As a western Slavic language of the Indo-European family, Czech is closest to Slovak and Polish. It is spoken as a native language by nearly 10 million people in the Czech Republic (Czech Statistical Office n.d.). About two million people living abroad, mostly in the USA, Canada, Austria, Germany, Slovakia, and the UK, claim Czech heritage (Ministry of Foreign Affairs of the Czech Republic 2009). However, it is not known how many of them are native speakers of Czech.
Sociolinguistically, the language situation in the Czech Republic bears diglossic features. There is a substantial gap between formal, highly codified language and the language used in everyday situations. Our aim is to describe the way most people speak most of the time rather than artificial orthoepic norms (for the latter see Palková Reference Palková1997: 320–345).
Geographically, in the western part of the country (Bohemia and western Moravia) pronunciation is relatively homogeneous compared to the greater dialectal diversity of the east (the rest of Moravia), where several dialectal areas can be distinguished (Cvrček Reference Cvrček and Cvrček2010: 24). Still, there are a number of features common to Moravian dialects, distinguishing them clearly from the pronunciation of Bohemia and allowing native speakers to identify someone as either Bohemian or Moravian. The present paper elaborates on the earlier illustration of Czech (Dankovičová Reference Dankovičová1997a) by describing the differences between Bohemian Czech (BC), spoken by more than six million Czech citizens, and Moravian Czech (MC), spoken by about three-and-a-half million Czech citizens. At the same time, our illustration provides additional information about what both varieties have in common.
The transcriptions of the sample text are based on recordings of a 32-year-old male native speaker of BC from the south of Bohemia, and a 44-year-old female native speaker of MC from the east of Moravia. Examples given throughout the text are recorded by the Bohemian speaker and, where MC differs from BC, also by the speaker from Moravia.
Consonants
The consonant inventories of BC and MC are the same. Phonemes /
/, /ɡ/ and /f/ came into Czech with borrowings. The youngest and most easily identified as foreign is /
/, whereas /f/, as the oldest of the three, occurs in a few native Czech words, such as foukat /foukat/ ‘blow'and doufat /doufat/ ‘hope’. Phonetic implementation of consonant phonemes does not differ in BC and MC either. In the stop series, /bd ɡ/ are prevoiced (i.e. voiced during closure) and contrast with voiceless unaspirated /ptk/. While /d/ is realized as apico-alveolar, /t/ is more likely to be lamino-dental (Machač Reference Machač2006: 139–141). A stop before another stop is generally released. The release of palatal stops is affricated (Machač & Skarnitzl Reference Machač, Skarnitzl and Vích2004: 30–31). Intervocalically, /d/ is sometimes reduced to [ɾ] (Machač & Skarnitzl Reference Machač and Skarnitzl2009: 37–38), and /v/ becomes a labio-dental approximant [ʋ] (Skarnitzl & Volín Reference Skarnitzl and Volín2005). Sibilants /ʃ ʒ
/ are non-retroflex post-alveolars articulated with tongue blade rather than with the tongue tip (Żygis Reference Żygis, Hall and Hamann2003). Both /r/ and /
/ are trills though commonly realized with a single contact. Phonetically, the sound /
/ is a period of friction interrupted at the beginning by a contact or contacts created by a retracted apico-alveolar gesture (see Figure 1). The approximant /l/ is mainly pronounced apico-alveolar, although a velarized pronunciation without a firm tongue tip contact is not unusual (Šimáčková Reference Šimáčková, Kügler, Féry and van de Vijver2009). Sonorants /r/, /l/, and marginally also /m/ and /n/, become syllabic between two consonants or after a consonant at the end of a word, e.g. vlk [v
k] ‘wolf’, kopr [kop
] ‘dill’, Rožmberkové [roʒ
bɛrkovɛː] ‘House of Rosenberg’, sedmnáct [sɛd
naː
t] ‘seventeen’. Alveolar /n/ as well as bilabial /m/ commonly undergo place assimilation. Before labiodentals, /n/ and /m/ change into [ɱ], e.g. konference [koɱfɛrɛn
ɛ] ‘conference’, tramvaj [traɱvaj] ‘tram’. The alveolar nasal is realized as velar before /k/ and /g/. Word-internally only sequences [ŋk] and [ŋɡ] are permitted, e.g. tenká [tɛŋkaː] ‘thin (f)’, while across the word boundary place assimilation is optional, e.g. ten kout [tɛŋ kout] ~ [tɛnkout] ‘the corner’. Sequences /ns/, /nz/, /nʃ/ and /nʒ/ are optionally pronounced with an epenthetic stop, e.g. ženská [ʒɛntskaː] ‘female’, Honza [ɦondza] ‘Johnny’, zmenšovat [zmɛntʃovat] ‘make smaller’, manžel [mandʒɛl] ‘husband’. Plosives followed by homorganic nasals may have a nasal release, e.g. pocestný [po
ɛstnniː] ‘traveler’. In fast casual speech, sequences /dn/ and /dɲ/ may be rendered [n˺n] and [ɲ˺ɲ], e.g. jednou [jɛn˺nou] ‘once’, hodně /ɦodɲɛ/ [ɦoɲ˺ɲɛ] ‘a lot of’ (Volín Reference Volín and Cvrček2010: 53). Stop–fricative sequences, especially if voiceless, may be simplified to affricates, e.g. dětská [ɟɛ
kaː] ‘children's (adj, f)’, větší [vjɛ
iː] ‘bigger’.
