Consonants

The system of consonants in Russian is characterised by the phonological opposition of palatalised (‘soft’) and non-palatalised (‘hard’) consonants that encompasses almost all consonants, with very few exceptions. Thus, /ʃ ʒ / have no palatalised counterparts, while / / and /ʃʲː/ have no non-palatalised counterparts. All non-palatalised consonants are realised with velarisation (e.g. Bolla Reference Bolla1981) which is particularly noticeable in /l/ [ɫ] and /ʃ ʒ/ [ʃˠ ʒˠ]. The table below shows the consonant phonemes of Russian. Only the palatalisation of consonants is marked in transcription.

Note that in the examples above the consonants are represented before non-front vowels. Only palatalised consonants and /j/ occur before [i], and in indigenous words only palatalised consonants and the non-paired consonants / ʃ ʒ j/ occur before /e/, whereas in loanwords non-palatalised consonants can occur before /e/, e.g. test [ˈ ɛˑs ] ‘test’, tire [ t ʲɪˈrɛˑ] ‘dash’.

Non-palatalised dental consonants are laminal denti-alveolar, velarised. The affricate [ ] has no palatalised counterpart in the system of consonants, and its palatalisation, although evident in some regional accents of Russian, is considered emphatically non-standard.

The realisation of the palatalised (‘soft’) consonants involves the secondary articulation of palatalisation in its purest form (the rising of the front of the tongue to the hard palate) only in bilabials and labiodentals. In other consonants, palatalisation is accompanied by further articulatory adjustments that affect both place and manner of articulation (Bondarko Reference Bondarko1998, Reference Bondarko2005). For instance, the point of constriction of /tʲ/ and /dʲ/ is retracted compared to non-palatalised laminal denti-alveolar /t/ and /d/ and they are normally affricated [ t ^sʲ] [ d ^zʲ]; /r/ is an alveolar trill in careful pronunciation, but its palatalised counterpart /rʲ/ is usually realised as a tap [ɾʲ]. In the palatalised counterparts of velar /k ɡ x/ the point of constriction is fronted so that they are realised as post-palatal [ ] (see Keating & Lahiri Reference Keating and Lahiri1993). Note that /kʲ ɡʲ xʲ/, while common in combination with front vowels (e.g. kislo [ˈkʲiˑs ə] ‘sour’, girja [ˈɡʲiˑrʲɪ] ‘weight’, xitryj [ˈxʲiˑ rɨ ] ‘cunning’; kepka [ˈkʲɛˑpk ] ‘cap’, gercog [ˈɡʲɛˑr k] ‘duke’, sxema [ˈsxʲɛˑmə] ‘scheme’) are rare before non-front vowels and occur in this position mainly in loanwords and foreign names, e.g. Gëte [ˈɡʲɔˑ ɛ] ‘Goethe’, Kjaxta [ˈkʲaˑx ə] ‘Kyakhta’ (a town in Buryatia, Russia), and a single indigenous verb: tkët [ˈ kʲɔˑ ] ‘(he) weaves’.

The combinations of non-palatalised velars and the /ɨ/ vowel /kɨ ɡɨ xɨ/ are rare and found only in a handful of loanwords and across word boundaries, e.g. kyš [ˈkɨʃ] ‘shoo’ (interjection), Arxyz [ʌrˈxɨs] ‘Arkhyz’ (a territory in Karachay-Cherkessia); k Igor’u [ˈk‿ɨˑɡərʲu] ‘towards Igor’, dvuxėtažnyj [ˌ vuxɨˈ aˑʒ ɨ ] ‘two-storeyed’.

Voicing is used contrastively in Russian; voiced consonants are fully voiced, voiceless plosives are always unaspirated, e.g. tok [ˈ ^ʊɔˑk] ‘current’, kot [ˈk ^ʊɔˑ ] ‘tomcat’. The distribution of consonants is such that only voiceless but no voiced obstruents occur word-finally, e.g. goda [ɡʌˈ aˑ] ‘years’, god [ˈɡ^ʊɔˑ ] ‘year’.

