Jim Davis
General Comments
The process of translation can be broadly divided into two distinct steps: comprehension (i.e., grasping the meaning) of the source text and expression (i.e., using appropriate words and phrases to render that meaning) in the target language. The relative strengths of an individual in these two areas vary from person to person and depend on a number of factors. In this commentary both comprehension and expression will be addressed, as needed.
All five finalists did a good job of conveying the overall meaning of the contest passage, and we wish all of them success in their translation careers. This section will be devoted to specific words and phrases in the passage that proved to be difficult for multiple finalists or to points of general interest in the context of translation.
When we create a translation, it is tempting to focus on individual words and to translate each word as accurately as possible. However, a translation that is “accurate” at the word level may not be truly “effective” at the document level. A translation that looks beyond the individual words to the message the writer is trying to convey—in other words, a “meaning-based” translation—is ultimately more faithful to the content the writer has prepared in the source language. (There are exceptions. In a legal context it is important to accurately convey what an individual actually said or wrote. In that context “intent” is not part of the equation.) One excellent example appears in the final sentence of the passage. In this sentence the writer introduces the term 点取りマシーン, which was variously translated as “point-scoring machines” (E05, E51), “scoring machines” (E20), “point-grabbing machines” (E42), and “test-taking robots” (E46). The first three options could be regarded as literally correct, but the reader of the translation might not fully grasp the point the writer is making. Given that the context here is Japan’s educational system, the fourth option (“test-taking robots”) is a more effective representation of the writer’s deep dissatisfaction with Japan’s current educational system.
On the other hand, there are instances in Japanese-into-English translation in which a single character conveys important information that must be incorporated into the translation. One example appears in the title of this article: “AI が可能にする「アイディア即プロダクト」の経済.” The character in question is 即. Here are the translations offered by the five finalists:
The economy of ‘idea to product’ made possible by AI E05
AI Enables “Instant Materialization” Economy E20
The AI-driven economy of ideas and instant products E42
The Instant
Idea-to-Product Economy and the AI Behind It E46
AI and the Birth of the Idea-to-Product Economy E51
(The italics were added for this discussion.) It is clear that three finalists accounted for the presence of 即 in the title of the article, while two finalists did not. The phrase “idea to product” does not reflect any revolutionary element of the product development process. Practically every product throughout human history originated as someone’s idea. In the body of this article the writer clearly articulates the “ease” or the “speed” with which that process may proceed in the era of AI. In that sense, the inclusion of “instant”—or some other term with equivalent meaning—is necessary in order to be faithful to the writer’s message. Other possibilities include translations such as “The Idea-Straight-to-Product Economy Engendered by AI” or “The Idea-Direct-to Product Economy that AI Makes Possible.” As a side note, the term “materialization” often appears in dictionaries and glossaries and may be appropriate in specialized fields but is seldom used in a context such as this one. In this instance retaining the terms “idea” and “product” is a more effective approach.
One distinction that is worth remembering is the distinction between a title and a job description. In English a title is usually placed in front of the person’s name and is capitalized. A job description frequently follows the person’s name and is normally not capitalized. For example, we might speak of “Professor Edward Seidensticker of Columbia University” or “Edward Seidensticker, professor at Columbia University.” In the first instance “Professor” is a title; in the second instance “professor” is a job description. The author information for this article was rendered as follows:
Economist Tomohiro Inoue E05, E42
Tomohiro Inoue, economist E20
INOUE Tomohiro, Economist E46
Tomohiro Inoue – Economist E51
Given the common practice in English, regardless of the word order that may appear in the source text, E20, E46, and E51 made the right choice. As an aside, it is true that the family name precedes the personal name when a Japanese person’s name appears in a Japanese text, but the standard practice in the English-language media (e.g., The New York Times or The Wall Street Journal) with regard to a Japanese person (e.g., the name of the author for an English translation of a Japanese novel) is to place the author’s personal name before the author’s family name (e.g., “Haruki Murakami”). In addition, rendering a family name completely in upper case may be requested by some translation clients in order to distinguish a family name from a personal name. However, this is not customary in the English-language media, and this should only be done upon special request.
