On Reading Japanese.

In a long and interesting TLS review (not paywalled) of Minae Mizumura’s The Fall of Language in the Age of English, Jay Rubin, a professor of Japanese Literature at Harvard and translator of Haruki Murakami, says this:

Even now, after some seventy post-war years of attempts to simplify and rationalize the Japanese writing system, its “appalling” mixture of Chinese characters and two supplementary phonetic scripts remains the single greatest stumbling block to foreigners who wish to become literate users of the language (to become literate in a language, you have to know its literature). Not even those few of us who survived boot camp and went on to read a good part of Japan’s literary canon in the original have it easy. As Minae Mizumura accurately (if somewhat ungraciously) observes in The Fall of Language in the Age of English, “Foreigners, even those who teach Japanese literature at a university, cannot read novels written in Japanese with any ease”.

Is that really true?

Comments

  1. My Japanese was mostly self-taught, so I’m very far from a university-level reader, but I spent a whole summer trying to read 1Q84 in Japanese. Not the whole work – the first volume. Of 6. And I tried very hard _not_ to stop and look things up, but read quickly and try to figure out things from context. 1/6 of the book still took me weeks to read. One problem was that my eyes wouldn’t move steadily from the bottom of one line to the top of the next line, so I kept re-reading the same line or skipping over lines.

  2. I taught myself Japanese while living in Japan and teaching English way back in Heisei Year One. Within 9 months I could get the gist of most newspaper articles without a dictionary, and within a year I was able to get through Abe’s Suna no onna fairly painlessly using a dictionary. A university lecturer in Japanese who is unable to read at least middle-brow Japanese fiction pretty easily is probably in the wrong profession.

    Of course it may well be that when Mizumura was at Yale most of the professors “teaching Japanese literature” were teaching in translation. That was back in the early 1970s. Americans who learned Japanese during the short moment in the 1980s/90s when Japanese culture was the next big thing may be more proficient (and had access to a much wider range of helpful learning materials).

    It is sadly true that it is fairly easy to forget how to read Japanese, because it does take more work to keep up. I find reading a French or Russian novel every three months or so is perfectly fine for maintaining French and Russian, but I have definitely let my Japanese lapse over the past decade and kind of dread the work it would require to learn it again. I still have some Murakami novels on my shelves I will probably never get to.

  3. I don’t think this is really true, or at least it’s certainly not universally true. I’d say I can read most modern Japanese novels without major difficulty, and I’m by no means the most skilled Japanese-reading foreigner. (I’m an American, and an almost purely self-taught Japanese learner.) Obviously many foreign learners don’t reach that level, but some certainly can and do. I would hope that includes most teachers of Japanese literature, although I can’t personally attest to that, since I haven’t studied Japanese in a university setting.

    The statement might be more applicable if you consider pre-modern works written in forms of the language other than modern kōgo. Such works certainly present further difficulties, since they’re written in a substantially different language from modern spoken and written Japanese. Some non-Japanese people definitely can read such works, but to the extent that they have difficulties with them, I don’t think it necessarily reflects very much on the difficulty of ‘Japanese’ for foreigners. Most Japanese people today can also read such works only with difficulty, if at all.

    Rubin’s reference to “some seventy post-war years of attempts to simplify and rationalize the Japanese writing system” also seems misleading to me. There was one period of a few years in the late 1940s when reformers were in a position of power following the war and were able to promulgate some orthographic reforms. There really haven’t been any more major simplifications since then. It’s true that there was something of a transitional period following the reforms where you might see a mix of old and new features between different works (and within individual works), but that transition was essentially complete decades ago.

    Personally, I like the existing Japanese writing system reasonably well, and certainly don’t consider it “appalling.” Learning the kanji and their various readings takes time, but I didn’t find it to be an insurmountable difficulty.

  4. @AG: I kept re-reading the same line or skipping over lines.

    Japanese is one of my native languages, but I grew up outside of Japan. Although I have been reading Japanese since I was very young, I have always found it slightly easier to read horizontal Japanese than vertical Japanese (especially vertical Japanese where the lines are long).

    I haven’t heard the same from people growing up in Japan, so it must be a matter of exposure. I would guess that 10-20% of what I’ve read in my lifetime is vertical Japanese.

    Still, I have a gut feeling that reading vertically is intrinsically (marginally) harder. The field of vision is wider than it is high, and I would guess that we humans are naturally more used to move our point of visual focus horizontally than vertically.

    Any research on this kind of stuff?

  5. Rubin’s entire review sets my teeth on edge. As Mizumura supposedly said, having an “orderly brain” is a trait common among American intellectuals but rare among speakers of Japanese. If that is the case, Rubin comes down squarely on the Japanese side. The review appears to be little more than a random collection of peeves, and if you don’t know what he is ranting against you could be forgiven for not understanding what is supposed to hold it together.

    With his entrenched but vaguely articulated views on Japanese language and culture, Rubin may not be the best person to review a book by someone like Mizumura. He appears to take exception to the author’s view that modern language and literature represent a degeneration from the Meiji and Taisho period — particularly as he is the translator of one of the most prominent modern Japanese authors. He does his best to paint the author as a weirdo for clinging to an older, narrower, more conservative model of Japanese culture and literature than the one he appears to embrace.

    Here and there, we get glimpses of what moves him. He is at war with strongly-held Japanese myths about language (which are very different in content but not so different in nature from prescriptivism in English), and he is highly impatient of views on the uniqueness of the Japanese language that imply that translation is impossible.

    Then he ‘funs’ us with a pseudo-literal translation of the original: Large rickshaw back floating, Middle Kingdom Han Country full crotch muscles bulge. Large rickshaw middle floating girl woman. Skull bottom sharp like Green Melon Smell 120 lunar cycles. My brain Vietnamese? Something else? Not know. Who exactly is he funning with? Not with anyone who has any acquaintance with Japanese, because if you knew Japanese you would realise that this is a ridiculous hodgepodge that is almost impossible to translate back into any semblance of Japanese. Did the author really use rickshaw (人力車) to mean bus (バス)? What exactly is Rubin talking about?

    I’m sorry, but that is really a poor excuse for a review.