In sequences of consonants, both within words and across word boundaries, various kinds of regressive assimilation take place. For example, if the second consonant is a voiced obstruent (other than /v vʲ/), the preceding consonant is also voiced, e.g. gorod [ˈɡ^ʊɔˑrə ] ‘city’ but gorod bol’šoj [ˈɡ^ʊɔˑrə ‿bʌlʲˈʃoˑ ] ‘(the) city is big’ (Verbickaja Reference Verbickaja2001). Under certain conditions, assimilation can also affect palatalisation or even the whole place and/or manner of articulation, e.g. bandit [bʌnʲˈ d ʲiˑ ] ‘bandit’, bez šuma [bʲɪˈʃːuˑmə] ‘without a noise’. In such cases we can also find sounds that otherwise represent gaps in the phoneme inventory, e.g. [ɣ] as a voiced allophone of /x/ in mox zelënyj [ˈm ^ʊɔˑɣ‿zʲɪˈlʲɔˑ ɨ ] ‘the moss (is) green’, [ɣʲ] as a voiced and palatalised allophone of /x/ in drugix gimnazij [ ruˈɡʲiˑɣʲ‿ɡʲɪmˈ aˑzʲɪ ] ‘of other grammar schools’, [ ] as a voiced allophone of / / in otec doma [ʌˈ t ʲɛˑ ‿ˈ ^ʊɔˑmə] ‘father is at home’, [ʃʲ] as an allophone of /s/ in s čaem [ʃʲˈ ʲæˑ ɪm] ‘with tea’ (Kasatkin Reference Kasatkin2006: 44), [ ʲ] as a voiced allophone of / / in doč bol’na [ˈ ^ʊɔˑ ʲ‿bʌlʲˈ aˑ] ‘(the) daughter is ill’. Sonorants can be realised as devoiced when word-initial and word‑final in the vicinity of voiceless obstruents, e.g. teatr [ t ʲɪˈaˑ ] ‘theatre’.

Labiodental fricatives /v/ and /vʲ/ are often weakly articulated [ ʲ] or are realised as approximants [ʋ ʋʲ], particularly in spontaneous speech. The palatal /j/ can be realised as an approximant [j] (especially in the onset of a stressed syllable), a semivowel [ ] (especially when unstressed), or emphatically as a fricative [ʝ] or even a devoiced fricative [ç].

Fricatives /ʃ ʒ/, as in šar /ˈʃar/ ‘ball’ and žar /ˈʒar/ ‘heat’, can be realised either as flat velarised postalveolars [ʃˠ ʒˠ] or as retroflexes [ʂ ʐ] (Ladefoged & Maddieson Reference Ladefoged and Maddieson1996, Hamann Reference Hamann2004) and tend to be slightly labialised even in the context of unrounded vowels. They have no palatalised counterparts in the system of consonant phonemes; their palatalisation is considered non-standard. This also precludes their assimilation to a following palatalised consonant, e.g. roždenie [rʌʒˈ d ʲeˑnʲɪ ɪ] ‘birth’, bašnja [ˈb ˑʃnʲɪ] ‘tower’. The long fricative /ʃʲː/, as in ščuka [ˈʃʲːuˑkə] ‘pike’, sčast’e [ˈʃʲːæˑsʲ t ʲjɪ] ‘happiness’, is a laminal palatalised post-alveolar (or alternatively, an alveolo-palatal [ɕː]). (The former bisegmental pronunciation of /ʃʲː/ as [ʃʲ ], often cited as a characteristic feature of the older St. Petersburg norm (e.g. Jones & Ward Reference Jones and Ward1969), is now clearly obsolete (see Comrie et al. Reference Comrie, Stone and Polinsky1996, Verbickaja Reference Verbickaja2001, Timberlake Reference Timberlake2004).) The /ʃʲː/ consonant has no voiced counterpart in the system of phonemes. However, in conservative Moscow standard and only in a handful of lexical items the combination /ʒʒ/ may be pronounced with palatalisation, e.g. drožži ‘yeast’ as [ˈ r ^ʊ oˑʒʲːɪ] instead of [ˈ r ^ʊɔˑʒːɨ], although this realisation is now also somewhat obsolete.

