1 Introduction
1.1 Location
Akita is spoken in Akita Prefecture—located in the northeast, that is, the Tohoku region of Japan (Figure 2.1). Sato (1982: 276–277; 2000: 14) stated that Akita has five subvarieties, as depicted in Figure 2.2. The description outlined in this chapter is based on fieldwork studies conducted by the author in Yokote City. Therefore, this chapter elucidates the prosodic system of Akita’s southern variety. However, as no significant difference exists between the prosodic data from previous studies on other subvarieties and this chapter’s data, the prosodic characteristics are expected to be generally similar across all subvarieties1,2.
Figure 2.2
Subvarieties of Akita and location of Yokote City
1.2 Speakers
According to the native speakers who participated in the author’s study, people in their 50s and younger speak a language closer to Standard Japanese, not Akita. Since about 420,000 people aged 60 years or older live in Akita per the prefectural population census of 2023 (Akita Prefecture Open Data 2023), the number of fluent Akita Speakers is estimated to be close to this number.
1.3 Classification
Based on Tojo’s (1954) classification of Japanese varieties, Akita belongs to the Tohoku group (see § 2.1 of Chapter 1, this volume).
1.4 Previous Studies
Previous studies on Akita’s prosodic systems include works by Sato (1982: 279–280), Ohashi (2000: 57–73), Kondo (2012, 2015, 2016a, b), and Suganuma (2022). Sato (1982: 279–280) described the tonal patterns of basic nouns in the variety spoken in Akita City. Ohashi (2000: 57–73) described the tonal patterns of basic nouns, verbs, and adjectives in the same variety. Kondo (2012, 2015, 2016a, b) analyzed patterns of various verb conjugations in the variety spoken in Oga City. All these studies focused on the central variety of Akita. By contrast, Suganuma (2022) analyzed the southern variety spoken in Yokote City and proposed phonological rules to elucidate the tonal patterns observed in various verb conjugations.
1.5 Data
Unless otherwise noted, the description in this chapter is based on data obtained from the three speakers listed in (1).
(1) NS1: female, born in 1935 in Yokote City, Akita Prefecture; has lived there since birth.
NS2: female, born in 1960 in Yokote City, Akita Prefecture; lived there until the age of 18, after which she moved outside the prefecture. (The study was conducted with her during her return to Yokote City in May 2019.)
NS3: female, born in 1962 in Yokote City, Akita Prefecture; has lived there since birth.
Although NS2 has lived outside the prefecture for a long time, no prosodic differences were observed between NS2 and the other speakers, at least vis-à-vis the features addressed in this chapter.
1.6 Overview of Segmental Phonology
Akita has five vowel phonemes (/a, i, u, e, o/) and the 15 consonant phonemes that are shown in Table 2.1, along with the consonant phoneme /N/ which is discussed below.
Sato (1982: 283) suggested that an additional vowel phoneme /ɛ/, historically deriving from the diphthongs /ai/ and /ae/, can be posited; however, he also highlighted that it is not stable. In this chapter, we do not posit this phoneme. Some notes are in order concerning the phonetic realization of the vowel phonemes /i/ and /u/. Both /i/ and /u/ are generally realized as centralized vowels, with /u/ typically pronounced as an unrounded vowel. Consequently, the phonetic forms of /i/ and /u/ are similar (i.e., /i/ [ɨ] vs. /u/ [ɯ̈]), yet distinctive. However, as Ohashi (2000) highlighted, when these vowels are combined with the apical consonants /s, z, c/, they tend to neutralize to [ɨ] (e.g., /sisi/ [sɨsɨ] ‘lion’, /susu/ [sɨsɨ] ‘soot’). Additionally, both /i/ and /u/ devoice when they are positioned between voiceless consonants and are followed by a non-high vowel (i.e., C[-voiced]__C[-voiced] V[-high], e.g., /sita/ [sɨ̥ta] ‘tongue’). The conditioning of devoicing by the following vowel’s height seems to be a prevalent feature of the Tohoku varieties (cf. Ohashi 2002: 170–172).
Voiceless plosive and affricate consonants become voiced when they are in intervocalic position (i.e., V__V), whereas voiced plosives and affricates become pre-nasalized in the same position (see (2) below, noting that /ɡ/ undergoes complete nasalization).3
Table 2.1
Consonant phoneme inventory and typical allophones. (Phonemes are indicated by slashes, whereas typical allophones are provided in parentheses. Symbols in square brackets are used for phonemic transcriptions wherein the phonetic realization may not be immediately apparent.)
|
Bilabial |
Alveolar |
Alveolo-palatal |
Palatal |
Velar |
Uvula |
Glottal |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Plosive |
/p/ |
/b/ |
/t/ |
/d/ |
/k/ |
/ɡ/ |
||||||||
|
Affricate |
/c/ (ts) |
/z/ (dz) |
(tɕ) |
(dʑ) |
||||||||||
|
Nasal |
/m/ |
/n/ |
(ŋ) |
(ɴ) |
||||||||||
|
Tap |
/r/ [ɾ] |
|||||||||||||
|
Fricative |
(ɸ) |
/s/ |
(ç) |
/h/ |
||||||||||
|
Approximant |
/w/ |
/j/ |
||||||||||||
|
(2) |
a. |
Voicing |
b. |
Pre-nasalization |
|
|
/kata/ [kada] ‘shoulder’ |
/hada/ [handa] ‘skin’ |
||||
|
/kuci/ [kɯ̈dzɨ] ‘mouth’ |
/kuzi/ [kɯ̈ndzɨ] ‘lottery’ |
||||
|
/kiku/ [kɨɡɯ̈] ‘chrysanthemum’ |
/kaɡe/ [kaŋe] ‘shadow’ |
Devoiced vowels do not induce consonant voicing. Describing this using a rule-ordering approach, we can state that the vowel devoicing rule and consonant voicing rule mutually bleed each other. Based on the author’s observations, which of the two rules presented in (3) first applies seems to be assessed from the word’s left edge. Specifically, it is the rule whose application environment is first encountered from the beginning of the word which applies, as presented in (4).
(3) a. Vowel devoicing rule (VD): V[+high]
→ [-voiced]/ C[-voiced]__C[-voiced]V[-high]
b. Consonant voicing rule (CV): C[+stop]
→ [+voiced]/ V[+voiced]__V[+voiced]
(4) Examples of derivation
a. /oki-ta/ ‘woke up’4
oɡita : CV is applied to /oki/ and bleeds the application of VD to /kit/.
oɡida : CV is applied to /ita/
[oɡɨ-da], *[okɨ̥ta]
b. /dasi-ta/ ‘put out (past form)’
dasi̥ta : VD is applied to /sit/, and bleeds the application of CV to /ita/.
