The Research Process of Translation: What 60 Translation Professionals Have to Say

Words translators use to describe their translation process.

In late 2019 I interviewed or surveyed 60 translation professionals (mostly freelance translators) about their translation research process and use of translation tools. I owe it to those I interviewed to share the insights, so I’ve decided to do so publicly.

Tips:

1. See the Results section to skip to the interesting insights.
2. See the Methods section and this template to do a similar analysis on your own data.

Who am I?

1. I studied Spanish Translation, Computer Science, and Computational Linguistics.
2. I founded a translation technology company but abandoned the idea months after finishing this survey to join another startup.

What’s different about this study?

1. It’s a translation market study focused on serving the translators themselves, as opposed to companies or clients.
2. There are no multiple choice answers. I combed through the open-ended answers of each interviewee with painstaking detail.
3. 65% of respondents were interviewed (not just surveyed).

A sneak peek:

Why do you seek information about a concept, word or phrase?

Purpose of the Study

This study was exploratory to understand the types of problems that translators face to see how technology can be used to solve those problems.

Starting 2 years before this study, I had planned out how to solve the translation problems I was aware of myself and had formalized a business model and even demo around such a solution.

As not all would be as interested in the business implications of this, I save this information to the end.

Types of Respondents

• Interviews: 39
• Answers by message or email: 21

For the below numbers, I considered what experience they had within the last 3 years before the interview. Note that the below numbers do not add up to 100% out of 60 as the same person may fit multiple categories.

Types of translation professionals who were surveyed

• Freelance translator: 39
• In-house translator: 8
• Translation services company leader: 7
• Technical translation professional: 4 (coder or translation technology expert)
• Translation Project Manager: 3
• Literary translator: 3
• Translation Professor: 1
• Senior/Regional translation manager: 1
• Interpreter: 1
• Only studied translation: 1
• Total respondents: 60

As you may surmise from the above numbers, my focus for this study was freelance translation, in addition to my existing network. It does also reflect to some extent what the most common roles are among translation professionals.

Interview Questions

Qualification and specific customer segment

1. How does your work involve translation?
2. How much time do you spend translating?

Understand the Experience

1. What is your process for translating?
2. What is your process for getting information about a word or phrase?
3. What translation tools do you use for your translation?

Problems

1. What parts of the process take too much time?
2. In a typical hour of translating, how much time do you spend getting information about a word or phrase?
3. What parts of the process are difficult?

Current work-around

1. What websites do you use to find information about words or phrases?
2. What other sources of information do you use when deciding how to translate a concept?
3. Do you use information from the CAT tool?

Unfulfilled Desires

1. Is there any type of information you wish was available to help you translate?

Motivation for Translation Research

1. Why do you seek information about a concept, word or phrase?
2. Does this information improve the quality of your translations? How do you know?
3. How do you know that translation information is trustworthy?
4. How do you know when your translation is done?

Method

Note that there were no multiple-choice questions in this survey! By carefully going through the free-form answers of interviewees and survey respondents, I identified categories from the data and then applied those categories to all the respondents where applicable.

Then, I took it a step further and generalized those categories into broad categories for each chosen area as described below at Results Summary. This was done on a Google Spreadsheet.

Advantages of this data organization:

1. You can at any point add new respondents and the summarizing numbers will update in real time.
2. You can identify the precise names associated with each category and go to the raw data (responses) to back up the numbers.
3. You can apply different splits of the data in the future without re-categorizing the responses, e.g. freelancers vs in-house translators.
4. You can group categorized responses any number of times without any duplication in the number of respondents (where 2 or more insights in a grouping apply to the same person).

Total amount of data:

1. People: 60
2. Insight categories: 637!
3. Categories x People they applied to = 1467 (on average 2.3 people per insight)

Due to the large number of categories, I created even groupings of groupings of categorized insights in order to better understand the results.

Here’s a spreadsheet to follow the same process yourself: Survey Analysis Categorization Template

Results Summary

I rank the major response groupings within each area of study, displayed as charts below showing the number of respondents.