Long consonants are found as realisations of biphonemic sequences particularly across morpheme boundaries, e.g. otdel [ʌˈ d ʲːɛˑ ] ‘department’ and rassada [rʌˈsːaˑ ə] ‘seeding’, and also in foreign words, e.g. massa [ˈmaˑsːə] ‘mass’, kolonna [kʌˈ ^ʊɔˑ ːə] ‘pillar’ (where the current general tendency appears to be for the Russian speakers to shorten them, see Cubberley Reference Cubberley2002).

Clusters of three or more underlying consonants are often simplified, e.g. pozdno [ˈp ^ʊɔˑz ə] ‘late’, peterburgskij [pʲɪ t ʲɪrˈbuˑrskʲɪ ] ‘of St Petersburg’. Consonants and consonant clusters before /o/ and /u/ are labialised, e.g. stul [ˈsʷ ʷuˑ ] ‘chair’.

Vowels

Russian has six vowels, /i ɨ e a o u/ (the above chart is based on Bondarko Reference Bondarko1998). Vowel quality varies substantially depending on whether the vowel occurs in stressed or in unstressed syllables: in unstressed syllables, all vowels are subject to reduction. Furthermore, the realisation of the vowels varies as a function of consonantal context: vowels are more fronted after or before and particularly between palatalised consonants than when surrounded by non-palatalised consonants. Thus, for example, the /a/ vowel gets progressively more front in CVCʲ, CʲVC, and CʲVCʲ contexts relative to CVC context. When adjacent to only one palatalised consonant (CVCʲ or CʲVC), it is a diphthongoid because it accommodates both to the velarisation and to the palatalisation of the adjacent consonants, e.g. sad [ˈs ˑ ] ‘garden’, brosat’, [brʌˈsaˑ^ɪ t ʲ] ‘to throw’, vprisjadku [fpɾʲɪˈsʲ^ɪ a ˑ ku] ‘in squatting position’, sjad’ [ˈsʲæˑ t ʲ] ‘sit down!’.

There are conflicting views on the phonological status of the [i] and [ɨ] vowels in Russian linguistics. As the two vowels [i] and [ɨ] are in near-complementary distribution, with [i] occurring after palatalised consonants and [ɨ] after non-palatalised consonants, they may be seen as one phoneme /i/ only, having two allophones [i] and [ɨ] (Avanesov Reference Avanesov1972, Reference Avanesov1974; Cubberley Reference Cubberley2002) or (as they are treated here) as two separate phonemes (Halle Reference Halle1959, Plapp Reference Plapp1996, Bondarko Reference Bondarko1998, Verbickaja Reference Verbickaja2001), see also discussion in Bernštejn (Reference Bernštejn and Leont’ev1996), Cubberley (Reference Cubberley2002) and Timberlake (Reference Timberlake2004). Both vowels can be realised word-initially in identical context, e.g. in the letter names i [ˈi] for и vs. y [ˈɨ] for ы or in the dialectological terms ikan’e [ˈiˑkənʲjɪ] ‘merger of unstressed /e/ and /i/ after palatalised consonants’ vs. ykan’e [ˈɨˑkənʲjɪ] ‘merger of unstressed /a/ and /ɨ/ after /ʃ ʒ /’. The /ɨ/ vowel tends to be diphthongised, with a glide towards a more front close vowel, particularly when word-final, e.g. my [ˈmˠɨⁱ] ‘we’, sady [sʌˈ ˠɨⁱ] ‘gardens’.

The /a/ vowel is an open central or back-advanced [ ] in the context of non-palatalised consonants and gets markedly fronted to [æ] between palatalised consonants, e.g. palka [ˈp ˑ kə] ‘stick’, pjal’cy [ˈpʲæˑlʲ ɨ] ‘embroidery hoop’. Similarly, the /e/ vowel is more retracted and centralised in the context of the non-palatalised consonants, e.g. šest [ˈʃ ˑs ] ‘pole’, and is realised as front in the context of the palatalised consonants, where it is also more close, e.g. čest’ [ˈ ˑsʲ t ʲ] ‘honour’.

The /o/ vowel is a diphthongoid, with a closer lip rounding at the beginning of the vowel that gets progressively weaker [^ʊ o] or even [^ʊɔ^ʌ], particularly when occurring word-initially or word-finally under the stress, e.g. očen’ [ˈ^ʊ oˑ ɪnʲ] ‘very’, okna [ˈ^ʊɔ^ʌ k ə] ‘windows’, moloko [mə ʌˈk ^ʊɔ^ʌ] ‘milk’.