  6. It sounds more like Chinese Pidgin English with several upgrades installed.

  7. I also found the review annoying, but for a different reason: the reviewer (and perhaps the author?) conflate the language with the script, or at least he seems
    to wander back and forth between them in a way that treats them as the same thing. There is the equivalence that moving to another script would render any past writings (that weren’t republished in a new form) largely inaccessible — but that is hardly the death of the language.

    My understanding is that the conversion of Turkish to the Latin alphabet was accompanied by enforced changes in language, so it might not be a good parallel: anybody know about this, or about examples of script changes that did leave the language intact? For that matter, what about acquiring a script for the first time?

  8. In Mongolia, change of the script to Cyrillic meant change of literary language from Classical Mongolian to Khalkha Mongolian. (In Buryatia, from Classical Mongolian to eastern Buryat dialect)

    So in a sense, we could talk about death of Classical Mongolian there as a result of script change.

    Linguistically, the closest analogy – loss of literary language status by Latin in Italy

  9. the reviewer (and perhaps the author?) conflate the language with the script, or at least he seems to wander back and forth between them in a way that treats them as the same thing

    My feeling is that it is Mizumura who is conflating the language and the script. The initial quote from Samson sets the tone of the article, and Rubin seems to be agreeing with Samson that the script is actually a disaster.

    Rubin appears to regard Mizumura as clinging to a Meiji-era view of Japanese, one which ties the script to the language in a mystical way and implies that Japanese is therefore difficult or impossible to translate into English — a view that he finds objectionable. I think he sees Mizumura’s jaundiced view of modern literature (which Rubin has made a career of translating) as reflecting this narrow-minded prejudice.

    Rubin feels that there is nothing mystical about it. Japanese can be translated (all that stuff about subjectless sentences, が meaning either ‘and’ or ‘but’, and ideographs is just obfuscation and nonsense), and as the translator of a modern “brainless writer of crap”, he finds Mizumura’s adherence to old-fashioned ideas about language and script objectionable. At least, that’s how I read him.

    To be honest, I tend to agree with Mizumura on this. Since I have a background in linguistics, I naturally take the view that the script is irrelevant; in formal terms it’s merely a way of representing the (spoken) language. But we’re talking about being “literate” here, and “literacy” is a lot more than representing language in a neutral way. Literacy involves access to older literature; it involves literary and graphical devices and how the writer makes creative use of them. It involves a whole lot of intangibles that we tend to ignore, like the way that texts are illustrated (e.g., reading Blake’s poetry in isolation from his drawings). Even looking at modern English, it’s interesting how writers obsess over trivial things like punctuation, which is actually quite a superficial aspect of language. Writers are criticised for using too many commas, or for putting full stops in the wrong place and producing “sentence fragments”. Compared with punctuation, the choice of writing systems to represent a language is a much more momentous decision. There are a lot of things that can be done with Chinese characters that can’t be done with English, and pretending that prizing the resources of the old script is simply clinging to something irrelevant, inferior, or outdated seems to me to be a kind of wilful blindness.

  10. But surely it’s going way too far to say that Japanese can’t be successfully translated. And you can’t blame a translator for objecting to such an idea.

  11. Trond Engen says

    I read the post title as On-reading Japanese.

  12. Ha!

  13. Of course it’s going too far to say that Japanese can’t be translated. But I don’t think you can deny that the obstacles to translation are far greater than from European languages, to the extent that an English translation from Japanese can often feel like a pale recreation of the original.

    My feeling on reading Rubin’s review is that he is merely expressing an allergic reaction as a translator to the quackery that is commonly trotted out in support of a mentality of Japanese linguistic exceptionalism. He appears frozen in a stance of permanent defensiveness towards the linguistic community whose works he is translating. Quackery it may be, but the objections of people like Mizumura do hold water.

    Mizumura claims that “As unbelievable as this may sound to the users of Western languages, Japanese sentences do not require a grammatical subject”. This is a red rag to a bull, as Rubin has elsewhere written at length demolishing the view that Japanese sentences do not have grammatical subjects. What in fact happens is that Japanese can, and frequently does, omit the grammatical subject of the sentence. Rubin is right in pointing this out. But this does not alter the fact that languages that mostly omit the subject have a very different feel from those that mostly express the subject. Mizumura is right in feeling that the many cultural, linguistic and orthographic differences between Japanese and English do add up to a very real distance between Japanese originals and English translations. You don’t need to exoticise (as in Rubin’s “funning” literal translation) to recognise this.

  14. Rubin has elsewhere written at length demolishing the view that Japanese sentences do not have grammatical subjects. What in fact happens is that Japanese can, and frequently does, omit the grammatical subject of the sentence.

    Nitpick: In that case, it has no grammatical (i.e. syntactic) subject. There may be something that would appear semantically as the subject if expressed, but what isn’t expressed can’t governed by syntactic rules. In any case, “subject” is not really a cross-linguistic concept at all: it can mean topic, or focus, or agent, or any number of other things that in many languages coincide but in others do not at all.

    But I agree with your main point. I once translated a short folktale in Hakka (from Ramsey’s The Languages of China) into Lojban, where any argument of a sentence predicate can be omitted if it is clear from context, but anglophones and other SAE-phones rarely omit the first (usually actor-like) argument. It definitely gave the story a different feel from other Lojban texts that were otherwise similar.

  15. I could understand most of the story by reading the Mandarin romanisation, but this one stumped me:

    – zhi1 zao2 kai1 lai2 – de ye3 – zon3 you3 (Note: zon3 is not a valid Mandarin syllable).

    Even with the English gloss (“where know bore open come something also not-exist assemble exist”) I had trouble understanding.

  16. @JC, isn’t that what the whole pro-drop business is about? Syntactic subjects existing in deep structure and getting suppressed on their way to the surface? (Shades of Lovecraft). So they do exist, you just can’t see them.

    (This should in no way or form be taken as an endorsement of pronouncements of universal truth emerging from just north of Boston).

  17. Well, yes, but personally I’d rather believe in Dagon than deep structure.

  18. Bathrobe: zon3 is a typo for zong3, I think. The fluent English translation, which I had not looked at when I made my Lojban version, says “How could she have known that when it [the hole] was bored open there would be nothing in it!” I think the word glossed “where” must belong to the previous sentence, anyhow.

  19. Well, but as Rubin points out, you do need a model of processing where the subjects that aren’t there can refer to entities that have been mentioned earlier.

