Show Summary Details

Page of

date: 23 August 2017

Accent in Japanese Phonology

Abstract and Keywords

The word accent system of Tokyo Japanese might look quite complex with a number of accent patterns and rules. However, recent research has shown that it is not as complex as has been assumed if one incorporates the notion of markedness into the analysis: nouns have only two productive accent patterns, the antepenultimate and the unaccented pattern, and different accent rules can be generalized if one focuses on these two productive accent patterns.

The word accent system raises some new interesting issues. One of them concerns the fact that a majority of nouns are ‘unaccented,’ that is, they are pronounced with a rather flat pitch pattern, apparently violating the principle of obligatoriness. A careful analysis of noun accentuation reveals that this strange accent pattern occurs in some linguistically predictable structures. In morphologically simplex nouns, it typically tends to emerge in four-mora nouns ending in a sequence of light syllables. In compound nouns, on the other hand, it emerges due to multiple factors, such as compound-final deaccenting morphemes, deaccenting pseudo-morphemes, and some types of prosodic configurations.

Japanese pitch accent exhibits an interesting aspect in its interactions with other phonological and linguistic structures. For example, the accent of compound nouns is closely related with rendaku, or sequential voicing; the choice between the accented and unaccented patterns in certain types of compound nouns correlates with the presence or absence of the sequential voicing. Moreover, whether the compound accent rule applies to a certain compound depends on its internal morphosyntactic configuration as well as its meaning; alternatively, the compound accent rule is blocked in certain types of morphosyntactic and semantic structures.

Finally, careful analysis of word accent sheds new light on the syllable structure of the language, notably on two interrelated questions about diphthong-hood and super-heavy syllables. It provides crucial insight into ‘diphthongs,’ or the question of which vowel sequence constitutes a diphthong, against a vowel sequence across a syllable boundary. It also presents new evidence against trimoraic syllables in the language.

Keywords: Japanese accent, pitch accent, unaccented word, default accent, loanword accent, compound accent, rendaku, emergence of the unmarked, syllable structure, branching condition

1. Basic Features of Japanese Pitch Accent

The word accent system of standard Tokyo Japanese has been studied for many decades from both phonetic and phonological perspectives. These studies have revealed the basic properties of the system, which are summarized here.

1.1. Pitch Accent System

Japanese dialects have pitch accent systems, systems where pitch, rather than intensity or other phonetic features, is used as the primary phonetic property of words (Beckman, 1986). Moreover, most dialects, including Tokyo Japanese, use pitch contrastively (see Haraguchi, 1977; Kubozono, 2012, 2015b; Uwano, 1999, 2012 among others, for the diversity of pitch accent systems in Japanese). Most of them are sensitive to an abrupt pitch fall and use it as the primary phonetic correlate of the phonological pitch accent (Hattori, 1960; McCawley, 1968; Pierrehumbert & Beckman, 1988). This can be demonstrated by the following data from Tokyo Japanese, where lexical meanings can be differentiated by the presence or absence of a pitch fall as well as the position of the pitch fall. In (1) and the rest of this article, apostrophes are used to denote word accent, or the position where a sudden pitch fall occurs. Words without an apostrophe are unaccented words that exhibit a rather flat pitch pattern, often beginning with a phrase-initial pitch rise—they are marked with superscript° in this article so that they can be clearly distinguished from accented words. Dots and hyphens indicate syllable and morpheme boundaries, respectively; /ga/ is a nominative particle (Nom).

(1)

Accent in Japanese Phonology

(2)

Accent in Japanese Phonology

Tokyo Japanese permits multiple accent patterns for nouns, which increase in number as the noun becomes phonologically longer. In contrast, adjectives and verbs exhibit only two patterns irrespective of the length of the word; they contrast only in the presence or absence of pitch accent, since accent is fixed on the penultimate mora and hence its position does not function contrastively.

1.2. Unaccented Words

Tokyo Japanese permits (n+1) patterns for nouns with n-syllable length: monosyllabic nouns exhibit two patterns, bisyllabic nouns three patterns, and trisyllabic nouns four patterns, etc.1 The additional one pattern in each case is attributable to the unaccented class of words, the feature that is attested in some pitch accent languages such as Japanese dialects (Kubozono, 2012), Northern Bizkaian Basque (Gussenhoven, 2004), and Nubi (Gussenhoven, 2006).

Interestingly, this class of words is not marginal in the lexicon but accounts for a majority of nouns and a good portion of verbs and adjectives. This can be seen from Table 1, which compares the ratio of trimoraic accented nouns with that of unaccented ones in the three major types of Japanese vocabulary—native, Sino-Japanese, and foreign (loanwords from languages other than Chinese).

Table 1. Accentedness in Three-Mora Nouns

Accented, %

Unaccented, %

Native words

29

71

Sino-Japanese words

49

51

Foreign words

93

7

Average

48

52

Note: Reprinted from H. Kubozono (2006). Where does loanword prosody come from? A case study of Japanese loanword accent. Lingua, 116, 1140–1170.

Not surprisingly, most minimal pairs that contrast in word accent contrast in terms of the accentedness, between accented and unaccented patterns. This is true not only in verbs and adjectives but also in nouns (see (1) and (2) above). This means that the contrastive function of word accent in Japanese is attributed primarily to the presence or absence of pitch accent rather than its position.

1.3. Default Accent

While Tokyo Japanese permits (n+1) accent patterns for nouns of n-syllable length, this does not mean that the multiple accent patterns are equally distributed across the vocabulary. First of all, nouns fall largely into two accent groups, those accented on the antepenultimate mora and those that are unaccented, while nouns accented in other positions are relatively rare. This can be seen from the data in Table 2, which shows how infrequent the final-accent and penultimate-accent patterns are in three-mora nouns. In fact, the two dominant accent patterns combined account for 95% of all three-mora nouns. The same is true of four-mora or longer words: the accent patterns other than the two dominant ones become rarer as the word becomes longer. This means that the antepenultimate accent represents the default accent pattern in accented nouns in Tokyo Japanese (see Katayama, 1998; Kubozono, 1996, 2006, 2008 for a new pre-antepenultimate pattern in nouns, as well as the crucial similarities between the antepenultimate rule and the Latin accent rule). The same observation led Kubozono (2008, 2011) to propose that nouns in Tokyo Japanese basically form a two-pattern system just like its verbs and adjectives, but with a certain number of lexical exceptions, especially in the native vocabulary.

