CAS-Gastvortrag

Prof. Phillips spricht auf Einladung der aktuellen CAS-Kohorte zum Thema „Specific and general mechanisms in real-time linguistic computation“. Sein Vortrag steht im Kontext des Projektes der aktuellen Kohorte mit dem Titel: „Knowledge acquisition, representation and application in human minds and machines“.

Abstract: (Psycho)linguists have known about individual examples of linguistic fragility and mis-perception for a long time, at least into the 19th century. Around 20 years ago I became interested in what we could learn by grouping them together under the broad heading of “linguistic illusions”. My students and I were impressed by the apparent fact that speakers seem not to be fragile across the board. Rather, there are specific kinds of mistakes that they make easily, and others that they make very rarely. Exploring this profile of “selective fallibility” proved to be even more fruitful than we suspected. Over the intervening years, we discovered more and more cases of fine-grained selectivity in where speakers do and do not mess up. This has been fun, as it has led to many new and unexpected discoveries. But it has also become increasingly worrying, as the attempt to come up with a unified account of linguistic illusions has yielded roughly the opposite: distinct detailed generalizations for each different phenomenon that we have looked at. A unified account has seemed ever more remote. In this talk I will survey some of this variability, especially in relation to negative polarity items (NPIs), agreement, and argument roles, three of the most thoroughly studied areas. I will offer some pointers towards reconnecting our understanding of the different phenomena.