This article is an non-exhaustive list of rules that govern the Tokyo pitch accent of numerals, numbers and number–counter sequences. For other rules, see Japanese pitch accent rules.
Numerals, numbers and counters follow significantly different rules and trends from those of other nouns. Number–counter sequences can also be deaccented when used as adverbs (see ).
Numerals
Native numerals All the native numerals from one to ten are initial-accented, except hito, futa and kokono which are second-accented: ,
, , , , , , , , , , , , ,
Sino-Japanese numerals All simplex Sino-Japanese numerals, except san and oku, are minimally final-accented; if they are unimoraic, or end in a special mora, they are initial-accented; go and ku used to be unaccented (large numbers are translated into English in short scale): , , , , ,
, , , , ,
However, when the numerals are enumerated, as in "one, two, three, …," the final-accented numerals and san can become initial-accented. The unimoraic numerals also have initial-accented bimoraic variants with lengthening morae for enumeration:
Numeral series and decimal parts Numeral series, made of numerals from zero to nine, as used in telephone and serial numbers and decimal parts, follow a specific set of rules: Numerals must be in groups of two if possible. Each group must form an exactly quadrimoraic minor phrase, which means only bimoraic numerals can be used in groups. Each group is penultimate-accented. A lone, ungrouped numeral, however, retains its original accent in its own minor phrase, or becomes dephrased if its nucleus is its final mora. There is disagreement as to whether a lone, originally unimoraic numeral should become bimoraic or remain unimoraic in a numeral series.
Some of these rules also apply to a group of a ones-place numeral and , the latter of which functions as the decimal point. This group, however, is minimally accented before ‑ten. 03-5400-▲▲21: 3.1416: 3.1: 0.1453: 0.145: 58.32412: 2.2: 2.22: 2.222: 5.3: 15.3:
These rules for the decimal point ‑ten are different from those for the counters ‑ten (see #Sample list). In addition, while the counters often trigger , the decimal point does not necessarily, especially in broadcasting where it is more desirable for clarity not to have . One survey finds that there was a slight preference for in () and no in (). Compare: 1 point: 1 item: 1.5: 8 points: 8 items: 8.5:
English numerals Most organically used numerals are either native or Sino-Japanese, with the sole exception of the Franco-English loanword zero, which can be used for special emphasis. The English numerals from one to ten can be used in exercise and sports, for example to count points under par in golf. English numerals all retain the original English accent: zero → one → two → three → four → five → six → seven → eight → nine → ten → eleven → million → billion →
English compound numbers behave just like any other type of compound:
Numbers Not all the numerals above are frequently used to create numbers productively (excluding collocations and proverbs). The numerals from one to nine can be grouped as follows, based on their behavior in compound numbers: Final-accented and Sino-Japanese four (G−1&Shi): Sino-Japanese seven (Shichi): Formerly unaccented, now initial-accented (GX0): Sino-Japanese two (Ni): Initial-accented and native two (G+1&Futa): Sino-Japanese three (G0):
Numeral + jū () Jū retains its original nucleus after: G−1&Shi: Shichi: GX0:
And is preaccenting after: Ni: G+1&Futa: G0:
All these combinations and jū itself can be grouped based on their behavior with following numerals, counters and counter-like numerals as follows (those marked with a superscript question mark are dubious due to inconsistent descriptions among sources): Pseudo-unaccented (GX0): Sino-Japanese ten (Jū): Initial- and second-accented (G+1+2):
Numeral + hyaku () Hyaku retains its original nucleus after: G−1&Shi: GX0: Ni:
And is preaccenting after: G+1&Futa: G0:
And either loses the nucleus or is preaccenting after Shichi:
All these combinations and hyaku itself can be grouped based on their behavior with following numerals, counters and counter-like numerals as follows (note that shichihyaku belongs to two different groups, and those marked with a superscript question mark are dubious due to inconsistent descriptions among sources): Pseudo-unaccented (GX0): Initial- and second-accented (G+1+2):
‑jū number + numeral () GX0 lose their nucleus before all groups. Take gojū as a representative example: G−1&Shi: Shichi: GX0: Ni: G+1&Futa: G0:
Jū works similarly, except before: GX0: G0:
G+1+2 retain their nucleus before all groups. The following numeral can form its own minor phrase: G−1&Shi: Shichi: GX0: Ni: G+1&Futa: G0:
‑hyaku number + number () GX0 lose their nucleus before all groups as well as larger numbers. Take happyaku as a representative example: G−1&Shi: Shichi: Ni: G+1&Futa: G0: Larger:
Additionally, the obsolete unaccented pronunciations of GX0 can resurface:
G+1+2 retain their nucleus before all groups as well as larger numbers. The following number can form its own minor phrase: G−1&Shi: Shichi: GX0: Ni: G+1&Futa: G0: Larger:
Numeral + sen () Short compounds with sen are all penultimate-accented:
Shisen is hypothetically pronounced , but it has become fossilized in the kabuki title .
Numeral + man () In short compounds, man behaves like an initial-accented counter: All groups except GX0: → → → GX0: → → → Larger:
Shiman is hypothetically pronounced , but it has become fossilized in the Buddhist holiday .
Long ‑jū/‑hyaku/‑sen number + man () In long compounds, man behaves like an initial-accented or preaccenting counter after GX0, GX0 and ‑sen numbers:
G+1+2 and G+1+2, additionally retain their nucleus:
Numeral or ‑jū/‑hyaku/‑sen number + oku/chō () Oku behaves like a preaccenting counter: All groups except GX0: → → → GX0: → → → Larger:
Chō behaves like an initial-accented or preaccenting counter: All groups except GX0: → → → GX0: → → → Larger:
‑sen or larger number + number () ‑sen and larger numbers normally combine with other numbers as separable minor phrase:
In some contexts such as sport, it is not impossible for the first component to lose its nucleus:
Indefinite + number There are three indefinites, which express "tens/hundreds/thousands (of people/things)," that can prefix multiples of ten as follows:
Number/indefinite + counter Counters are usually dominant and determine the accent pattern of numeral–counter compounds. Each counter follows a specific set of numerals and has a specific accent type with specific exceptions.
