Japanese

Source: Labrune, Laurence. 2012. The Phonology of Japanese. New York: Oxford University Press.

Comments: "Japanese /i/, /e/, and /o/ are slightly less tense than the corresponding cardinal vowels... /u/ is generally rather unrounded, especially in Tôkyô Japanese. Its phonetic quality varies between [ɯ], [ɯ̈], [ᶶ], and [ᶤ] (p. 26). The high vowels /i/ and /u/... have devoiced allophones in certain contexts, for example hiku° [çi̥kɯ] ‘to draw’, de su [desɯ̥] (p. 29). /a/ and /o/ also undergo devoicing but in a more occasional manner (p. 35). [T]he pronunciation of long e as [ei] is now spreading among some speakers of the standard language (p. 41). Sequences of two different vowels are very frequent in Modern Japanese. All the possible combinations between the five vocalic elements are attested, namely iu, ie, io, ia, ui, ue, uo, ua, ei, eu, eo, ea, oi, ou, oe, oa, ai, au, ae, ao. Each vowel in these sequences represents one mora, and is distinctly articulated... Sequences of three or four vowels, or even more, are easy to find (p. 54). I consider that it is mistaken to regard V1V2 sequences as diphthongs since I know of no phonetic or phonological evidence which would prove that they are. They are simply a succession of two distinct vocalic nuclei, each with its own prosodic weight representing one mora (p. 55). The alveolar plosive /t/ has three allophones: the apico-alveolar (or apico-dental) [t] before a,e, and o; two affricate consonants: [ʨ] before i and y and [ts] before u (p. 63). The consonant /d/ displays even more varied realizations: [d] before a,e, and o in any position, and, inside words, [ʑ] before i and y, and [z] or [dz] before u. Word-initially, it is frequently realized as [dʑ] before i and y, and as [dz] before u (p. 64). Some speakers realize /g/ as a plosive [ɡ] word-initially but as a nasal [ŋ] in intervocalic position under certain conditions... The fricative /s/ has two main allophones: [s] before u,e,o, and a, and a pre-dorso-palatal [ɕ] (or dorso-palatal [ç]) before i and y. The /z/ consonant has two or even three realizations: [z] before u, e, o, and a inside words; [ʑ] or [dʑ] before i and y, and finally, for certain speakers, [dz] before u, e, o, a word-initially or after the mora nasal /N/. Before u, the affricate consonant [dz] sometimes occurs even in word-internal position (p. 65). Although sh, j, and ch are sometimes transcribed by means of the IPA symbols [∫], [Ʒ], and [t∫], the closest IPA transcription of these sounds is [ɕ], [ʑ], and [tɕ] (or [ç], [ʝ], and [tç] before i and y), since they are actually alveolo-palatals or pre-dorso-palatals. Note in addition that they are articulated with no marked lip-rounding, and a flattened tongue blade (p. 67). [I]t is preferable to consider that they correspond phonologically to simple consonants followed by the palatal element y: sh [ɕ] = /sy/, j [ʑ] = /zy/ or /dy/, ch [tɕ] = /ty/ (p. 68). The fricative /h/ has the following realizations: [h] before a,e, and o, [ç] or [ɕ] before i and y for many speakers (especially in Tôkyô speech), and [ɸ] before u (p. 70). A rather peculiar segment realized as a moraic m (IPA [m̩]) is sometimes heard in the speech of older speakers, or in certain forms of the traditional Japanese theatre (p. 78). The nasal consonant /n/ is apico-dental or apico-alveolar. A majority of Japanese phoneticians mention a palatal realization [ɲ] before the vowel i and the glide y, but it seems to me that palatalization is not very marked... The phonological status of the dorso-velar nasal consonant [ŋ]... in Tôkyô Japanese has been a much debated issue (p. 79). The labiovelar glide /w/ is slightly less rounded than its English counterpart (for instance in way). Its phonetic realization is between that of the symbols [ɰ] and [w] of the IPA (p. 91). The prototypic realization of the only Japanese liquid /r/ is [ɾ], the apico-alveolar flap... Outside [ɾ], the following phonetic (social or regional) realizations are attested: [l], [ᶩ], [r], [rː], [d], [ɽ], [ɮ] (p. 93). [Consonants f [ɸ] and v [β]] result from the phonologization of sounds already existing in the language but with no phonemic status [prior to the influx of borrowings] (p. 98). The term “special segments” refers to the three moraic segments of Japanese, which constitute a rather unique feature of the language and have been granted special status in traditional Japanese analyses, namely the mora nasal /N/, the first part of an obstruent geminate /Q/ and the second part of a long vowel /R/ (p. 133). /N/, the mora nasal, is a generic nasal contoid, with no definite place of articulation. By default, in slow speech and before a pause, /N/ is a uvular realized with no dorsal occlusion and transcribed as [N]... Before oral or nasal labial stops /p/, /b/, /m/, it is realized as [m].., before the alveolars /t/, /d/, /n/, as [n].., and before the velars /k/ and /g/, as [ŋ]... Before the fricatives /h/, /s/, and /z/, be they palatalized or not, phoneticians disagree about its place of articulation: /N/ is either realized as the nasalized version of the preceding vowel.., as a fricative nasal, or even a nasalized high vowel [ɯ̃] or [ĩ]... Before the semiconsonants /w/ and /y/ and before vowels, the special segment /N/ is phonetically a nasal vowel whose quality is said to be that of the preceding vowel (p. 134). Another notable fact about /N/ is that it can never be linked to an onset position before a vowel. It can never be resyllabified either (p. 136). [T]he syllable is not a relevant unit in the phonology of Japanese (p. 143). The accent of Tôkyô Japanese consists of a distinctive lexical pitch accent (sometimes called musical accent). It is marked phonetically by the change from a high pitched mora (noted H henceforth) to a low pitched mora (L). The last mora of the word carrying the H tone before the drop towards L is regarded as the accented mora of the word, its prosodic peak, or accent kernel... The lexicon is divided into tonic words... and atonic words... A tonic word contains an HL sequence (in other words, a pitch drop), an atonic word does not (p. 180)."

Contributed by: Anton Kukhto (kukhto@mit.edu)