation with ('shaft'), this instance of belongs to the morphophoneme . Meanwhile, ('oxen') alternates with ('ox'), showing that this instance of belongs to the morphophoneme . If there are no alternations between stressed and unstressed syllables for a particular morpheme, then no assignment is made, and existence of a hyperphoneme is postulated. For example, the word ('dog') is analysed as , where is a hyperphoneme.
*Some linguists prefer to avoid making the decision. Their terminology includes strong vowel phonemes (the five) for stressed vowels plus several weak phonemes for unstressed vowels: thus, represents the weak phoneme , which contrasts with other weak phonemes, but not with strong ones.

Diphthongs

Russian diphthongs all end in a non-syllabic , an allophone of and the only semivowel in Russian. In all contexts other than after a vowel, is considered an approximant consonant. Phonological descriptions of may also classify it as a consonant even in the coda. In such descriptions, Russian has no diphthongs.

The first part of diphthongs are subject to the same allophony as their constituent vowels. Examples of words with diphthongs: ('egg'), ('her' dat.), ('effective'). , written or , is a common inflexional affix of adjectives, participles, and nouns, where it is often unstressed; at normal conversational speed, such unstressed endings may be monophthongized to .

Consonants

denotes palatalization, meaning the center of the tongue is raised during and after the articulation of the consonant. Phonemes that have at different times been disputed are enclosed in parentheses.

; Notes
* Most consonant phonemes come in hard–soft pairs, except for always-hard and always-soft and formerly or marginally . There is a marked tendency of Russian hard consonants to be velarized or uvularized, though this is a subject of some academic dispute.: "Note that though Russian has traditionally been described as having all consonants either palatalized or velarized, recent data suggests that the velarized gesture is only used with laterals giving a phonemic contrast between and (...)." Velarization is clearest before the front vowels and , and with labial and velar consonants as well as the lateral. As with palatalization, it results in vowel colouring and diphthongisation when stressed, in particular with , realized approximately as or . Its function is to make the contrast between hard and soft consonants perceptually more salient, and the less salient the contrast is otherwise (such as labial consonants being universally the most resistant to palatalization), the higher the velarization degree.
** and are always hard in native words (even if spelling contains a "softening" letter after them, as in , , , and ). A few loanwords are spelled with or ; authoritative pronunciation dictionaries prescribe hard pronunciation for some of them (e.g. , , , ) but soft for other ones (e.g. , ); may be pronounced either way. The letter combinations , , , , , and also occur in foreign proper names, mostly of French or Lithuanian origin. Notable examples include ( Gölcük, Kocaeli), (Jeune Afrique), (Jules Verne), (Gerhard Schürer), (Šiauliai), and (Šešuvis). The dictionary of prescribes soft pronunciation in these names. However, since the cases of soft and are marginal and not universally pronounced as such, and are generally considered always-hard consonants, and the long phonemes and are not considered their soft counterparts, as they do not pattern in the same ways that other hard–soft pairs do.
** is generally listed among the always-hard consonants; however, certain foreign proper names, including those of Ukrainian, Polish, Lithuanian, or German origin (e.g. , , , ), as well as loanwords (e.g., , from Chinese) contain a soft . The phonemicity of a soft is supported by neologisms that come from native word-building processes (e.g. , ). However, according to , really is always hard, and realizing it as palatalized is considered "emphatically non-standard", and occurs only in some regional accents.
** and are always soft.
** is also always soft. A formerly common pronunciation of indicates the sound may be two underlying phonemes: and , thus can be considered as a marginal phoneme. In today's most widespread pronunciation, appears (instead of ) for orthographical where starts the root of a word, and -з/-с belongs to a preposition or a "clearly distinguishable" prefix (e.g. , 'without a clock'; , 'to rule'); in all other cases is used ( , , , , , , etc.)
** The marginally phonemic sound is largely obsolete except in the more conservative standard accent of Moscow, in which it only occurs in a handful of words. Insofar as this soft pronunciation is lost, the corresponding hard replaces it. This sound may derive from an underlying or : , modern . For most speakers, it can most commonly be formed by assimilative voicing of (including across words): . For more information, see alveolo-palatal consonant and retroflex consonant.
* and are somewhat concave apical postalveolar. They may be described as retroflex, e.g. by , but this is to indicate that they are not laminal nor palatalized; not to say that they are subapical. They also tend to be at least slightly labialized, including when followed by unrounded vowels.
* Hard are laminal denti-alveolar ; unlike in many other languages, does not become velar before velar consonants.
* Hard has been variously described as pharyngealized apical alveolar , cited in and velarized laminal denti-alveolar .
* Hard is postalveolar, typically a trill .; cited in
* Soft is an apical dental trill , usually with only a single contact.
* Soft are laminal alveolar . In the case of the first two, the tongue is raised just enough to produce slight frication as indicated in the transcription.
* Soft is either laminal alveolar or laminal denti-alveolar .
* are dental , i.e. dentalized laminal alveolar. They are pronounced with the blade of the tongue very close to the upper front teeth, with the tip of the tongue resting behind the lower front teeth.
* The voiced are often realized with weak friction or even as approximants , particularly in spontaneous speech.
* A marginal phoneme occurs instead of in certain interjections: , , , , , , . (Thus, there exists a minimal pair of homographs: 'aha!' vs ' agha'). The same sound can be found in (spelled , though in , is ), optionally in and in a few other loanwords. Also optionally (and less frequently than a century ago) can be used instead of in certain religious words (a phenomenon influenced by Church Slavonic pronunciation): , ... (declension forms of 'God'), 'Lord' (especially in the exclamation 'Oh Lord!'), 'good'.
* Some linguists (like I. G. Dobrodomov and his school) postulate the existence of a phonemic glottal stop . This marginal phoneme can be found, for example, in the word . Claimed minimal pairs for this phoneme include 'narrowed' (a participle from 'to narrow', with prefix and root , cf. 'narrow') vs 'betrothed' (originally a participle from 'to judge', now an adjective; the root is 'court') and 'with Ann' vs '(by) Alex'.