    You can phrase it in terms of the verb presupposing a subject (topic / agent / experiencer) that will be filled in from context, but in so far as that role can be also supplied by an explicit argument I don’t see what is gained by claiming that sentences leaving it implicit are modelled differently — that actually leads to the ‘inscrutable Japanese mind’ claim that there is no intended subject.

    And if the thought process behind the production of the sentence, however it is realized, does hold awareness of the subject relation even when it’s redundant and not going to be expressed overtly — what’s wrong with analyzing the spoken utterance as containing a zero subject?

  20. I don’t doubt that sentences without a subject (modulo the problems with that term) are about something; I just don’t think that the something they are about is featured in the syntax, that’s all. Claiming otherwise seems to me like the old Latin grammarians saying that the subject of Pluit ‘It’s raining’ was Iūppiter “understood”!

  21. I tend to agree with Lars on this, but I’m having more and more trouble accepting any fixed or universal concepts of grammar.

    According to Y. R. Chao, the subject of a Chinese sentence is the topic.

    ‘The grammatical meaning of subject and predicate in a Chinese sentence is topic comment, rather than actor and action… The subject is literally the subject matter to talk about, and the predicate is what the speaker comments on when a subject is presented to be talked about.’

    Not everyone agrees that subject = topic, and many regard “subject” and “topic” as grammatically distinguishable notions in Mandarin. But proving that subject and object are grammatically distinct will require some deft argumentation of the hair-splitting type that linguists are so good at.

    Since the old subject-predicate division that formed the bedrock of grammatical theory seems to be increasingly under attack, it seems that nothing is really set in stone any more.

  22. But I don’t think you can deny that the obstacles to translation are far greater than from European languages, to the extent that an English translation from Japanese can often feel like a pale recreation of the original.

    Well, since I don’t know Japanese, I can neither confirm nor deny, but to me any translation is a pale recreation of the original, and I’m always suspicious of claims that a particular language (always, of course, one that the claimant has a strong attachment to) is particularly resistant to translation. I would prefer to say that in any language some texts will be relatively easy to translate and others relatively hard.

  23. I’m always suspicious of claims that a particular language (always, of course, one that the claimant has a strong attachment to) is particularly resistant to translation

    I agree with this attitude.

    My point is that Rubin’s frustration at Japanese exceptionalism has left him so hung up that he can’t help producing a disjointed dirge of denial when reviewing a Japanese author who embodies what he dislikes. Of course Mizumura is being frustratingly “Japanese”; that doesn’t make Rubin right. Railing against kanji for making Japanese more different than Rubin feels it is — or should be — doesn’t detract from the importance of kanji in Japanese language and culture. Rubin would be a lot more convincing if he gave some recognition to the actual barriers that exist rather than charging out to deny everything that niggles at him.

  24. It looks like it’s impossible to delete my preceding comment.

    It’s difficult to know where I’m coming from unless you’ve been through the mill of Japanese exceptionalism (Japan is unique, Japanese is unique, etc.). Reading his review, the scars of his experience and his obsession with Japanese exceptionalism are very apparent to me. If you get too obsessed with it you can’t see straight — and that’s why I think he’s written such a poor review.

  25. January First-of-May says

    Well, since I don’t know Japanese, I can neither confirm nor deny, but to me any translation is a pale recreation of the original, and I’m always suspicious of claims that a particular language (always, of course, one that the claimant has a strong attachment to) is particularly resistant to translation. I would prefer to say that in any language some texts will be relatively easy to translate and others relatively hard.

    A lot depends on how precise of a translation you want (this seems to correlate with how short the text is); on the more precise side of the range, languages that mark a lot of categories explicitly can be hard to translate into languages that don’t (because you need to figure out where to put the extra info), and vice versa (because you need to figure out where to get the extra info – this is harder, and thus shows up for slightly less precise translations). For example, “he stood” and “it had been standing” can easily be the same phrase in Russian (and many other languages), only disambiguated by context (which is not necessarily obvious).

    From what I’ve heard of Japanese, it’s pretty light on explicit marking, so a translation from Japanese would hit the “where to get the extra info” part fairly often. But then the English tense system is rather elaborate (I think I’ve even heard somewhere that no other natural language has a more elaborate tense system, though I doubt that this is actually true).

    But aside from effects like this (which, as I’ve mentioned, only occur for fairly precise translation attempts, and are thus mainly irrelevant for literary translation, which is not particularly precise), yes, I agree fully.

  26. @JC, there’s a not-so-subtle difference between the Japanese and Latin cases.

    If you join a Japanese conversation and the first thing you hear is ikimashita, you know you missed a bit of context because you don’t know who or what did the walking.

    If the first thing you hear in a Latin conversation is pluit you also don’t know who or what did the raining, but that’s fine because you don’t think about raining as something that has to have an agent.

    I would be surprised if Japanese doesn’t have similar impersonal verbs where no topic / agent / patient / experiencer is expected. And I won’t complain if your description of the syntax doesn’t have a subject ‘slot’ for sentences with such verbs — the verb would ‘select’ that sentence shape lexically.

    I’m complaining about ikimasu being marked ‘impersonal if we already know who does it.’ I don’t like automaton-like theories of utterance production either, but neither are we cogitating out each sentence from first principles as we speak, a lot is happening at levels below conscious thought. It seems simpler to have ikimasu marked for ‘sentence with a subject slot’ and just leaving that slot empty.

    Consider also that you can repersonalize rain by replacing the dummy subject in English (I’m gonna rain on your parade!) or adding an explicit subject in Latin (sanguinem pluisse senatui nuntiatum est). But you can’t depersonalize walk by using a dummy subject, because it in it walks will necessarily be taken as a ‘real’ pronoun with some sort of coreference; similarly for ire where leaving out the subject means the same. You have to use a passive construction of some sort.

    I think that proves that there is a syntactical difference between the cases. You can call it something other than syntax if you want, but then I’ll probably think you’re just being ornery.

  27. In general, I find it’s better to avoid commenting on something too close to home, but:

    (1) For *anyone* out there who may have googled their way here wondering if it really is true that “[f]oreigners, even those who teach Japanese literature at a university, cannot read novels written in Japanese with any ease,” please rest assured: as a principle it is definitely not true. Not in the slightest. One can only guess what sort of foreign Japanese experts led Mizumura to form this opinion, but whoever they were, if her impressions are at all accurate, it only means they simply hadn’t read enough. To me, the fact that Rubin himself seems to assent (“As Minae Mizumura accurately (if somewhat ungraciously) observes…”) is shocking if not scandalous, and I strongly suspect it’s just false modesty. At any rate it is false: if you read enough Japanese novels with enough persistence, you will most certainly get good enough to read them with some ease and pleasure.

