Innocence Seekers: The Black Rose – Namari pitch accent

This post will focus on the pitch accent of Namari dialects. Much of this post will focus on the Yaezora dialect, as it is the most complex of the major Namari dialects with regards to pitch accent. However, I will talk a bit about the pitch accent of Chiyohara dialect.

The pitch accent of Yaezora Namari is best described as a register-locus system, with elements of stress. Each phonological word has two inherent tones (i.e. registers), high and low, and has either one or two loci where the pitch changes abruptly. Only words with inherent low tone can have two loci; words with inherent high tone must have only one locus.

It is important to note that the register-locus system only applies to nominals. Verbs and verbal adjectives only have a register system, with either high or low inherent tone, and the location of the accent is dependent on conjugation.

Note that inflection can alter the pitch accent of a word. Thus, for nominals the nominative represents the original pitch accent. In general, suffixes and clitics may have their own loci, which interact with words in their own ways.

A full list of possible accent classes is beyond the scope of this article (and theoretically there are infinite). However, I can mention some of the accent classes, and rules when combining words.

A Yaezora accent class is given as X.Y.Z, where:

  • X is the number of morae in the stem of the word.
  • Y is either L or H, depending on the inherent tone of the word, or if the word has two loci, the location of the first.
  • Z is the location of the (second) locus.

A locus in Yaezora Namari is defined as the mora and its following mora boundary. The mora itself is considered “stressed”, meaning its vowel cannot be devoiced or elided at all, and the following mora boundary is where the pitch changes, either from low to high, or high to low. Stressed morae are traditionally pronounced with the lowest or highest pitch. Unlike in Tokyo Japanese, the syllable does not have any bearing on pitch accent (Tokyo Japanese actually prohibits downstep at the end of a heavy syllable, hence wakánnai versus wakaránai, both meaning “one does not understand”; note the difference in the placement of the accent).

Nominals with monomoraic stems actually only have four of the five possible bimoraic accent classes; class 2.1.2 is impossible. On the other hand, bimoraic stems can have any of all five accent classes.

Also important to note is the accent class of Sino-Namari borrowings. In general, they tend to follow the Middle Chinese tones, by the following correspondences:

  • Level: 2.H.2, 2.L.2
  • Rising: 2.L.1, 2.L.2
  • Departing: 2.H.1, 2.1.2, 2.H.1
  • Entering: 2.H.1, 2.L.1

The first class given is for voiceless initials, while the second class is for voiced initials (including sonorants). The third alternant for the departing tone is also for voiced initials, and it is used if and only if the borrowing is monomoraic.

(A side note: borrowings of entering tone words are guaranteed to be bimoraic, due to their final consonant.)

For reference, here are the pitch contours of the bimoraic accent classes, and their Namari names (L = low, H = high, parentheses represent following suffixes and clitics):

  • 2.H.1 (ote “falling”): HL(L)
  • 2.H.2 (kami “high”): HH(L)
  • 2.L.1 (agai “rising”): LH(H)
  • 2.L.2 (shimo “low”): LL(H)
  • 2.1.2 (take “peaking”): LH(L)

The same Namari terms are used for words with more morae, albeit used with qualifiers (e.g. 3.H.2 is kamiote, while 3.H.1 is shimoote; 3.1.2 is atamadake, 3.1.3 is nakadake and 3.2.3 is odake).

It’s unlikely, for now, that I’ll post a comprehensive list of words and their accent classes, largely due to lack of resources (Wiktionary does have pitch accent data, but only for the Tokyo dialect).

What I will go over, on the other hand, is how pitch accent interacts in compounds. First, prefixes have neither loci nor inherent tone; they take the tone of the word to which it is attached. Suffixes and enclitics do not have inherent tone, but may have loci. If a suffix with a locus attaches to a low tone word with a single locus, the locus of the suffix becomes the second locus of the entire phonological word. If, instead, the word has high tone, the word’s original locus is removed and the locus of the suffix becomes the new locus of the word. Words with two loci lose their second locus, with the new second locus being that of the suffix.

Compounding two words, on the other hand, is more complex. The following rules are observed:

  • If both words have identical inherent tone and single loci, the first word loses its locus.
  • If the first word has two loci and the second an inherent high tone, the first word loses its second locus.
  • If the first word has two loci and the second an inherent low tone with a single locus, not only is the second locus of the first word lost, but also the locus of the second word.
  • If both words have two loci, the second locus of the first word and the first locus of the second word are lost.
  • If the first word has an inherent low tone with a single locus and the second an inherent high tone, then the two loci remain, forming a “peaking” accent class.
  • If the first word has an inherent high tone and the second an inherent low tone, then not only does the first word lose its locus, but also becomes a low tone word.

The only declensional suffix with a locus is the ablative suffix -kaya, whose locus is on its second mora. On the other hand, many conjugational suffixes have their own loci.

While the locus of a verb is typically on the antepenultimate (for ku verbal adjectives), the penultimate (for consonant-stem verbs) or ultimate (for vowel-stem verbs) mora of the conclusive, or a specific mora (the -shi- of shiku verbal adjectives), some forms place the locus at a different location, relative to the suffix itself:

  • Imperative: ultimate
  • Future: penultimate
  • Indirect evidential: penultimate
  • Desiderative: antepenultimate
  • Participle: ultimate (low tone)
  • Attributive (as adnominal): removed (inherent tone also removed)
  • Negative: penultimate (high tone), ultimate (low tone)
  • Causative, passive, potential: ultimate
  • Reciprocal: penultimate

Note that derived forms are not listed (such as the past, from the participle). And also, the attributive lacks not only a locus, but its own inherent tone; rather, it takes the tone of the word following it (not necessarily the head noun).

Anyway, that will be it for Yaezora pitch accent.

Chiyohara pitch accent is different. Largely, it resembles the Tokyo accent (meaning no registers), and indeed there is a huge amount of lexical similarity in terms of the location of the accent between the two dialects (at least when native vocabulary is considered). However, there are things to note about Chiyohara pitch accent:

  • Where two words are combined, the second word’s accent is used.
  • Sino-Namari words have their own accents, based on Middle Chinese tones (level: unaccented, rising: second, departing: first, entering: first/second)
  • Due to extensive changes in how vowel sequences are pronounced, the syllable plays a larger role than even in the Tokyo dialect. Again, like in the Tokyo dialect, downstep in heavy syllables must be in the middle.
  • The final mora of an unaccented word has mid-low pitch, rather than mid-high in the Tokyo dialect.
  • Morae between the initial and the accented mora are pronounced with mid pitch.

And for Haguya dialect, well… there isn’t much to talk about. It is a nikei dialect, where there are no loci, only two registers. The accented register has a downstep after the first syllable (not mora), while the unaccented register causes the pitch to steadily rise from low to high. Inflectional suffixes and endings do not affect accent.

That will be all for now.

Edit (2019-03-26): A bit off-topic, but for reference, Uzuki pronounces her given name as [ù.ðỳ.kʲí] in her native language, with accent class 3.2.3.

Edit (2019-04-10): I realised that I forgot to mention the default loci of verbal adjectives.