Figure 1 A devoiced allophone of Czech /
/ in the BC speaker's production of the word kouř ‘smoke’.
Obstruent voicing
The key source of differences between pronouncing consonants in Bohemian and in Moravian Czech are voice assimilations. In addition, there is a difference in how voice assimilation interacts with word-final devoicing.
Both in BC and MC, obstruents devoice word-finally, e.g. závod [zaːvot] ‘race’, rez [rɛs] ‘rust’. In connected speech, BC and MC differ with respect to final obstruent devoicing when the next word begins in a vowel, e.g. závod aut ‘car race’. Speakers of BC insert a glottal stop before the initial vowel and devoice the word-final obstruent: [zaːvot ʔaut]. In MC the final obstruent becomes the onset of the following syllable and keeps its voicing [zaːvo.daut]. In both varieties, word-final devoicing usually does not occur if the following obstruent is voiced (compare bez dechu [bɛzdɛxu] ‘out of breath’ and bez tebe [bɛstɛbɛ] ‘without you’). This is because Czech prefers adjacent obstruents to agree in voicing.
Underlying mismatches in the voicing of adjacent obstruents are usually resolved by voice assimilation which is generally regressive. The voicing agreement is obligatory word-internally, thus prosba ‘plea’ is pronounced as [prozba] and lebka ‘skull’ as [lɛpka]. Across a word boundary, voicing of adjacent obstruents varies within and across speakers. In BC, regressive voicing assimilation in consonant clusters applies in fewer contexts than in MC. First, a BC exception to the regressive direction of voice assimilation is the cluster /sɦ/ pronounced as voiceless [sx] (e.g. shoda [sxoda] ‘agreement’, shnít [sxɲiːt] ‘rot’), although in a small number of words voiced [zɦ] occurs (e.g. shora [zɦora] ‘from above’, shluk [zɦluk] ‘cluster’). More notably, in MC but not in BC is word-boundary regressive assimilation triggered by sonorants as well as obstruents, yielding MC pronunciations such as k lesu [ɡ lɛsu] ‘to the forest’.
In this context, it is important to discuss the classification of sounds /v/ and // as sonorant or obstruent. The labiodental ‘fricative’ /v/ behaves as an obstruent in that it undergoes voice assimilation (e.g. in vplout [fplout] ‘sail into’, krev teče [krɛf tɛɛ] ‘blood flows’) but at the same time it has some properties of sonorants. First, like other sonorants it does not trigger voice assimilation within a word, e.g. tvar [tvar] ‘shape’, sval [sval] ‘muscle’. Second, in MC always, but in BC only about half of the time, /v/ triggers voice assimilation across a word boundary, e.g. napsat větu [napsatvjɛtu] ~ [napsadvjɛtu] ‘write a sentence’ (Volín & Skarnitzl Reference Volín, Skarnitzl, Palková and Janoušková2006). And third, in both MC and BC, it is phonetically often weakened to the approximant [ʋ] (Skarnitzl & Volín Reference Skarnitzl and Volín2005). The sound // is usually paired off with the alveolar trill /r/ in tables of Czech consonant phonemes. However, it is in fact an obstruent. Phonetically it is a trilled fricative (see above), phonologically, it does not behave as a sonorant in that it cannot occupy the position of a syllable nucleus, and unlike /r/ and other sonorants which do not devoice contextually, it loses voicing word-finally and when it is adjacent to a voiceless obstruent, e.g. kouř [kou
] ‘smoke’, dvířka [dviːka] ‘little door’, přes [pɛs] ‘over’. In addition, // triggers voicing agreement across a word boundary in both Czech varieties (e.g. až řekneš ‘when you say’ pronounced as [aʒ ɛknɛʃ] and jak řekneš ‘how you say’ as [jaɡ ɛknɛʃ]), although in BC only obstruents do that.