    The key word is enough. It will take a lot of time. Language learning difficulty is mostly a question of relative distances. And for native speakers of most Western languages–the clear target audience for Rubin’s review–the learning curve is definitely much steeper than it is for Korean speakers (a lot of similar grammar, a large shared lexicon, intertwined histories), or Chinese speakers (some key grammatical similarities, a large shared lexicon, a shared writing system). But with enough effort and time spent actually, well, reading, Japanese is no different than any other language, and fluency is definitely within reach.

    (2) As far as vertical vs. horizontal, dainichi’s comment above is fascinating. I tend to find vertical writing much faster and easier to read (especially to scan), so I’m inclined to agree with him that it’s a matter of habituation, but I too would love to know if some concrete research exists. At times I’ve speculated that for alphabetic monoglots, native visual habits–used to horizontal parsing according to a very different visual grammar–perhaps interfere when reading Japanese horizontally, but not vertically. Still, I have no evidence whatsoever, so take it with a grain of salt.

  28. I agree with everything Bathrobe has said here. Especially the scars: Rubin’s obsession with the Japanese uniqueness bugaboo is indeed obvious and all too familiar. I admit to being really impatient with the whole thing. Academics especially should really have stopped taking it at face value long ago. “Unique” is just the positive way to spin “alone.” At the very least, anyone as familiar with Japan as Rubin knows these sentiments are not in the slightest confined to right-wing nationalists (though that’s where they go to curdle), and his insistence in seeing it all as a relic of prewar nationalism is warping. Take the following:

    Her readers were initially Japanese – lots and lots of Japanese, who made the book one of the most widely discussed non-fiction titles ever published when it first appeared in 2008, because they instantly identified with the author’s discomfort among other languages and speakers of other languages.

    Note the generalizing plural: a strong statement. And yet, his quoted evidence suggests discomfort with only a very specific subgroup of “other languages and speakers of other languages”:

    Mizumura finds herself surrounded by “the kind of people who would have been called ‘Aryan’ . . . . People who I secretly thought would look perfect in Nazi propaganda films as members of the Hitler Jugend”.

    The original title, too, hardly suggests a message of generalized other-generated discomfort: Nihongo ga horobiru toki: Eigo no seiki no naka de (lit. “When Japanese Collapses: In The Century of English”, though “Life in the Century of English” might capture the latter half better). Rubin must know that fear of decline, at a stretch of forced assimilation, is closer to the title’s mark. There’s no need to agree with her, of course, but on the surface he does seem to be misrepresenting her argument.

    Hat, I wonder (if I’m reading you right) that you think different languages are all equally translatable into any given target language. The point of subtracting for personal enthusiasm (“inexpressible!”) is well-taken, but surely there’s an opposite extreme to be avoided. Doesn’t it remind you just a little of the Chomskyan sense that basically all languages are the same, so English grammar is a suitable-enough universal proxy? It might be impossible to determine if German or Norwegian is easier to translate into English, but German or Japanese…?

  29. elessorn: Thanks for a pair of thoughtful and enlightening comments!

    Hat, I wonder (if I’m reading you right) that you think different languages are all equally translatable into any given target language. The point of subtracting for personal enthusiasm (“inexpressible!”) is well-taken, but surely there’s an opposite extreme to be avoided.

    Well, obviously from one point of view it’s silly to even talk about whether different languages are “equally translatable” when there’s no way to measure translatability and every translation is a different problem. My point is not that when measured against the Universal Translation Difficulty Scale Ibo-to-Mongolian and Japanese-to-English both measure 67.8 (which is in fact the Universal Translation Difficulty Constant), it’s that it makes more sense to start from the presumption that translation is going to be pretty difficult no matter which languages you’re going from/to (ease in one aspect being balanced by difficulty in others) than from the presumption that certain forms are easy and others hard (which is inevitably going to lead to the conclusion that the form one has personally sweated over is particularly hard). It’s about useful attitudes rather than scientifically measurable facts.

  30. Lars: My example was a bad one, because it dragged in the idea of truly impersonal verbs, which is irrelevant to my point.

    If you join a Japanese conversation and the first thing you hear is ikimashita, you know you missed a bit of context because you don’t know who or what did the walking.

    True. But if you join an English conversation and the first thing you hear is He walked … you are no better off (except you know the walker is male). In each case, there is an act of walking being mentioned, with a walker whose identity is not being stated. But in one case, a subject must be expressed and in the other case it need not be. This is a syntactic difference, and saying that all Japanese sentences really do have syntactic subjects, but they are sometimes suppressed, is just trying to paper over the distinction. Better, I think, to say that all acts of walking inherently have walkers, but in order to talk about such acts, some languages (the minority, I think) require a linguistic representative of the walker and some don’t.

  31. Maybe it just boils down to preference — I think it’s more complicated to allow two ‘syntaxes’ for ikimasu than to allow zero subjects. Nobody is papering over anything, allowing zero subjects is an interesting fact about the language.

    (Also where is your CS mindset? 0 is a number, {} is a set, ” is a string).

    Possibly relevant is that some ‘pro-drop’ languages allow the subject that isn’t there to be the antecedent of a relative clause. I believe that ivit qui potuit is good Latin (and I know too little Japanese to say if there’s equivalent constructions there).

    Seems to me that that must be simpler to handle in the zero subject narrative.

  32. No need for two syntaxes, just say that subjects can appear or not appear depending on context (whereas in English and other pro-require languages, they must appear).

    Likewise, I account for ivit qui potuit by saying that a relative clause in Latin needn’t attach to a NP, but can float free. What I don’t know is whether a missing transitive object can also have such a relative clause, or if it has to be about the semantic subject.

  33. The motto of Berea College is Vincit qui patitur. But i see I’ve been anticipated.

  34. Something similar seemed to be in vogue in English in the first half of the 20th century – the SAS’s “Who dares, wins”, or Orwell’s “Who controls the past controls the future; who controls the present controls the past”.

