Before, I wrote about different purposes for test development. Given those different test functions an implication is that the way we then develop tests for these should be different.
For language proficiency, we need to think about how well someone knows a language. I think here, we need to focus on tests that test mastery of a given domain or domains. So, the comparison group should be highly proficient native speakers of a language. Domains can include speaking, listening, reading, and writing. For me, it’s not so much about accent. We all have accents. Some of us have regional accents; others of us have foreign accents. But, accents in and of themselves usually don’t interfere with comprehension. Rather, proficiency for me is about how well the person can understand what they read, how well they have mastered the grammatical forms and so on. They need to be able to communicate well in different situations. The target domain should somehow represent the situation for which the person will know the language. For example, if the target context is elementary school, then the child needs to know the language well enough to meet the social and academic demands of that grade.
Language dominance is going to be different. If this is relative proficiency, then this implies two tests that are compared. So, the two tests have to be of equal difficulty and target the same areas. So, if we’re testing reading, then there needs to be an equivalent reading test in the other language. The challenge is that similar items may have different levels of difficulty (I talk about this in Peña, 2007). Also items that have the same level of difficulty aren’t necessarily the same. So, here I think that the difficulty has to be the same and that the two need to be compared to each other. But, I think that the items may end up being different in the two languages or be in a different order depending on what’s important for that language. This is why I prefer to look at time in each language and then to relate that to performance in each language.
In terms of testing language ability, the kinds of difficulties that people with language impairment have depend on the language they are speaking. A test of ability isn’t going to sample everything in a language. This kind of test is going to sample items that are typically those that are difficult for persons with language impairment. For English this will likely focus on verb morphology such as past tense –ed, or progressive -ing. For Italian on the other hand, items might focus on articles and clitics. This means that ability tests of language are going to be even more different from those for dominance or proficiency. This doesn’t mean that these are the ONLY problems that children with language impairment may have—just that these are the kinds of items that most reliably differentiate impairments from typical development.
So, here I think it’s clear that the purpose of a test drives what kinds of items it will contain. These differences further highlight the reasons why these tests cannot be used interchangeably.