Two obstruent phonemes lack an opposite voicing counterpart at the same place of articulation – the voiceless velar fricative /x/ and the voiced glottal fricative /ɦ/. The two fricatives are connected through processes of final devoicing and voice assimilation. In devoicing contexts /ɦ/ is substituted by [x], e.g. in sníh taje [sɲiːxtajɛ] ‘snow is melting’, lehká [lɛxkaː] ‘light (f)’. In voicing contexts, /x/ has the voiced allophone [ɣ], e.g. prach země /praxzɛmɲɛ/ [praɣ zɛmɲɛ] ‘dust of the earth’, but there are speakers who pronounce [ɦ] here instead, i.e. [praɦ zɛmɲɛ]. Another allophone arising from voicing assimilation is [
] substituted for //, e.g. moc dobrá /modobraː/ [mo dobraː] ‘very good (f)’.
Vowels
The Czech vowel inventory contains ten monophthongs and three diphthongs. The monophthongal vowel system has been described as consisting of five different vowel qualities /i ɛ aou/ occurring in two quantities, i.e. as short and long vowels. More recently, it has been accepted (see e.g. Dankovičová Reference Dankovičová1997a) that the short–long distinction in the high front vowel pair is not realized on the basis of duration only but entails a qualitative difference as well. This is especially clear in BC where the short counterpart of the ‘long’ high front vowel is realized as mid-high [ɪ]. It can be seen in Figure 3 that for BC speakers the two (mid-)high front vowels differ in their F1 and F2 values to a much larger extent than they do for MC speakers. As a consequence, BC speakers rely less on the durational difference between the short and the long high front vowel. Podlipský et al. (Reference Podlipský, Skarnitzl and Volín2009) show that the long high front vowel is only about 1.3 times longer than its short counterpart, while the long:short ratio is about 1.7 for the other four vowel pairs. This is why in this illustration the high front vowels are transcribed as [iˑ ɪ] for BC. In MC, the high front vowels are spectrally more similar, the primary difference between them being duration, hence the transcription [iː i]. The front mid vowels in both varieties are realized as mid-low, i.e. as [ɛ] and [ɛː], while the corresponding mid-back vowels /o/ and /oː/ remain mid, resulting in less vowel height symmetry between front and back vowels (see Figures 2 and 3).
Figure 2 Bohemian Czech (left) and Moravian Czech (right) monophthongs in the IPA chart.
Figure 3 F1 and F2 values of Bohemian (left) and Moravian (right) monophthongs produced by three speakers per gender per variety. The vowels were embedded in a number of different consonantal contexts in monosyllabic Czech words. Larger grey symbols represent average F1 and F2 of vowels produced by women, smaller black symbols represent average values of vowels produced by men. Although the marks are in Hz, the axes have logarithmic scaling.
In Czech, short vowels are about 3.5 times more frequent than their long counterparts (Ludvíková Reference Ludvíková and Těšitelová1987: 93). In both varieties, long /oː/ occurs very rarely and almost exclusively in loanwords, e.g. tón /toːn/ ‘tone’, óda /oːda/ ‘ode’. In BC, the vowel /ɛː/ is also infrequent because in many native morphemes it is replaced by /iː/ (e.g. ‘small (n)’: malé /malɛː/ in MC and malý /maliː/ in BC; ‘length’: délka /dɛːlka/ in MC and dýlka /diːlka/ in BC).