  35. That’s presumably the result of immersion in Latin.

  36. The construction applied to an indefinite subject, equivalent to one who or whoever, is probably as old as English. The OED quotes Hali Meiðhad (1230): “Hwa þat sehe þenne hu þe engles beoð isweamed … stani were his heorte ȝif ha ne mealte i teares”, i.e. “Who that sees then how the English are grieved … stony were his heart if he has not melted into tears.” A century later, Wyclif was writing, “All this shall be bought, as who buyeth an ox or a cow.” Shakespeare has Iago tell us that who steals his purse steals trash, and Jane Austen says “When a young man, be he who he will, … promises marriage, he has no business to fly off from his word.”

    The OED does say that its use with a definite (but unexpressed) subject probably is a Latinism; in Macbeth we hear “Who was the thane [of Cawdor] lives yet”, though no longer the thane. Still, there is of course no period of English writing when Latin influence is entirely absent.

  37. @JC, I’ll just repeat that to me zero subjects seem a small price to pay for the simplification in accounting for coreference and the like. But like someone said, this is a matter that reasonable people may agree to disagree about.

  38. Japanese doesn’t have relative pronouns, and you can’t really have free-floating relative clauses in modern Japanese. The Japanese equivalent of a relative clause is formed by an attributive phrase, with the antecedent serving as the phrase head:

    ki-wo mi-ta hito
    tree-OBJ see-PAST person
    “The person who saw the tree”

    Eliminating the phrase head turns this phrase into a complete sentence (with zero subject):

    ki-wo mi-ta
    tree-OBJ see-PAST
    “[implied subject] saw the tree.”

    In Classical Japanese, however, the attributive form of the verb was morphologically distinct from the sentence-final form, so you could have something like a free-floating relative clause; the implied headword would usually be “fact” or “act”:

    ki-wo mi-ker-u hito
    tree-OBJ see-PAST-ATTRIB person
    “the person who saw the tree”

    ki-wo mi-ker-u koto
    tree-OBJ see-PAST-ATTRIB fact
    “the fact of [implied subject] having seen the tree”

    ki-wo mi-ker-u
    tree-OBJ see-PAST-ATTRIB
    “the fact of [implied subject] having seen the tree”

    (cf. ki-wo mi-ker-i
    tree-OBJ see-PAST-FINAL
    “[implied subject] saw the tree”)

  39. Chinese would be a better test case than Japanese for the “intrinsic” virtues of horizontal and vertical scripts, since you have communities of readers who are accustomed to reading essentially the same script in horizontal (PRC) and vertical (Taiwan) orientations. You could therefore test in both directions, although you can never really eliminate the effect of prior exposure. (Taiwanese readers have almost all had some exposure to horizontal Chinese, but many PRC readers will never have read vertical Chinese except on signage.) In practice, prior exposure effects are likely to dominate, so it is easier for people just to keep doing whatever they have always been doing.

  40. @DMT, if I read you right it doesn’t look like Japanese (modern or classical) allows an unexpressed subject in a main clause to be the antecedent for a restrictive phrase (relative or attributive).

    Maybe Latin (and its descendants?) are unusual in that way. (I have no idea beyond Romance and Germanic languages, and WALS doesn’t seem to have a chapter for it).

  41. elessorn says

    …it makes more sense to start from the presumption that translation is going to be pretty difficult no matter which languages you’re going from/to (ease in one aspect being balanced by difficulty in others) than from the presumption that certain forms are easy and others hard (which is inevitably going to lead to the conclusion that the form one has personally sweated over is particularly hard). It’s about useful attitudes rather than scientifically measurable facts.

    I guess I have to disagree. Well, not disagree, exactly, but insist that your perspective, while very reasonable, rational, even natural to me as a cultural compatriot, still elides what I wager is the most important reason Mizumura’s book sold like hotcakes.

    The god’s-eye perspective you offer, where Ibo, Mongolian, Japanese, English are available as interchangeable objects of thought, like so many distinct yet intercompatible letter-forms in a single human-cultural font, is not how I think most non-Westerners experience the current arrangement. True enough that in a Platonic vacuum, English and Ibo are just two kinds of human language, and I think a lot of Hatters probably do experience language this way to some degree. But as modes of living that for most practical purposes are tied up undisentwinably with specific cultures, one of those four languages is simply not like the others, and translation into or out of it from non-filiate language communities is going to inevitably involve some of those differences.

    I still think, personally, that even in an absolutely objective sense, it is just easier as a rule to translate between similar languages (e.g. JapaneseKorean and GermanEnglish). But here I’m referring to something else–the preexisting cultural infrastructure that a translator can or can’t count on to support his translation’s reception. Because there is a huge existing market for translations from English into Japanese, a built-up base of readers interested in them, and a deep history of successful reception of Anglophone cultural products, it is effectively far “easier” to translate from English to Japanese than from Japanese to English.

    As someone who focuses on texts prior to the Meiji period, I’m not personally that sympathetic to many of Mizumura’s enthusiasms for the era. But when she waxes about what it is impossible to get across the bridge of translation, I think it’s worth considering that without being aware of it she’s thinking more of the cultural infrastructure aspect. She knows that much of what she appreciates in the best Japanese novels is at least not likely to strike readers untrained to appreciate it. And if we take her in this sense, she is clearly right, and Rubin wrong, about Japanese novels’ (effective) translatability. It surprises no one, for example, that translator-of-The-Great-Gatsby Murakami is popular in the West, but transient Japanese best-sellers are usually popular in translation only in East Asia, where a lot of similarities in everyday cultural attitudes act as supporting infrastructure.

  42. David Marjanović says

    But then the English tense system is rather elaborate (I think I’ve even heard somewhere that no other natural language has a more elaborate tense system, though I doubt that this is actually true).

    If you mean the whole tense/aspect system, AAVE has all the Wonderbread Englishes beat by a factor of 2 or so.

    If you just mean the tenses – present, past, future, pluperfect, “future perfect” –, that’s just the SAE complement (with the two rarest tenses likely calqued from Latin).

    Even looking at modern English, it’s interesting how writers obsess over trivial things like punctuation, which is actually quite a superficial aspect of language. Writers are criticised for using too many commas, or for putting full stops in the wrong place and producing “sentence fragments”.

    Some of these cases may be criticism of what the resulting intonation sounds like.

    the SAS’s “Who dares, wins”, or Orwell’s “Who controls the past controls the future; who controls the present controls the past”.

