Controlled Judgement Experiments


Language is constantly changing. Over
time, new meanings can emerge for words and phrases and older meanings sometimes
drop out of use. If you want to study how the meaning and syntax of particular words
are changing, then you need a way to collect data that allows you to track
the trajectory of their meanings over time. A commonly used source of data is a corpus, where others have already collected and organized a large amount
of relevant data, and you can make use of them easily. But there are occasions when a corpus is not the ideal source of data, even when studying language change. For example, if someone said this sentence here, in a corpus, Jo could
actually have said it out loud, or it could have been just the way the speaker
reported how Jo felt at the time. Either is plausible, which makes it really
difficult for the researcher to decide one way or another. So even though there
can be many examples like this in a corpus, a lot of them may be too
ambiguous. If a corpus cannot help us then how can we tell if something has a
certain meaning? Well, we can simply go to other people and ask what they think. The
idea is if something is used in a sentence, in a way this meaning doesn’t
allow, people would judge the sentence as ungrammatical. With experiments like
these you can, by design, control what materials to present to people for
judgment. By giving them sentences that ensure or encourage a certain
interpretation, you can minimize their ambiguity. Traditionally, grammaticality
judgment is conducted in a binary manner. Is a sentence grammatical or is it not?
Sometimes however, things are not so clear-cut. A speaker could feel that a
sentence falls in between. This is usually the case with non-standard forms,
which are less often a case of grammatical or ungrammatical, but a case of more acceptable or less. This is where controlled judgment experiments come in. In these experiments binary judgment is replaced by a scaled acceptability.
Speakers can simply score on a scale how good they think it sounds, and so with
marginal cases, in emerging forms, this method allows for more nuance in the way the researcher assesses their acceptability.

Leave a Reply

Your email address will not be published. Required fields are marked *