The diphthongs are /au/, /ɛu/ and /ou/ in both Czech varieties, but /au/ and /ɛu/ occur only in loanwords and interjections, e.g. euro /ɛuro/ ‘euro’, leukémie /lɛukɛːmɪjɛ/ ‘leukemia’, autor /autor/ ‘author’, au /au/ ‘ouch’.
Phonetically, short vowels in both varieties may be subject to articulatory undershoot induced by increased speech tempo or casual speech style (see Volín Reference Volín and Cvrček2010: 44).
Phonotactics
Phonotactically, Czech allows as many as four consonants in a syllable onset (pstruh /pstrux/ ‘trout’) and three in a coda (pomst /pomst/ ‘revenge (gen pl)’), although this is rare and about half of all syllables are in fact CV (Palková Reference Palková1997: 272). A number of Czech onset clusters violate the sonority hierarchy (e.g. mzda /mzda/ ‘wage’, rtut ˈ /rtuc/ ‘mercury’, pták /ptaːk/ ‘bird’) but the complexity of these structures is often reduced in actual pronunciation. One strategy is insertion of a short epenthetic vowel, e.g. lpět [l
pjɛt] ‘adhere to’, dbát [dbaːt] ‘observe (a rule)’. Another strategy is deletion, e.g. která [kɛraː] ‘which (f)’. Most words beginning with the cluster /j/ + C are lexicalized also without the /j/ (jméno /jmɛːno/ ~ /mɛːno/ ‘name’, jde /jdɛ/ ~ /dɛ/ ‘go (3sg)’, jsem /jsɛm/ ~ /sɛm/ ‘I am’). The preference for the canonical CV structure is also evident in strategies repairing onsetless syllables. In MC, a coda of the preceding syllable will usually be resyllabified into the missing onset (závod aut ‘car race’ [zaːvo.daut], see above), otherwise a glottal stop is used, e.g. nejupřímější [nɛjʔup
iːmɲɛjʃiː] ‘the most sincere’, Sezame, otevři se. [sɛzamɛǀʔotɛvisɛ] ‘Open, Sesame.’. In BC, glottal stop insertion is preferred and resyllabification is relatively infrequent. When it does occur the resyllabified obstruent is never voiced as it is in MC, e.g. závod aut is in BC [zaːvot ʔaut], possibly [zaːvo.taut], never *[zavo.daut]. In BC frequent native words beginning with /o/ are also lexicalized with a prothetic /v/, e.g. oči /oɪ/ ~ /voɪ/ ‘eyes’.
Suprasegmentals
Czech has stress fixed to the first syllable of a word, with the exception that a monosyllabic preposition mostly forms a single metrical unit with the following word and bears stress, e.g. moře [ˈmoɛ] ‘sea’ vs. do moře [ˈdomoɛ] ‘into the sea’. As is common in fixed-stress languages (see Cutler Reference Cutler, Pisoni and Remez2005: 273), phonetic realization of Czech stress is weak (Volín Reference Volín and Cvrček2010: 57). Vowel duration does not function as a cue to stress, it is reserved for marking vowel quantity contrasts which occur independently of stress, e.g. uhel [ˈʔuɦɛl] ‘charcoal’ vs. úhel [ˈʔuːɦɛl] ‘angle’, plocha [ˈploxa] ‘area’ vs. plochá [ˈploxaː] ‘flat (f)’. Nevertheless, vowel duration is variable to some extent: final syllables in intonation phrases are subject to substantial lengthening (e.g. Dankovičová Reference Dankovičová1997b).
Czech rhythm has proved difficult to classify using acoustic measures (e.g. Dankovičová & Dellwo Reference Dankovičová and Dellwo2007). As is typical of syllable-timed languages, Czech exhibits little vowel reduction. On the other hand, compared to syllable-timed languages, Czech allows more complex consonant clusters, which contributes to relatively greater variability of consonant-interval durations.
It should be noted that prosodic structure exerts influence on word-boundary phenomena such as assimilations, glottal stop insertion and resyllabification (as described in the previous section). In both dialects, the preference for glottal stop insertion as opposed to resyllabificiation increases across stronger prosodic boundaries.