    For what that’s worth, this is perfectly natural everyday German, dialects included. That doesn’t necessarily mean it isn’t Latin influence there, too, but I suppose it makes it less likely…

    Wer wagt, gewinnt is even a proverb. (One of many that I only know from reading, but still.) I should perhaps add that in such sentences I’ve always interpreted the relative pronoun as the subject.

    I believe that ivit qui potuit is good Latin

    See also French sauve qui peut.

    I wonder if the German version rette sich, wer kann is calqued or just irregularly shortened: as a 3rd-person present subjunctive, rette does require a subject (dummy subject: es rette sich…). Perhaps confusion with the imperative (also rette – but rette dich) played a role in making either of these options possible.

  43. @David Marjanović: It’s fascinating how you, as a native speaker of a different Germanic language, have different associations about what are Latinisms than a native speaker would. I think it’s highly unlikely that any of the English tenses are Latinate calques. My brother, who is fluent in Old English (at least in written form) assures me that all the English tenses are as old as the hills. For that matter, I learned all of them (except the present progressive, obviously) as standard German, although some of them do not see much use in modern German. (“Ich werde schlafen zu gehen angefangen haben” was my favorite sentence in high school German.)

    On the other hand, “Who controls the past controls the future; who controls the present controls the past,” is transparently Latinate to me as an educated English speaker. Starting sentences with “who” just feels like an obvious imitation of Latin. Of course, I am familiar with the same construction in German, it nonetheless feels like a Latinism. Realistically, I suppose, all the grammatical structures under discussion here are probably as old as the hills in English and German (and Latin). But those that seem stilted and not particularly natural in one’s native tongue are easily interpreted as Latinisms.

  44. David Marjanović says

    Starting sentences with “who” just feels like an obvious imitation of Latin.

    How about the Biblical whosoever? Does that feel obsolete and foreign to you, or obsolete but native?

    assures me that all the English tenses are as old as the hills

    Well, certainly not if the youngest hills are 10,000-year-old moraines… 🙂

    I forgot that the plain perfect, which pretty much has to be traceable to some kind of Proto-West-Germanic and is now the only kind of past tense left in southern German (including Yiddish this time, right?), is thought to be a calque from Romance, where in turn it’s from Greek. Of course it doesn’t feel foreign to me at all, but it was apparently still sort of marginal in Old High German, and Gothic had no trace of it. I should have written “the two rarest tenses and one of the most common ones”.

  45. January First-of-May says

    If you mean the whole tense/aspect system, AAVE has all the Wonderbread Englishes beat by a factor of 2 or so.

    If you just mean the tenses – present, past, future, pluperfect, “future perfect” –, that’s just the SAE complement (with the two rarest tenses likely calqued from Latin).

    I meant the set {Past, Present, Future} times {Simple, Continuous, Perfect, Perfect Continuous}. The first component can also include Future-in-the-Past (as in “he said that he would”), and, depending on the textbook, might also include some passive varieties.

    Though that might be just the EFL version of English, as opposed to the actual English system as studied by actual linguists (though the “English makes all the tense distinctions that are made in any natural languages” assertion came from a linguist; that version counted eight, from three binary choices, though I cannot recall now what they were).

    EDIT: And, sadly, I’m not fluent enough in linguistic terminology to be reasonably sure what an aspect is (and how it differs from a tense).

  46. Eli Nelson says

    “English makes all the tense distinctions that are made in any natural languages” is clearly not true. If you include aspect in that category, you don’t even have to go to exotic languages to find counterexamples: French and Spanish both have a perfect-imperfect distinction in the past tense that is not present in English (in some cases, a similar distinction in meaning is expressed in English by the plain-progressive contrast, but in many other cases English simply doesn’t mark the distinction).

    If you exclude aspect, there are still plenty of other tenses aside from {Past, Present, Future} times {Simple, Continuous, Perfect, Perfect Continuous}. There are distant or mythical past tenses, and various types of tenses that specify the day an event occured or will occur: hesternal = occured yesterday, hodiernal past tense = occured today, hodiernal future tense = will occur today, crastinal = will occur tomorrow. English doesn’t have any of these.

    “English has the most elaborate tense system overall” is not as obviously false, but it seems very unlikely to be true from a statistical standpoint.

  47. David M: As the head of a fused relative (one that both introduces the relative clause and is its antecedent), whosoever is archaic but not foreign; whoever is normal; who is either archaic or foreign. Geoffrey Pullum has a table in his paper “The Truth About English Grammar: Rarely Pure and Never Simple” that shows how the simple wh-words can and cannot be used in English. Put into prose form (and replacing Puddlestonese with more ordinary terminology), it says:

    Introducing a question in indirect discourse: who, whom, whose (of persons), what, which, where, when, how, why grammatical; whose (of things), while ungrammatical.

    Introducing a restrictive relative clause: who, whom, whose, which, where, when, why grammatical; what, how, while ungrammatical (but grmmatical in non-standard dialects).

    Introducing a non-restrictive relative clause: who, whom, whose (of persons), which, where, when grammatical; whose (of things) dubious; what, how, why, while ungrammatical.

    Introducing a fused relative clause: what, where, while grammatical; who, when, how dubious; whom, whose, which ungrammatical.

    As Pullum says, there is no hope of explaining these brute facts even within a theory of English only, never mind any sort of universal theory of syntax.

  48. @JC: What I don’t know is whether a missing transitive object can also have such a relative clause

    Yep: e.g. misit qui cognoscerent “he sent (those) who would find out, sent people to find out”. This is fairly common in Latin, and more so in Greek.

  49. For what it’s worth, Danish cannot fuse subject and relative in these constructions: redde sig hvem/den som kan is the only option. Old Norse was full of sá/sú er too, or sas/sus in less normalized versions, so I think that’s the original state of affairs.

    hrørnar þöll
    sú er stendr þorpi á
    hlýrat henni börkr né barr
    svá er maðr
    sá er mangi ann
    hvat skal hann lengi lifa?

    (This stanza bothers me, by the way, or rather the traditional translation of verse 5 — mangi looks like accusative which would make it ‘who likes no man,’ but it’s always ‘whom no man likes’. Makes a difference).