[dasɨ̥ta], *[dasɨda]
The possible syllable structures in Akita can be generalized as (C1)(G)V1(V2)(C2), where “C”, “V”, and “G”, stand for a consonant, vowel, and glide, respectively. The G slot can be filled only with /j/. Superheavy syllables, such as C1VVC2, are found only in a limited number of cases, such as the emphatic form of onomatopoeia. The variety of coda consonants in Akita is limited to nasal or voiceless obstruents. In this chapter, /N/, which is not displayed in Table 2.1, is used as the phoneme representation for nasal codas. Utterance-final /N/ is typically articulated by the closure of the tongue dorsum against the palate (i.e., [ŋ] ~ [ɴ]). Word-internal nasal codas are realized as homorganic nasals, according to the following segment:5 Voiceless obstruent codas appear only in word-internal position and not in word-final position. Voiceless obstruent codas are always realized as the first half of a geminate. Voiceless codas are traditionally represented by /Q/ in studies of Japanese phonology; however, in this chapter, the phonetic form is used directly for its phoneme representation (e.g., /kappa/ [kappa] ‘raincoat’). The rhyme slots—namely, V1, V2, and C2—are one mora each.
2 Word-Prosodic System
2.1 Classification Based on Uwano (1989, 1999)
Akita’s prosodic system is an instance of Uwano’s (1989, 1999) multipattern system, defined as a system wherein the number of tonal contrasts increases proportionally with word length (see § 2.4 of Chapter 1, this volume).
2.2 Number and Nature of Tonal Contrasts
Akita exhibits a multipattern prosodic system, which renders it difficult to determine the number of contrastive tonal patterns. In this section, we examine the realization of tonal patterns in Akita, focusing primarily on basic vocabulary nouns as examples.
In Akita, lexical words, including nouns, are divided into accented and unaccented words depending on whether they possess a lexically determined accent kernel. The accent kernel in Akita is a lowering kernel (H*+L) (see § 2.3 of Chapter 1, this volume), which is associated with one of the syllables within an accented word. (Throughout this chapter, the terms accent and accent kernel are used interchangeably.) Consequently, the kernel can be analyzed as a word-level tone. This chapter uses an apostrophe (’) as a shorthand for the lowering kernel. According to Uwano (2012), varieties throughout Akita Prefecture have an ascending kernel (L+H*) (see § 2.3 of Chapter 1, this volume). However, the author’s fieldwork data do not necessarily support this claim. Uwano (2012: 1430) demonstrated that dialects with an ascending kernel exist in the northern Tohoku region, including Aomori, Iwate, and Akita prefectures. However, he provided examples of Hirosaki (a subvariety of Aomori Prefecture) and Shizukuishi (a subvariety of Iwate Prefecture), but not Akita; hence, the idea that Akita and its subvarieties have an ascending kernel must be re-examined. Furthermore, prosodic data within previous studies of Akita (Sato 1982: 279–280; 2000: 28, and so on) exhibit tonal patterns which can be explained by assuming a lowering kernel, not an ascending kernel. These facts suggest that Akita has a lowering kernel.
For nouns, the accent kernel position is lexically determined and contrastive.6 For words ending in an accented light syllable, the kernel’s L tone is not realized in isolated pronunciation (see (5b, e, i) below).7
Figure 2.3
Pitch pattern examples of three-syllable words pronounced by NS1
Regarding phrase-level tone, if a phrase contains more than one syllable, the phrase-initial syllable has a phrase-level low tone, unless that syllable is accented. The following syllables exhibit a phrase-level high tone up to the accented syllable, or—if there is no accent kernel—up to the end of the phrase. The examples in (5) illustrate underlying, citation, and nominative forms with the additive particle =mo ‘also’ for one- to three-syllable nouns. The diacritics “⸢” and “⸣” indicate a pitch rise and fall, respectively, between morae, whereas “⸣⸣” denotes a pitch fall within a mora (see § 2.2 of Chapter 1, this volume). Figure 2.3 depicts the fundamental frequency (fo) contours of three-syllable nouns with the particle =mo, indicating the location of the fo peak corresponding to the accent kernel’s high tone using the “H*+L” notation. The interaction between word- and phrase-level tones results in the tonal patterns presented in (5ii), (5iii), and Figure 2.3.
|
i. Underlying form of lexical word |
ii. Citation form |
iii. Nominative form with the additive particle =mo ‘also’ |
||||
|---|---|---|---|---|---|---|
|
(5) |
Tonal patterns in Akita |
|||||
|
1- |
||||||
|
a. |
Unaccented |
/na/ |
‘name’ |
⸢na |
na=⸢mo |
|
|
b. |
Accented |
/na’/ |
‘greens’ |
⸢na |
⸢na⸣= mo |
|
|
2- |
||||||
|
c. |
Unaccented |
/ame/ |
‘candy’ |
a⸢me |
a⸢me=mo |
|
|
d. |
First syllable accented |
|||||
|
/sa’ru/ |
‘monkey’ |
⸢sa⸣ru |
⸢sa⸣ru=mo |
|||
|
e. |
Last syllable accented |
|||||
|
/ame’/ |
‘rain’ |
a⸢me |
a⸢me⸣=mo |
|||
|
3- |
||||||
|
f. |
Unaccented |
/janaɡi/ |
‘willow’ |
ja⸢naɡi |
ja⸢naɡi=mo |
|
|
g. |
First syllable accented |
|||||
|
/ho’taru/ |
‘firefly’ |
⸢ho⸣taru |
⸢ho⸣taru=mo |
|||
|
h. |
Second syllable accented |
|||||
|
/ino’ci/ |
‘life’ |
i⸢no⸣ci |
i⸢no⸣ci=mo |
|||
|
i. |
Last syllable accented |
|||||
|
/takara’/ |
‘treasure’ |
ta⸢kara |
ta⸢kara⸣=mo |
|||
In accented words where the accent kernel is on a syllable other than the first, as well as in unaccented words, the beginning of these words exhibits a pitch rise from low to high when in phrase-initial position. As previously mentioned, this pitch rise is the result of a phrase-level tone. Consequently, when an unaccented element precedes these words, the pitch rise is not observed in those words; instead, the pitch rise occurs in the preceding element located in phrase-initial position (e.g., ko⸢no ame ‘this candy’, ko⸢no janaɡi ‘this willow’).