Table 2. Accent Patterns and Their Frequencies in Three-Mora Accented Nouns (N = 3,788 words)

Accent Position Word type

Antepenultimate Mora, %

Penultimate Mora, %

Final Mora, %

Native words

59

33

9

Sino-Japanese words

95

2

3

Loanwords

96

2

2

Average

89

7

4

A second noteworthy asymmetry found in Japanese accentuation is that the balance between accented and unaccented words varies greatly depending on the lexical strata. As Table 1 shows, the unaccented pattern accounts for a good majority of three-mora native nouns and about a half of Sino-Japanese nouns, but it accounts for less than 10% of loanwords. The general tendency observed here is that words become unaccented as they exist for longer periods in the language. Moreover, the extremely low ratio of the unaccented pattern in loanwords can be accounted for by positing that native speakers of Japanese are sensitive to the sudden pitch fall in the citation forms of English words, just as they are when they perceive their own language (Kubozono, 2006). Namely, Japanese is faithful to the input of English words, which accounts for 84% of all loanwords in modern Japanese (Sibata, 1994).

Most accented loanwords are accented on the antepenultimate mora or, more precisely, on the syllable containing the third mora from the end of the word: /ba’.na.na/ ‘banana,’ /o.re’n.zi/ ‘orange,’ /wa.si’n.ton/ ‘Washington’ (McCawley, 1968). As Table 2 suggests, this loanword accent rule comes from the native phonology of Japanese where it accounts for a majority of accented native nouns (Kubozono, 2006). In other words, the dominant accented pattern in non-loanwords is simply applied to words that have entered the language recently. This is an instance of TETU, or the emergence of the unmarked (McCarthy & Prince, 1994).

The antepenultimate effect in accented nouns can be attributed to the interactions of several independent constraints, notably the Nonfinality Constraint penalizing phonological prominence at the very edge of the word and the Edgemostness Constraint (or any analogous constraint) that prohibits the prominence from occurring too far from the same edge (Kubozono, 2006, 2008; Shinohara, 2000).

1.4. Compound Accent

Unlike morphologically simplex words, morphologically complex words, or compounds, are more or less rule-governed with respect to their accentuation. They are similar to morphologically simplex words in that they may be either accented or unaccented. However, they are different from the morphologically simplex words in that they are mostly accented rather than unaccented.

Putting aside unaccented compounds for a moment, the basic compound accent rule in Tokyo Japanese is to preserve the lexical accent of the final member in compounds with the accent (or any other phonological feature) of non-final members playing little or no role. This is illustrated in (3).

(3)

Accent in Japanese Phonology

This right-dominant rule is subject to the NonFinality Constraint by which the accent on the very final syllable (and often on the very final bimoraic foot) cannot be preserved. In such a case, a default compound accent emerges on the syllable immediately before the final member if it is ‘short’ (monomoraic or bimoraic), as in (4a), or on the initial syllable of the final member if it is ‘long’ (three-mora or four-mora long), as in (4b).2

(4)

Accent in Japanese Phonology

This is another instance of the emergence of the unmarked, a case where the default accented pattern of the language has surfaced. Two points must be noted here. First, the two patterns in (4) can be generalized if the notion of bimoraic foot is introduced: compound accent is placed on the rightmost, non-final foot (Kubozono, 1995a, 1997). Moreover, the accent patterns in (4) can be further generalized with the antepenultimate rule responsible for the accentuation of accented simplex nouns (Kubozono, 2008, 2011). These accent rules place phonological prominence as close to the end of the word as possible, while avoiding the final syllable or bimoraic foot at the right edge of the word. This is illustrated in (5), where the foot structure is minimally shown.

(5)

Accent in Japanese Phonology

The idea of the emergence of the unmarked also explains why the same pattern emerges in compounds whose final member is lexically unaccented: /to/ in (6a) changes into /do/ by rendaku voicing (see Section 3.1).3

(6)

Accent in Japanese Phonology

While the accentuation of accented compounds is largely rule-governed, that of unaccented compounds is more or less lexical, since it can be attributed largely to the class of morphemes called ‘deaccenting morphemes’ (McCawley, 1968). As illustrated in (7), these morphemes exert a deaccenting effect on the nouns to which they are attached. They are constrained by both morphological and phonological structures. Morphologically, they are either native or Sino-Japanese morphemes in origin. Phonologically, they are either monomoraic or bimoraic in length. Stated conversely, neither trimoraic or longer morphemes, nor foreign (loan) morphemes can be deaccenting morphemes. Furthermore, almost all deaccenting morphemes are accented on their final syllable when they are pronounced in isolation although not all finally accented elements are deaccenting morphemes (Kubozono, 1988; Poser, 1984).

(7)

Accent in Japanese Phonology

It is worth adding here that the deaccenting effect of deaccenting morphemes is sensitive to the internal structure of the compounds. For example, the following compound nouns bear a compound accent in the default position, although they end in a deaccenting morpheme. The difference between the unaccented compounds in (7) and the accented ones in (8) is that the deaccenting morphemes are the final constituents of the entire compounds in the former, whereas they form an internal compound with the immediately preceding elements in the latter.4

(8)

Accent in Japanese Phonology

2. Unaccentedness

One of the most crucial questions—and the biggest mystery—regarding Japanese pitch accent concerns the emergence of the unaccented pattern, the pattern that accounts for a majority of native and Sino-Japanese words. This peculiar accent pattern emerges due to various factors. We have already seen one factor, the morphological category of deaccenting morphemes, which operates in compounds. Unaccented words emerge for other reasons, too, as will be discussed in this section.

2.1. Unaccented Loanwords

Generally speaking, the unaccented pattern in morphologically simplex words is difficult to explain, especially in native words. For example, it is not possible to explain why /a’.me/ ‘rain’ is lexically accented, while /a.me°/ ‘candy’ is unaccented. However, this is not true of loanwords, where the distinction between the two accent patterns is by and large predictable.

While the average unaccentedness ratio is only 10% in loanwords on the whole (Sibata, 1994), the ratio rises to 19% in four-mora loanwords (Table 3), and goes up even higher in those that end in a sequence of two light, i.e., monomoraic, syllables (Table 4). Relevant statistics are given in Tables 3 and 4, which are based on NHK (1985) and NHK (1998), respectively: H and L in Table 4 denote heavy (bimoraic) and light syllables, respectively. Typical examples are given in (9).

Table 3. Word Length and the Percentage of the Unaccented Pattern in Loanwords (N = 1,863 words)

Three-Mora, %

Four-Mora, %

Five-Mora, %

Average, %

5

19

8

13

Note: Reprinted from H. Kubozono (2006). Where does loanword prosody come from? A case study of Japanese loanword accent. Lingua, 116, 1140–1170.

Table 4. Prosodic Structure and the Percentage of the Unaccented Pattern in Four-Mora Loanwords (N = 963 words)

Prosodic Structure

LLLL

HLL

LHL

LLH

HH

Unaccentedness ratio

54

45

24

19

7

Note: Reprinted from H. Kubozono (2006). Where does loanword prosody come from? A case study of Japanese loanword accent. Lingua, 116, 1140–1170.