For grouping insights, I use my best judgment given the purpose of the grouping. Note that insights are not mutually exclusive! The same sentence from a participant in some cases falls under more than one item on the same graph. The purpose of these comparisons is to understand what type of responses were most common.

Reminder: The numbers and percentages are out of 60 total respondents.

Themes

Themes are common topics of conversation. Note that “identify the best term” includes any mention of translation dictionaries or term lookup.

Common themes among translators.

Descriptors, with the number of respondents (only the top or most informative responses are listed):

1. Understand context or topic around terms: Linguee (19, included as it provides contextual examples); examples online (15); understanding the topic (8); examples from competitors of clients or similar websites (8); examples from the same company (5).
2. Identify the best term: Monolingual dictionaries (30); specialized dictionaries (15); determine the best word (15); no easy equivalents (9); chemical or medical terms (8); term for a specific country (8); specialized terminology (7); slang, clichés (5); newer terms (3); legal (3); create new terms (transliterate) (3).
3. Specialization/field/locale: Specialized dictionaries (15); chemical or medical terms (8); term for a specific country (8); understanding the topic (8); specialized terminology (7); difficulty with a different specialization than usual (7); lexicon by locale/country (6); cultural knowledge or references (4).
4. Ask others: Ask friends or community, specialists (22); ProZ (14); verify difficult terms/parts of translation with the client (8); review from natives (e.g. friends) who aren’t translators (6).

Observations:

1. Translators talk about looking up terms or words with more consistency than anything else in this whole study.
2. Context and specialization were also very common topics (more than 70% of respondents).
3. A surprising number of people spoke about asking others for feedback or advice on translations, including natives who aren’t translators.
4. 2 people said they wished for more live interaction with other people, and 1 said that translation is a lonely process and isolating at times.

Problems

The problems listed include answers to what translators find to be either difficult or time-consuming.

What do you find difficult or time-consuming?

Descriptors:

1. Specialized terminology (21): It’s hard to translate terminology specific to the field (e.g. medical).
2. Understanding the topic (11): It’s time-consuming to understand the topic of the translation.
3. Finding the latest terms (10): Resources don’t have the latest terms.
4. Appropriate style (10): Hard to use an appropriate style for the translation (based on the translation’s purpose).
5. Client communication (9): Not enough time to ask the client (2); understanding the client’s instructions (1); persuading the client that the translation accomplishes the client’s purpose; the client doesn’t understand how long translation takes; must understand how the client perceives quality; clients will use the subjectivity of translation against you; the proofreader may be less qualified but cause you to have less pay due to mistakes.
6. Bad source text (7): Bad source, confusing source, the source itself is translated (6); handwritten source text (1).
7. Maintaining consistency (7): Apply accurate religious doctrine/business principles/legal principles of the organization or location (3); consistency is hard with multiple translators (1).
8. Researching (7): Any mention of researching as a problem (7).
9. Transcreation (6): Writing something new in the target that wasn’t directly in the source text (6); creative/marketing translation(1).

Observations:

1. Specialized terminology is mentioned about two times as often as the next most common problem (understanding the topic).
2. The third problem, finding the latest terms, is also related to terminology.
3. Translators had a large variety of insights about client communication and work relationship, yet almost all insights were mentioned by only one person (but likely apply to more translators).

Current Work Arounds

These are methods translators use to research a specific translation.

What do you use to inform your translations?

Descriptors:

1. Monolingual dictionaries (30)
2. Google search (31): Use the number of results on Google (7); use of quotation marks in Google (5), use another word for context in addition to quotation marks (4), search by region (1), search in the company’s website, Google thesaurus, Google ngrams.
3. Ask other people (30) (Asking people other than the client): Review from supervisor or proofreader (4); prefer to work with a senior translator/mentor (2); asking others helps when blocked or overwhelmed (1); get help from an engineer who knows the subject matter (1); research hard terms together as a team; prefer to work with a translator who is native to the other language.
4. Context site (25): Linguee (19); Google search for context (4); Reverso (3); Ludwig Guru (1).
5. Q&A site (18): ProZ (14); Quora (2); Viva (1); Yahoo; Reddit; Facebook groups; Interpals; Translator’s Café.
6. Ask client (11): Verify difficult terms/parts of translation with the client (8); ask the client if they want images, etc. translated (1).

