feat(glossary): add Japanese (ja) translation glossary

[mmcky]

Adds a Japanese built-in glossary (glossary/ja.json, 357 terms), mirroring the English term set and context fields of the existing zh-cn / fa glossaries so coverage is identical across languages.

What's included

• glossary/ja.json — 357 terms (economics, macro/micro, finance, linear algebra, probability/statistics, time series, network science, economist & mathematician names).
• dist-action/glossary/ja.json — the bundled copy produced by npm run build, so the action loads it automatically when target-language: ja (the CI dist-action freshness guard passes).
• Docs updated: Japanese row in docs/user/glossary.md and the "Current Glossaries" entry in glossary/README.md (🚧 Planned → 🟡 Draft).

Conventions followed

• Same English keys + contexts as the other glossaries (verified programmatically — identical order, no missing/extra terms).
• Abbreviations expand to the full Japanese term with the abbreviation in parentheses, e.g. GDP → "国内総生産（GDP）", OLS → "最小二乗法（OLS）".
• Personal names transliterated to katakana, e.g. Robert Solow → "ロバート・ソロー".

⚠️ Review note

This is an initial draft produced for native-speaker review (per the glossary quality guidelines), marked 🟡 Draft in glossary/README.md. A Japanese-speaking economist should confirm terminology before promotion to ✅ Complete. Two areas to prioritise:

• CJK-origin researcher names are currently katakana placeholders (e.g. Hiroki Kondo → ヒロキ・コンドウ, Yoshi Fujiwara → ヨシ・フジワラ, and the Chinese-origin names). The authors should supply preferred kanji.
• A few term choices worth a second look: 連結状態 (communicating states), 残余費用関数 (cost-to-go function), レイクモデル (lake model).

Note: this PR and #68 (French) both add a row to the same glossary table; whichever merges second will need a trivial rebase.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR adds a new built-in Japanese translation glossary to the action-translation repository, alongside documentation updates intended to expose target-language: ja support and ensure the bundled dist-action/ artifact contains the new glossary.

Changes:

• Added a new Japanese glossary file (glossary/ja.json) and bundled copy (dist-action/glossary/ja.json) with 357 terms.
• Updated glossary documentation to list Japanese as a draft glossary and reflect current status/metadata.
• Updated user docs to include Japanese in the built-in glossary table and add a draft status note.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
glossary/README.md	Documents Japanese glossary status/contents and updates “Last updated” metadata.
glossary/ja.json	Adds the new built-in Japanese glossary (357 terms).
docs/user/glossary.md	Adds Japanese to the built-in glossaries table and notes draft status.
dist-action/glossary/ja.json	Bundled (built) copy of the Japanese glossary for runtime use in the GitHub Action distribution.

[jstac]

Thanks @mmcky !

I think that

• "lake model" should stay in English --- the katakana looks strange to me, and Japanese readers are probably happy if 90% of the sentence is in Japanese.
• non-Japanese economists names should stay in English --- easy for a Japanese reader to parse if the rest of the sentence is in Japanese; not-so-famous people won't have standard katakana and it's not obvious from the English spelling what the katakana should be. (Claude translates my name as スタチャースキー based on the guess that it's pronounced "Sta-chur-ski", but in fact a more accurate translation, which I use for banking etc., is スタハースキー).
• I would even say that Japanese economists could be rendered in English at a first pass, not much harm done, better than trying to guess the Kanji.

[xuanguang-li]

Hi @jstac,

Thanks for your suggestions.

"lake model" should stay in English

I don't have a good sense of how this term is translated, but I couldn't find anything relevant by searching for レイクモデル on Google.

non-Japanese economists' names should stay in English

I agree with this point. References to non-Japanese economists are often written in English, which makes it easier for readers to look up the relevant literature. This convention is generally followed even for economists whose names have kanji equivalents, such as Chinese economists.

I would even say that Japanese economists could be rendered in English at a first pass

I think this approach is fine as long as the paper is written in English. However, in presentations, it's also very common to write Japanese economists' names in kanji on the slides.

[Chihiro2000GitHub]

Thanks @jstac — I'm a native Japanese speaker, so here are my thoughts on the three points.

1. lake model — I agree that レイクモデル looks odd, but I would write it as lake モデル: keep the model name in English, and translate only model. Writing the name in English and the rest in Japanese is common in Japanese academic writing. If a Japanese term is already widely used, we should use it — for example 世代重複モデル (OLG model) or ソロー・スワン成長モデル (Solow-Swan growth model). But there is no such term for lake.

