The Improve Group has had enlightening experiences working across projects requiring multiple languages and leading us to use all manner of data collection – from one-on-one interviews, to focus groups, to our own Image Grouping participatory techniques. We’ve worked in a variety of contexts and regions, with a wide variety of participant groups. (See accompanying article by Sara Stalland McGarraugh on Working Across Multiple Languages)

These are a few of the lessons we have learned over the years:

1.    Match your outreach and data collection techniques to your population

Be sure that the outreach and data collection methods you use reach the population you are targeting. This might mean including multiple languages on one tool (such as an invitation letter or a poster), or giving people options to respond in different languages. You could create a phone response system for people in multiple languages, or send a bi-lingual survey administrator into a community to gather responses. Because some languages may be widely spoken but not widely written or read, people may be more literate in English, but more comfortable speaking a different language—it’s important to consider whether your survey is more likely to be read or heard.

2.    Be prepared to answer questions in other languages at any time

There’s nothing worse than sending off parental consent letters in a certain language  and then getting a call in that language with no ability to respond. If your staff people are not multi-lingual, there are language services that allow you to dial in a third caller who can act as a translator. This can work best if you’ve learned beforehand how to ask callers to please wait in the respective languages. Enter the language line number into your contacts or speed dial, so that you’ll be ready if a call comes in.

3.    Create a single point of entry

For maximum accessibility, we always have a single link for a survey. The landing page has links in each respective language that say “click here to take the survey in [your language].” Each link leads to the corresponding version of the survey. This is particularly helpful when a respondent may be exposed to marketing in a different language than they primarily speak. For example with a five country survey we did with Global Citizen Corps, participants received invites in their native language, but the Facebook page for the program also advertised the survey to everyone who liked their page specifying which countries we were seeking participation from. This also allows multi-lingual respondents to select which language they feel most comfortable responding in. Some sites use flags or other country symbols to denote language, although there are some problems with that practice.

4.    Meaning is more important than literal translation in instruments

Consistency in meaning across language and culture when designing instruments can be a challenge. Direct translation of the specific words can result in losing some nuance. Surveying is particularly difficult when you are asking about things that did or will happen: languages deal with past and future tense differently. For example, in several languages it is nonsensical to say that your phone died, the logic being that an inanimate object that never had life cannot die. Using native speakers as translators can help you avoid losing nuance and meaning. Ideally, they will also steer you away from making blunders like asking about taboo subjects.

5.    Careful consideration of meaning in analysis is important as well

Cross-verification of data is both highly valuable and time-intensive. Having two people interpret data and compare findings is a worthwhile practice as it strengthens the validity of conclusions drawn and sheds light on differences in interpretation. While this is true of most scenarios in which data is being analyzed, it is particularly important when working across multiple languages, as the act of translating adds an additional layer of personal bias to the process of analyzing of qualitative data. One instance of when we relied on cross-verification was while analyzing data from interviews with doctors and patients in Spanish for an evaluation for Physicians for Peace. Both native speakers and our staff who served as translators reviewed findings to affirm results and deepen understanding.

(For more on this topic, see Danielle’s previous article: Considering Cultural Context in Evaluation)