Per-Word, Per-Character, and the Risk Factors a Rate Card Doesn't Always Show

# Is Pricing Based on the Source Language or the Target Language?

Q. When a quote comes back with a different rate per language, is that based on the language being translated from, or the language being translated into?

A. This is the question most clients should ask first, and it usually gets skipped. In most professional translation and localization projects, volume is counted on the source language, not the target language.

If an English source document of 10,000 words is translated into Korean, Japanese, Chinese, and Arabic, the standard basis for the quote is "10,000 English source words," regardless of which target languages are involved. Translating into Japanese or Chinese does not automatically mean the quote shifts to a character-based count.

The relationship reverses depending on which language the content originates in. If the source document is Japanese, which has no spacing between words, the quote is typically based on the Japanese character count instead.

So the real question a client needs answered upfront is simple:

Is this project quoted on a source-language basis or a target-language basis?

Most global LSPs and CAT-tool-based workflows default to source-based counting.

Why Are Chinese and Japanese Usually Priced Per Character?

Q. Why do Chinese and Japanese rate cards usually show a per-character price instead of a per-word price?

A. Because Chinese and Japanese don't use spaces between words the way English does, which makes word counting structurally unreliable for these languages.

In English, a sentence like "The product is available now." can be counted as five words simply by counting the spaces. But in Chinese or Japanese, sentences such as 产品现已上市 or 製品は現在利用可能です have no built-in word boundaries. Where one "word" ends and the next begins is something a segmentation algorithm has to infer, and different tools can draw that line differently.

Character count, by contrast, is unambiguous and produces the same number regardless of which tool or vendor measures it. That stability is exactly why it became the standard:

If Chinese or Japanese word counts were forced through a word-based conversion anyway, different vendors would likely produce different numbers for the exact same source text. So rather than asking "why isn't Chinese priced per word like everything else," it's more accurate to see character-based pricing as the more objective and reproducible standard for these specific languages.

Korean Is an Asian Language, So Why Is It Priced Per Word?

Q. Korean is often grouped with Chinese and Japanese under "CJK." Why doesn't it follow the same per-character pricing?

A. Because Korean uses spacing, which makes word-count analysis in CAT tools reliable in a way it isn't for Chinese or Japanese.

A sentence like 이 제명은 현재 사용할 수 있습니다 (roughly, "This product is currently available") splits cleanly along spaces into units that can be analyzed as words or word-like tokens in standard CAT workflows.

This isn't a perfect one-to-one match with how English counts words, Korean particles and verb endings attach to a base word and carry more grammatical information per unit than an equivalent English word does. Even so, for quoting and CAT-analysis purposes, Korean behaves much closer to a European language than to Chinese or Japanese.

So the common assumption that "Asian languages are priced per character" isn't quite accurate. A more precise way to put it:

Chinese and Japanese lend themselves naturally to character-based counting. Korean's spacing makes word- or token-based counting viable.

How Are Arabic and Persian Priced?

Q. Arabic and Persian use a completely different script and run right-to-left. Does that mean they're priced per character too?

A. No, somewhat counterintuitively, Arabic and Persian are usually priced per word. Both languages use spacing between words, even though the script itself is unfamiliar to readers of Latin-based languages, so standard source-word counting applies in most quotes, just as it would for English-to-Arabic or Persian-to-Korean projects.

What makes Arabic and Persian projects genuinely different isn't the counting method, it's the operational risk that sits alongside the word count:

In practice, the more realistic model for Arabic and Persian is per-word pricing plus a separately scoped DTP/QA risk line, rather than treating the per-word rate as the full picture. Plain business text in a Word document may need nothing more than a standard per-word quote. But Arabic or Persian content inside a PPT, brochure, packaging design, app UI, or webpage should carry separate DTP, LQA, and in-context review costs on top of the translation rate itself.

Does Content Type Change the Pricing Model, Even Within the Same Language?

Q. Once the language-based rate is set, is that the end of the pricing conversation?

A. Not quite. Content type often matters as much as language. The same Japanese-language project, for instance, can fall under very different billing models depending on what's actually being translated:

Marketing copy, brand text, taglines, and slogans are the clearest example of why word count alone is a poor proxy for cost. A ten-word slogan can take far longer to get right than a ten-word instruction sentence, because the translator isn't moving words across languages, they're redesigning a phrase to function naturally in an entirely different market. This is exactly where the question "it's only ten words, why does it cost this much?" tends to come from.

What Should a Client Actually Ask When Requesting a Quote?

Q. Beyond "how many words is this," what should a client be confirming before accepting a quote?

A. A short checklist goes a long way toward avoiding cost surprises later in the project:

1. Is this quoted on a source-language or target-language basis?
2. Are Chinese and Japanese being counted by character?
3. Is Korean being counted by word, or by a token/어절 unit?
4. Do Arabic or Persian require RTL DTP review?
5. What's the source file format, Word, Excel, PPT, PDF, InDesign?
6. Is there an existing TM, glossary, or style guide?
7. Is this MTPE, human translation, or transcreation?
8. Does the final deliverable need to be reviewed in its actual on-screen or design context?

The clearer the answers to these questions are upfront, the more accurate the quote, and the less likely the project is to run into cost increases or schedule slips later on.

Case Study: A Global SaaS Company's Simultaneous APAC and MENA Localization

Q. Can you walk through how this plays out on an actual multi-language project?

A. Consider a global SaaS company localizing a 50,000-word English help center and UI string set into Korean, Japanese, Chinese, Arabic, and Persian.

The client initially assumed every language could be quoted at the same flat per-word rate. The actual breakdown looked different once each language's basis and risk profile were applied:

Because the source language was English throughout, the base quote could be unified on English source word count across all five target languages. Arabic and Persian, however, carried a separately scoped RTL layout review cost on top of translation, and Japanese and Chinese added in-context review to account for UI string length limits and the lack of context in short strings.

The end result was that the client stopped seeing "different rates per language" as a pricing inconsistency, and started seeing it as a direct reflection of differences in quality risk and scope of work.

Conclusion: What Does a Rate Card Actually Need to Reflect?

Q. If you had to summarize what actually drives translation pricing across languages, what would it be?

A. Translation pricing is never just a function of character or word count. It reflects a language's writing system, spacing convention, file format, CAT-tool analyzability, DTP requirements, and review scope, all at once.

For a client, the more useful question isn't "why does this language cost differently," it's "what additional work risk does this specific language actually introduce." For an LSP, explaining pricing well means going beyond handing over a rate card, and walking the client through the counting basis and risk factors behind each language. That's what makes a quote transparent, and it's what helps a client understand the real relationship between quality and cost.