Replies: 2 comments 9 replies
-
|
Hi Kimeiga! Thank you so much! I think @enellis would have a much better idea about the feasibility of repurposing those rules. My impression is that unfortunately it might be difficult for the reasons you mentioned. |
Beta Was this translation helpful? Give feedback.
-
|
Hey Kimeiga! First of all, I have to agree with you that @birtles really is some kind of magician, who puts so much work into the little details, listening patiently to every feedback he gets, so this tool works out of the box for everyone without any problems. Thanks to him for maintaining this project in such a friendly and attentive manner! I am not really sure what is meant by "conjugation explainer" in #1970, but I assume this idea goes in the same direction. As a first step of working towards something like this, I will summarize some of my thoughts about Japanese inflections. I was working on the In my opinion, the biggest mistake you can make, is trying to learn Japanese inflections as you would do with western languages through conjugation tables. Japanese verbs don't really conjugate. There are the stems and the auxiliary verbs that attach to them. It's a very regular and straightforward system, far more simple than the irregular conjugation mess you have to deal with when learning any western language. Don't learn Japanese from resources that are "westernizing" the language! I think, the best approach of learning the language is in a way Japanese people themselves learn about their language. Of course, in the beginning, you are not able to approach Japanese resources, but you should try to get there as soon as possible. To do that, you should learn the following simplified version of the grammatical structure of Japanese, which helps you becoming being able to "parse" and break down Japanese text in a very short amount of time. As long as you have a firm grasp of the grammatical structure, you will be able to understand any Japanese text with the help of 10ten quite quickly. 五十音 - The heart of Japanese(Read top-to-bottom / right-to-left. あ, い, う, え, お, か, き, く, け, こ, ...)
Learning Hiragana and Katakana should always be the very first step when learning Japanese. Knowing Hiragana and the system behind it by heart is essential when trying to get behind the logical structure of Japanese grammar. Don't learn Japanese using Rōmaji, you will become confused! Everyone learning Japanese should already know this, so this is giving just a very brief overview. Most Japanese syllables, or, to be more precise, a mora starts with a consonant and ends with a vowel. The rows of the table represent the 5 vowels of the Japanese language and the columns represent the consonants. The exceptions are the first column's あ, い, う, え, お, which are just the single vowel sounds without a preceding consonant, and the last letter ん, which is a little special in that it cannot be at the beginning of a word and that it has several different pronounciations depending on what follows. ん doesn't really belong into the Excluding Yōon moras (e.g. しゃ, きゅ, りょ), which are not really of interest when looking at inflection, this table represents all sounds of the Japanese language. What word types to worry about?%%{init: {'themeVariables': { 'fontSize': '12px', 'fontFamily': 'Inter'}}}%%
graph TD;
di[No inflection] --> id1["Nouns / 名詞"];
Inflection --> id2["Verbs / 動詞"];
Inflection --> id3["Adjectives / 形容詞"];
Inflection --> id4["Copula / だ"];
As a start it is best to categorize Japanese words into 4 basic categories. Nouns, verbs, adjectives and the copula だ. We don't really have to worry about nouns. They don't inflect. Verbs, adjectives and the copula do. Let's have a closer look at verbs first. Verbs%%{init: {'themeVariables': { 'fontSize': '12px', 'fontFamily': 'Inter'}}}%%
graph TD;
id5["Verbs / 動詞"];
id5 --> id2["Godan / 五段"];
id5 --> id1["Ichidan / 一段"];
id5 --> id3["suru / する"];
id5 --> id4["kuru / くる"];
First, a few premises:
Godan verbsExamples: The stems of Godan verbs can easily be determined by looking at the specific column of the ending kana of the verb in dictionary form. Remember how there are 5 rows in the Japanese syllabary. These are the five stems!
