Cross-linguistic adaptations of the Comprehensive Aphasia Test (CAT)
As a first step in meeting the needs for comparable assessment tools across the languages of Europe and thus paving the way for more robust research on aphasia and aphasia rehabilitation, this project aims to adapt the Comprehensive Aphasia Test (CAT) (Swinburn, Porter, and Howard, 2005) into as many European languages as possible. Once completed, the various language versions of the CAT will facilitate a basic assessment of monolingual speakers from different language backgrounds as well as the assessment of multilingual speakers in at least one of their languages.
The languages currently included in the project are: Basque, Catalan, Croatian, Finnish, French, Greek, Hungarian, Lithuanian, Norwegian, Portuguese, Serbian, Spanish, Swedish, and Turkish. These fourteen languages belong to a variety of language families.
For each language, local, often interdisciplinary, working groups have been established, and the adaptation process is informed by guidelines that have been jointly established and agreed upon in WG2. Each language group follows a committee approach in the adaptation. Following the adaptation, each language version of the CAT will be standardized and normed based on a common set of instructions.
Central to the linguistic part of the CAT is the variety of underlying variables reflected in the test items, including phonological, semantic (e.g. imageability), syntactic, and usage-based (e.g. frequency) variables. A key feature of the adaptation process is to ensure comparability across the various languages on the level of these underlying variables. Several of the languages involved lack the necessary background data on variables such as frequency and imageability. Hence, as part of the adaptation process these data also have to be established.
So far, we have divided the work as follows: Manual (5 chapters, Appendix and Scoreforms) and Stimulus book (27 tasks). Each one translates one subpart and sends it to the other in order to check the vocabulary used in standard Basque. Currently, only the Introduction of the Manual is in progress. Much remains to be done but should go faster in the coming months.
The Basque language or Euskara is spoken in both Southern France and Northern Spain, which represent respectively the Northern and Southern parts of the Basque country. Several dialects are spoken in the Basque country and from the late 1960s, the Basque Language Academy has developed a standard Basque to be commonly used in the media, administration and education. In the CAT Basque version, we are using the standard Basque so that it can be used in all the Basque country. However, the terminology related to clinical linguistics does not have any tradition in Basque. Therefore, this takes time and necessarily involves consultation between us to agree on some terms.
To adapt the CAT into Catalan, we borrowed the structure and test items from the original version that were relevant to assess Catalan-speakers of aphasia, and constructed new items when necessary. We came up with nonwords that ascribe to our morphophonological rules, considered syntactic structures that our speakers have difficulties with, relied on psycholinguistic databases to obtain frequency ratings, and administered a questionnaire to obtain imageability ratings from native Catalan speakers
The Croatian team secured the funding for new pictures from two projects (Adult language processing, Croatian Science Foundation HRZZ-2421 and Interdisciplinary approach to language model of dyslexia in adult, HR.3.2.01-0247 EU-ESF). 330 new pictures were drawn and downloaded to our shared Dropbox file. All languages are welcome to use them.
Next phase is pre-standardisation, i.e. piloting of the test on sample of health speakers with aim to test validity of items and pictures. First data of this phase will be present on out next Meeting in Tampere in February 2016.
The greatest challenges in the adaptation of CAT test are mostly related to tasks including words that had to be presented by the use of pictures. The target words had to have a certain level of imageability and frequency and therefore the tasks were not always simple to match with the original version of the test. For example, spoon and pineapple are marked as having low frequency and low imageability in English, which is not the case in Croatian. That is why we could not include those words in the same part of the task. Moreover, it is logical that most of the low imageable words are not that simple to draw. Consequently, it was a great challenge to do so, especially for the Naming objects task. Some examples of low imageable words in Croatian language are time, wind and galaxy.
Another challenge includes the Repetition of sentences task. For some items it was somewhat difficult to find sentences involving same number of content and functional words and still to preserve the exact meaning of the original sentences.
