Talk:湯桶読み

"multiword terms" is a debatable concept when it comes to Japanese. In Western languages, you have spaces so you can make claims about word counts, and even so, they vary among languages. "Today" used to be spelled "to day". There are an awful lot of English terms that may be spelled with spaces or not, as one word or multiple words. "Rorschach test" in English is "Rorschachtest" in German. Spaces are sparsely used in Japanese, so what criteria are based your claims on? Also, "no sources" is a gigantic claim and you better have a solid comprehensive corpus of all sources there exist to back it up. I do hope you, as an admin, will try your best to remain subjective and neutral and not to project your own mere viewpoint and conjecture (or "deduction" or whatever) as you have done very often. ばかFumiko￥talk 12:01, 1 December 2017 (UTC)
 * The Japanese language has a concept of "word", despite the lack of whitespace in running text. There are even many words to describe different kinds of "word".  There may be disagreement on whether a given string constitutes one word or more, but there is no disagreement on the existence of single and discrete "words" in Japanese.
 * Monolingual Japanese sources specifically describe and  as applying only to single words:
 * -- Daijisen for 湯桶読み, Daijisen for 重箱読み, Britannica International for 湯桶読み, Britannica International for 重箱読み
 * -- Mypedia for 湯桶読み, Mypedia for 重箱読み
 * -- Sekai Dai Hyakka Jiten for 重箱読み
 * -- Shogakukan's Kokugo Dai Jiten
 * more restrictedly, limited explicitly to just -- Daijirin for 湯桶読み, Daijirin for 重箱読み, Encyclopedia Nipponica for 湯桶読み, Encyclopedia Nipponica for 重箱読み, Shin Meikai Kokugo Jiten
 * No description anywhere that I can find describes these reading patterns as applying to more than one word. If anyone can find such a description in a respectable source, I am happy to have "no sources" changed to "almost no sources".
 * Please do not remove content from entries without discussion; at a bare minimum, add an edit summary. If you have questions, I will work to answer them.
 * I do hope you, as someone who doesn't read Japanese, will try your best to remain subjective and neutral and not to project your own mere viewpoint and conjecture, in disagreement with cited sources. ‑‑ Eiríkr Útlendi │Tala við mig 18:57, 1 December 2017 (UTC)

RFV discussion: November 2017
lists three-kanji compounds, not just two like the definition claims. It's not an ideal source, so please provide with better ones. ばかFumiko￥talk 13:00, 20 November 2017 (UTC)


 * Daijirin specifically defines these reading patterns as applying to two-character compounds (「漢字二字でできている熟語」 → "compounds formed of two characters"). C.f. Daijirin entry for 湯桶読み, Daijirin entry for 重箱読み. However, Shogakukan adds a note that these labels can be used more broadly for any single compound term (for 重箱読み: 「また、広く、一語の漢字熟語を音訓まぜて読むことにもいう. 」 → "Also, broadly, used to describe readings of single-term kanji compounds read with a mixture of on-kun.").
 * Notably, the example terms with three kanji listed in the JA WP articles for ja:w:湯桶読み and ja:w:重箱読み all appear to be instances of an existing two-kanji compound read with on or kun and either prefixed or suffixed with another term with the opposite reading pattern. Some cases are what I would consider a multi-word term, like or, and as multi-word terms, these would not be either  or.
 * (The portion of  is itself read with the  pattern, but the entire term  cannot be considered as either  or  -- especially so given the inclusion of borrowed katakana term, which by very definition cannot be either on or kun).
 * However, some of the example terms include rendaku, indicating that these three-kanji compounds could be considered as integral words and not multi-word terms, such as or, and as such, the reading patterns for these could be considered as  or.
 * I will rework the and  entries to clarify the definitions and to add usage notes. ‑‑ Eiríkr Útlendi │Tala við mig 18:39, 20 November 2017 (UTC)


 * First, most sources you frequently cite also make faulty claims, such as a Portuguese term as *Olanda, so I would take them with a huge grain of salt, and if I spoke Japanese, I would seek something, well, more linguistic than some dictionaries that might favor prescriptiveness over descriptiveness, or be outdated and therefore not reflect the true current status of the language (which they do seem like they do and are). Second, the way you divide words into smaller parts seems arbitrary; I've read a romanization guideline that would do very differently based on kanji count, but then with various exceptions. It doesn't help that Japanese doesn't use spaces to separate words, so it's very tricky to determine whether a morpheme is free or bound, whether it should be separated from other parts with spaces or not. I've been following a way that more or less resembles that guideline I've read (based on kanji count), factoring word-medially only processes such as rendaku or renjo. ばかFumiko￥talk 19:23, 20 November 2017 (UTC)


 * You asked for better sources than Wikipedia. I provided several widely published monolingual Japanese dictionaries: Shogakukan's Kokugo Dai Jiten, Daijirin, Daijisen, and Shinmeikai.
 * As you note, these sources sometimes include mistakes. Importantly, mistakes such as the derivation of 🇨🇬 arise from misunderstandings of non-Japanese languages.  These sources are quite solid when it comes to describing the Japanese language itself.
 * By your own self-description, you don't read Japanese. I'm not sure how you'd be qualified to judge the quality of monolingual Japanese resources.
 * Regarding romanization and word chunking, you're correct that are both important factors to consider.  However, in the absence of these, I'm not sure how kanji count would factor into things, unless one is combining a simple count of kanji with an awareness of the underlying vocabulary.  Probably most kanji-spelled integral terms are two characters in length.  However, some are three characters long, and some are only one character long .  In , for instance, it helps to know that  is an independent term, and  is an independent term, but that * is not a term.  With this knowledge, we can tell that this is a compound of  and .  This compound exhibits no  (rendaku or renjō), the two portions have different reading types, the two portions are also used as independent terms, and the semantics are also clear as the two concepts put together as “hand” + “explosive shell / grenade”.  Given all of these factors, it makes sense to render this in romaji with the space as two separate terms.
 * If you have a link to that romanization guideline, I'd be interested in reading it. ‑‑ Eiríkr Útlendi │Tala við mig 01:35, 23 November 2017 (UTC)


 * I agree with Eirikr (except for the romaji part, I think, but I'm not sure how romaji is relevant to this RFV discussion). —suzukaze (t・c) 01:53, 23 November 2017 (UTC)