Japanese compound word morphology: intra-word order influences verbal argument type

Terry Joyce, in an paper about something else, mentions this neat datum that I had never noticed.

Japanese is a head-last language, meaning that the arguments of a verb precede it:

Yama wo noboru
mountain OBJ climb
“Climb the mountain”

In English, some compound words are built like shortened phrases: “spoil a sport” → “spoilsport”. The same happens in Japanese:

“hiking, mountaineering”

However, Japanese also has a large number of Chinese loans, and in Chinese, the verb comes first (as in English). This results in a number of Japanese words with “reversed” order, i.e. compounds in which the argument comes after the verb:

“hiking, mountaineering”

Yamanobori and Tozan are synonymous; but the first follow Japanese syntax (verb last), and the second, Chinese syntax (verb first).

The bit I’d never noticed is that the “Chinese” syntax in Japanese words almost always uses direct objects (the ones that would be marked with を wo). By contrast, the “Japanese” order allows all sorts of arguments and modifiers, including adverbs:

  • 外食 Gai-shoku “outside-eat” = to eat outside, eating out
  • 毒殺 Doku-satsu “poison-kill” = to kill by poison
  • 女装 Jo-sō “woman-dress” = to crossdress, to dress as a woman, female-attired
  • 長引く Naga-biku “long-draw” = to drag on

Notice that this doesn’t depend on whether the components themselves are on’yomi (of Chinese origin) or kun’yomi (native Japanese); only on the morphosyntax.

Source: Terry Joyce, Modeling the Japanese mental lexicon: Morphological, orthographic and phonological considerations, who credits:

  • Kageyama, T. (1982) Word formation in Japanese. Lingua, 57, 215–258.
  • Ozaki et al. (1992) Daijigen [Great Etymological Dictionary]. Kadokawa Shoten.

2 thoughts on “Japanese compound word morphology: intra-word order influences verbal argument type

  1. I’m having trouble understanding this one. The so-called “Japanese syntax” is also found in Chinese. For instance, a Google search finds 毒杀 in Chinese texts. Or try the Chinese term equivalent to 外注, which is 外包. Neither of these requires an を in normal syntax…

  2. Ok so first let’s distinguish phrasal syntax from morphological (intra-word) order. When we say that OV is “Japanese order” while VO is “Chinese order”, we’re talking about the normal order of sentences. Even though there may be sentences in other orders (cf. English “What light from yonder window breaks?”), I think it’s uncontroversial that the normal, everyday order of verb and argument in Japanese is OV, while Chinese is VO.

    Now let’s consider intra-word order. You can think of “Japanese” and “Chinese” as convenient labels for OV and VO. However, the choice of these labels imply an hypothesis; that, in Chinese, VO words are typical; and moreover, that in Japanese they used to be atypical, until Chinese influence (post-Heian). I don’t have the means to provide data for either of these hypotheses right now. Instead, let’s try testing something easier, viz., Kageyama/Osaki’s proposal, which could be paraphrased like this:

    In Japanese two-morpheme compounds, if the compound is VO (quote-unquote “Chinese order”), then O will usually be a direct object (i.e. if we rephrase it as a sentence, the O will be marked by を).

    You mentioned 外注 gaichū “outsourcing”; but, here, 注 is not an object (one doesn’t 外 the 注, the way that one 登s the 山 in 登山; instead, one 注ぐ to (from?) the 外); so this isn’t a VO compound. Rather, it is OV, argument-verb, just like English “out-sourcing”. So it’s a quote-unquote “Japanese-order” compound, and it’s within the prediction that its argument can be non-を-marked.

    Let’s get a few more examples. The following are words from EDICT with the following characteristics:

    1. They’re two-kanji compounds;
    2. They’re suru-verbs;
    3. The first kanji, when alone, doesn’t begin a verb;
    4. The second kanji, when alone, does begin a verb.

    That is, these words are good candidates for OV compounds. Here’s a random sampling; let’s consider if the first kanji would be marked by を:

    完成 – non-を
    窃盗 – non-を
    晩成 – non-を
    兌換 – non-を
    篆刻 – を
    予表 – non-を
    族滅 – を
    九拝 – non-を
    公聴 – non-を
    特集 – non-を?
    露見 – non-を (assuming 露 is the Chinese adverb, “clearly, exposedly”)
    概言 – を

    And now, found with the same method but reversed, a sampling of probable VO compounds:

    指南 – を
    発電 – を
    離党 – を
    転宅 – non-を? (に, I suppose)
    改号 – を
    滞京 – non-を (ibid.)
    筆誅 – を
    廃園 – を
    排菌 – を
    接吻 – を
    吐露 – non-を (adverbial 露 again?)
    出炭 – を

    So from a casual glance, it seems to be the case that, in VO compounds, there’s a clear tendency for the O to be a direct object of the V (when it’s not, it seems to often be a locative, and locatives sometimes can accept を – I don’t know Chinese, so I don’t know whether that’d count as “direct objects” in those languages); while, in OV compounds, there’s relatively more freedom on the relationship between the argument and verb.

Leave a Reply

Your email address will not be published. Required fields are marked *