2. Non-Japanese economists — I fully agree. Please keep them in English.

3. Japanese researchers — if we are citing a paper written in English, English is better. But when the name appears in a normal sentence, kanji reads much better. If we know the person's research area and university, we can almost always find the correct kanji, so we are not guessing here. I am happy to do this myself.

Before we decide, though, I would like to check how these names are actually used in the lecture text.

[Chihiro2000GitHub]

@jstac and @xuanguang-li -- I just had a look at glossary/ja.json, and there are almost no Japanese economists in it — only two names (Hiroki Kondo and Yoshi Fujiwara).

So thinking about it again, I'm fine with keeping all of the names in English. One rule for everything is simpler.

What do you think?

[jstac]

Many thanks @xuanguang-li @Chihiro2000GitHub , nice comments.

So thinking about it again, I'm fine with keeping all of the names in English. One rule for everything is simpler.

Yes, one rule is much simpler, I agree.

We have to think about efficiency and, at least in this case, it doesn't cost us much quality.

[xuanguang-li]

Hi @Chihiro2000GitHub and @jstac,

Thanks for your feedback. I also agree with the principle of keeping the rules simple.

I read through the json file and looked up the terms flagged by @mmcky.

• communicating states: I think we can use コミュニケートする状態, which I found in a lecture slide.

状態 𝑥から出発し、状態 𝑦 に到達する確率が正、すなわち𝜌_𝑥𝑦 = 𝑃_𝑥(𝑇𝑦 < ∞) > 0であるとき𝑥 は𝑦にコミュニケートするといい、 𝑥 → 𝑦と表す.

• cost-to-go function: I think cost-to-go can remain in English, as it is used in the title of this paper.

近似的最適cost-to-go関数を用いて3D誘導性能をマッピングする

I also found the following terms a bit unusual, although I'm not sure whether they should be changed:

• forcing variable (強制変数)
• forward-looking difference equation (先見的差分方程式)

I would appreciate any suggestions on these terms.

[sayaikegawa]

Hi everyone, (cc @jstac )

John has just asked me for my thoughts on translating technical terms. (I'm Japanese and a former RA and student of his.)

I think it is better to keep technical terms in English, partly because this makes it easier for students to transition to materials taught entirely in English later on. Personally, I also find technical terms more difficult to learn in Japanese than in English, since the Japanese translations often use quite complicated kanji.

I also looked at the teaching slides by Professor Abe at Hitotsubashi University, whose lecture notes I used when I first learned dynamic programming. He uses the English terms “value function” and “Bellman equation.”

Here is his lecture note link.

[jstac]

Thanks @sayaikegawa , super helpful!

[Chihiro2000GitHub]

@xuanguang-li — I completely agree with your suggestions.

On forcing variable and forward-looking difference equation, I would like to look through some Japanese textbooks and papers myself and think about them a bit more. I'll come back with suggestions.

Hi @sayaikegawa, thanks for joining!

Professor Abe's lecture note is very well known. I think it is probably the first thing a Japanese student finds when they search for dynamic programming in Japanese, so using it as a reference is a really good idea.

The examples you gave are dynamic programming terms — "value function" and "Bellman equation" — and I agree that those should stay in English.

But I would like to note that a general rule of "keep technical terms in English" would make some terms feel unnatural. For example, 行列 is much more familiar than "matrix", and 予算制約 is the normal way to write "budget constraint". Budget constraint → 予算制約 is in the glossary, and 行列 appears in many entries such as 隣接行列 (adjacency matrix) and 確率行列 (stochastic matrix).

In fact, the same lecture note does exactly this:

予算制約を使って、今期の資本ストックと来期の資本ストックの可能な組み合わせ、ここでは1600×1600の行列の各要素に消費水準を割り当てる。

Here it writes 予算制約 rather than "budget constraint", and 行列 rather than "matrix".

Of course, there may be a question of whether words like 予算制約 and 行列 count as "technical terms" in the first place...

[jstac]

Thanks @Chihiro2000GitHub for your additional comments.

Summarizing: We can use Japanese when there's a well-known and stardard Japanese counterpart, as in 予算制約 for "budget constraint", and English otherwise.

Even an AI might be able to make these judgements...

[Chihiro2000GitHub]

Hi @jstac, @mmcky, @xuanguang-li and @sayaikegawa,

Sorry this took a while — I wanted to go through the whole glossary carefully rather than just skim it. Here is what I found.

I have split it into three parts: changes I would suggest, terms I think should stay in English, and a few questions I am not confident about.

