The problem
I was recently talking with some colleagues about the following question about structural ambiguity from an assignment in an introductory syntax course:
There is more than one way to form the word “unrecognizable” with the three morphemes un-, recognize, -able. True or false?
The intended answer was false: the only way to form “unrecognizable” is by first combining “recognize” and “-able” to make “recognizable”, then attaching “un-”. However, a student who had answered true wrote to the instructor to complain, insisting that “unrecognizable” was structurally ambiguous. This argument was based on the purported acceptability of the word “unrecognize” in the context of someone suffering from memory problems and no longer being able to recognize someone (e.g., “After developing Alzheimer’s I unrecognized my own child.”)
As an L1 English speaker, to me “unrecognize” is clearly not a valid construction except perhaps in a political context (e.g., “The Trump administration unrecognized the Maduro administration”), and I would be hard-pressed to find a context where I would ever then say “The Maduro administration was unrecognizable”. If anything, it sounds like language play. But the student’s argument for their answer clearly indicated that they understood the concept of structural ambiguity: if “unrecognize” is a valid construction, we should be able to get “unrecognizable” from it by adding “-able”. If we accept that they thought “unrecognize” was possible, their only error was in their knowledge of English vocabulary. The point of the question was structural ambiguity, not English vocabulary. So should they lose the mark? I’ve run into this dilemma many times as a teaching assistant in syntax courses. It can be very tricky to write and grade syntax problems which depend on English lexical or syntactic knowledge, especially when many of our students are not L1 English speakers.
My position on these issues as a teaching assistant and grader has generally been that we should be generous with our students and grade them as if their judgments about English are correct. But that can be problematic in cases like the example above, which is supposed to be a 1-point true-or-false question, with no space for students to justify their answer. Unless they tell us after the fact, we can’t know if their answer was wrong because they didn’t understand structural ambiguity or because of their English vocabulary — giving everyone the benefit of the doubt would mean giving everyone full marks, no matter how they answered. The question therefore does not cleanly distinguish between students who understand structural ambiguity and those who do not.
Solutions
One solution is to modify the question. If a question depends on some knowledge of English vocabulary, can it be rewritten so that it doesn’t? This may require changing the type of the question from a simple true/false to a more complex short-answer one:
Explain how we can determine if a word is structurally ambiguous.
Or we could ask it as multiple-choice:
A word is structurally ambiguous if:
- The word can be formed by combining its constituent morphemes in multiple different ways.
- The word has multiple different meanings.
- …
Or can we provide students with the necessary vocabulary knowledge in the question itself?
There is more than one way to form the word “unrecognizable” with the three morphemes un-, recognize, -able. True or false? Assume that “unrecognize” is not a valid construction in English, but “recognizable” is.
If providing this information makes the question feel too easy, that should inspire some reflection about what your question is really testing!
Another is to make the instructor and/or TA(s) available to provide judgments for such problems, and state this explicitly in the question (or in the assessment instructions) so students know what to do.
There is more than one way to form the word “unrecognizable” with the three morphemes un-, recognize, -able. True or false? You can ask your TA for a judgment if you aren’t sure if a construction is valid in English.
All of these options even out the playing field for L1 and L2 speakers by removing or deemphasizing the required English knowledge. These all make the question better, but I think a better way to solve this problem is to just avoid asking questions about the language of instruction at all. This stops us from relying on the (often false) assumption that our students are comfortable enough in English to make acceptability judgments about it and forces us to write questions that deal exclusively with linguistics knowledge as opposed to language knowledge.