One small exception: The a-stem of verbs from the first column of the Japanese syllabary (single vowel moras) is not あ but rather わ for a better pronunciation. That means that the a-stem of 買う is not 買あ but rather 買わ. We end up with 5 stem types -> 五段. The te/past-stemThere are two essential verb forms that cannot be constructed using the previously mentioned five stems: the past tense and the te-form. These forms share a specialized stem, which varies depending on the verb's final kana. The final kanas of Godan verbs can be categorized into "hard" sounds (う, つ, る, し, く) and "soft" sounds (ぐ, ぬ, ぶ, む). For verbs with hard ending sounds, the te-form and past tense are created by adding て and た to this specialized stem. For soft sounds, the suffixes て and た become softened to で and だ. To construct the stem, replace the final
Ichidan verbsExamples: Every Ichidan verb ends in -iru or -eru, but not every verb ending in -iru or -eru is an Ichidan verb. The stems of Ichidan verbs are super simple. They are all the same, i.e. there is only one: the dictionary form without the final る. Let's again look at all the examples above:
We end up with 1 stem type -> 一段. Be prepared though! Ichidan verbs utilize different auxiliaries in some cases. AuxiliariesThis is an overview of what auxiliary word attaches to which stem type.
Irregularities of 来るThe verb
Examples: こない, こられる, こさせる, こよう, きます, きて, きたら, くれば, こい. Irregularities of 為るWe can think of
Examples: しない, される, させる, しよう, します, して, したら, すれば, しろ. Irregularities of ますThe auxiliary verb
Examples: 見ました - did see, 食べません - does not eat, 読みませんでした - did not read, 行きましょう - let's go, いらっしゃいませ - Come! (used as a welcome in shops, restaurants, etc.) AdjectivesExamples: There are two types of adjectives in Japanese: i-adjectives (形容詞) and na-adjectives (形容動詞). Na-adjectives are fundamentally nouns and do not inflect on their own. To use them, a special form of the copula だ is attached, which governs their inflection. The copula, however, is covered in a different section. This section focuses on i-adjectives. The dictionary form of i-adjectives ends with the kana い. By removing the い, you get the stem, to which you can attach the various forms shown in the table below.
Examples: 嬉しく - happily, 新しくない - not new, 可愛かった - was cute Other auxiliaries, such as すぎる and そう, attach directly to the stem. Examples: 高すぎる - too high, 難しそう - seems difficult. Playing LegoThis is the most fun part of Japanese: You can play Lego with it – auxiliaries attach other auxiliaries. Let's say for example, you want to negate the adjective 嬉しい in the past tense. The first step is to build the negative form: 嬉し As another example, how would you politely express that you were drinking too much last night? First, remember the word type of This is very much work in progress. To be continued and updated... My understanding of Japanese is heavily based on Cure Dolly's Course On Organic Japanese. Even though the videos are not really that accessible, as the teacher takes the form of a V-Tuber and the audio quality can be quite rough and hard to understand sometimes, I think Cure Dolly's knowledge of Japanese was quite profound and admireable. Regrettably, she died in 2021 because of an illness. Her understanding itself was based on Dr. Jay Rubin's Making Sense of Japanese: What the Textbooks Don't Tell You (1998). |
Beta Was this translation helpful? Give feedback.
-
|
@birtles I'm wondering if it might be a good idea to start a wiki and explain the deinflection done by 10ten in a similar way to how I've tried here. We could also create individual articles for the various auxiliary words, providing detailed explanations of how they work, with the long-term goal of eventually integrating this information somehow into 10ten. What do you think? I can understand if you feel it might stretch the scope of the project a bit. |
Beta Was this translation helpful? Give feedback.
-
|
@enellis That sounds awesome. Stretching things a bit further, would there be some sort of human readable format we can synchronize with the code? Or perhaps the other way around—e.g. having markdown code comments we can extract along with the corresponding rules to generate a human readable summary of what is going on? |
Beta Was this translation helpful? Give feedback.
-
|
@birtles, That's a good idea—the explanations for the auxiliaries could be added as markdown comments for each member of the Reason enum. These comments could be parsed by a tool like TypeDoc, allowing us to display the generated information when, for example, hovering over a reason in the reason chain. I'm not a big fan of using submodules, but another option could be to parse the markdown pages from the Wiki Git repository, added as a submodule, using something like markdown-it. A GitHub Action syncing a |
Beta Was this translation helpful? Give feedback.