Observations:

1. While 50% of translators use some sort of feedback from others, 4 respondents specifically described asking others as time-consuming. While feedback from others is highly desired, it’s also a difficult part of the process.
2. Translators do use frequency information (number of search results), but want to ensure that frequency is in the right context for their translation.
3. Translators appreciate mentorship by a senior translator.
4. It’s most effective for at least two translators to collaborate where both are a native of the two different languages involved in the translation.
5. 9 respondents spoke to using a physical dictionary. For those who are deep in the nuances of words, the Internet hasn’t completely replaced physical reference books!

Desires

These are desires for information to aid translation.

Is there any type of information you wish was available to help you translate?

Descriptors:

1. Terms by locale or field (13): Lexicon by locale/country (6); examples of the word in context, like Linguee but by field or language (5); highly specialized terms or knowledge, compiled thesauri for a specific field; frequency within the country or domain (2); explanatory definitions for terms in the source text (financial, technical concepts) (1).
2. Updated databases/dictionaries (7): Urban dictionary for other languages (new or slang terms) (4); updated translation memories (2); updated abbreviations (1).
3. Words in similar context (6): Examples of word in context, like Linguee but by field or language (5); idioms or proverbs that a word is part of (1).
4. Research history (personal or as a team) (2): Keep track of where I found a term or information about it, remind me what I did last time (1); share research history on terms between team members (1).

Observations:

1. The main theme among desires is not that a completely new tool needs to be created, but that the information in the currently existing tools isn’t comprehensive enough, up-to-date enough, or specific to the field or locale desired. Specialized dictionaries do exist, but not comprehensively or conveniently for everything.

Motivation

Why do you seek information about a concept, word or phrase?

Descriptors:

1. Sound natural: I want the most natural language for the given context and audience (not a root motivation, but shown separately due to frequency).
2. Client satisfaction: The client is satisfied; my reputation; avoid embarrassment (5); avoid complaints (5); get more work (3); be professional (2); the translation achieves its purpose (1).
3. Feel confident: I want to feel good (9) or confident (8) about it; I could be ignorant (4); my intuition (4); not familiar with the field (1); not familiar with the term (1); be able to justify my decision (1).
4. Seek perfection (This was a catch-all for any motivation related to the translation correctness, which includes all the results for “sound natural”): best translation possible (6); accurate translation (5); accuracy is important for the project (5); I could be ignorant (4); give an honest translation (2); the translation isn’t good enough; verify that others have used the phrase; I don’t know the equivalent term; help the author look good; important people will read it.

Observations:

1. A very high percentage of translators expressed an innate desire for the translation to be correct (“feel confident”, “seek perfection”), as opposed to client satisfaction. This doesn’t mean most translators don’t care about client satisfaction; the results reflect what they’re thinking in the moment of translating a term.

The below 2 questions provide key insights on translation quality.

How do you know that translation information is trustworthy?

Total respondents: 12

1. Official source (e.g. academic, government, Google Books as opposed to a Google web search): 5 respondents
2. Written by a native or checked by a native: 4 respondents

How do you know that your translation is done?

Total respondents: 38

1. The client is satisfied: 12 respondents
2. Someone else has reviewed it: 10 respondents
3. Self-review (no more errors are identified): 9 respondents
4. Seen a corresponding image in both languages: 9 respondents
5. My own intuition: 8 respondents

The common use of images (point 4) was the most surprising insight to me in this whole study! But it makes sense when you consider the fact that specialized terminology is the #1 translation problem.

Translation Tools

What translation tools do you use?