(9)

Accent in Japanese Phonology

The data in Tables 3 and 4 raise an interesting question of why the unaccented pattern is so popular in four-mora loanwords with the specific prosodic structure. This question can be answered, at least in part, by the fact that the unaccented pattern tends to appear under the same conditions in native and Sino-Japanese words. As shown in Table 5, it is generally most popular in four-mora nouns as opposed to nouns of other lengths. Moreover, Light-Light sequences represent the most typical word-final structure in native words. This suggests that the two phonological conditions triggering unaccentedness in loanwords come from the native phonology.

Table 5. Unaccentedness Ratios in Native and Sino-Japanese Nouns Combined as a Function of Word Length

Three-Mora, %

Four-Mora, %

Five-Mora, %

Average, %

53

66

30

54

2.2. Unaccented Acronyms

The two phonological conditions illustrated in Tables 3 and 4 can be observed more clearly in alphabetic acronyms such as BS (broadcast satellite) and FM (frequency modulation), which are commonly used in contemporary Japanese. In prosodic terms, alphabetic acronyms are mostly accented rather than unaccented: when accented, they behave like compounds in preserving the lexical accent of their final members although they are not constrained by the NonFinality Constraint (Kubozono, 2003, 2010).

(10)

Accent in Japanese Phonology

However, alphabetic acronyms permit the unaccented pattern in some phonological contexts. Typically, they tend to become unaccented if they are four moras long and end in a sequence of light syllables, as illustrated in (11). This observation can be borne out statistically, too, as shown in Table 6.

(11)

Accent in Japanese Phonology

Table 6. Unaccentedness Ratios in Alphabetic Acronyms

Length

Four-Mora Long, %

Five-Mora or Longer, %

Prosodic structure

LLLL

HLL

LLH + HH

Unaccentedness ratio

97

70

0

0

80

Note: Adapted from H. Kubozono (2003). Accent of alphabetic acronyms in Tokyo Japanese. In T. Honma, M. Okazaki, T. Tabata, & S. Tanaka (Eds.), A new century of phonology and phonological theory (pp. 356–370). Tokyo: Kaitakusha; and from H. Kubozono (2010). Accentuation of alphabetic acronyms in varieties of Japanese. Lingua, 120, 2323–2335.

It is interesting to find that alphabetic acronyms show the unaccented pattern in exactly the same phonological contexts as morphologically simplex loanwords like those in (9). On the other hand, it remains a mystery why they exhibit non-compound behavior in these contexts, although they otherwise behave like compounds as shown in (10).

2.3. Truncated Compounds

We have seen that four-mora nouns tend to be unaccented in morphologically simplex words and alphabetic acronyms. The same tendency is found in truncated compounds, too. Japanese displays two major patterns of compound truncation: one is to retain the initial element and delete the second, e.g., /kee.tai°/ for /kee.tai-de’n.wa/ ‘mobile phone,’ while the other is to combine the initial two moras of the first element with those of the second one (Ito, 1990). The second of these generally yields four-mora truncated words, most of which are unaccented.

(12)

Accent in Japanese Phonology

Note that these truncated outputs are unaccented regardless of the prosodic structure involved: they are unaccented not only when they end in a sequence of light syllables as in (12a, b) but also when they end in a heavy syllable as in (12c, d). Truncated compounds are thus free from the prosodic structure condition imposed on unaccented loanwords and alphabetic acronyms. The reason for this difference is not clear, but it can probably be related to the fact that native compounds also tend to be unaccented if they consist of two bimoraic nouns. For example, /hi.to-ka.ge°/ ‘person + shadow; silhouette’ and /wa.go-yo.mi°/ ‘Japanese word + reading; Japanese kun reading’ are unaccented as opposed /hi-to’.ka.ge/ ‘fire + lizard; fire lizard (a Pokémon’s name)’ and /wa-go’.yo.mi/ ‘Japan + calendar; traditional Japanese calendar,’ which attract a compound accent on the second element. Truncated compounds are hence similar to native compounds consisting of two bimoraic nouns in that they do not obey the regular compound accent rule and become unaccented instead.

2.4. Deaccenting Pseudo-morphemes

Another source of unaccented nouns in Tokyo Japanese is a class of loanwords ending in a particular sound sequence. Typically, loanwords tend to be unaccented if they end in /ia/, /in/, or /ingu/ (see Giriko, 2009 for exceptions).

(13)

Accent in Japanese Phonology

Two points must be noted here. First, these loanwords do not fulfill the phonological conditions described in Tables 3 and 4 and are hence different from the ‘ordinary’ unaccented loanwords sketched above. Second, the unaccented loanwords in (13) are conditioned not only morphologically but also semantically: those in (13a) are place names, usually those of the country or area, those in (13b) are medical terms typically referring to medicines, and those in (13c) are deverbal nouns denoting a particular action. These semantic conditions are necessary for the deaccenting effect since the same loanwords do not become unaccented if they have different meanings: those in (13b) would be accented if they were used as personal names, for example. These facts led Giriko (2009) to propose that /ia/, /in/, and /ingu/ function as deaccenting foreign morphemes comparable to the native and Sino-Japanese morphemes that exert the same effect as shown in (7). This analysis implies that the unaccented loanwords in (13) are pseudo-compounds involving a pseudo-morpheme boundary.

(14)

Accent in Japanese Phonology

2.5. Compound Nouns With an Overlong Final Member

Another interesting source of unaccented nouns in Tokyo Japanese is compounds involving a long unaccented noun as their final member. Generally speaking, compound nouns with an unaccented final member are accented by the compound accent rule, as illustrated in (6): namely, they attract a default compound accent immediately before the final member if it is ‘short’ and on the initial syllable of the final member if it is ‘long.’ However, this rule admits a notable class of exceptions. If the final member is an unaccented noun with five or more moras, the unaccented nature of this member is preserved in the entire compound. This is illustrated in (15): compounds in (15a) have long loanwords as their final members and exhibit a contrastive accent pattern with /mi.na.mi-a’.me.ri.ka/ and /nyuu-me.’ki.si.ko/ in (6), which display a regular compound behavior. Compound nouns in (15b) involve an unaccented native or Sino-Japanese compound as their final members. In both cases, the accent, if any, of the first member is lost, like ordinary compounds in (6), but the second member does not attract a default compound accent.

(15)

Accent in Japanese Phonology

These observations led Kubozono, Ito, and Mester (1997) to propose a foot-based account of compound accentuation, according to which compound nouns preserve the accentuation—both accentedness and accent position—of their final member if this member is longer than two feet. Under this analysis, the compounds in (6) and those in (15) differ in prosodic structure as shown in (16), which gives rise to the difference in accent patterns (parentheses denote foot boundaries). In other words, compounds with a short or long final member achieve the accentual integration of the two members, as in (16a), while those with an overlong second member in (16b) respect the accentual independence of this second member.

(16)

Accent in Japanese Phonology

Moreover, this analysis allows us to generalize the unaccented pattern in (15) with the accented pattern in (17), where the final member is more than four moras long and keeps its original accent in compounds.