2.3 Domain of Lexical Tone Assignment
In Akita, the domain to which lexical tone is assigned is primarily the lexical word. Additionally, accented particles form their own domain so that one bunsetsu—a unit comprising a lexical word and any following particles (see § 2.6 of Chapter 1, this volume)—is divided into two tonal domains when the preceding lexical word is accented. However, specific instances exist where the bunsetsu can be interpreted as the domain. Whether the bunsetsu serves as the domain in these cases depends on the interpretation of how the particles’ accent kernel is realized. Numerous particles have no accent kernel and do not form their own domain. However, at least the particle =jori ‘than/from’ has its own accent kernel, and a pitch fall on the particle is observed when =jori is attached to an unaccented word (see (6a) and Figure 2.4-i below). When =jori is attached to an accented word, the pitch fall on =jori is not clearly observed, and a flat pitch is frequently observed at =jori. However, this is not always the case, and sometimes, a pitch fall is observed in =jori (see (6b), Figure 2.4-ii, and iii below).
(6) The pitch patterns of the construction “lexical word + =jo’ri”
a. Unaccented lexical word: kuruma=jo’ri ‘than cars’
b. Accented lexical word: a’bura=jori ~ a’bura=jo’ri ‘than oil’8
If the accent kernel of =jo’ri in the phrase a’bura=jori is obligatorily deleted in this environment, it indicates a constraint prohibiting more than one accent kernel within the bunsetsu, which would lead us to regard the lexical tone domain as the bunsetsu. However, it cannot be definitively asserted that such a constraint exists for the following two reasons. First, deleting the accent kernel of a particle is not obligatory. As depicted in Figure 2.4-iii, the kernel in the particle can be realized even when preceded by an accented lexical word. In this case, the lexical word and particle each have their own domain. Second, even in instances where the kernel of a particle can be considered deleted, whether it is indeed deleted is questionable. Instead, the kernel may merely be phonetically suppressed. Even if suppressed, insofar as the kernel is not completely deleted, the lexical word and particle each form a separate domain. In this chapter, we assume that the accent kernel of the particle remains realized in the given environment. Consequently, the lexical tone domain is not considered the bunsetsu.9
Figure 2.4
Pitch patterns of the construction “lexical word plus =jo’ri” pronounced by NS3 (All figures mark the location of the pitch fall with arrows.)
2.4 Tone-Bearing Units
In Akita, the tone-bearing unit is the syllable, while the “counting unit”—a unit for measuring phonological length or distance (McCawley 1978; see § 2.7 of Chapter 1, this volume)—is the mora, as can be demonstrated by investigating loanword accentuation.10 Most loanwords in Akita are realized as accented words; thus, they have a pitch fall. In loanwords of four or more morae, the pitch fall begins from the antepenultimate mora (see (7a–e)). However, if the antepenultimate mora corresponds to a syllable’s coda or the second half of a long vowel, the pitch fall begins from the mora immediately preceding the antepenultimate mora (7f–h).11 These patterns are generalized by assuming that the syllable is the tone-bearing unit and the mora is the counting unit in Akita and proposing a rule that places the accent kernel on the syllable containing the antepenultimate mora.
(7) Tonal patterns of loanwords of four or more morae in Akita (Period and
μ denote syllable boundary and mora, antepenultimate morae are bolded, and phrase level-tone is omitted.)
|
Number of morae contained within the word-final three syllables |
Examples |
|||
|---|---|---|---|---|
|
a. |
… |
pu |
‘program’ |
|
|
| |
||||
|
H*+ L |
||||
|
b. |
… |
su |
‘straw’ |
|
|
| |
||||
|
H*+L |
||||
|
c. |
… |
bi |
‘vinyl’ |
|
|
| |
||||
|
H*+L |
||||
|
d. |
… |
ba |
‘barbecue’ |
|
|
| |
||||
|
H*+ L |
||||
|
e. |
… |
hu |
‘Friendoll (Name of local pastry shop)’ |
|
|
| |
||||
|
H*+L |
||||
|
f. |
… |
ka |
‘calendar’ |
|
|
| |
||||
|
H*+L |
||||
|
g. |
… |
a |
‘AX (Name of local supermarket)’ |
|
|
| |
||||
|
H*+L |
||||
|
h. |
… |
ka |
‘carnation’ |
|
|
| |
||||
|
H*+ L |
Notably, as demonstrated in (7c) and (7e–h), if the accented syllable is a heavy syllable, the position of the accent kernel “ ’ ” and the beginning of the pitch fall are displaced.12 However, for the remainder of this chapter, the accent kernel is shown in the position where the pitch fall occurs (that is ka.reN’.daa is represented as ka.re’N.daa).
The patterns presented in (7) are the same as those in Tokyo (cf. McCawley 1968: 134); a possibility exists that most loanwords have simply been borrowed from Tokyo without any change in accent location.13 However, loanwords of three or fewer morae exhibit a pattern that partially differs from those of Tokyo, and this pattern also supports the idea that the tone-bearing unit is the syllable, while the counting unit is the mora in Akita. In three-mora loanwords beginning with a light syllable (i.e., CV.CV.CV, CV.CVC, and CV.CVV), we observe a variation between two different patterns. That is, either the first or second syllable is accented, as presented in (8a, b). This variation is particularly prevalent in the case of familiar loanwords. By contrast, three-mora loanwords beginning with a heavy syllable and two-mora loanwords exhibit only one pattern. That is, the word-initial syllable is accented, as presented in (8c, d).
|
Syllable structure of words and presence or absence of variation |
Examples |
|||
|---|---|---|---|---|
|
(8) |
Tonal patterns of loanwords with three or fewer morae |
|||
|
a. |
CV |
te |
‘television’ |
|
|
ba |
‘banana’ |
|||
|
b. |
CV |
me |
‘melon’ |
|
|
pu |
‘pudding’ |
|||
|
ze |
‘jerry’ |
|||
|
c. |
CV |
ma |
‘mantle’ |
|
|
da |
‘dance’ |
|||
|
so |
‘soda’ |
|||
|
ke |
‘cake’ |
|||
|
d. |
Two-mora loanwords |
ɡa |
‘gum’ |
|
|
CV |
cu |
‘two’ |
||
|
ka |
‘can’ |
|||
In Tokyo, the accentuation of the second syllable in (8a, b) is not attested: It is accentuated on the word-initial syllable, regardless of syllable structure. Currently, we have no explanation for why we observe the variation presented in (8a, b) in three-mora loanwords with a specific syllable structure in Akita. Nevertheless, positing a single rule elucidating all patterns presented in (8) in a unified manner is still possible. However, considering that loanwords of the same mora length but different syllable counts (i.e., CV.CV.CV and CV.CVC, CV.CVV) share the same variation in pattern, such a rule can only be formulated by referring to the mora. Evidently, this observation provides additional support for positing the mora as the counting unit in Akita. The rule is presented in (9), and its application is exemplified in (10).