Some finalists had difficulty deciding whether “AI” should be treated as singular or plural. To resolve this question, we might ask, “What is AI?” Fundamentally, AI is a concept. In contrast, ChatGPT or Stable Diffusion may be regarded as a tool, a computer program, or a model that incorporates this concept. Thus, AI should generally be treated as singular (e.g., “AI is ...” or “AI makes it possible to ...”). Consequently, the pronoun that is used in place of “AI” should be “it,” not “they.” Of course, a phrase such as “AI tools” or “AI models” should be treated as plural.
E05
E05 produced a solid translation, but a number of changes could be made to sharpen the focus of certain words and phrases or to make the scope of other words and phrases better match the intent of the writer. For example, in the second paragraph E05 translated 電子書籍 as “electronic documents.” In the sentence in question the writer cited 絵本 and 漫画 as specific examples. Thus, “e-book,” which was the choice of the other four finalists, would be a better translation.
In the third paragraph the writer uses the phrase アイディアをすぐ形にできる to describe one attribute of the new economy made possible by AI. E05 expressed this attribute as follows: “allowing us to immediately give form to our ideas.” This rendering suggests that the ideas in question have no form. That is not really the writer’s intended meaning. It is true that 形 literally means “shape,” “form,” or “style,” but in this article the writer is focusing on the process of turning ideas into products quickly. Given the context, this attribute could be rendered, “allowing people to immediately put their ideas into physical form.” This translation would allow the attribute to better align with the writer’s image of this new economy. E05 handled the fifth paragraph very well.
The term 潮干狩り in the seventh paragraph is a bit tricky. Literally, this combination of characters seems to refer to “hunting (for something) at low tide” or “gathering (something) on a beach.” In that sense, the general term “beachcombing” would seem to be a perfectly good choice. However, online research suggests that in practice 潮干狩り is used specifically for “clamming,” “clam digging,” or “digging for clams.”
No matter how hard we may try, the English translation of a Japanese document usually ends up being significantly longer than the Japanese original. However, E05 was able to find a very concise translation for 美醜や善悪: “aesthetics or morality.” The eighth and ninth paragraphs were well done.
In the eleventh paragraph the writer states the following: 油絵具もカメラもテクノロジーであり、絵画も写真も一種のメディアである. E05 produced this translation: “Oil paints and cameras are technologies, and paintings and photographs are one kind of media.” The writer used であり in the first half of this sentence, but the translator may well ask, “Are oil paints really a technology? Are cameras really a technology?” In fact, most people probably regard oil paints as a product that embodies the technology of a certain era. Similarly, most people would regard a camera as a product that incorporates the technology of a different era. This understanding should appear in the translation. The second half of the sentence is problematic for a different reason: there are two topics (絵画も写真も), but the information presented about each topic is the same (一種のメディアである). In fact, paintings and photographs are two different kinds of media. One option would be to make the second half of the sentence plural, matching the first half of the sentence. Such a translation could read, “Both oil paints and cameras are embodiments/products of technology, and paintings and photographs are both types of media.”
E05 wrapped up the article on a strong note. Phrases such as “In order to flourish in the AI era” (thirteenth paragraph) and “creativity that harnesses technology” (thirteenth paragraph) demonstrate E05’s potential in the realm of Japanese-into-English translation.
E20
E20 exhibited good comprehension of the source text but struggled with expression in a number of instances. In the first paragraph E20 referred to ChatGPT as “a language-generation AI.” As mentioned in the General Comments section, “AI” is a concept, while ChatGPT is a tool, a product, or a model. Thus, it would be more accurate to describe ChatGPT as “a language-generation tool,” “a language-generation product,” or “a language-generation model.” In the following sentence E20 employed the phrase “image-generation AIs such as Stable Diffusion.” Online research indicates that some people do use the term “AIs,” but “an image-generation AI tool/product/model” would be a much better description of Stable Diffusion.