(17)

Accent in Japanese Phonology

Under the Kubozono et al. (1997) analysis, the crucial difference lies between compounds whose final member is up to two feet long and those with a longer final member. Those in the former class tend to preserve the accent of their final member, whereas those in the latter class preserve the accentuation (accentedness and accent) of the same member. This analysis further accounts for the mysterious unaccented pattern found in compound nouns, such as those in (18), whicht have an unaccented four-mora final member made up of three (Sino-Japanese) morphemes. Adopting the general idea that each morpheme forms at least one foot (Kubozono, 1997), the final members in question are comprised of more than two feet just like the final members of the compounds in (16b) and, hence, keep their unaccented status in compounds.

(18)

Accent in Japanese Phonology

2.6. Idiosyncratic Compounds

Finally, compound nouns often become unaccented in certain phonological contexts even if they do not end in a deaccenting morpheme. These are more or less idiosyncratic compounds that do not obey the regular compound accent rule. Specifically, these compounds differ from ordinary compound nouns in being sensitive to the phonological structure of their initial members.

Personal names ending in /zi.roo/ and /ta.roo/, denoted as X-ziroo and X-taroo here, are typical examples. These compound names yield three different accent patterns depending on the phonological length of their first member, that is, X. X-ziroo (i) takes the unaccented pattern if X is monomoraic, (ii) attracts a compound accent on the final syllable of X if it is bimoraic, and (iii) preserves the original accent of /zi’roo/ if X is three moras long or longer. The last of these is the regular compound accent pattern shown by ordinary compounds, as in /bii.ti-ba’.ree/ ‘beach volleyball’. X-taroo exhibits similar patterns to X-ziroo, the only difference being that it yields the unaccented pattern if X is monosyllabic rather than monomoraic (Kubozono, 1999). For example, /kin-ta.rooo/ is unaccented, while /ki’n-zi.roo/ is accented.

These three patterns are summarized in (19). It remains unclear why the unaccented pattern emerges in these personal names if the first member is short and, moreover, why X-ziroo and X-taroo exhibit the observed difference.

(19) Accentuation of X-ziroo and X-taroo

Accent in Japanese Phonology

Place names ending in the Sino-Chinese morpheme /tyo’o/ ‘town’ also display similar idiosyncratic patterns as they tend to be unaccented only if their initial member is three moras long. These compounds are exemplified in (20a), which contrasts with the regular compound accent pattern shown by other place names with the same morpheme given in (20b). Note that the place names in (20a) are five moras long but are nevertheless unaccented.

(20)

Accent in Japanese Phonology

The native morpheme /mo.no’/ ‘thing’ displays more complex accent patterns. This morpheme combines with other morphemes, especially deverbal nouns, to produce nouns very productively: /no.mi’-mo.no/ ‘to drink + thing; drink,’ /ta.be’-mo.no/ ‘to eat + thing; food’. Interestingly, the resultant compound nouns split into three accent groups, those that are accented as in (21a), those that are unaccented as in (21b), and those that fluctuate between the two patterns as in (21c).

(21)

Accent in Japanese Phonology

The general tendency is that accented first members yield accented compounds, while unaccented first members produce unaccented compounds (Kawahara, 2015). A careful examination of the data further reveals that the unaccented pattern is expanding its domain as in (21c), while the accented pattern readily occurs in very restricted contexts. In other words, the accented pattern occurs only if the initial member is derived from an accented two-mora verbal stem as in (21a), while the unaccented pattern now emerges elsewhere—if the initial member comes from a verb that is lexically unaccented as in (21b), or if it has an accented stem that is not two moras long as in (21c). Obviously, this situation is different from all other cases that have been discussed so far, where the unaccented pattern emerges in certain restricted contexts and accented patterns occur elsewhere.

Like /mo.no/ in (21), the suffixal morpheme /ya/ ‘shop’ also yields unaccented compounds except in some restricted environments: it generally yields unaccented compounds except when it is attached to two-mora nouns: /ko.me’-ya/ ‘rice shop,’ /ma.me’-ya/ ‘bean shop,’ /ku.tu’-ya/ ‘shoe shop’ vs. /o.ko.me-ya°/ ‘rice shop,’ /sa.ka.na-ya°/ ‘fishmonger,’ /ni.ma.me-ya°/ ‘cooked beans shop,’ /ka.ree-ya°/ ‘curry restaurant,’ /na.ga.gu.tu-ya°/ ‘boot shop’.

In sum, it is difficult to generalize the contexts where the unaccented pattern emerges in idiosyncratic compounds in Tokyo Japanese. What one could do at best is to define the conditions for each morpheme, that is, the conditions where each morpheme yields the unaccented pattern when it is combined with other morphemes.

3. Interactions With Other Processes and Structures

Having seen how accent patterns are determined in Tokyo Japanese, let us now consider how compound accentuation interacts with other phonological processes and other linguistic structures such as syntactic and semantic ones.

3.1. Interaction With Rendaku Voicing

Compound accentuation is related to rendaku, another phonological process characteristic of compounds. Rendaku is a sequential voicing process by which the first consonant of the non-initial member of compounds is voiced (/h/ alternates with /b/ in Japanese for historical reasons) (Ito & Mester, 2003; Vance, 2015).

(22)

Accent in Japanese Phonology

This process interacts with compound accentuation in that it appears to trigger deaccenting of the entire compound. This effect is most clearly observed in compounds ending in the morpheme /ta/ ‘rice field’, which attaches to other morphemes productively to yield family names, such as Shibata, Honda, Yamada, Yoshida. Interestingly, those names with rendaku voicing tend to be unaccented, as exemplified in (23a), whereas those without it tend to be accented, as in (23b) (Sato, 1989; Sugito, 1982).5 The other combinations do also exist, as shown in (24a, b), but they are much rarer.

(23)

Accent in Japanese Phonology

(24)

Accent in Japanese Phonology

A similar correlation can be found in personal names involving /si.ma’/ ‘island’: /na’.ka + si.ma’/, for example, is pronounced without an accent when it undergoes rendaku voicing, i.e., /na.ka-zi.ma°/, whereas its counterpart without rendaku tends to be accented, i.e., /na.ka’-si.ma/.

It is not clear whether rendaku voicing triggers deaccenting or accent patterns determine the application of rendaku. In either case, the correlation observed here presents an interesting case where segmental features interact with suprasegmental ones.

3.2. Interactions With Syntactic Structure

So far, we have restricted ourselves to compound words that are pronounced as one prosodic word, with at most one accent. However, Japanese permits many expressions that are compounds both morphologically and semantically but behave like phrases in phonological terms. In these compound expressions, the compound accent rule is blocked between their components, with the result that they are realized in more than one prosodic word. These multi-phrasal compounds are discussed in this and the next subsection.

One of the major sources of multi-phrasal compounds consists of three or more words and involves a right-branching structure. To take one pair of words, Boston Red Sox and New York Yankees exhibit contrastive accent patterns in Japanese, with the former manifested in two prosodic words. This prosodic difference can be accounted for if the internal morphosyntactic structure of the compounds is considered. Namely, the compound accent rule is blocked in the right-branching structure as in (25a), but not in its left-branching counterpart as in (25b): { } denote prosodic word boundaries.