(9) Accent rule for (familiar) loanwords with three or fewer morae in Akita
Assign the accent kernel to the syllable containing the antepenultimate or penultimate mora.
|
(10) |
Examples of the application of rule (9)14 |
|||
|
a. |
te |
Penultimate mora: re |
→ te |
|
|
Antepenultimate mora: te |
→ te |
|||
|
b. |
me |
Penultimate mora: ro |
→ me |
|
|
Antepenultimate mora: me |
→ me |
|||
|
c. |
ma |
Penultimate mora: N |
→ ma |
|
|
Antepenultimate mora: ma |
→ ma |
|||
|
d. |
ɡa |
Penultimate mora: ɡa |
→ ɡa |
|
|
Antepenultimate mora: None |
||||
2.5 Directionality
Akita exhibits no absolute directionality—defined as the direction in which word-level prominence is calculated (see § 2.8 of Chapter 1, this volume)—that dominates the entire vocabulary. As observed in (5), the position of the accent kernel of basic nouns is lexically determined and is not based on any specific direction. However, the position of the accent kernel in loanwords is determined by its position from the end of the word, indicating right-to-left directionality (see (7), (8), and (10)).
For complex forms containing a morpheme boundary, such as compounds or verbal forms, the kernel’s position is not determined by a specific directionality but is rather computed to place the kernel as close as possible to the morpheme boundary. For example, in compound tonal patterns where the accent kernel is placed in a different position from the underlying form, it is invariably placed on a syllable adjacent to the morpheme boundary, as discussed in § 2.7 and § 3 (see (12), a multiple pattern in (13e–g), and (20)). Furthermore, compound words in (13a, b) and (18) in § 2.7 and § 3 preserve the right element’s accent kernel, and both have in common that they originally had an accent kernel in the syllable adjacent to the morpheme boundary and satisfy the requirement that brings the accent kernel as close to morphological boundaries as possible (e.g., /ɡoma+a’bura/ → ɡoma+a’bura ‘sesame oil’ and /roseN + ba’su/ → roseN+ba’su ‘scheduled bus’).
Although some compounds do not exhibit patterns where the accent kernel is placed on the syllable adjacent to morpheme boundaries (see (13c, d)), considering that most compounds do position the accent kernel on the syllable adjacent to a morpheme boundary, it can be concluded that the computation mentioned earlier is at play in compounds.
Noteworthily, a similar computation operates in compounds and verbal accentuation in Tokyo, as stated by Kubozono (2008: 182) and Yamaguchi (2010: 4). However, when comparing the verbal accentuation in Akita and Tokyo, this computation seems more pronounced in Akita than in Tokyo. It seems to work in a manner that brings the prominence of accent kernels (i.e., H* of H*+L), rather than the accent kernel itself, to the morpheme boundary in Akita’s verbal accentuation. For example, the accentuation of the past-form oki-ta ‘woke up’, joN-da ‘read (past form)’, and mat-ta ‘waited’ of Akita and Tokyo are as follows:
(11) Akita’s verbal accentuation (data from Suganuma 2022): H*+L is also shown with “ ’ ”.
|
Akita |
cf. Tokyo |
|||
|---|---|---|---|---|
|
a. |
o |
a. |
o |
|
|
| |
| |
|||
|
H*+L |
H*+L |
|||
|
b. |
jo |
b. |
jo |
|
|
| |
| |
|||
|
H*+L15 |
H*+L |
|||
|
c. |
ma |
c. |
ma |
|
|
| |
| |
|||
|
H*+L |
H*+L |
|||
In the past form in Tokyo, the accent kernel is typically placed on the syllable containing the antepenultimate mora, allowing it to be positioned away from the morpheme boundary, as observed in o
In sum, Akita exhibits right-to-left directionality in loanwords. However, in compounds and verbs, the accent’s position is determined by a computation aiming to bring the accent kernel (or its prominence) as close as possible to the morpheme boundary instead of adhering to a specific directionality.
2.6 Obligatoriness and Culminativity
Obligatoriness requires at least one prominence within the word domain, whereas culminativity requires at most one prominence within the word domain (Hyman 2006; see § 2.5 of Chapter 1, this volume). The Akita prosodic system does not satisfy the obligatoriness requirement but satisfies the culminativity requirement.
As presented in (5), Akita includes unaccented words, which lack word-level prominence and, thus, fail to satisfy the obligatoriness requirements.
Akita fulfills this criterion regarding culminativity requirements. This is supported by the absence of lexical words with more than two prominences in the data obtained from the author’s fieldwork and the data presented in previous studies.
2.7 Prosodic Compound Rule
The prosodic compound rules in Akita are right-dominant as the tonal pattern of compound words is always determined by the number of morae and accentual information of the right element. When the right element of compounds is one or two morae long, the prosodic compound rules are somewhat complex; they have lexically determined aspects, and the tonal patterns are not easily predictable. However, when the right element is three or more morae long, the tonal pattern can easily be predicted from the right element’s accentual information. In this section, we examine only cases where the right elements are three or more morae long. Instances involving one- or two-mora right elements are discussed in § 3.
Firstly, if the right element comprises three or four morae and its tonal pattern is unaccented or accented on the final syllable (e.g., tokaɡe ‘lizard’, kotoba’ ‘word’ etc.), the accent kernel is placed on the first syllable of the right element in compound words, as presented in (12).
(12) The tonal pattern of compounds whose right element comprises three or four morae and is either unaccented or accented on the final syllable.
|
a. |
tokaɡe ‘lizard’ |
||||
|
niho’N + tokaɡe |
→ |
nihoN + to’kaɡe |
‘Japanese lizard’ |
||
|
eri’maki + tokaɡe |
→ |
erimaki + to’kaɡe |
‘frilled lizard’ |
||
|
b. |
tomodaci ‘friend’ |
||||
|
otoko’ + tomodaci |
→ |
otoko + to’modaci |
‘male friend’ |
||
|
oNna’ + tomodaci |
→ |
oNna + to’modaci |
‘female friend’ |
||
|
c. |
kotoba’ ‘word’ |
||||
|
inaka + kotoba’ |
→ |
inaka + ko’toba |
‘rural dialect’ |
||
|
a’kacjaN + kotoba’ |
→ |
akacjaN + ko’toba |
‘baby talk’ |
||
|
d. |
hukuro’ ‘bag’ |
||||
|
bini’iru + hukuro’ |
→ |
biniiru + bu’kuro17 |
‘plastic bag’ |
||
|
ɡomi + hukuro’ |
→ |
ɡomi + bu’kuro |
‘garbage bag’ |
Secondly, if the right element comprises three or four morae and is accented on a non-final syllable (e.g., a’bura ‘oil’, hiko’oki ‘airplane’), the accent is preserved as such in compounds. However, in compounds whose right element is accented on a light penultimate syllable, we observe some degree of instability in their tonal patterns. That is, some of these compounds exhibit a variation between the accent-preserving pattern and the pattern whereby the accent falls on the first syllable of the right element (see (13e–g)).