1. Suggested changes

Term	Current	Suggestion	Reason
Annuity	年金（アニュイティ）	年金	年金 alone is enough. The katakana gloss is not needed.
Eigenvector decomposition	固有ベクトル分解	固有値分解	固有値分解 is the phrase actually used in Japanese, even though "vector" drops out of it.
Exercise	演習	演習問題	演習問題 sounds more natural in teaching material.
Fiscal theory of price levels	物価水準の財政理論	物価水準の財政理論（FTPL）	FTPL is fairly widely used in Japan.
Flow utility function	フロー効用関数	各期の効用関数	More familiar to me, though this may be a matter of taste.
Gross Domestic Product	国内総生産	国内総生産（GDP）	GDP is more familiar to Japanese readers than 国内総生産 alone.
Kernel	核	カーネル	カーネル is the normal word here.
Marginal utility of wealth	富の限界効用	資産の限界効用	資産 is more familiar, if "wealth" means assets in this context.
Matrix inversion	逆行列	逆行列の計算	逆行列 means "inverse matrix" — it is a noun, not the name of an operation.
Multiple consumer economy	複数消費者経済	複数の消費者からなる経済	Reads more naturally.
Projected corporate tax revenue	予測法人税収	法人税収の見通し	The current one reads like a literal translation.
Real bills theory	真正手形理論	真正手形説 (or 真正手形学説)	I did not know this term, so I looked it up. 説 / 学説 seems to be more common.
Trigonometric form	三角形式	三角関数表記	I could not find 三角形式 in any search. See for example this page.
Unpleasant monetarist arithmetic	不愉快なマネタリストの算術	マネタリストの不愉快な算術	This word order is the one that has settled in Japanese.

One further note on the table:

Recessions and Expansions. These two are a pair, but the glossary has 景気後退 for one and 景気拡大局面 for the other. Both are fine, but we should add 局面 to both or to neither.

2. Terms I would keep in English

Term	Current	Reason
Authority centrality	オーソリティ中心性	オーソリティ is not a familiar word in Japanese. I would treat this as a technical term and leave it as authority centrality.
Cost-to-go function	残余費用関数	Agreeing with @xuanguang-li's earlier suggestion.
Forcing variable	強制変数	@xuanguang-li flagged this. I looked into it as well and could not find a good Japanese equivalent.
Forward-looking difference equation	先見的差分方程式	Same as above.
Survey of Consumer Finances	消費者金融調査（SCF）	消費者金融 means the consumer lending business in Japanese, so 消費者金融調査 is misleading. This is also the proper name of a survey, so I would leave it in English — the same as FRED and NBER, which the glossary already keeps as they are.

3. Questions for @xuanguang-li and @sayaikegawa

I am not confident about these, and I would like your views.

Market clearing → 市場清算. I have heard 市場清算（条件）before, but honestly it does not sound natural to me as Japanese. I would expect something like 市場均衡（条件）. What do you think?

State-contingent claims → 状態依存請求権. This is the Arrow-Debreu context, I assume. 請求権 does not feel like the word normally used here. In derivatives, "contingent claim" is usually 条件付き請求権, so 状態条件付き請求権 might be closer, but I am not sure.

Counter CDF → 相補累積分布関数. I have not studied this area much, so I cannot judge it. Does this look right to you?

Reform and liberalization → 改革と自由化. I first thought this meant China's 改革開放, but that is usually "Reform and Opening Up" in English, so I may be wrong. It is worth noting that the zh-cn glossary renders this same entry as 改革开放, which suggests it is the China context. Could someone check which lecture this comes from?

＝ versus ・ in compound names. The file mixes the two: コブ＝ダグラス and ギアリー＝ケイミス use ＝, while ペロン・フロベニウス, コルモゴロフ・スミルノフ and ソロー・スワン use ・. Claude suggested that the convention is ＝ for one person's compound name and ・ for joining two different people, which would make コブ・ダグラス the consistent choice. But when I searched, I found コブ・ダグラス, コブ＝ダグラス and コブダグラス all in use, so I am not sure Japanese writers distinguish these very carefully in practice. Do you have a preference? Either way I think we should pick one and apply it consistently.

[Chihiro2000GitHub]

One clarification on my previous comment, in case it was confusing.

Section 2 ("terms I would keep in English") was not meant as a complete list of the terms that should stay in English. Those were just the ones I found borderline, or that had already come up in this thread. There are many others — Bellman equation, as @sayaikegawa mentioned — where English is clearly the right choice, and I did not list them.

So I am not proposing anything different from @jstac's summary: Japanese where there is a well-known and standard Japanese counterpart, English otherwise. I was applying that rule, not making exceptions to it — please take my list as suggestions rather than conclusions.

[jstac]

Thanks @Chihiro2000GitHub .

I propose the following rule: if in doubt, keep it in English. Most (maybe all?) Japanese undergraduate students will be comfortable reading Japanese sentences that have some English words.