In the fifth paragraph the writer introduces several categories of employment that he believes will be significantly impacted by AI at some point: 事務職や会計士・税理士などの士業、私のような教員・研究者など. Dealing with lists in Japanese can be difficult, particularly when the particle や, a comma, and the “nakaten” (・) occur in the same list. E20 rendered this list as follows: “including clerical staff, certified professionals such as public accountants and tax accountants, academics like myself and researchers.” E20 treated会計士 and 税理士, which are linked by the “nakaten,” as examples of 士業, but kept 事務職, which is separated from 会計士 by や, as a separate category. This is reasonable. However, E20 chose to separate 教員 and 研究者 into distinct categories, even though these two professions are also linked by the “nakaten.” If 教員and 研究者 are linked in the same way that 会計士 and 税理士 are linked, we may assume that 私のような modifies both 教員 and 研究者. If the writer is an academic in the field of economics, it is likely that he is also a researcher. (Online research confirms that this is true.) If so, the translation could read, “including clerical staff, certified professionals such as public accountants and tax accountants, and academics and researchers like myself.” (It is worth emphasizing that the presence of an “Oxford comma” following “tax accountants” is essential in order to make clear to the reader that there are three categories of employment in this list.)
The sixth paragraph includes several awkward phrases. E20 stated that, “There are three things the present AI is void of.” The meaning is certainly correct, but the phrase “the present AI is void of” seems unnatural. If we think about the meaning behind the words, it is the current AI tools (such as ChatGPT) that are lacking these elements. Better options could include phrases such as “the current AI tools lack ...” and “missing from the current AI tools are ....” E20 expressed 体験 as “real experience.” The reader can certainly guess what “real experience” might be, but “real-world experience” would be a better choice.
In the seventh paragraph the phrase “indulge in web surfing or reading” is an excellent choice for the corresponding phrase that includes the verb かまける. However, the first clause in the following sentence reads, “Never having experienced activities like clamming or campfires.” “Clamming” is a good choice for 潮干狩り, but a campfire itself is not an activity. The phrase “activities like clamming or sitting around a campfire” sounds more natural. Following “clamming” with another “-ing” word—in this case, “sitting”— adds a pleasing sense of rhythm to the translation. (After all, it is clear from the context that the writer has consciously selected enjoyable activities that people would remember fondly.)
In the eighth paragraph the writer describes an additional limitation of current AI tools. E20 translated the first sentence as follows: “Furthermore, their inability to make a subjective value judgment on things like beauty and ugliness or good and evil prevents themselves from doing more than following judgments made by humans.” The first half of the sentence is right on the mark. However, use of the phrase “prevents themselves” in the second half of this sentence is unfortunate for two reasons. First, if a pronoun is desired in this context, that pronoun should be “them,” not “themselves.” Second, and more importantly, Japanese is a high-context language. Thus, Japanese writers are free to use pronouns without clear antecedents, relying on the reader to make the desired linkage. In contrast, English is a low-context language. Thus, good writing in English requires that each pronoun be clearly linked to the noun that pronoun is replacing. The Japanese writer is free to use “自分で” to refer back to the AI tools he mentioned in the previous paragraph, but an English translation must be more specific. The use of “their” near the beginning of the sentence is fine, so long as some specific noun appears later in the sentence to let the reader know who or what the word “their” refers to. In this context the phrase “prevents AI tools” (rather than “prevents them”) would provide the necessary linkage. E20 expressed the phrase 人間の判断を真似る as "following judgments made by humans.” The verb “following” is too vague for this context. Phrases such as “mimicking judgments made by people” or “copying judgments made by people” would make this criticism of AI tools as pointed in English as it is in Japanese.
The twelfth paragraph is very well phrased. This paragraph shows the kind of translation E20 can produce, if sufficient attention is paid to the kind of details cited here.
E42
E42 made a number of excellent choices in the process of expressing the writer’s intended meaning. However, a few tweaks to the wording could improve the overall translation. For example, as mentioned in the General Comments section, the abbreviation “AI” should be treated as singular, not plural. In contrast, applications or models that incorporate AI could be singular or plural, depending on the structure of the sentence in question.
In the discussion of the ratio of jobs to applicants in the fourth paragraph E42 expressed the sentence この倍率は今後ますます低くなるだろう as, “This figure will only lower more and more in the future.” The choice of “lower” as the verb in this sentence seems odd. “Lower” is frequently used as a transitive verb (e.g., “Please lower your voice.”), but it is seldom used as an intransitive verb. An alternative such as “This ratio is likely to decline even further in the future” would be more natural.
E42 did a good job of grouping the various professions cited by the writer in the fifth paragraph, but the phrase “academics and researchers like me” is too colloquial for an article published in a magazine with national circulation. “Academics and researchers like myself” is the phrase an American economist would probably use if (s)he were writing this article in a comparable English-language magazine. It is worth remembering that one goal of a professional translator is the production of a document that is indistinguishable from a document originally written in the target language by a native speaker of the target language. The tone of the translation is one important factor in the achievement of this goal.