(13) Tonal pattern of compounds whose right element comprises three or four morae and is accented on a non-final syllable.
|
a. |
a’bura ‘oil’ |
||||
|
ɡoma + a’bura |
→ |
ɡoma + a’bura |
‘sesame oil’ |
||
|
kika’i + a’bura |
→ |
kikai + a’bura |
‘machine oil’ |
||
|
b. |
ra’ameN ‘ramen’ |
||||
|
mi’so + ra’ameN |
→ |
miso + ra’ameN |
‘miso ramen’ |
||
|
sjooju’ + ra’ameN |
→ |
sjooju + ra’ameN |
‘soy-sauce ramen’ |
||
|
c. |
hiko’oki ‘airplane’ |
||||
|
kami + hiko’oki |
→ |
kami + hiko’oki |
‘paper airplane’ |
||
|
zjetto + hiko’oki |
→ |
zjetto + hiko’oki |
‘jet plane’ |
||
|
d. |
hure’Ndo ‘friend (loanword)’ |
||||
|
pe’N + hure’Ndo |
→ |
peN + hure’Ndo |
‘pen pal’ |
||
|
be’suto + hure’Ndo |
→ |
besuto + hure’Ndo |
‘best friend’ |
||
|
e. |
nezu’mi ‘mouse’ |
||||
|
dobu + nezu’mi |
→ |
dobu + nezu’mi |
‘rat’ |
||
|
~ dobu + ne’zumi |
|||||
|
hacuka’ + nezu’mi |
→ |
hacuka + nezu’mi |
‘house mouse’ |
||
|
~ hacuka + ne’zumi |
|||||
|
f. |
hasa’mi ‘scissors’ |
||||
|
se’ejoo + hasa’mi |
→ |
seejoo + basa’mi |
‘western scissors’ |
||
|
~ seejoo + ba’sami |
|||||
|
seNtaku + hasa’mi |
→ |
seNtaku + basa’mi |
‘clothes peg’ |
||
|
~ seNtaku + ba’sami |
|||||
|
g. |
kamiso’ri ‘razor’ |
||||
|
de’Nki + kamiso’ri |
→ |
deNki + kamiso’ri |
‘electric razor’ |
||
|
~ deNki + ka’misori |
|||||
|
teezi + kamiso’ri |
→ |
teezi + kamiso’ri |
‘T-shaped razor’ |
||
|
~ teezi + ka’misori |
|||||
Finally, if the right element comprises five or more morae, its tonal pattern is maintained in compounds, as presented in (14).
(14) Tonal pattern of compounds whose right element comprises five or more morae, as follows:
|
a. |
monoɡa’tari ‘story’ |
|||
|
jume’ + monoɡa’tari |
→ |
jume + monoɡa’tari ‘dreamy story’ |
||
|
iso’ppu + monoɡa’tari |
→ |
isoppu + monoɡa’tari ‘Aesop’s Fables’ |
||
|
b. |
oriNpi’kku ‘olympics’ |
|||
|
ri’o + oriNpi’kku |
→ |
rio + oriNpi’kku ‘Rio de Janeiro Olympics’ |
||
|
tookjoo + oriNpi’kku |
→ |
tookjoo + oriNpi’kku ‘Tokyo Olympics’ |
||
|
c. |
taNkeNtai ‘expedition team’ |
|||
|
naNkjoku + taNkeNtai |
→ |
naNkjoku + taNkeNtai ‘South Pole expedition team’ |
||
2.8 Merger and Split of Historic Tonal Classes
The general correspondences in tonal classes between Middle Japanese (MJ) (see § 2.10 of Chapter 1, this volume) and Akita are presented in Table 2.2. Examples are provided in (15–16).
Table 2.2
Correspondences in tonal Classes between MJ and Akita
|
MJ tonal classes |
Akita tonal classes |
|
|---|---|---|
|
1.1 |
Type U |
|
|
1.2 |
Type U |
|
|
1.3 |
Type A1 |
|
|
2.1 |
Type U |
|
|
2.2 |
Type U |
|
|
2.3 |
Type A2 |
|
|
2.4 |
Word-final vowel [+high] |
Type A1 |
|
Word-final vowel [-high] |
Type A2 |
|
|
2.5 |
Word-final vowel [+high] |
Type A1 |
|
Word-final vowel [-high] |
Type A2 |
|
|
(15) |
Examples of one-mora words |
||
|
a. |
1.1: Type U |
ci ‘blood’, to ‘door’, ko ‘child’ |
|
|
b. |
1.2: Type U |
na ‘name’, ha ‘leaf’, hi ‘sun’ |
|
|
c. |
1.3: Type A1 |
su’ ‘vinegar’, ha’ ‘tooth’, ju’ ‘hot water’ |
|
|
(16) |
Examples of two-mora words |
||
|
a. |
2.1: Type U |
ame ‘candy’, kubi ‘neck’, ebi ‘shrimp’, |
|
|
kane ‘money’, niwa ‘garden’ |
|||
|
b. |
2.2: Type U |
isi ‘stone’, oto ‘sound’, nacu ‘summer’, |
|
|
hata ‘flag’, tera ‘temple’ |
|||
|
c. |
2.3: Type A2 |
asi’ ‘foot’, kusi’ ‘comb’, uma’ ‘horse’ |
|
|
kaɡi’ ‘key’, iro’ ‘color’ |
|||
|
d. |
2.4, word final vowel [+high]: Type A1 |
||
|
ki’nu ‘silk’, o’bi ‘belt’, u’mi ‘sea’, ka’zu ‘number’ |
|||
|
e. |
2.4, word final vowel [-high]: Type A2 |
||
|
kata’ ‘shoulder’, hune’ ‘ship’, naka’ ‘inside’, awa’ ‘millet’ |
|||
|
f. |
2.5, word final vowel [+high]: Type A1 |
||
|
sa’ru ‘monkey’, cu’ru ‘crane’, ha’ru ‘spring’, a’ki ‘autumn’ |
|||
|
g. |
2.5, word final vowel [-high]: Type A2 |
||
|
ame’ ‘rain’, kumo’ ‘spider’, huna’ ‘crusian carp’, mado’ ‘window’ |
|||
However, some exceptions exist to the correspondences listed in Table 2.2, as illustrated in (17).
|
(17) |
a. |
2.2: Type A2 |
uta’ ‘song’ |
|
b. |
2.3: Type U |
semi ‘cicada’, tani ‘valley’ |
|
|
c. |
2.3: Type A1 |
ka’mi ‘god’ |
|
|
d. |
2.4, word final vowel [-high]: Type A1 |
ka’ma ‘sickle’ |
|
|
e. |
2.5, word final vowel [-high]: Type A1 |
ka’ɡe ‘shadow’ |
|
|
f. |
2.5, word final vowel [+high]: Type A2 |
cuju’ ‘dew’ |
The tonal class of the words presented in (15), (16), and (17) matches the tonal class reported for the same words in Sato (1982: 280) for the central variety of Akita, except uta’, which is reported as regularly belonging to type U. This aligns with Sato’s (1982: 279) observation that only slight differences exist among the subvarieties of Akita.