In standard pronunciation, /e/ and /o/ do not occur in unstressed syllables; /e/ is replaced with /i/ or /ɨ/, and /o/ is replaced with /a/ (with exceptions in only a handful of loanwords, e.g. radio [ˈr ˑ d ʲɪɔ] ‘radio’, ėmbolija [ɛmbʌˈlʲiˑ ɪ] ‘embolism’). Vowels in unstressed syllables are subject to reduction. Generally, there are two degrees of vowel reduction, depending on the location of the vowel relative to the stressed syllable (see Cubberley Reference Cubberley2002: 68). The first-degree reduction is realised in the syllable immediately before the stressed syllable and when the word begins with the unstressed vowel. It is also found (variably) in phrase-final open syllables. The second-degree reduction applies to all other unstressed syllables. This is most striking for the /a/ vowel, which is realised as [ʌ] or [ɐ] in the first degree of reduction (the former is characteristic of St. Petersburg and the latter of Moscow pronunciation, see Kasatkina Reference Kasatkina2005), and as [ə] in the second, e.g. moloko [mə ʌˈk ^ʊɔ^ʌ] ‘milk’, katastrofa [k ʌˈstr ^ʊɔˑfə] ‘catastrophe’. Unstressed /a/ after palatalised consonants merges with /i/ and is realised as [ ] or [ɪ], e.g. djadja [ˈ d ʲæˑ d ʲɪ] ‘uncle’, časy [ ɪˈsɨˑ] ‘clock’. The qualitative differences between the respective allophones of /i ɨ u/ in stressed and unstressed syllables are less perspicuous.

Furthermore, unstressed vowels in Russian tend to be shorter than stressed vowels (and the second-degree unstressed vowels are shorter than first-degree ones), e.g. govorit’ [ɡ vʌˈɾʲiˑ t ʲ] ‘to speak’, particularly under phrasal stress. (Outside that context lexically stressed vowels are not necessarily longer than unstressed ones, see Knjazev Reference Knjazev2006.)

Stress and intonation

The prominence of the stressed syllable in Russian is achieved primarily through the duration and quality of the stressed vowel; the vowels in the stressed syllables are full quality /i ɨ e a o u/ and usually half-long whereas the unstressed vowels (only /i ɨ a u/ are possible in this position) are subject to various degrees of qualitative and quantitative reduction (see above). The stress is free and can fall on any syllable in a word. In the majority of cases, the stress is stable, that is it falls on the same syllable in the word within its paradigm or in its derivatives, e.g. brat’ja [ˈbraˑ t ʲjɪ] ‘brothers’, brat’jami [ˈbraˑ t ʲjɪmʲɪ] ‘brothers (instrumental case)’, bratskij [ˈbr ˑ kʲɪ ] ‘brotherly’, bratstvo [ˈbr ˑ və] ‘brotherhood’. There is, however, a large number of common words where the stress moves within the word's paradigm or in derived forms, e.g. gorod [ˈɡ^ʊɔˑrə ] ‘city’ but goroda [ɡ rʌˈ aˑ] ‘cities’, gorodskoj [ɡ rʌˈ k ^ʊ oˑ ] ‘of (the) city (adj), urban’ (Bondarko Reference Bondarko1998).

There are several descriptions of Russian intonation. The classic is by Bryzgunova (Reference Bryzgunova1977), who impressionistically differentiates five basic ‘intonational contours’. Further descriptions include Odé (Reference Odé1989) and Svetozarova (Reference Svetozarova, Hirst and Di Cristo1998). Odé's (Reference Odé, Houtzagers, Kalsbeek and Schaeken2008) ToRI (Transcription of Russian Intonation) project is an attempt at a comprehensive description of Russian intonational phonology within the autosegmental-metrical framework. One of the main functions of Russian sentence intonation is to mark the information structure of a sentence. A conspicuous feature of Russian is that wh-questions have a falling contour similar to statements, and even yes/no-questions are not characterised by a final rise but rather a rise-fall (H*L) on the focally accented syllable.

Transcription of the recorded passage

In the transcriptions below, stressed syllables are marked, but intonation is not marked.