The first sentence of the sixth paragraph (“What the current AI lack is will, first-hand experience and judgement of value.”) is problematic for several reasons. First, the subject should be “current AI tools,” “current AI models,” or some comparable alternative. The problem the writer is addressing in this sentence lies not in the concept of AI itself but in the AI applications that are currently available. Second, the pattern “what [subject] lacks is ...” or “what [subjects] lack is ...” is normally combined with a single noun—not a series of nouns (e.g., “What he lacks is patience” or “What they lack is time”). Given that the writer intends to introduce three elements, a different pattern is called for. Finally, the phrase “judgement of value” seems odd. Perhaps this is a British expression. From an American point of view, “value judgment” is the term we need. Putting everything together, this sentence could read, “Current AI tools/models are lacking/deficient in the following three areas: volition/will/initiative, first-hand experience, and value judgment.”
The first half of the second sentence of the same paragraph reads, “Although AI like ChatGPT have high deviation scores indicating great academic ability, they are also likened to humans that wait for instructions and don’t take the initiative.” E42 translated 偏差値 literally as “deviation score,” but that doesn’t really make sense in this context. If we think about the meaning of the entire phrase スーパー偏差値エリート, we realize that the writer is comparing generative AI tools (such as ChatGPT) to people with a particular skill (i.e., the ability to obtain significantly above-average scores on exams). The first portion of this sentence could read like this: “A generative AI model, such as ChatGPT, could be regarded as a superb test taker.” The remainder of this paragraph and the following four paragraphs are very well done.
The first sentence in the eleventh paragraph reads, “Oil paint and cameras are forms of technology, and pictures and photographs are forms of media.” Inclusion of the word “forms” in both the first half and the second half of the sentence accurately captures the writer’s intent and adds a pleasing sense of rhythm to the sentence. However, “oil paints” would probably be a better choice than “oil paint.”
In the final sentence E42 wisely translated 今の教育 as “Current Japanese education.” The source text does not explicitly state that the writer is focused on the Japanese educational system, but that is certainly the writer’s intent, and it is that intended message that a professional translator is tasked to deliver. E42 is on the right path.
E46
E46 produced an excellent translation. However, a few sentences could be improved.
In the fifth paragraph the writer introduces several categories of employment that he believes will be significantly impacted by AI at some point: 事務職や会計士・税理士などの士業、私のような教員・研究者など. Dealing with lists in Japanese can be difficult, particularly when the particle や, a comma, and the “nakaten” (・) occur in the same list. E46 rendered this list as follows: “From certification-bearing professionals such as office workers, accountants and tax specialists to educators and researchers (such as myself).” E46 correctly recognized that 私のような modifies both 教員 and 研究者, which are linked by the “nakaten.” However, E46 chose to include 事務職 in the same category as 会計士 and 税理士 (which are linked by the “nakaten”), even though 事務職 and 会計士 are linked by the particle や (not the “nakaten”). The key question here is whether or not “office workers” are 士業 in the same way that “accountants and tax specialists” are. The writer probably used や to indicate that he regards 事務職 as a category of employment that is distinct from 士業 (such as 会計士 and 税理士). If that understanding is correct, the translation could read, “From office workers to certified professionals, such as public accountants and tax accountants, to academics and researchers like myself.” This translation makes clear to the reader that this list includes three categories of employment.
In the twelfth paragraph E46 translated 人間の革新的なクリエイティヴぃティ as “radical creativity.” This is an overstatement of the writer’s intended meaning. (E46 may have been thinking of 革命的な, which often means “revolutionary.”) In a political context the adjective 革新的な could be expressed as “liberal,” “progressive,” or perhaps “reformist.” However, in this context the best choice for 革新的な would simply be “innovative.”