  50. @David Marjanović: My understanding is that the simple past survives in Yiddish in exactly the same places as in other southern German dialects: in certain fixed idioms (although I can’t think of any at the moment) and for modal or auxiliary verbs (where it is needed to supply the past perfect, for example). However, you should keep in mind that my fluency with Yiddish is based pretty much entirely on treating the language as southern German with a lot of Hebrew vocabulary; so if there are real grammatical differences between the German and Yiddish, I may never have learned them.

  51. It’s worth noting, however, that the be-perfect is down to about 30 verbs in Yiddish, and in Litvak Yiddish it doesn’t exist any more, being completely replaced by the have-perfect.

  52. David Marjanović says

    in certain fixed idioms (although I can’t think of any at the moment) and for modal or auxiliary verbs (where it is needed to supply the past perfect, for example).

    That’s interesting about the past perfect; not only have all southern German dialects lost the simple past even for this purpose, replacing it with the compound past (so that the past perfect is now expressed with three words: “I had done” > “I have had done”), but – to my great surprise – this construction has recently been spreading northward and even reached the coast!

    In my dialect the simple past survives for “be”, where it’s used rather more often than the compound past, and for “want” except the 2pl (where, of the two obvious options for what that form could be, one would be identical to the present and the other to the subjunctive). No fixed idioms, no modal verbs, and “was/were” isn’t used for the pluperfect. In Switzerland it’s apparently gone entirely; I’m sure yet more variation exists. And that’s apart from the fact that each verb has its own isogloss for this in white-sausage-equatorial latitudes 🙂

    It’s worth noting, however, that the be-perfect is down to about 30 verbs in Yiddish

    Off the top of my head, that sounds rather high for German, actually. “Be” is used with verbs of motion, in the south also with “sit”, “stand” and “lie” and increasingly with “sleep”, but that’s it, I think.

    if there are real grammatical differences between the German and Yiddish

    Adjective declension: groser verterbukh fun der yidisher shprokh – the German expectation would be n.

  53. Trond Engen says

    spreading northward and even reached the coast!

    In Denmark well before 1989, reportedly.

  54. @David Marjanović: I think the double auxiliary for the past perfect exists in Yiddish as well (“ikh hab gehat getan”), but it is not obligatory in the Yiddish I am familiar with. However, I don’t know if my Yiddish knowledge corresponds to any authentic old-country dialect, or whether it’s just a mishmosh.

    Since modals in Germanic languages have a special grammar, they obviously do not always fit so well into the usual paradigms. What I would call the simple past for the modals, you would probably consider the imperfect subjunctive. But with German können, könnte, how can you tell which form the second one is? Of course, the meaning of könnte is not a simple past; it is effectively another modal of its own, with a slightly different grammar. The meanings of the two forms match pretty much exactly those of their cognates in English can, could; but English has no subjunctive form, so could must be the grammatical past. In the case of Yiddish, the pair is kenen, ken, and I was interpreting ken as the simple past, although as in German and English, it is effectively a separate modal.

    In addition to the differences in adjective declension in Yiddish, there are also some differences in verb forms for verbs of Hebrew origin. Going the other directly, some Germanic nouns can be pluralized with -im, although this is not obligatory in any cases (so far as I know).

  55. David Marjanović says

    In Denmark well before 1989, reportedly.

    …Fascinating.

    But with German können, könnte, how can you tell which form the second one is?

    That’s entirely unambiguous: the simple past is konnte, the subjunctive (Konjunktiv II) is könnte. Pretty much throughout the strong verbs, the subjunctive is formed from the simple past by umlaut (…sometimes with additional ablaut complications).

    There’s one trace of this left in conservative English: if I were – compare German simple past war, subjunctive wäre.

  56. Trond Engen says

    Lars:

    svá er maðr
    sá er mangi ann

    (This stanza bothers me, by the way, or rather the traditional translation of verse 5 — mangi looks like accusative which would make it ‘who likes no man,’ but it’s always ‘whom no man likes’. Makes a difference).

    Isn’t it “the-one who no-man loved”? Being a fusion of a noun and a distributive particle turned negator, manngi is not a regular nominal paradigm.

  57. Trond Engen says

    David: If havde is simple past, we must be looking either at an independent development or at a rather distant reinterpretation.

    Well, both har haft gjort (double perfect) and havde haft gjort (plusquam-double-perfect) are mentioned in the source. I agree that it might be an independent development, but the Northern German development could well be related to the Danish one.

  58. David Marjanović says

    I didn’t edit fast enough. 🙂 Most of the examples on p. 6 do indeed fit what I mean.

  59. North Sea Germanic not under the Standard German Dachsprache went to have-perfects long ago. In the King James Version the only remaining be-perfects are go and (be)come. Dutch still keeps be-perfects for intransitive change of state plus the six verbs zijn, blijven, beginnen, kwijtraken, naderen, tegenkomen. The exception is Shaetlan, where the be-perfect is the only one: I’m read [rɛd] my Bible, bit aftener [more often] da [New] Testament.

    I got that example from this paper, which says that in the Standard the be-perfect belongs to verbs of coming into or going out of existence (erscheinen, sterben), defined post-state (erbeben, erklingen), change of location (fallen, steigen), change of state (brechen, erblinden), and change over locations (laufen, schwimmen).

  60. In the King James Version the only remaining be-perfects are go and (be)come.

    I think there are a few more – rise and spring, at least. Genesis 19:23: “The sun was risen upon the earth when Lot entered into Zoar.” Matthew 13:26: “But when the blade was sprung up, and brought forth fruit, then appeared the tares also.”

    Also, though go mostly takes be, there are a few instances where it takes have. Leviticus 17:7: “And they shall no more offer their sacrifices unto devils, after whom they have gone a whoring.” Numbers 13:32: “The land, through which we have gone to search it, is a land that eateth up the inhabitants thereof…”

  61. It’s hard to know whether these are perfects or predicate adjectives. In contemporary English, we have the door has closed, but also the door is closed.

  62. the-one who no-man loved — well maybe, who is doing the non-loving? Is who or no-man the subject? That is my question. Since Old Norse has cases, it should be easy to find out.

    I haven’t been able to find a resource with the paradigm for manngi online and haven’t had time to visit a library. But you give manngi as the citation form, so that would be the nominative — then how can mangi (presumably accusative) be the subject? (I seem to remember that I did look it up 30 years ago and that it was accusative then. But I had no online fora to complain in).