Transcription
Bohemian Czech
‖ ˈsɛvɛɾaːka ˈslunɛ sɛ ˈɦaːdalɪ | ɡdo ˈzɲɪxjɛ ˈsɪlɲɛjʃiˑ ‖ ˈftom ˈspatɪlɪ ˈpoəst˺nɛːɦo ‖ ˈktɛɾiˑ ˈkɾaːəl ˈzaɦalɛn ˈplaːʃcɛm ‖ ˈʔujɛdnnalɪ ˈtɛdɪ ‖ ʒɛ ˈtɛnsɛ maː ˈpovaʒovad ˈzasɪlɲɛjʃiˑɦo ‖ ɡdo ˈpvɲiˑ ˈdokaːʒɛ | ˈʔabɪ sɪ ˈpoɛstniˑ ˈsʋlɛːk ˈplaːʃc ‖ ˈtu ˈzaal ˈsɛvɛɾaːk ˈfoukad ˈzɛ fʃiˑ ˈsiˑlɪ ‖ ˈʔalɛ ˈiˑmviˑ ˈfoukal | ˈciˑm ˈviˑsɛ ˈpoɛstnniˑ ˈzaɦaloʋal ˈdosvɛːɦo ˈplaːʃcɛ ‖ ˈkonɛɲɛ sɛ ˈsɛvɛɾaːɡ ˈvzdal ˈmarnɛːɦo ˈʔuːsɪliː ‖ ˈpaɡ ˈzaalo ˈslunɛ ˈsviˑcit ʔa ˈɦaːt ‖ ʔa ˈza ɲɛjakiˑ ˈʔokamʒɪk ˈpoəst˺niː ‖ ˈktɛɾɛːmu ˈbɪlo ˈɦoɾko ‖ ˈsxoɟɪl ˈplaːʃc ‖ ˈtakmusəl ˈsɛvɛɾaːk ˈʔuznad | ʒɛ ˈslunɛ ˈjɛ ˈsɪɫɲɛjʃiː ‖
Moravian Czech
‖ ˈsɛʋɛɾaːk ʔa ˈslunɛ sɛ ˈɦaːdali ‖ ɡdo ˈzɲiɦ jɛ ˈsilɲɛjʃiː ‖ ˈftom ˈspatili ˈpoɛstnnɛːɦo ‖ ˈktɛɾiː ˈkraːɛl ˈzaɦalɛn ˈplaːʃcɛm ‖ ˈʔujɛdnali ˈtɛdi ‖ ʒɛ ˈtɛnsɛ maː ˈpoʋaʒovad ˈzasilɲɛjʃiːɦo ‖ ˈɡdopvɲiː ˈdokaːʒɛ ‖ ˈʔabisi ˈpoɛstnniː ˈs
lɛːkˈplaːʃc ‖ tu ˈzaal ˈsɛʋɛɾaːk ˈfoukad ˈzɛ fʃiː ˈsiːɫi ‖ ˈʔalɛ ˈiːm ʋiː ˈfoukal ‖ ˈciːm ˈviːsɛ ˈpoɛstnniː ˈzaɦaloʋaldo ˈsvɛːɦo ˈplaːʃcɛ ‖ ˈkonɛɲɛ sɛ ˈsɛʋɛɾaːɡ ˈvzdal ˈmarnɛːɦo ˈʔuːsiliː ‖ ˈpaɡ zaalo ˈslunɛ ˈsviːcida ˈɦaːt ‖ ʔa ˈza ɲɛjakiː ˈokamʒik˺ ˈpoɛstnniː | ˈktɛɾɛːmu ˈbəlo ˈɦorko ‖ ˈzɦoɟil ˈplaːʃc ‖ ˈta
ˈmusɛl ˈsɛʋɛɾaːk ˈʔuznat ‖ ʒɛ ˈslunɛ jɛ ˈsilɲɛjʃiː ‖
Orthographic version
Severák a Slunce se hádali, kdo z nich je silnější. V tom spatřili pocestného, který kráčel zahalen pláštěm. Ujednali tedy, že ten se má považovat za silnějšího, kdo první dokáže, aby si pocestný svlékl plášt'. Tu začal Severák foukat ze vší síly, ale čím víc foukal, tím víc se pocestný zahaloval do svého pláště. Konečně se Severák vzdal marného úsilí. Pak začalo Slunce svítit a hřát a za nějaký okamžik pocestný, kterému bylo horko, shodil plášt'. Tak musel Severák uznat, že Slunce je silnější.