Obversations:

1. SDL Trados has a large market share, and as it’s considered a standard, some jobs require it. Some translators have the expensive SDL Trados for those jobs that require it and then use another tool for other jobs.
2. I was surprised that 9 respondents use Microsoft Word, since it’s not a translation-specific tool so offers no translation memory, machine translation, etc. My observation is that there are two reasons: 1. Microsoft Word is required in some cases to ensure proper formatting, and 2. some prefer Microsoft Word as a familiar and simpler tool.

Machine Translation

Machine translation is a sensitive subject among translators. I noticed at a translation conference that some translators are offended when required to use machine translation in their job and won’t accept such a job. What do the survey respondents have to say?

Not all respondents were asked about machine translation (MT), but the following insights emerged:

1. Use MT and post-edit: 14 respondents
2. MT is sometimes wrong: 6 respondents
3. Don’t use MT: 3 respondents
4. MT saves time (e.g. from typing): 2 respondents
5. MT reminds you of what you know: 2 respondents
6. Use custom MT, not generic MT (e.g. Google Translate): 2 respondents
7. MT doesn’t have a good future: 1 respondent
8. Some companies don’t use MT for data privacy: 1 respondent
9. I wish I could train my own MT: 1 respondent

Overall, post-editing is common practice, but some will not use machine translation at all.

Secondly, customization of MT is a desire but not very accessible to freelance translators unless their client or translation services company provides this.

Note that the respondents weren’t specifically asked these insights, so further study may find they apply to more people.

Translation Process Insights

1. 15 respondents said they will identify difficult terminology and then focus their time on those.
2. 14 respondents said they read the document completely first, and 5 that they learn about the topic first.
3. 7 respondents said they put the document in the translation tool as the first step.
4. 4 said they make sure to come back to the translation after taking a break.
5. Don’t just use the first search result (4).
6. Ask the client early about purpose, timeline, etc (2).
7. Use a tool/site that looks in multiple databases at once (2).

Miscellaneous Insights

There’s more data than what I’m including in this article, but here are a few interesting points:

1. 50% (30) of participants mentioned using a monolingual dictionary (term definitions).
2. 1 person mentioned going to the store or a physical location to learn about the concept being translated.
3. 1 person mentioned using Bible translations as a reference since they’re well researched.

Research Notes — A Business Perspective

The following were my hypotheses before carrying out the interviews.

Note: This translation technology business idea was paused, but if you’re interested in the idea or the research, reach out at https://discord.gg/AxVfGRB.

Problem hypothesis:

Current translation tools and sources of information are

1. Difficult to use ✅
2. Waste translator’s time ✅
3. Not accessible (available on the web, cheap) ✅ (Not comprehensive enough)

Value hypothesis (how the problem applies to business):

1. What task — Text translation ✅
2. What context — A translation interface or workspace ✅
3. What is needed or offered — Information relevant to the translation ✅
4. What’s improved — Speed, Awareness, Applicability of information to the context of the translation ✅
5. Value add — Faster translations, Higher-quality translations ✅

Faster translations:

1. Freelance translators: Better pay due to more translations. ✅
2. Businesses: Lower cost of translations due to less work hours required. (Not verified)
3. Businesses: Faster roll-out of new texts or new languages, leading to more growth. (Not verified)

Higher-quality translations:

1. Freelance translators: Improved professional reputation. ✅ (11 respondents)
2. Businesses: Higher customer satisfaction, or the value depends on the use of the translation. (Not verified)

Note: I did sum up responses that indirectly matched 4 particular solutions I wanted to validate, but most of this analysis duplicates the Desires section.

A more thorough analysis on the hypotheses would show the survey responses or number of people validating each hypothesis. I do recommend this approach for any business idea!

Conclusion

First off all, translators are incredible people. Thank you for speaking with me and sharing your story!

In terms of the translation process itself, reliably and confidently identifying the right translation for specialized terminology stood out among all problem-related insights.

Finally, to be an effective translator, the most important skills are terminology research and communication with clients.

What additional questions do you have about the data or the research process of translation? Ideas for future studies? Add your comments below.