Native numeral + native counter Short counters tend to be preaccenting. Futa may make final-accented compounds, although they may sometimes be regularized by younger speaker:
Long counters may be initial-accented after unimoraic numerals, and preaccenting after bimoraic numerals:
‑tsu The native numerals from one to nine are usually used with the generic native counter ‑tsu. Of these combinations, all those containing are final-accented, while the rest are second-accented; however, the -less variants can be final- or initial-accented as well: , ,
, , ,
When these ‑tsu words are used as modifiers in compounds, ‑tsu tends to be the nucleus:
However, hitotsu meaning tends to keep its original accent:
Other combinations Idiomatic or high-frequency compounds may be unaccented: , , , , , , , , , , , etc.
Long native counters tend to form regular long compounds:
Number groups The aforementioned number groups can be reorganized based on their behavior with counters as follows: Sino-Japanese one (Ichi): Sino-Japanese eleven (JI): Sino-Japanese ten (Jū): Final-accented (G−1): 10 + G−1 (J−1): Sino-Japanese two, four and seven (GNSS): 10 + GNSS (JNSS): Initial-accented and less than ten (G+1): 10 + G+1 (J+1): Formerly unaccented and unimoraic native four (GX0&Yo): 10 + GX0&Yo (JX0&Yo): , jū yo‑ Sino-Japanese three (G0): Sino-Japanese thirteen (J0): Pseudo-unaccented (GJHX0): Initial- and second-ccented and divisible by ten (GJH+1+2): Half-native and half-Sino-Japanese seventy (Nanajū): Sino-Japanese one thousand (Sen): Compound numbers divisible by one thousand (GSC):
Counter groups Just like compound heads (see ), counters can be short or long, and be preaccenting, deaccenting, final-accented, initial-accented or retain their original accent. They may behave uniformly with all numbers, or have specific patterns with specific numbers.
Short counters Of the number groups, GSC in particular tends to form unaccented combinations with short counters, despite how their uniform behavior with other numbers. GX0&Yo forms two separable minor phrases, the first of which is jū with its original accent, the second is go, ku or yo with the counter, unless the counter in question uniformly results in unaccented or final-accented sequences with all numbers, in which case GX0&Yo behaves exactly like other groups. In some cases, JI, J−1, JNSS, J+1 and J0 may be more uniform compared to ichi, G−1, GNSS, G+1 and G0.
GJHX0, GJH+1+2 and nanajū forms the first minor phrase before any numeral–counter sequence, in the same ways shown in the #Numbers section. When GJH+1+2 connects directly to a counter, they traditionally retain their original accent, and innovatively follow whichever pattern is more dominant, usually the preaccenting one. does not shown any instance where nanajū behaves similarly, while lists some examples where nanajū and nanahyaku also retain their original accent.
The following are groups of counters with some representative examples: Preaccenting after all number groups except GSC (CS_P). Eg: : Ichi, JI: Jū: G−1, J−1: GNSS, JNSS: G+1, J+1: GX0&Yo: JX0&Yo: G0, J0: GJHX0: GJH+1+2: Sen: GSC: Deaccenting after GX0&Yo, JX0&Yo and GSC, and preaccenting after all other number groups (CS_P_DX0). Note that JX0&Yo still forms two separable phrases. Eg: : Ichi, JI: Jū: G−1, J−1: GNSS, JNSS: G+1, J+1: GX0&Yo: JX0&Yo: G0, J0: GJHX0: GJH+1+2: Sen: GSC: Deaccenting after GX0&Yo, JX0&Yo, GSC and G0, and preaccenting after all other number groups except GSC (CS_P_D0X0). Note that JX0&Yo still forms two separable phrases, and G0 additionally becomes regularized after GJHX0 and GJH+1+2. Eg: : Ichi, JI: Jū: G−1, J−1: GNSS, JNSS: G+1, J+1: GX0&Yo: JX0&Yo: G0: J0: GHX0 + G0: GJH+1+2 + G0: GJHX0: GJH+1+2: Sen: GSC: Initial-accented after ichi, jū, G−1, GNSS, GX0&Yo, JX0&Yo, G0 and sen, and preaccenting after all other number groups except GSC; in other words, these counters are initial-accented after all single numerals except G+1 (CS_P_+1S!+1). G0 additionally becomes regularized after GJHX0 and GJHX+1+2, and some of these counters may be completely regularized into CS_P, or the counters can be initial-accented after all single numerals without any other preceding numerals, as well as JI, J−1, JNSS, J0 and J+1 (CS_P_+1SJ). Eg: : Ichi: JI: Jū: G−1: J−1: GNSS: JNSS: G+1, J+1: GX0&Yo: JX0&Yo: G0: J0: GHX0 + G0: GJH+1+2 + G0: GJHX0: GJH+1+2: Sen: GSC: Initial-accented after ichi, jū, G−1, GNSS, GX0&Yo and JX0&Yo, and preaccenting after all other number groups except GSC; in other words, these counters are initial-accented after all single numerals except G+1 and G0 (CS_P_+1S!+1!0). Some of these counters may be completely regularized into CS_P, or the counters can be initial-accented after all single numerals without any other preceding numerals, as well as JI, J−1, JNSS, J0 and J+1 (CS_P_+1SJ). Eg: and : Ichi: ; JI: Jū: ; G−1: ; J−1: GNSS: ; JNSS: G+1, J+1: ;