Have a Question? Contact the Humanities Office or an Academic Unit

November 1, 2017

Title: Decomposing the Frequency by Skill Interaction

Speaker: Sascha Schroeder , Max Planck Institute for Human Development, Germany

Time: 11:30am-1:00pm

Location: CNH -103

In this talk, we discuss two alternative approaches to explain the commonly observed frequency by skill interaction in visual word recognition, which refers to the fact that frequency effects are typically stronger in less-skilled readers than in more-skilled/ individuals. The first approach assumes that, due to fact that low-frequency words are underrepresented in smaller language samples, low-frequency words are encountered disproportionally less often by low-skill individuals than by low-skill individuals. The second approach also assumes that low-frequency words are less familiar for low-skill than for high-skill individuals, but not disproportionally so. Instead, the relationship between the previous encounters with a word – which we refer to as “individual frequency” – and response accuracy/latency in visual word recognition tasks is non-linear.

In order to evaluate these two approaches, we analyzed lexical decision data for high- and low-frequency words in three different age groups (4th grade, 6th grade, and young adults). Using simulations based on corpus data from different age groups and different languages, we first show that the exposure to both low- and high-frequency words increases linearly with increasing reading exposure/age, thus ruling out the first approach to explain the frequency by sill interaction. Next, we introduce a new method to estimate individual frequencies, which is based on the previous reading exposure of an individual. Using this measure, we are able to estimate the (individual) frequency effects in different age groups. Combing the frequency trajectories of different age groups, we show that there is indeed a single underlying function relating (individual) frequency and lexical decision accuracy/latency, which is non-linear. This is in line with the second approach to explain the frequency by skill interaction. We discuss different potential forms for the individual frequency function and their theoretical implications for theories of visual word recognition.