There is some dispute over the phonemicity of soft velar consonants. Typically, the soft–hard distinction is allophonic for velar consonants: they become soft before front vowels, as in ('short'), unless there is a word boundary, in which case they are hard (e.g. 'to Ivan'). Hard variants occur everywhere else. Exceptions are represented mostly by:
* Loanwords:
** Soft: , , , , , , , ;
** Hard: , , , (), .
* Proper nouns of foreign origin:
** Soft: , , , , , , , , , , , ;
** Hard: , , , , , .

The rare native examples are fairly new, as most them were coined in the last century:
* Soft: forms of the verb 'weave' (, etc., and derivatives like ); /, /; and adverbial participles of the type , , , , , , (it is disputed whether these are part of the standard language or just informal colloquialisms);
* Hard: the name of letter , acronyms and derived words (, ), a few interjections (), some onomatopoeic words (), and colloquial forms of certain patronyms: , , (where is a contraction of standard language's patronymical suffix -ович rather than a continuation of ancient ).

In the mid-twentieth century, a small number of reductionist approaches made by structuralists put forth that palatalized consonants occur as the result of phonological processes involving (or palatalization as a phoneme in itself), so that there were no underlying palatalized consonants. Despite such proposals, linguists have long agreed that the underlying structure of Russian is closer to that of its acoustic properties, namely that soft consonants are separate phonemes in their own right.

Voicing

Final devoicing

Voiced consonants (, and ) are devoiced word-finally unless the next word begins with a voiced obstruent. In other words, their voiceless equivalent will be used (see table on the right).Russian language course "Russo Sem Mestre" (Portuguese for ''Russian without Master''), by Custódio Gomes Sobrinho

Examples:
* (story, tale) sounds like расскас
* (knife) sounds like нош
* (Ivanov) sounds like Иваноф ; and so on.

also represents voiceless word-finally in some words, such as ('god'). This is related to the use of the marginal (or dialectal) phoneme in some religious words .

Voicing elsewhere

Basically, when a voiced consonant comes before an voiceless one, its sound will shift to its voiceless equivalent (see table).

* Example: (spoon) sounds like Лошка .

That happens because ж is a voiced consonant, and it comes before the voiceless к.

The same logic applies when a voiceless consonant comes before a voiced one (except в). In this case, the sound of the former will change to its voiced equivalent.

* Example: (to do) sounds like зделать �zʲdʲeɫətʲ

Russian features general regressive assimilation of voicing and palatalization. In longer clusters, this means that multiple consonants may be soft despite their underlyingly (and orthographically) being hard. The process of voicing assimilation applies across word-boundaries when there is no pause between words.
Within a morpheme, voicing is not distinctive before obstruents (except for , and when followed by a vowel or sonorant). The voicing or devoicing is determined by that of the final obstruent in the sequence: ('request'), ('vodka'). In foreign borrowings, this isn't always the case for , as in ('Adolf Hitler') and ('the count is ill'). and are unusual in that they seem transparent to voicing assimilation; in the syllable onset, both voiced and voiceless consonants may appear before :
* ) ('the creature')
* ('two')
* ('of light')
* ('star')

When precedes and follows obstruents, the voicing of the cluster is governed by that of the final segment (per the rule above) so that voiceless obstruents that precede are voiced if is followed by a voiced obstruent (e.g. 'to the widow') while a voiceless obstruent will devoice all segments (e.g. 'without an admission').

, , and have voiced allophones (, and ) before voiced obstruents, as in ('a daughter would'), ('bridge-head') and ('peas are ready').

Other than and , nasals and liquids devoice between voiceless consonants or a voiceless consonant and a pause: ) ('buttress').

Palatalization

Before , paired consonants (that is, those that come in a hard-soft pair) are normally soft as in ('I drink') and ('I hit'). However, the last consonant of prefixes and parts of compound words generally remains hard in the standard language: ('departure'), (' Min stry_ofJust[ice.html"_;"title="ce.html"_;"title="stry_ofJust[ice">stry_ofJust[ice">ce.html"_;"title="stry_ofJust[ice">stry_ofJust[ice);_when_the_prefix_ends_in__or__there_may_be_an_optional_softening:___('to_travel').

Paired_consonants_preceding__are_also_soft;_although_there_are_exceptions_from_loanwords,_alternations_across_morpheme_boundaries_are_the_norm._The_following_examples_show_some_of_the_morphological_alternations_between_a_hard_consonant_and_its_soft_counterpart:

Velar_consonants_are_soft_when_preceding_,_and_never_occur_before__within_a_word.

Before_hard_dental_consonants,_,_labial_and_dental_consonants_are_hard:___('eagle'_Genitive_case.html" "title="ce">stry_ofJust[ice.html" ;"title="ce.html" ;"title="stry ofJust[ice">stry ofJust[ice">ce.html" ;"title="stry ofJust[ice">stry ofJust[ice); when the prefix ends in or there may be an optional softening: ('to travel').

Paired consonants preceding are also soft; although there are exceptions from loanwords, alternations across morpheme boundaries are the norm. The following examples show some of the morphological alternations between a hard consonant and its soft counterpart:

Velar consonants are soft when preceding , and never occur before within a word.

Before hard dental consonants, , labial and dental consonants are hard: ('eagle' Genitive case">gen.