    “You will perish if noone likes you” is a somewhat pedestrian and predictable sentiment, but the translations I’ve found go on this line unambiguously. “You will perish if you keep yourself from the company of men” is much sharper. And good advice. And fits the image of the tree standing in a barren place.

    EDIT: Found it now, Wimmer’s Oldnordisk Læsebog has a reference to his grammar, it was under indefinite pronouns and says that both nom. and acc. may be mangi or manngi indifferently. I may be in trouble here. On the other hand it is truly ambiguous and I can read it the way I like.

  63. Well, Bible Gateway finds 51 occurrences of risen in the KJV, all of them preceded by a form of be – with a singular exception in Acts, found in a strange syntactic spot.

  64. David Marjanović says

    I got that example from this paper, which says that in the Standard the be-perfect belongs to verbs of

    All the examples there are correct. Never trust an uneducated native speaker with statements about what exists and what doesn’t, especially not when they’re made late at night. 🙂

    Unrelatedly, for me, er hat den Arm gebrochen can only mean “he broke the arm I’m talking about” – “there’s that arm, and this guy broke it”; if it’s his own arm that’s broken, it has to be er hat sich den Arm gebrochen. Also, I seem to use the “have”-passive a lot less than the author; I find it really hard to contrive a situation where I wouldn’t understand er muss das Auto gewaschen haben as “he must have washed the car” – I’d replace the passive use by er muss das Auto sauberkriegen, “he needs to get the car clean”.

  65. I find it really hard to contrive a situation where I wouldn’t understand er muss das Auto gewaschen haben as “he must have washed the car” – I’d replace the passive use by er muss das Auto sauberkriegen, “he needs to get the car clean”.
    Same here. In my dialect it’s possible to replace haben by kriegen even in the original construction, i.e. er muss das Auto gewaschen kriegen, although I’d use and expect that kind of passive with kriegen only in situations where colloquial German is appropriate – in more formal contexts I’d say something like er muss es schaffen, dass das Auto gewaschen ist or er muss es schaffen, das Auto gewaschen zu haben. (For that reason, I also wouldn’t use bekommen instead of kriegen in that construction – bekommen is too formal to be used in such a colloquial cosntruction.)

  66. @Hans: in my dialect … er muss das Auto gewaschen kriegen

    Is that Ruhrpott, or gar a kind of Kölsch ?

    Cf. “he has to get the car washed”. Which, in the absence of further context, just like the German, could mean either that he intends to do it himself, or to have it done.

  67. In Denmark well before 1989

    But that’s not a past perfect, that’s some sort of perfect perfect. As I understand it, the dialects that David’s talking about have simply lost the past tense forms of ‘have’ and are using ‘I have had said’ for ‘I had said’.

    Jeg har haft gået i to timer means that yes, I have walked for that long on some specific occasion(s), but it wasn’t right now and I’m not making a habit of it, thank you very much. Very distinct from jeg havde gået i to timer [før jeg mødte nogen] which has to be part of a sequence of events where something happens after the walking. And from jeg har gået i to timer which is what you say if someone asks why you you look tired.

    For me even something like Hun har haft haft marsvin sounds fine in a context like ‘she already tried keeping guinea pigs but decided to stop’. As opposed to Hun har haft marsvin where she did it successfully but they died of old age.

  68. Is that Ruhrpott, or gar a kind of Kölsch ?
    It may be influenced by those. My father comes from the Rheinland / Ruhrgebiet “border” zone (he grew up in Essen and Oberhausen), I was born on the Lower Rhine (Mönchengladbach) and my mother grew up there (although she was born in now-Polish Elbing (West Prussia), from where her father came, too, while her mother was from Thuringia), but then I grew up in Northern Germany (Ostfriesland). So my idiolect is somewhat of a mixture of Rhein/Ruhr and Northern features.

  69. David Marjanović says

    As I understand it, the dialects that David’s talking about have simply lost the past tense forms of ‘have’ and are using ‘I have had said’ for ‘I had said’.

    Yes. So the Danish manifestation is a whole new aspect that didn’t exist before?

  70. Bathrobe says

    I read elessorn’s comments on translation with interest.

    Most major modern languages seem to have been “Westernised” to the extent that translation among them is no longer an insuperable challenge. Vocabulary equivalents for Western concepts have taken root and are in common use everywhere — witness newspaper reports which use similar language around the world to describe the same events. Even Western-style grammatical devices have been widely adopted in non-Western languages.

    Since elessorn focuses on texts prior to the Meiji period (I’m not sure which languages, but I assume they are East Asian), I’d be curious to hear his views on the ease of translation of pre-modern texts compared with modern texts. My feeling is that it’s far more difficult to translate Classical Chinese or Japanese texts into English than modern Chinese or Japanese tracts dealing with modern issues. I’ve never read “translator-of-The-Great-Gatsby Murakami” but I wouldn’t be surprised if he were among the easiest of them all.

  71. a whole new aspect: Whether it’s new (in Danish) is not clear to me. The 1989 treatment was probably triggered by peevery on the lines of ‘why use a double modal when it means the same as a single’ — or a descriptive query about what the difference in meaning would be. There were references to dialect surveys, but I didn’t look at the dates — for all I know it’s a feature that has been living a quiet life of obscurity for hundreds of years.

    It’s clearly not the same as the German development where the loss of the preterite of ‘have’ leads to the reshaping of the past perfect. This is a present stative construction for conveying present knowledge of perfective events in the past. At least an extra twist on one of the usual aspects.

    But I can hardly believe that nothing like it has ever been described, applying the periphrastic perfect construction twice is easy to come up with. Or if you have other verbal constructions to convey perfective aspect — maybe “I have been after walking that far” works in Irish English?

    (Hmm, there’s an antikvariat in Gothenburg that will sell me Skautrup’s “Det danske sprogs Historie” for 200 SEK per volume. I’m tempted. It’s too new to be on archive.org).

  72. David Marjanović says

    Fascinating.

  73. John Cowan says

    Lameen trying to translate Alfred North Whitehead into French (easy), Modern Standard Arabic (hard), and Darja Arabic (impossible). It’s about vocabulary, mostly.

  74. See Lola Seaton’s New Statesman review of Mizumura’s An I-Novel (excerpts in this comment).

Speak Your Mind

*