As Table 2.2 indicates, the merger of MJ tonal classes 1 and 2 in both one- and two-mora words categorizes Akita within the so-called Gairin type of Japanese varieties (see (15a, b) and (16a, b)).
3 Detailed Description of the Prosodic System
As mentioned in § 2.7, in this section, we examine the tonal patterns of compounds with one or two morae right elements.
When the right element comprises one or two morae, the tonal patterns observed in compounds can be categorized into the following three types: accent-preserving (18), de-accenting (19), and pre-accenting (20).18
(18) Accent-preserving type: Accentual information of the right element is preserved.
|
a. |
ba’su ‘bus’ |
niho’N + ba’su |
→ |
nihoN + ba’su |
‘Japan bus’ |
|
|
roseN + ba’su |
→ |
roseN + ba’su |
‘scheduled bus’ |
|||
|
pa’Nda + ba’su |
→ |
paNda + ba’su |
‘panda bus’ |
|||
|
b. |
sa’ru ‘monkey’ |
niho’N + sa’ru |
→ |
nihoN + za’ru |
‘Japanese macaque’ |
|
|
teNɡu + sa’ru |
→ |
teNɡu + za’ru |
‘proboscis monkey’ |
|||
|
ki’cune + sa’ru |
→ |
kicune + za’ru |
‘Lemuridae’ |
|
(19) |
Deaccenting type: The entire compound word becomes unaccented.19 |
||||
|
iro’ ‘color’ |
nezu’mi + iro’ |
→ |
nezumi + iro |
‘grey color (like the color of mouse)’ |
|
|
kohaku + iro’ |
→ |
kohaku + iro |
‘amber color’ |
||
|
ku’ro + iro’ |
→ |
kuro + iro |
‘black color’ |
||
(20) Pre-accenting type: The accent kernel is placed on the syllable immediately before the right element.
|
a. |
kawa ‘river’ |
||||
|
kita’kami + kawa |
→ |
kitakami’ + ɡawa |
‘Kitakami River’ |
||
|
a’mazoN + kawa |
→ |
amazo’N + ɡawa |
‘Amazon River’ |
||
|
jokote + kawa |
→ |
jokote’ + ɡawa |
‘Yokote River’ |
||
|
b. |
e’ki ‘station’ |
||||
|
de’rii + e’ki |
→ |
deri’i + eki |
‘Delhi Station’ |
||
|
jokote + e’ki |
→ |
jokote’ + eki |
‘Yokote Station’ |
||
|
uɡoho’Nzjoo + e’ki |
→ |
uɡohoNzjo’o + eki |
‘Ugo-Honjo Station’ |
||
|
c. |
pa’N ‘bread’ |
||||
|
kuri’imu + pa’N |
→ |
kuriimu’ + paN |
‘custard-cream bread’ |
||
|
kokutoo + pa’N |
→ |
kokuto’o + paN |
‘brown-sugar bread’ |
||
|
bu’doo + pa’N |
→ |
budo’o + paN |
‘raisin bread’ |
||
|
d. |
uta’ ‘song’ |
||||
|
hajari + uta’ |
→ |
hajari’ + uta |
‘popular song’ |
||
|
komori + uta’ |
→ |
komori’ + uta |
‘lullaby’ |
||
|
e. |
si’ ‘city’ |
||||
|
jokote + si’ |
→ |
jokote’ + si |
‘Yokote City’ |
||
|
kita’kami + si’ |
→ |
kitakami’ + si |
‘Kitakami City’ |
The examples presented in (18–20) indicate that we cannot fully predict the type of tonal pattern observed in compounds given the tonal class of the right element. Indeed, the pairs sa’ru and e’ki and iro’ and uta’ belong to the same tonal class, but the compounds they contain are of different tonal types (compare (18b) vs. (20b) and (19) vs. (20d)). Furthermore, the pre-accenting type includes a variety of right elements belonging to different tonal classes. As compounds containing the right elements belonging to the same tonal class do not exhibit the same tonal pattern, the tonal type to which a compound is assigned must be analyzed as a lexically determined property of the right element that is not predictable from its tonal class. Nevertheless, considering that the accent-preserving type of compounds is confined to instances wherein the right element is accented on the first syllable, the tonal pattern of compounds is not entirely independent from the right element’s tonal class.
4 Phrase-Level Prosody
4.1 Interrogative Intonation
Owing to the lack of studies and data on phrase-level prosody in Akita, this chapter offers only a broad overview of the intonation patterns found in interrogative sentences.
Yes-no questions are obligatorily marked with a rising boundary pitch movement (hereafter, BPM is denoted by ‘↗’) and the sentence-final question particle =ka. Yes-no questions lacking either of these features are considered ungrammatical, as presented in (21a, d) vs. (21b, c, e, f).
|
(21) |
e.g., nuke ‘hot’ and joNda ‘need’ |
||||||
|
a. |
nuke=ka↗ |
b. |
*nuke↗ |
c. |
*nuke=ka |
‘Do you feel hot?’ |
|
|
d. |
joNda=ka↗ |
e. |
*joNda↗ |
f. |
*joNda=ka |
‘Do you need (something)?’ |
|
Wh-questions (WHQ) are formed with a distinct pattern from yes-no questions and are never marked with =ka. Wh-questions may be i. unmarked by any sentence-final particle, ii. marked by the complementizer =na, or iii. marked by the complementizer =na and copula =da, as presented in (22a–c), respectively. In none of these cases can WHQ be marked with BPM, which is considered ungrammatical by the speakers. However, another possibility arises when =na or =na=da is followed by the sentence-final particle =jo, resulting in the presence of the BPM, as demonstrated in (22d).20 No data are available concerning the obligatoriness of BPM when =jo is present; however, it can be inferred that this BPM arises because of the presence of the sentence-final particle =jo rather than the WHQ structure itself, considering that WHQ s without =jo cannot be marked with BPM, as presented in (22a–c).
