Desu is weird. In textbook introductions to Japanese, it is often described as the polite (teinei) equivalent to da, and this description works fine for “nouny” words:
|1a.||Otoko desu.||“[It's a] man.” [+pol]|
|1b.||Otoko da.||“[It's a] man.”|
|2a.||Hansamu desu.||“[He’s] handsome.” [+pol]|
|2b.||Hansamu da.||“[He’s] handsome.”|
However, we soon learn that the supposed equivalency breaks for “verby” words, such as “i-adjectives” (keiyôshi):
|3a.||kawaii desu.||“[He’s] cute.” [+pol]|
|3c.||*kawaii da.||ungrammatical (but see below)|
Of course, this doesn’t apply to verbs themselves, since verbal politeness is marked by the bound morpheme -masu, never by desu. …Or is it? (dun dun duuun):
|4a.||mi ni iku yo.||“[I'm] going to see!”||4b.||mi ni iku desu yo.||“[I'm] going to see!” [+col, +pol]|
|4c.||mi ni iku da yo.||“[I'm] going to see!” [+col]|
Let’s forget this anomaly for a moment. At any rate, any grammar of Japanese has to account for asymmetries like 2a/2b vs. 3a/3b/*3c, and this often introduces unseemly kludges. In traditional school grammar (Hashimoto-style), it’s said that complete utterances need to mark “assertion” (dantei) somewhere. The copula da is an assertion particle, dantei no joshi, and nouny words have no assertion, so they need it; while verby words (including i-adjectives) have assertion built-in and can stand alone. Then desu is explained awkwardly as being a dantei-teinei no joshi “politeness-assertion particle” when following nounies (“Ringo desu”), and as a non-assertive teinei no joshi “politeness particle” when following verbies (“oishii desu”). In other words, “there are two desu”. (Morikawa, in a more sophisticated argument, proposes two da too.)
However, this analysis is mixing description with prescription, as shown by examples 4b, 4c above (which might be considered “incorrect Japanese” but occur often enough, even by native speakers). What’s more, even the “needs assertion” rule is prescritive; utterances like hontō yo are common. Considering similar cases, Daniels has no choice but to conclude that “the function of da, desu, de aru, etc., in many of their uses, is not copulative” (in school-grammar terms, they’re often not dantei, and this happens in contexts that school grammar doesn’t predict). This is particularly the case for deshō/darō, and Daniels argues that darō shouldn’t even be considered an inflection of da at all—unlike da, it’s not hard to find examples of it following a verby like in 3c (oishii darō?, kawaii darō?).
Rather than being a sign of the decay of the Japanese language, I think these complications are rather a consequence of the muddy origins of desu. It’s much like keigo, a modern, artificial category which the Japanese have been perpetually complaining of “being in a state of disarray” and that “young people can’t speak it properly anymore” literally since its inception (see Wetzel, Keigo in modern Japan; Miller, Levels in Speech (Keigo) and the Japanese Linguistic Response to Modernization). Just like keigo, the “rules” about da/desu usage were somewhat impositive, an attempt to unify what previously depended on varied norms of specific social groups and contexts (not to open the can of worms of dialects…). So we find the following in the prescriptive “Keigo from Now On” (Kore kara no Keigo), published by the government agency Kokugo Shingikai in 1952:
From now on we would like for desu/masu to set the tone of conversation.
Note: This has been decided as setting the tone for general conversation among members of society and does not limit forms such as the de arimasu of lectures, the de gozaimasu of formal occasions, or the familiar da.
We now recognize the acceptable ending for [verby] adjectives [keiyôshi], which until now had been a longstanding problem [Kore made hisashiku Mondai to natteita Keiyōshi no Musubi-kata]—for example, simple, clear forms like ōkii desu, chiisai desu.
I wonder what was the longstanding problem. Are these forms “clear and simple” in comparison to the euphonic (onbin) honorifics, e.g. oishū gozaimasu, now in disuse?
Wieger cites Twine, Language and the Modern State, who says that, before the twentieth-century deliberate standardization, there were no less than six copulas in common use:
- de gozaru
- de gozarimasu
- de arimasu
- de aru
De aru and da were felt as nonpolite by the reformers. De gozaimasu was associated with the elites, and de gozaru with the fashionable women of the pleasure quarters, from which it spread to popular writers. De arimasu was also common in fiction, and in general society. When the movement for vernacular literature (Genbun-Itchi) started getting traction, there was explicit competition between forms. In a period of four years we find (from Sanada, Hyôjungo wa ika ni Seiritsu shita ka) :
- Da, in Futabatei’s Ukigumo (1887);
- De arimasu, in the same Futabatei’s Nozue no kiku (1889);
- Desu, in Yamada Kochô (1889);
- De aru, in Ozaki’s Ninin Nyôbô (1891).
Da likely arised as a contraction of de aru, and desu of de arimasu or de sô or the like; but even if this is true historically, it might not be “subjacently” (see Morikawa, above). Meanwhile goza(ru) was a spelling-pronunciation of 御座, which in turn was originally a kun spelling of (now extinct) owasu. Some even say that desu was originally a feature of the speech of Yoshiwara sex workers. To this day de aru remains typical of literature.
This principle of linguistic variants as a mark of social position survives in Yakuwari-go “role speech”, the trope in contemporary Japanese fiction where certain stereotypical characters speak with certain marks, which don’t really correspond to real-world usage. So, for example, a grandpa- or grandma character is likely to use ja, a comic-relief “weird” character might use the Kansai-ben ya, an evil overlord would prefer da, and of course samurai and ninja choose de gozaru.
The contextual implications of these marks can be counter-intuitive. Once I noticed that some bloggers write in da-level and others in desu/masu. According to standard accounts, politeness imply distance, and so I thought I should write in “plain” da to make it more intimate and casual. Then I’ve been told the opposite would happen: because written da is used in newspapers and formal texts, it would feel heavier and colder, while desu would make it conversational and therefore friendlier.
It’s this kind of stuff that makes me want to learn corpus linguistics. I’d like to see some large-scale statistics on desu patterns…