(25)

Accent in Japanese Phonology

This structural condition, which is called ‘Branching Condition’ or ‘Right-branching Condition’ (Kubozono, 1988), accounts for many instances such as those in (26).

(26)

Accent in Japanese Phonology

These compounds are readily fused into one prosodic word if the right-branching structure is removed when they are shortened. This means that it is the morphosyntactic configuration, and not the semantic content of the compound, that is relevant to prosodic phrasing.

(27)

Accent in Japanese Phonology

The Branching Condition illustrated in (25a) and (26) accounts for the two phrasing patterns found in segmentally homophonous but structurally different pairs of compounds. For example, the compound expressions in (28) are structurally ambiguous, but they can be disambiguated by different prosodic patterns, biphrasal and monophrasal.6

(28)

Accent in Japanese Phonology

Given the Branching Condition, one may naturally wonder why the right-branching structure exhibits marked prosodic behavior, or, in other words, why right-branching compounds behave differently from their left-branching counterparts. A descriptive generalization is that a sequence of two elements in compounds, [A B], do not undergo the prosodic compounding rule if A does not modify B.

This generalization can be captured more formally in the following two ways. One is an analysis based on the syntactic notion of ‘c-command,’ which Otsu (1980) proposed to account for a similar branching effect in rendaku voicing. Under this analysis, [A B] undergoes the compound accent rule if B ‘c-commands’ A, that is, if the first branching node that dominates B also dominates A. In the two compound nouns in (25), for example, the second element c-commands the first in (25b), but not in (25a). The third element c-commands the second in (25a) and (25b), thus yielding biphrasal and monophrasal prosodic structures, respectively, in the entire compounds.

A second, and more functional, explanation is that [A B] undergoes the prosodic unification if B is its semantic head. This analysis posits that prosodic compounding is not merely a juncture phenomenon as is often assumed in the literature (Kubozono, 1995a), but is a head-marking process manifesting the modifier-head relation in compounds. Under this analysis, prosodic compounding is blocked between the first two elements in (25a) since the second element is not the head of the sequence. In contrast, it readily applies to the second and third elements in (25a) because the third element is the semantic head of the two-element sequence. The same analysis correctly predicts that compounding occurs between the first and second elements as well as between the second and third ones in (25b).

The two formal accounts proposed here explain the prosodic asymmetry between right-branching and left-branching structures equally well. However, the semantic analysis based on the the notion of semantic head turns out to be better and more general than the syntactic one involving the notion of c-command, as we will see shortly (Section 3.3).

Finally, it is worth emphasizing that the Right-branching Condition in question is a constraint of a general nature. For one thing, this condition constrains many phonological processes in Japanese including rendaku voicing, intonational phrasing by which two adjacent phrases are intonationally amalgamated into one ‘minor phrase,’ and downstep, by which word accent of one phrase lowers the pitch range of the following phrases in the same utterance (Kubozono, 1988). The effect of the Branching Condition on rendaku voicing is illustrated below: the second element /si’.ro/ ‘white’ undergoes voicing in (28b) but not in (28a) (Otsu, 1980; Sato, 1989).7

(29)

Accent in Japanese Phonology

The Right-branching Condition is also effective in languages other than Japanese. English, for example, exhibits a marked prosodic pattern in right-branching compounds such as [evening [computer class]], where the main stress appears on the second element, computer (Chomsky & Halle, 1968; Liberman & Prince, 1977). This contrasts with the regular compound stress pattern found in left-branching compounds such as [[computer class] instructor], where the main stress is placed on the initial element, computer. A similar right-branching effect is found across languages such as Ewe (Clements, 1973), Italian (Napoli & Nespor, 1976, unpublished manuscript), and Mandarin Chinese (Chen, 2000).

In sum, it is highly interesting to find that the same constraint is at work in different types of prosodic systems across languages as well as in different phonological processes within the same system.

3.3. Interactions With Semantic Structure

Prosodic phrasing in compounds is also constrained by the semantic structure involved. A most typical semantic structure that blocks the compound accent rule is the coordinate structure found in dvandva compounds (see Kubozono, 1988, 1995b for other types of semantic constraints). Thus, the two-element compounds in (30) fail to undergo the compounding rule and are hence manifested in two prosodic words (Kubozono, 1988, 1995b).

(30)

Accent in Japanese Phonology

The same constraint is at work in three-element or longer compounds, as exemplified below. What is interesting here is that prosodic compounding is blocked between two elements constituting a coordinate structure, but not in other positions. In three-element compounds in (31a–c), for example, the first and second elements are not unified into one prosodic word due to the semantic constraint, but the second and third elements are readily fused into one prosodic word: /to’o/ in (31a) is a deaccenting morpheme, as discussed in (7b). Note, furthermore, that this semantic constraint does not work any longer if the coordinate structure is resolved due to shortening, as illustrated in (32).

(31)

Accent in Japanese Phonology

(32)

Accent in Japanese Phonology

The Semantic Condition illustrated in (30) and (31) raises several important questions. The first one is why the coordinate structure blocks the prosodic process and how it can be formalized. A plausible account is based on the notion that prosodic compounding is a head-marking process, an idea that was proposed to account for the branching constraint above. Namely, prosodic compounding is blocked between the two elements constituting a coordinate structure since [modifier + head] relation does not hold there. In the case of (30a), for example, the right-hand member, /su.ro.ba’.kia/, is not the semantic head of the whole compound. This analysis has an advantage of being capable of providing a principled account for the Branching Condition and the Semantic Condition alike: the two seemingly different constraints can thus be generalized by the notion of semantic head.

A second question that arises regarding the semantic constraint in question is how general it is in natural languages. Previous studies have shown that the same constraint accounts for the lack of rendaku voicing in dvandva compounds in Japanese (Ito & Mester, 2003; Sato, 1989). This is illustrated in (33), where dvandva compounds are compared with non-dvandva ones.8

(33)

Accent in Japanese Phonology

The coordinate structure blocks prosodic processes in other languages, too. In English, for example, it is well known that the compound stress rule is blocked in dvandva compounds such as coca-cola, producer-director, Czecho-Slovakia, and parent-teacher association (Fudge, 1984; Ladd, 1984). Prosodic compounding is blocked in dvandva compounds in other languages such as Malayalam and Vedic Sanskrit (Han, 1994). These facts suggest that the Semantic Condition illustrated in (30) and (31) is a constraint of a rather general nature.

4. Implications for Other Phonological Structures

Japanese pitch accent sheds new light on other phonological structures of the language, particularly its syllable structure. A long-standing question regarding syllable structure in Japanese has been which vowel sequence forms a diphthong, a sequence that belongs to one syllable, and which does not. Equally important is the question of whether Japanese permits superheavy, i.e., trimoraic, syllables. Evidence from word accent provides invaluable insights into these questions.