The lone sentence of the thirteenth paragraph (“What is truly needed to excel in the age of AI is volition, experience, and judgment, as well as creativity that makes full use of technology.”) is problematic for several reasons. Generally speaking, the pattern “what is truly needed is ...” is more effective when it is followed by a single noun—not a series of nouns (e.g., “What is truly needed is patience” or “What is truly needed is time”). Given that the writer intends to introduce multiple elements, a completely different pattern is called for. Also, “first-hand experience” would be a better equivalent to 体験 than simply “experience.” (If we think about the difference between 体験 and 経験, “experience” would be fine for 経験, but a somewhat stronger word is needed for 体験.) Finally, the third element listed in the source text is 価値判断, which should be “value judgment,” not merely “judgment.” One option for this paragraph would be the following: “Volition, first-hand experience, and value judgment will be essential in order to excel in the age of AI. In other words, we will need creativity that makes the best use of technology.”
On the other side of the coin, E46 included several turns of phrase that enlivened the text. Phrases such as “a servant awaiting its next command” (指示待ち人間 in the sixth paragraph), “a style all its own” (そもそもの「〇〇風」in the ninth paragraph), and “test-taking robots” (点取りマシーン in the final paragraph) make the text interesting and help the reader visualize the writer’s intended message.
E51
E51 did well in many respects but missed the mark in a few instances. In the opening paragraph E51 included the phrase “take in a simple verbal prompt” for 言葉を入力する. The word “prompt” is exactly what the writer of an English-language article on this topic would use. However, the word “verbal” could be misleading. It is true that “verbal” could indicate that the prompt consists of words (rather than images), but “verbal” could also suggest that the prompt is spoken, rather than written. In principle, the prompt could be spoken or written. Thus, “inputting a simple prompt” would be a better choice.
The second paragraph ends with the clause 小物や家具といったものも個人でたちどころに作れるようになるだろう. E51 expressed this clause as follows: “individual users may even be able to craft tools, accessories, and household items on the spot.” The source text lists two examples of products; E51’s translation includes three. The noun 小物 could be “accessories” or perhaps “small items.” The noun 家具 usually refers to “furniture,” but “household items” is not a bad choice. However, it is not clear where “tools” came from.
In the third paragraph the writer uses the phrase アイディアをすぐ形にできる to describe one attribute of the new economy made possible by AI. E51 expressed this attribute as follows: “in which ideas can instantly be given form.” This rendering suggests that the ideas in question have no form. That is not really the writer’s intended meaning. It is true that 形 literally means “shape,” “form,” or “style,” but in this context the writer is focusing on the process of turning ideas into products quickly. Given the context, this attribute could be rendered, “in which ideas can instantly assume physical form.” This translation would better align with the writer’s image of this new economy.
In some portions of the article the writer refers to drawing; in other portions of the article the writer refers to painting. In the second sentence of the third paragraph the writer makes reference to people for whom 絵が苦手. Rather than taking a chance on “regardless of drawing ability” or “regardless of painting ability,” E51 chose the phrase “regardless of artistic ability,” which covers all bases.
E51’s opening sentence in the sixth paragraph reads, “Current AI systems are lacking in three crucial elements: intention, experience, and judgment.” E51’s use of “AI systems” rather than just “AI” is commendable. However, in this context “will,” “volition,” or “initiative” would be a better translation of 意志than “intention.” In addition, “first-hand experience” or “real-world experience” would be better than “experience” for 体験. (Please see the comments for E46 on this point.) Finally, the term 価値判断 refers specifically to “value judgment,” not “judgment” in the general sense.
The phrase “building campfires” is a good choice in the seventh paragraph. Although the text simply reads キャンプファイヤー, the writer is clearly focused on human actions and human experiences. Either “building campfires” or “sitting around a campfire” would emphasize the experiential element that the writer is trying to convey.
In the eighth paragraph E51 employed the phrase “create works” for 作品を生み出す. A 作品 could certainly be a “work” (in the artistic sense), and the phrase “works of art” is widely used, but the phrase “create works” seems awkward. If we focus on the writer’s intended meaning, rather than the individual words used by the writer, we may be able to find a different pattern that is faithful to the writer’s intended meaning but avoids the awkward wording that arises from a more literal translation. In this instance, “engage in creative endeavors” would be one option.
For the opening clause of the eleventh paragraph E51 wrote, “Paintbrushes and cameras are both technologies.” The term 油絵具 actually refers to “oil paints,” not “paintbrushes.” In addition, as mentioned in the comments for E05, both oil paints and cameras are “products that embody technology” or “products that incorporate technology.” These products are not themselves technologies. E51 clearly has potential for a career in Japanese-into-English translation, and we wish E51 success in this field.