2

Older grammar books tend to rely on a lot of romanization to teach the nitty-gritty of morphology, which is also reflected in some of @snail's answers like this one. I was reading Samuel Elmo Martin's 1975 book Reference Grammar of Japanese and noticed how verbs are described (P. 308):

  1. Transitivization, in which an underlying intransitive verb (such as kawak- 'get dry') is converted into a transitive verb (kawakas- 'dry it') by the addition of a suffix, here -as-.
  2. Intransitivization, in which an underlying transitive verb (such as hasam- 'interpose') is converted into an intransitive (hasamar- 'is interposed') by the addition of a suffix, here -ar-.
  3. Polarization, in which both transitive and intransitive are to be derived from some hypothetical basic form : e.g. naor- 'get improved' and naos- 'improve it' seem to be derived, by the suffixes -(a)s- and -(a)r- respectively, from a nonexistent verb *nao­- (etymologically to be found in the adverb nao 'yet, rather').

Just as @snail did in her answer, Martin also breaks up syllables/morae and splits consonants from their pairing vowels when illustrating morphological changes. Also he calls "-as-" in the word kawakasu 乾かす a suffix. In his theoretical framework, a word like 乾かす has two suffixes, -as- being one of them.

Hence my questions:

  1. Why is morphological analysis done at this level? Does it have to be done this way? Or is this kind of morphological analysis—anchored in romanization—a thing of the past?

  2. How is morpheme defined in Japanese? Nothing about its clear definition can be found on either the English morpheme Wikipedia page or the Japanese 形態素 Wikipedia page. Has there been more than one definition? Morphemes are supposed to be the smallest meaningful lexical item in a language, but in Martin's theoretical framework morphemes are not bound by morae or 仮名. This kind of morphology, it seems to me at least, would make sense in a phonetic and alphabetically based language, but the Japanese language had long existed before the language, the land, and the people had contact with alphabets and way before romanization attempts were made. It's almost as if no morphology is possible without alphabets and phonemic letters.

  3. How is suffix defined in Japanese morphology or morphologies I should say? The information available here on 接尾辞 seems more intuitive and 仮名-based.

Eddie Kal
  • 11,332
  • 5
  • 19
  • 40
  • Isn't it similar to English (and probably all the languages) where morphemes are not bound by syllables? – Yaroslav Fyodorov Apr 27 '22 at 08:25
  • 1
    Morpheme seems to mean a variety of things https://linguistics.stackexchange.com/questions/39066/the-classification-of-morphemes Also kind of related https://linguistics.stackexchange.com/q/17121/34148 – sundowner Apr 27 '22 at 08:31
  • 1
    There is no such thing as alphabetically based language - the spoken language comes first. But I feel that you are aware of that. Japanese is special in a way but not that special. Sorry, if I am writing obvious things. – Yaroslav Fyodorov Apr 27 '22 at 08:33
  • @YaroslavFyodorov I get your point: oral tradition always precedes written language. True. But morphology has to be done based on (or let's say with the help of) some form of written text, right? By definition morphology is the study of forms (what things _look_ like). Claiming it is solely about sounds (phones) and is irrelevant to the form of written representation is in itself taking a very bold position IMHO. And my question is: is that the only position we have to take? If so, where's the proof and theoretical basis? – Eddie Kal Apr 27 '22 at 15:41
  • 2
    @EddieKal erm. Oral-only (correct English term escapes me at the moment) languages has morphology as well. Often much richer than written languages (such as English that has almost no morphology from the point of view if someone coming from highly inflecting language ;) So, no, morphology is study of word form, it doesn't matter whether it has written representation. Moreover, written representation often diverge from the phonetics (English I am looking at you) and only interferes with word analysis. Also, think about language such as Hebrew which doesn't mark it vowels (at least originally) – Yaroslav Fyodorov Apr 27 '22 at 15:45
  • 1
    @EddieKal Would you analyze it (Hebrew) using only consonants of the word - of course not. – Yaroslav Fyodorov Apr 27 '22 at 15:51
  • @YaroslavFyodorov I'm not arguing against your position. I'm summarizing it, scrutinizing it, teasing out the core, and asking for its theoretical basis. The Hebrew example may be a red herring. I understanding you are trying to draw an analogy, but also in analogy it's like I'm asking for the proof of 1+1=2, and you showed me 1+1≠3 which is true and related but doesn't really prove 1+1=2. Again, I'm not arguing 1+1=3. – Eddie Kal Apr 27 '22 at 16:23
  • 2
    @EddieKal never mind Hebrew. My position is that written language is a nice addition, sometimes convenient, very important for historical analysis, invaluable for dead languages, but entirely secondary when analyzing a living language. I more or less take it for granted and my understanding is that is a consensus in contemporary linguistics (putting aside language freaks, and I am not implying anything), but I can try finding some citations. I find it hard to think otherwise - take any amazonian language - does it has morphemes? But of course – Yaroslav Fyodorov Apr 27 '22 at 16:43
  • I don't always agree with either analysis approach (using kana or using romaji) -- both have their places. I think the romaji approach arose for two reasons -- 1) writing for an audience that may not necessarily read Japanese, and 2) indicating morpheme divisions that do not align with a syllabaric approach. In 書【か】く, for instance, the core of the verb is clearly //kak-//, which cannot be clearly written in kana (without coming up with some new way of using kana -- which some authors have indeed done). – Eiríkr Útlendi Apr 27 '22 at 16:46
  • In Martin's analysis above, his theorized //-as-// breaks down when looking at bigrade verb conjugations, which clearly have no //a// component preceding the //s// in the causative / transitive forms -- they have just the //-s-//. Modern //tabesaseru// has //-saseru// as the suffix, which is from older //-sasu//, and that //-sasu// is from older //-su//. Following Martin's approach to decomposition, this //-su// would be something like //-s-// (causative / transitive) + //-u// (terminal). – Eiríkr Útlendi Apr 27 '22 at 16:48

0 Answers0