Tokyo Japanese is basically a mora-counting language where phonological distances are measured in terms of the number of moras (McCawley, 1978). The famous antepenultimate rule sketched in Section 1.3, for example, places an accent on the third mora from the end of the word. However, Tokyo Japanese does not permit an accent on the non-head mora of heavy syllables; if an accent is placed on such a marked mora by rule, it automatically shifts one mora to the left, i.e., onto the head mora of the same syllable. Because of this, the antepenultimate rule is formulated as in (34) (McCawley, 1968).

(34)

Accent in Japanese Phonology

Similarly, the default compound accent is placed (a) on the syllable immediately before the final member if this member is ‘short’ (monomoraic or bimoraic) and (b) on the initial syllable of the final member if it is ‘long’ (three or four moras long).

(35)

Accent in Japanese Phonology

These rules can be used to examine which vowel sequence forms a diphthong as opposed to a heterosyllabic vowel sequence. This accent test shows that /ai/, /oi/, and /ui/ function as diphthongs in Tokyo Japanese, whereas other vowel sequences do not (Kubozono, 2005, 2015a). This is demonstrated in (36) and (37): the compound accent placed at the end of the first member shifts one mora to the left in (36) but not in (37).

(36)

Accent in Japanese Phonology

(37)

Accent in Japanese Phonology

It is interesting to note here that /au/ behaves differently from /ai/; that is, there is an asymmetry between /ai/ and /au/. Moreover, /ao/, /ae/, and /oe/ do not form diphthongs although they are legitimate vowel sequences in native Japanese morphemes just like the three vowel sequences in (36) (see Kubozono, 2015b for similar evidence from other dialects of Japanese).

The same accent test also gives a clue as to the status of superheavy syllables in Japanese. Although the language appears to permit several trimoraic syllables such as /ain/, /ein/, and /oin/, the accent test shows that these trimoraic sequences are, in fact, composed of two syllables, one monomoraic syllable plus a bimoraic one. This is illustrated in (38), where the inputs are underspecified with respect to syllabification.9

(38)

Accent in Japanese Phonology

Interestingly, /ai/, /oi/, and /ui/ pattern with /au/ in these instances, thus splitting into two syllables when followed by a moraic nasal. In other words, the three vowel sequences form diphthongs on their own, but not when they are followed by another non-head mora. On the other hand, /au/ does not form a diphthong in any context. What appears to form a superheavy syllable does not actually function as such. One potential exception to this is the past tense form of the verb /tooru/ ‘to pass,’ i.e., /toot-ta/, where the accent placed at the end of the stem moves to the left by two moras rather than one: /toot’-ta/ → /to’ot-ta/, */to.o’t-ta/ ‘passed’.

In summary, phonological tests using lexical pitch accent provide crucial insight into ‘diphthong-hood,’ or the question of which vowel sequence constitutes a diphthong as against a vowel sequence across a syllable boundary in Japanese. It also presents new evidence against trimoraic syllables in the language.

Acknowledgments

The work reported in this article was supported by the JSPS KAKENHI grants (Grant no. 25580098 and no. 26244022) and by the NINJAL collaborative research project ‘Cross-Linguistic Studies of Japanese Prosody and Grammar.’

Further Reading

Akinaga, K. (1985). Kyōtsūgo no akusento. In NHK (Ed.), NHK hatsuon akusento jiten (Appendix). Tokyo: Nihon Hōsō Shuppan Kyōkai. (Accent of standard Japanese: NHK pronunciation and accent dictionary.)Find this resource:

Beckman, M. E. (1986). Stress and non-stress accent. Dordrecht, The Netherlands: Foris.Find this resource:

Haraguchi, S. (1977). The tone pattern of Japanese: An autosegmental theory of tonology. Tokyo: Kaitakusha.Find this resource:

Haraguchi, S. (1991). A theory of stress and accent. Dordrecht, The Netherlands: Foris.Find this resource:

Hayata, T. (1999). Onchō no taiporojī Tokyo: Taishukan. (Prosodic typology.)Find this resource:

Hirayama, T. (Ed.). (1960). Zenkoku akusento jiten. Tokyo: Meiji Shoin. (All-Japan accent dictionary.)Find this resource:

Kawahara, S. (2015). The phonology of Japanese accent. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 445–492). Berlin: De Gruyter Mouton.Find this resource:

Kawakami, S. (1995). Nihongo akusento ronshū. Tokyo: Kyūkoshobō. (Papers on Japanese accent.)Find this resource:

Kubozono, H. (1988). The organization of Japanese prosody (PhD diss.). University of Edinburgh. Published by Kurosio, Tokyo, 1993.Find this resource:

Kubozono, H. (2006a). Where does loanword prosody come from? A case study of Japanese loanword accent. Lingua, 116, 1140–1170.Find this resource:

Kubozono, H. (2006b). Akusento no hōsoku. Tokyo: Iwanami. (Laws of accent.)Find this resource:

Kubozono, H. (2008). Japanese accent. In S. Miyagawa & M. Saito (Eds.), The handbook of Japanese linguistics (pp. 165–191). Oxford: Oxford University Press.Find this resource:

Kubozono, H. (2011). Japanese pitch accent. In M. van Oostendorp, C. Ewen, E. Hume, & K. Rice (Eds.), The Blackwell companion to phonology, Vol.5 (pp. 2879–2907). Malden, MA: Wiley-Blackwell.Find this resource:

Kubozono, H. (2012). Varieties of pitch accent systems in Japanese. Lingua, 122, 1395–1414.Find this resource:

Kubozono, H. (2013). Japanese word accent. In M. Aronoff (Ed.), Oxford Bibliographies in Linguistics. New York: Oxford University Press.Find this resource:

Matsumori, A., Nitta, T., Kibe, N., & Nakai, Y. (Eds.). (2012). Nihongo akusento nyūmon. Tokyo: Sanseidō. (Introduction to Japanese accent.)Find this resource:

McCawley, J. D. (1968). The phonological component of a grammar of Japanese. The Hague, The Netherlands: Mouton.Find this resource:

McCawley, J. D. (1978). What is a tone language? In V. Fromkin (Ed.), Tone: A linguistic survey (pp. 113–131). New York: Academic Press.Find this resource:

NHK (Ed.). (2010). NHK hatsuon akusento jiten. Tokyo: Nihon Hōsō Shuppan Kyōkai. (NHK pronunciation and accent dictionary.)Find this resource:

Poser, W. J. (1984). The phonetics and phonology of tone and intonation in Japanese. (PhD diss.), MIT.Find this resource:

Poser, W. J. (1990). Evidence for foot structure in Japanese. Language, 66, 78–105.Find this resource:

Sugito, M. (1982). Nihongo akusento no kenkyū. Tokyo: Sanseidō. (Studies on Japanese accent.)Find this resource:

Tokugawa, M. (Ed.). (1980). Akusento (Ronshū nihongo kenkyū 2). Tokyo: Yūseidō. (Accent: Papers on Japanese studies 2.)Find this resource:

Uwano, Z. (1999). Classification of Japanese accent systems. In S. Kaji (Ed.), Cross-linguistic studies of tonal phenomena: Tonogenesis, typology, and related topics (pp. 151–186). Tokyo: ILCAA.Find this resource:

Uwano, Z. (2012). Three types of accent kernels in Japanese. Lingua, 122, 1415–1440.Find this resource:

References

Beckman, M. E. (1986). Stress and non-stress accent. Dordrecht, The Netherlands: Foris.Find this resource:

Chen, M. (2000). Tone sandhi: Patterns across Chinese dialects. Cambridge Studies in Linguistics 92. Cambridge, U.K.: Cambridge University Press.Find this resource:

Chomsky, N., & Halle, M. (1968). The sound pattern of English. New York: Harper & Row.Find this resource:

Clements, G. N. (1973). Tone ad syntax in Ewe. In D. J. Napoli (Ed.), Elements of tone, stress, and intonation (pp. 21–99). Washington, D.C.: Georgetown University Press.Find this resource:

Fudge, E. (1984). English word-stress. London: George Allen & Unwin.Find this resource:

Giriko, M. (2009). Nihongo ni okeru go-ninshiki to heibangata akusento (PhD diss.). Kobe University. (Word recognition and unaccented pattern in Japanese.)Find this resource:

Giriko, M., Oshita, T., & Kubozono, H. (2011). On the nature of the morpheme [zu] in Japanese. NINJAL Research Papers, 2(2), 1–18.Find this resource:

Gussenhoven, C. (2004). The phonology of tone and intonation. Cambridge, U.K.: Cambridge University Press.Find this resource:

Gussenhoven, C. (2006). The word prosody of Nubi: Between stress and tone. Phonology, 23, 193–223.Find this resource:

Han, E. (1994). Prosodic structure in compounds (PhD diss.). Stanford University.Find this resource:

Haraguchi, S. (1977). The tone pattern of Japanese: An autosegmental theory of tonology. Tokyo: Kaitakusha.Find this resource:

Hattori, S. (1960). On’inron kara mita kokugo no akusento. In S. Hattori (Ed.), Gengogaku no hōhō (pp. 240–275). Tokyo: Iwanami Shoten. (Japanese accent seen from phonological perspectives.)Find this resource:

Ito, J. (1990). Prosodic minimality in Japanese. In M. Ziolkowski, M. Noske, & K. Deaton (Eds.), CLS 26-II: The parasession on the syllable in phonetics and phonology (pp. 213–239). Chicago: Chicago Linguistic Society.Find this resource:

Ito, J., & Mester, A. (2003). Japanese morphophonemics: Markedness and word Structure. Cambridge, MA: MIT Press.Find this resource:

Katayama, M. (1998). Optimality theory and Japanese loanword phonology (PhD diss.). University of California at Santa Cruz.Find this resource:

Kawahara, S. (2015). The phonology of Japanese accent. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 445–492). Berlin: De Gruyter Mouton.Find this resource:

Kawahara, S. (2016). Japanese has syllables: A reply to Labrune. Phonology, 33, 169–194.Find this resource:

Kawahara S., & Wolf, M. (2010). On the existence of root-initial-accenting suffixes: An elicitation study of Japanese [-zu]. Linguistics, 48(4), 837–864.Find this resource:

Kubozono, H. (1988). The organization of Japanese prosody (PhD diss.). University of Edinburgh. Published by Kurosio, Tokyo, 1993.Find this resource:

Kubozono, H. (1995a). Constraint interaction in Japanese phonology: Evidence from compound accent. Phonology at Santa Cruz (PASC), 4, 21–38.Find this resource:

Kubozono, H. (1995b). Gokeisei to on’in kōzō. Tokyo: Kurosio. (Word formation and phonological structure.)Find this resource:

Kubozono, H. (1996). Syllable and accent in Japanese: Evidence from loanword accentuation. The Bulletin (The Phonetic Society of Japan), 211, 71–82.Find this resource:

Kubozono, H. (1997). Lexical markedness and variation: A non-derivational account. In B. Agbayani & S-W Tang (Eds.), WCCFL 15: The proceedings of the fifteenth West Coast Conference on Formal Linguistics (pp. 273–287). Stanford, CA: CSLI.Find this resource:

Kubozono, Haruo (1999) Mora and syllable. In N. Tsujimura (Ed.), The handbook of Japanese linguistics (pp. 31–61). Oxford: Blackwell.Find this resource:

Kubozono, H. (2003). Accent of alphabetic acronyms in Tokyor Japanese. In T. Honma, M. Okazaki, T. Tabata, & S. Tanaka (Eds.), A new century of phonology and phonological theory (pp. 356–370). Tokyo: Kaitakusha.Find this resource:

Kubozono, H. (2005). [ai]-[au] asymmetry in English and Japanese. English Linguistics, 22(1), 1–22.Find this resource:

Kubozono, H. (2006). Where does loanword prosody come from? A case study of Japanese loanword accent. Lingua, 116, 1140–1170.Find this resource:

Kubozono, H. (2008). Japanese accent. In S. Miyagawa & M. Saito (Eds.), The handbook of Japanese linguistics (pp. 165–191). Oxford: Oxford University Press.Find this resource:

Kubozono, H. (2010). Accentuation of alphabetic acronyms in varieties of Japanese. Lingua, 120, 2323–2335.Find this resource:

Kubozono, H. (2011). Japanese pitch accent. In M. van Oostendorp, C. Ewen, E. Hume, & K. Rice (Eds.), The Blackwell companion to phonology, Vol. 5 (pp. 2879–2907). Malden, MA: Wiley-Blackwell.Find this resource:

Kubozono, H. (2012). Varieties of pitch accent systems in Japanese. Lingua, 122, 1395–1414.Find this resource:

Kubozono, H. (2015a). Dipthongs and vowel coalescence. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 215–249). Berlin: De Gruyter Mouton.Find this resource:

Kubozono, H. (2015b). Japanese dialects and general linguistics. Gengo Kenkyu, 148, 1–31. (Journal of the Linguistic Society of Japan.)Find this resource:

Kubozono, H., Ito, J., & Mester, A. (1997). On’in kōzō kara mita go to ku no kyōkai In Onsei Bunpō Kenkyūkai (Ed.), Bunpō to onsei (pp. 147–166). Tokyo: Kurosio. (The phonological boundary between the word and the phrase.)Find this resource:

Ladd, D. R. (1984). English compound stress. In D. Gibbon & H. Richter (Eds.), Intonation, accent, and rhythm (pp. 253–266). Berlin: de Gruyter.Find this resource:

Liberman, M., & Prince, A. (1977). On stress and linguistic rhythm. Linguistic Inquiry, 8(2), 249–336.Find this resource:

Lyman, B. S. (1894). The change from surd to sonant in Japanese compounds. Philadelphia: Oriental Club of Philadelphia.Find this resource:

McCarthy, John J., & Prince, Alan S. (1994). The emergence of the unmarked: optimality in prosodic morphology. In M. Gonzàlez (Ed.), Proceedings of NELS 24 (pp. 333–379). Amherst, MA: GLSA.Find this resource:

McCawley, J. D. (1968). The phonological component of a grammar of Japanese. The Hague, The Netherlands: Mouton.Find this resource:

McCawley, J. D. (1978). What is a tone language? In V. Fromkin (Ed.), Tone: A linguistic survey (pp. 113–131). New York: Academic Press.Find this resource:

NHK (Ed.). (1985). NHK hatsuon akusento jiten. Tokyo: Nihon Hōsō Shuppan Kyōkai. (NHK pronunciation and accent dictionary.)Find this resource:

NHK (Ed.). (1998). NHK hatsuon akusento jiten. Tokyo: Nihon Hōsō Shuppan Kyōkai. (NHK pronunciation and accent dictionary.)Find this resource:

Otsu, Y. (1980). Some aspects of rendaku in Japanese and related problems. In Y. Otsu & A. Farmer (Eds.), Theoretical Issues in Japanese Linguistics (pp. 207–227). MIT Working Papers in Linguistics 2. Cambridge, MA: Department of Linguistics and Philosophy, MIT.Find this resource:

Pierrehumbert, J. B., & Beckman, M. E. (1988). Japanese tone structure. Cambridge, MA: MIT Press.Find this resource:

Poser, W. J. (1984). The phonetics and phonology of tone and intonation in Japanese. (PhD diss.), MIT.Find this resource:

Sato, H. (1989). Fukugōgo ni okeru akusento kisoku to rendaku kisoku. In M. Sugito (Ed.), Kōza nihongo to nihongo kyōiku 2: Nihongo no onsei, on’in (jō) (pp. 233–265). Tokyo: Meiji Shoin. (Accent rules and rendaku rules in compound words.)Find this resource:

Shinohara, S. (2000). Default accentuation and foot structure in Japanese: Evidence from Japanese adaptations of French words. Journal of East Asian Linguistics, 9(1), 55–96.Find this resource:

Sibata, T. (1994). Gairaigo ni okeru akusento kaku no ichi. In K. Sato (Ed.), Gendaigo hōgen no kenkyū (pp. 338–418). Tokyo: Meiji Shoin. (On the location of accent in loanwords.)Find this resource:

Sugito, M. (1982). Nihongo akusento no kenkyū. Tokyo: Sanseidō. (Studies on Japanese accent.)Find this resource:

Uwano, Z. (1999). Classification of Japanese accent systems. In S. Kaji (Ed.), Cross-linguistic studies of tonal phenomena: Tonogenesis, typology, and related topics (pp. 151–186). Tokyo: ILCAA.Find this resource:

Uwano, Z. (2012). Three types of accent kernels in Japanese. Lingua, 122, 1415–1440.Find this resource:

Vance, T. J. (2015). Rendaku. In H. Kubozono (Ed.), Handbook of Japanese phonetics and phonology (pp. 397–441). Berlin: De Gruyter Mouton.Find this resource:

Napoli, D. J., & Nespor, M. (1976). The syntax of raddoppiamento sintattico. Unpublished manuscript.Find this resource:

Notes:

(1.) This syllable-based rather than mora-based generalization is attributed to the fact that heavy (bimoraic) syllables can bear an accent only on their initial moras (see Section 4 for details). See Kawahara (2016) for a summary of evidence for the syllable in Tokyo Japanese.

(2.) This rule is also responsible for the accentuation of verbs and adjectives, all of which are morphologically complex, i.e., stem + ending. This explains why most verbs and adjectives are accented rather than unaccented (Kubozono, 2008, 2011).

(3.) The idea of the emergence of the unmarked also explains the seemingly peculiar accent patterns shown by a group of words derived from unaccented nouns. For example, proper names such as /ra’i.on-zu/ ‘the Lions,’ /to’n.ne.ru-zu/ ‘the Tunnels,’ and /do.ra.e’.mon-zu/ ‘the Doraemons’ become accented, although the base forms from which they are derived are unaccented, /rai.ono/ ‘lion,’ /ton.ne.ruo/ ‘tunnel,’ and /do.ra.e.mono/ ‘Doraemon, an anime character.’ The accent patterns of these derived words can be explained if one assumes that the covert accent in the base has emerged in the derived forms without being interfered by the extrametrical ending /zu/: e.g., /ra’i.on + zu/ → /ra’i.on-zu/ (Giriko, Oshita, & Kubozono, 2011; see Kawahara & Wolf, 2010 for a different analysis).

(4.) These compounds are not sensitive to the right-branching condition to be discussed in Section 3.2, since the second and third elements are not independent ‘words’; in other words, they are compound nouns consisting of two morphological ‘words,’ the second of which consists of two morphemes.

(5.) Many words that resist rendaku voicing have a voiced obstruent in the final mora of the first member: e.g., /siba-ta/. Lack of rendaku in these cases is due to Lyman’s Law, by which a voiced obstruent in one mora blocks the voicing of another obstruent within the same word (Ito & Mester, 2003; Lyman, 1894; Sato, 1989; Vance, 2015).

(6.) Left-branching compounds often split into two prosodic words if they consist of four or more elements: e.g. [[[too.nan° + a’.zi.a] syo’.ko.ku] ren.goo°] → {too.nan-a’.zi.a}{syo.ko.ku-re’n.goo} ‘(((south-east + Asia) + countries) + association); Association of South-East Asian Nations, ASEAN.’ This is due to a rhythmic constraint by which a long prosodic word consisting of four (or more) elements splits into two prosodic words each consisting of two elements (Kubozono, 1988).

(7.) These compounds are integrated into one prosodic word since they are made up of short, semi-bound morphemes. This analysis implies that rendaku is not merely a juncture phenomenon linking two morphemes, as is often assumed in the literature (Ito & Mester, 2003), but a head-marking process denoting the modifier-head relationship in compounds.

(8.) The dvandva compounds presented here are manifested in one prosodic word despite the coordinate structure because they consist of short native (and mostly bound) morphemes. In contrast, the non-dvandva compounds are unaccented because they are four-mora compounds consisting of two bimoraic nouns (Section 2.3) and also they undergo rendaku voicing (Section 3.1).

(9.) Note that the trimoraic sequence in (38f.), i.e., /donk/, splits into /do/ and /nk/, the latter being a bimoraic syllable headed by the moraic nasal /n/. This syllabification is superior to the other bisyllabic structure, i.e., /don.k/, in which the first part of a geminate consonant forms an independent syllable on its own. The same syllabification is observed in other dialects of Japanese, too (Kubozono, 2015b).

Was This Useful?