(22) Example of a WHQ sentence: ‘What do you need?’ with nani ‘what’ and joNda ‘need’
|
a. |
nani joNda |
nani joNda=*ka/*↗/*ka↗ |
|
|
b. |
nani joNda=na |
nani joNda=na=*ka/*↗/*ka↗ |
|
|
c. |
nani joNda=na=da |
nani joNda=na=da=*ka/*↗/*ka↗ |
|
|
d. |
nani joNda=na(=da)=jo↗ |
Figure 2.5
Pitch pattern examples of interrogative sentences pronounced by NS3
Acknowledgements
This work was supported by the Leading Initiative for Excellent Young Researchers (MEXT, Japan); the NINJAL collaborative research projects “Research on the Conservation of Endangered Languages”, “Evidence-based Study on the Intonational Diversity of Japanese and Ryukyuan”. I would like to express my heartfelt gratitude to the native speakers who kindly cooperated with the fieldwork. Their contributions were invaluable to this research. Any remaining errors or inaccuracies are, of course, solely the responsibility of the author.
Original map data for Figures 2.1 and 2.2 are downloaded from
As far as is dealt with in this chapter, there are only two differences between this chapter’s prosodic data and the data from in previous studies: i. the difference between the pitch patterns of unaccented words and word-final syllable accented words in isolated pronunciation, which is mentioned in footnote 7 below; and ii. the tonal class of uta ‘song’ mentioned in § 2.8.
This description is based on an analysis that largely posits the same phonemes in the same positions as Standard Japanese. This view has been found in many previous studies dealing with Tohoku varieties, and this chapter also follows this view (cf. Ohashi 2002). However, some previous studies have pointed out that such an analysis is problematic (e.g., Inoue 1968).
In this chapter, an asterisk (*) preceding a form indicates that it is ill-formed.
Noteworthily, although found in only some words, word initial /N/ is allowed in Akita (e.g., /Nda/ ‘yes’, /Nɡa/ ‘you’).
This section uses nouns as examples; however, here is an overview of verbs and adjectives. In verbs and adjectives also, a distinction exists between accented and unaccented (e.g., kake’-ru ‘to hang’ vs. kake-ru ‘to lack’). However, unlike nouns, the accent kernel’s position is not contrastive and is determined by rules established for each conjugation form. For example, in the non-past affirmative, the accent kernel is placed on the penultimate syllable for any accented verb; conversely, in the non-past negative, the accent kernel is placed on the ultimate syllable (e.g., mi’-ru ‘look’, oki’-ru ‘wake up’, tano’m-u ‘ask’, mi-ne’ ‘do not look’, oki-ne’ ‘do not wake up’, tanom-ane’ ‘do not ask’).
This suggests that the pitch patterns of unaccented words and words accented on the word-final syllable in isolated pronunciation are identical (see the citation forms of 1-
Notably, the fo peak in the word a’bura ‘oil’, which has an accent kernel on the initial syllable, is observed within the second syllable; however, the auditory impression is that lowering begins with the first syllable. This type of apparent peak delay is sometimes observed in Akita. The details of the conditions for the occurrence of this peak delay are unclear, but in this section, it is important that this word is an accented word.
Moreover, the same behavior is observed in Tokyo: A pitch fall in particles is frequently not observed when they are attached to an accented word, but this is not always the case (Maekawa and Igarashi 2006).
Akita is referred to as a syllabeme dialect, wherein the isochrony of the mora is not readily apparent and special morae (tokushu haku) are frequently pronounced significantly shorter than independent morae (jiritsu haku), as proposed by Sibata (1962). For instance, the cognate of [doːteɴ] ‘being surprised and stunned’ in Tokyo is pronounced [dodeɴ̆] in Akita. Additionally, Maekawa (1984) reported that in three-mora CVCCV words, the final CV (e.g., [to] in [mat.to] ‘mat’) is pronounced extremely briefly in Akita, precipitating an overall word duration nearly identical to that of a two-mora CVCV word (e.g., [ma.to] ‘target’). Thus, while the presence of a “phonetic mora” may not be prominent in Akita, the existence of a “phonological mora” and its function as a counting unit are evident, as demonstrated in this section.
Noteworthily, as in Tokyo, certain four-mora loanwords are realized as unaccented words in Akita (e.g., amerika ‘America’ and koNsome ‘consommé’). The conditions for the occurrence of unaccented loanwords are probably identical to those of Tokyo described by Kubozono (1996) and Kawahara (2015: 457–459), among others.
An anonymous reviewer questioned the necessity of assuming that the accent is first put on the antepenultimate mora and subsequently “displaced” to an independent mora. We can exemplify this by stating that the second half of long vowels and diphthongs and codas cannot be pronounced alone basically but are segmentally dependent on the vowel immediately preceding them. Moreover, an element that is segmentally dependent on the preceding vowel is suprasegmentally dependent on the preceding vowel. Hence, when they would have an accent kernel, kernel shifts to the vowel immediately preceding it (cf. Kubozono and Homma 2002: 37–38). In this manner, one could think that the tone-bearing unit is not a syllable but a mora, and that when an accent kernel is assigned to a special mora, the accent kernel shifts to the immediately preceding independent mora. However, this idea has problems in that it explains the phenomenon by introducing the special concepts of special and independent mora, which are assumed only in the Japanese language.
Additionally, this point is underscored by the anonymous reviewer. Notably, however, though trivial, the store names hu.reN.doo.ru and ak.ku.su in (7e, g) exist only in Akita, and it is unlikely that these store names were borrowed from Tokyo.
An anonymous reviewer highlighted that because the pronunciation variant which has the accent kernel on the ultimate mora is not attested, a constraint that prohibits placing the kernel on the ultimate mora may exist. If such a constraint exists, it further reinforces the idea that the mora is the counting unit.
As mentioned in § 2.2, if the accent kernel is associated to the word-final syllable, and the word is pronounced in isolation, the L-tone of the kernel is not realized unless a particle follows that word.
Considering that in o
As in Tokyo Japanese, compounds in Akita exhibit the phenomenon of sequential voicing known as rendaku, wherein the initial voiceless consonant of the right element becomes voiced under certain conditions.
Although further research is needed, certain bimoraic right elements exhibit a deaccenting pattern when they combine with a bimoraic left element (e.g., kumo + zaru, *kumo + za’ru ‘Atelidae’ and mame + paN, *mame + pa’N ‘bean bread’). Moreover, this phenomenon is observed in Tokyo (Kubozono and Fujiura 2004).
Currently, only one example iro ‘color’ has been determined to be of the deaccenting type. However, because several deaccenting type words containing iro have been reported in Standard Japanese (see Akinaga 1998:180, e.g., tera ‘temple’, mura ‘village’, and so on), these could also be deaccenting types in Akita, though this has not yet been investigated.
Hidaka (2000:117, 119) stated that /=kja/ derives from /=ka=e/, which comprises the question particle /=ka/ and the sentence-final particle /=e/, and can be used in especially polite WHQ. However, NS2 and 3 commented that they do not use /=kja/ frequently and cannot make grammatical judgments. No study has yet been conducted with NS1 on /=kja/ in WHQ s.