One more thing that has to be mentioned involves all the tasks in which the main imperative had to be put on word structure (e.g. phonological properties of the words; number of syllables; number of distinctive features etc.). As English and Croatian language differ significantly, in most cases the words in the adaptation of the test are different from the ones in the original form.
Considering all the above, it is reasonable that the English and the Croatian version differ significantly in certain tasks, but it is important to mention that all the given criteria have been met and fulfilled. Accordingly, it would be finally possible to carry out cross-linguistic studies analysing accomplishments of patients with various type of aphasia in different language
At the moment we are gathering words for the imageability testing of the Finnish version of the CAT;we will then start the imageability testing soon. We are also checking through the existing CAT pictures “with Finnish eyes” to make sure they have the same relevance abd applicability. Then we begin to test the imageability.
We have adapted the “Language Part” of the CAT. We carried out this adaptation taking into account phonological and lexical properties of items in order to follow the same principles that underlie the design of the English version of the CAT. In particular, we controlled for word frequency, word and nonword length (in terms of syllables or phonemes depending on the task), word imageability and orthographic transparency. Regarding sentences, we adapted them in order to respect as much as possible the syntactic structure of the English version. We are currently doing the adaptation of the “Cognitive Part” and the translation of the instructions
No More Details
We have finished the translation and adaptation of the language part of the CAT. We managed to make our subtests very similar to the original version. This was very challenging for several reasons. In the subtests focusing on words, in line with the original test, we had to control for a number of factors, such as frequency, animacy, imageability, and length. Although we have a web-based frequency database that we could use for item selection, we had to find a way to transform the Hungarian values into the unit ‘word per million’ to be able to compare our potential items to the English ones. Also, we have a limited number of nouns with one syllable, therefore finding words and minimal word-pairs with this length, that also meet all other criteria, took a lot of time and effort. Because there are substantial differences in morphology between English and Hungarian, we had to face some challenges also in the subtests aiming to examine the effects of morphological complexity and regularity. There is no imageability database in Hungarian, so we are preparing an imageability study to collect ratings for our potential items, after which we can finalize our wordlists. We have started to work on the adaptation of the cognitive screening as well.
No More Details
For the adaptation of the CAT into Norwegian, we have prepared first versions of all subtests, with suggetsions for possible items. They will be further decided upon through a two-step-process: 1) through a complemtary imageability study of the words (to fill in those words that are not already included in our imageability database) and 2) through a naming agreement study. When all items are in place, the norming can be done.
Challenges: Norwegian is relatively similar to English, so that the adaption is less complicated than for more typologically different languages. However, finding enough words with high imageability and high frequency that are also long has been challenging for Norwegian. Another possible challenge is to present the linguistic background of the test in a way that is understandable and relevant for clinicians.
No More Details
Serbian adaptation of CAT has followed the general adaptation guidelines, and as a result, we have managed to create a translation of the CAT that is comparable to CAT in other European languages. Besides the usual challenges of crafting and adapting a complex multidimensional test, Serbian adaptation faced some language-specific challenges. Namely since Serbian has two letters, Cyrillic and Latin, which are used interchangeably, the issue of incorporating both letters in the test has arisen. Is order to overcome this issue, Serbian CAT will have additional items, some tasks will include both Cyrillic and Latin items, and some test will be administered with letters that the person feels more comfortable with.
In these preliminary stages, we have adapted the semantic fields according to our culture. We have also controlled for aspects such as word length and word frequency according to pan-Spanish measurements. Additionally, our stimuli list has been controlled for minor differences that exist in Spanish from Argentina, Uruguay, Colombia, Chile and Spain. We have ready our final version, which will be translated back to English to check percentage differences with the pan-Spanish adaptation. We are currently defining the final list of drawings that will be drawn here in Chile.
No More Details
Intial translation of the CAT into Turkish has been completed. The test words have been arranged in terms of imagebility, frequency, length and according to category. Alternatives have also been identified. Almost 200 words have been translated and are now being validated for mental imagery, familiarity and age of acquisition. After this phase, the test words will be identified.