When do people know about the grammatical gender and sounds of words they are preparing to say?

What is it about?

Part of our ability to speak requires that we retrieve information from long-term memory about the words we are preparing to utter. This includes information about a word's semantics (its meaning), its grammatical properties such as a noun's grammatical gender, and the sounds that make up the word (its phonology). An open question in research on speech production is when these different types of information about words are retrieved in time, and whether they are retrieved in a serial manner (i.e. first one, then the next, then the next) or whether retrieval happens in a parallel fashion (i.e. multiple information sources being retrieved at the same time). In this paper, we show that the grammatical gender of a noun is not always retrieved prior to information about a noun's phonology (i.e. its sounds), which contrasts with what prior work had shown. We also show that the relative order with which information about a word is retrieved depends on the task people are performing. These findings are not consistent with speech production models that assume a serial order of information retrieval. We therefore conclude that our data support parallel models of speech production.

Featured Image

Why is it important?

It's easy to take speech for granted given the relative ease with which we are able to speak. However, the ability to speak can and does break down due, for example, to traumatic brain injuries, strokes etc. In order to effectively treat these break-downs and help patients recover, it is necessary to understand both how speech breaks down and how speech operates under normal conditions. Our work helps to refine our understanding of how speech works under normal conditions, and can thus help to inform our understanding of how speech can break down, and how speech disorders can, in turn, be treated.

Perspectives