References
Akinaga, Kazue (1998) Kyōtsūgo no akusento [Accent of Standard Japanese]. In NHK (ed.) Nihongo hatsuon akusento jiten [The Japanese language pronunciation and accent dictionary], 174–221. Tokyo: Nihon Hoso Shuppan Kyokai.
Akita Prefecture Open Data (2023) Akita-ken nenrei-betsu jinkō ryūdō chōsa [Akita Prefecture population flow study by age group]. https://ckan.pref.akita.lg.jp/dataset/050008_toukeika_036. Last accessed 12 November 2024.
Hidaka, Mizuho (2000) Akita hōgen no bunpō [Grammar of Akita dialect]. In Akita Prefectural Board of Education (ed.) Akita no kotoba [Language in Akita], 74–132. Akita: Mumyosha Shuppan.
Hyman, Larry M. (2006) Word-prosodic typology. Phonology 23(2): 225–257.
Inoue, Humio (1968) Tōhoku hōgen no shiin taikei [Consonant system of the Tōhoku dialect]. Gengo Kenkyu 52: 80–98.
Kawahara, Shigeto (2015) The phonology of Japanese accent. In Haruo Kubozono (ed.) Handbook of Japanese phonetics and phonology, 445–523. Berlin, Boston, Munich: Walter de Gruyter.
Kondo, Sugaye (2012) Akita-ken Oga hōgen dōshi no meireikei kateikei ni tsuite [On the hypothetical/imperative forms of the Oga dialect verbs]. Seirei Joshi Tanki Daigaku Kiyo 40: 47–60.
Kondo, Sugaye (2015) Akita-ken Oga hōgen no dōshi katsuyōkei to akusento: ichirui dōshi hen [Verbal conjugation and its prosody of the Oga subdialect in Akita: Category I verbs]. Seirei Joshi Tanki Daigaku Kiyo 43: 46–61.
Kondo, Sugaye (2016a) Akita-ken Oga hōgen no dōshi katsuyō to akusento: Nirui oyobi fukisoku dōshi hen [Verbal conjugation and its prosody of the Oga subdialect in Akita: Group II and irregular verbs]. Seirei Joshi Tanki Daigaku Kiyo 44: 30–46.
Kondo, Sugaye (2016b) Akita-ken Oga hōgen no dōshi katsuyō to akusento: Keiyōshi oyobi kopyura hen [Verbal conjugation and its prosody of the Oga subdialect in Akita: The adjective and the copulae]. Tohoku Daigaku Gengogaku Ronshu 24: 31–42.
Kubozono, Haruo (1996) Syllable and accent in Japanese: evidence from loanword accentuation. The Bulletin of the Phonetic Society of Japan 211: 71–82.
Kubozono, Haruo (2008) Japanese accent. In Shigeru Miyagawa and Mamoru Saito (eds.) Handbook of Japanese linguistics, 165–191. Oxford: Oxford University Press.
Kubozono, Haruo and Takeru Homma (2002) Onsetsu to mōra [Syllables and morae]. Tokyo: Kenkyusha.
Kubozono, Haruo and Yayoi Fujiura (2004) Morpheme-dependent nature of compound accent in Japanese: An analysis of “short” compounds. Phonological Studies 7: 9–16.
Maekawa, Kikuo (1984) Akita hōgen sokuon no jizoku jikan—‘Sunzumari’ no jittai to seiin—[Duration of geminate in Akita: the reality and cause of short sawed off pronunciation]. Hogen Kenkyu Nenpo 27: 231–247.
Maekawa, Kikuo and Yosuke Igarashi (2006) 2-mōra joshi no inritsu-jō no dokuritusei: Nihongo hanashi kotoba kōpasu no bunseki [Prosodic independence of bimoraic accented particles: Analysis of the Corpus of Spontaneous Japanese]. Onsei Kenkyu 10(2): 33–42.
McCawley, James D. (1968) The Phonological component of a grammar of Japanese. The Hague: Mouton.
McCawley, James D. (1978) What is a tone language? In Victoria A. Fromkin (ed.) Tone: A linguistic survey, 113–131. New York: Academic Press.
Ohashi, Jun’ichi (2000) Akita hōgen no on’in akusento [The phonology and accent of Akita dialect]. In Akita Prefectural Board of Education (ed.) Akita no kotoba [Language in Akita], 32–73. Akita: Mumyosha shuppan.
Ohashi, Jun’ichi (2002) Tōhoku hōgen onsei no kenkyū [A study of Tōhoku dialect phonetics]. Tokyo: Ofu.
Sato, Minoru (1982) Akita-ken no hōgen [Dialects in Akita Prefecture]. In Kiichi Iitoyo, Sukezumi Hino and Ryoichi Sato (eds.) Kōza hōgengaku 4: Hokkaidō Tōhoku chihō no hōgen [Dialectology course 4: dialects in Hokkaidō and Tōhoku region], 271–295. Tokyo: Kokusho Kankokai.
Sato, Minoru (2000) Akita no kotoba gaisetu [An overview of the Akita language]. In Akita Prefectural Board of Education (ed.) Akita no kotoba [Language in Akita], 2–31. Akita: Mumyosha Shuppan.
Sibata, Takesi (1962) On’in [Phonology]. In The Society for Japanese Linguistics (ed.) Hōgengaku gaisetsu [An introduction to dialectology], 137–208. Tokyo: Musashino Shoin.
Suganuma, Kentaro (2022) Akita-ken Yokote-shi hōgen no yūkaku dōshi ni kansuru akusento kisoku—jakkyō futto wo mochiita bunseki—[On accentual pitch patterns of accented verbs in the Yokote dialect: An analysis based on iambic feet]. Gengo Kenkyu 161: 63–89.
Tojo, Misao (1954) Nihon hōgengaku [Japanese dialectology]. Tokyo: Yoshikawa Kobunkan.
Uwano, Zendo (1989) Nihongo no akusento [Accent in Japanese]. In Miyoko Sugito (ed.) Kōza Nihongo to Nihongo kyōiku 2: Nihongo no onsei on’in jō [Series on Japanese language and Japanese language education 2: Japanese phonetics and phonology volume 1], 178–205. Tokyo: Meiji Shoin.
Uwano, Zendo (1999) Classification of Japanese accent systems. In Shigeki Kaji (ed.) Proceedings of symposium cross-linguistic studies of tonal phenomena: tonogenesis, typology, and related topics, 151–178. Tokyo: ILCAA.
Uwano, Zendo (2012) Three types of accent kernels in Japanese. Lingua 122(13): 1415–1440.
Yamaguchi, Kyoko (2010) The difference in accentuation between the present and the past tenses of verbs in Japanese. Journal of the Phonetic Society of Japan 14: 1–10.