-
i wrote a long answer to this and then closed the tab but basically in that case it would just be hanas + -(i)ta(i) = hanasitai where "si" is pronounced "shi" in modern japanese. The point of the latin letters is just to be able to bisect the mora and think of godan verbs as having a stem that ends in a consonant and ichidan verbs as having their stem ending in a vowel. Indeed it seems there is some evidence that linguists analyse japanese in this manner https://books.google.com.br/books?id=SszxbMtHbs8C&lpg=PP1&pg=PA308&redir_esc=y#v=onepage&q&f=false you can see that the linguist in the book linked is discussing transitive-intransitive verb pairs in japanese and uses this structure like here: 
where he says -as- makes intransitive verbs transitive e.g. 乾く kawak- + -as = kawakas- 乾かす where 乾く is intransitive and 乾かす is transitive. Using this structure kinda merges the -a of the irrealis with the s of す with the knowledge that if another suffix was to be added to the end of this word, the す (or whatever kana is the final mora of the suffix) would have to be inflected to the correct verb base anyways to accept the next suffix. i dont really know a whole lot about linguistics here but i can see this as an interesting opportunity to cut down on the amount of things to remember (which verb stem accepts which suffix -> just memorize them as a pair put onto a stem). however I wouldn't know a whole lot about if this method of thinking would produce any incorrect conclusions |
Beta Was this translation helpful? Give feedback.
-
Oh no, I feel you, so frustrating when this happens.
It's an interesting perspective, and in the end, there really is no right or wrong. If you are able to remember and comprehend Japanese grammar analyzed this way best, then you should learn Japanese from resources that do so. What I'm not seeing, though, and this is the most important part when it comes to 10ten, is how that way of thinking about inflections could reduce the ruleset further than the stem system. Isn't remembering, for example, that たい attaches to the i-stem and せる attaches to the a-stem the same amount of information (especially from a programmer's standpoint) as remembering (i)tai and (a)seru? I'm sorry for being so opinionated toward the traditional stem system; it's just that it has proven very useful (we were able to reduce the deinflection ruleset by ~60% after integrating abstractions for the te/past-stem (連用形の音便), i-stem (連用形), and a-stem (未然形) for the first time). It could very well be that I just haven't fully grasped the benefits of that kind of analysis of Japanese yet. |
Beta Was this translation helpful? Give feedback.
-
Hi guys I just want to start by saying thank you so dearly for this project. Actually all of the chinese popup readers that I know of have been killed by the move to manifest v3 by chrome and I noticed right away that 10ten reader is still working, not sure how you guys pulled off this magic.
I am trying to study Japanese verb and adjective conjugations and noticed a lot of different ways to study them as indicated online. There's sort of a traditional approach of learning the different verb bases (Mizenkei, Kateikei, etc) (https://en.wikipedia.org/wiki/Japanese_conjugation) and then learning which auxiliary verbs can attach to which bases.
There also seems to be a way to understand it using "modern linguistic analysis" where you are willing to break up the mora in the case of godan verbs and imagine the stem of 書く (kaku) as kak-, and the "conjugations" would then include the vowel.
https://en.wiktionary.org/wiki/Appendix:Japanese_verbs
I've watched a lot of youtube video teachers as well and there's some dissonance about which way to teach it.
However when using 10ten reader I feel like I discover new conjugations I didnt know previously. For instance yesterday while using 10ten reader I discovered the conjugation: 待たされる which is 待たせられる where せら〜 contracts into さ〜, where I hadn't been taught that explicitly anywhere online (albeit I did find it on wikipedia later).
Given you guys have done so much work with deinflect.ts and word-search.ts to figure out japanese verb conjugations completely,
I was wondering if it made any sense for one to try to convert the knowledge contained in those files into some kind of human readable understandable learning tool for how to conjugate verbs?
Or would it be too difficult to create something human understandable out of it and it's better to learn it a different way? If this is true, is it because the rules are somewhat over-redundant for the purpose of ensuring correctness, as the deinflect.ts file is quite long? For instance, I see
['ませんでした', '', Type.Initial, Type.MasuStem, [Reason.PolitePastNegative]],['ません', '', Type.Initial, Type.MasuStem, [Reason.PoliteNegative]],mentioned separately. I suppose as a learner you might not learn these as separate rules but rather learn ません and learn です's past tense and put them together.
Any thoughts on this? And if it wouldn't make much sense, do you guys have any grammar resources that you would think are the best to learn the rules from? Again thank you so much!!
Beta Was this translation helpful? Give feedback.
All reactions