Guidelines for Developing Translations
Translation and Standardization Project Director
The initial step in the development of translations of the Press's instruments is the identification of a Translation and Standardization Project Director, a researcher who will supervise the development of the translation and ultimately declare it ready for review, and will supervise the collection of normative and clinical data. This person should be from the country or culture for which the translation is being developed and must be a credible authority in the area of personality assessment, diagnosis, and psychopathology, preferably a bilingual academic psychologist. He or she must be fluent in the target language and must be at least reasonably conversant with English. He or she should also be knowledgeable about the current English language assessment literature and about issues and methods relating to translations of psychological tests.
Development of the Translation
The translation should be developed by a team of translators (no fewer than two) fluent in the target language and in English, and knowledgeable about the English-language assessment literature and about issues and methods relating to the translation of psychological tests. The translators will independently translate the items and then compare the results, negotiating differences in the translation of items to obtain the most equivalent item.
Development of a Back-Translation
The translated items should then be back-translated into English by someone other than the translators to determine whether they are equivalent in meaning to the English original. Any substantive differences between an original and back-translated item should be considered by the translation team and revised as appropriate.
Evaluation by the Language Service Employed by the University of Minnesota
The language service reviews the translation to determine whether each translated item is equivalent in meaning to the English-language original. In assessing equivalence, attention is paid to vocabulary, idiom, syntax, and tone. The review will also note any errors in spelling, punctuation, and grammar. The University sends the service's report to the translation team, with the requirement that they respond to every suggestion in the report, either accepting the suggested change or providing a reason for not doing so. After the suggestions in the language service's report have been incorporated into the translation, it should be submitted to the project director for final approval.
Conducting a Bilingual Test-Retest Study
To evaluate the adequacy of the translation, the original English-language version and the translation should be administered to a minimum of 35 bilinguals who are fluent in both languages and familiar with both cultures. Each subject must take the test in both languages in two separate sessions, counterbalanced so that roughly half take the English-language version first and the remaining subjects take the translated version first. Their responses should then be used to examine scale score differences and discrepancies at the item level. The baseline for evaluating differences should be the normative test-retest correlations reported in the U.S. test manual and supporting documents.
Collection of Normative and Clinical Data
At a minimum, a normative sample of 350 men and 350 women demographically representative of the country of the target language should be collected, as well as a demographically representative clinical sample of 100 men and 100 women. The US norms are to be used for comparative research purposes only. They are not to be included in published materials on translations.
Scales That May Require Adaptation
It is often the case that in developing translations of the MMPI instruments the validity indicators require adaptation from their original form in the English language versions developed for use in the United States.
As a result of the translation, modifications in the choice and/or scoring of items may be indicated in some cases. In particular, the translated item pairs that comprise inconsistency scales VRIN and TRIN need to be carefully checked to ensure that each translated item remains suitable for its intended use and scoring method as a VRIN or TRIN item pair (Chapter 2 of the MMPI-2-RF Technical Manual describes in detail the criteria for evaluating VRIN and TRIN item pairs and the corresponding steps for selecting suitable pairs).
The same criteria and procedures apply to the MMPI-2 and MMPI-A inconsistency scales. The software used to select the MMPI-2-RF VRIN-r and TRIN-r item pairs can be made available to the translation team for implementation of this procedure. Alternatively, the normative and clinical data can be sent to the Press for this purpose.
Each item pair is examined to determine that it retains the properties necessary to function adequately as an inconsistent response indicator.
Pairs that do not function properly are deleted and, if possible, replaced by alternative pairs with proper characteristics.
Once the vetted and approved inconsistency scales are available, the distribution of scale scores for subjects comprising the normative and clinical samples are examined and outliers removed from the samples.
When outliers have been removed, item endorsement frequencies of the infrequency scale items (e.g., F, Fp) are examined and items that do not meet the requirements used to assign them to an infrequency scale are deleted and, if possible, replaced with appropriate/properly functioning items.
The distribution of scores on the infrequency scales for subjects in the normative and clinical samples are then examined and outliers are removed from the samples.
Developing Standard Scores
Two types of T scores are used with the MMPI instruments- linear and uniform. The same type of T score will be developed for the scales of translated instruments. Development of Uniform T scores requires use of a software package provided by the Press.
A final report on Data Collection and Standardization of the Translation is to be submitted to the Press before approval for publication is granted. This report will include a description of how the steps just outlined were implemented, descriptive statistics, reliability estimates, standard errors of measurement, and validity findings, as well as scoring keys and T-score lookup tables for all scales scored on the U.S. version of the instrument.
University of Minnesota Press, Test Division
Permissions and Translations Coordinator
111 3rd Ave S Ste 290
Minneapolis, MN 55401-2520