Other things

Here are some projects that did not fit under the other headings.

Ongoing projects

How many observations do we need for stable RT estimates in reading aloud studies?

You can read a preprint here.

 

In single-word reading aloud studies, the researcher normally analyses the Reaction Time (RT): the time between target presentation and the onset of the participant's response. RT measures are inherently noisy. For noisy measures, a large number of observations are sometimes required to obtain reliable estimates. Reducing the noisiness of a measure by including many observations leads to more precise estimates of an effect, and to increased power, as it diminishes unsystematic variance and thus leads to higher standardised effect sizes. Here, a set of simulations aims to establish how the variability of the observed mean RT for a single item changes as a function of the number of participants who respond to this item. A rule-of-thumb cannot be established, because the researcher needs to decide on a range of expected values which is acceptable for their design and research question.

Can we measure individual differences in lexical processing?

Developing a sound and well-functioning mental orthographic lexicon is important for fluent reading. Some children struggle to shift from reliance on a letter-by-letter decoding strategy to fluent, automatic activation of a word's pronunciation and meaning, which characterises reading in adults. This raises the questions: Can we measure the efficiency of a given person's lexical processing? What tasks and statistical methods would be most suitable for this?

 

In an ongoing project, we collected data from >80 Italian native speakers on 4 tasks and 9 psycholinguistic effects that are used interchangeably in the literature as markers of lexical processing. We aimed to determine whether the size of the effects correlates across individuals. As it turns out, it doesn't; now I'm trying to figure out why. I presented some preliminary analyses at Macquarie University in October 2017, slides here [starting from Slide 15].

Projects on hold

Grapheme saliency

In English, multi-letter graphemes (i.e., graphemes consisting of more than one letter, e.g., th, a_e, oo) are required to represent the phonology. In the Dual Route Cascaded Model of Single-Word Reading Aloud (DRC), all two-letter graphemes are processed in the same manner. This means that the cognitive system is assumed to take the same time and effort to process, for example, aw --> /oː/, oo --> /ʉː/, or ar --> /ɐː/ (Australian English).

 

In a series of experiments using a letter detection task, we explored whether there are any differences at an early, visual-orthographic stage of the visual word recognition process. Basically, we found that some graphemes appear to be more salient than others, but we couldn't figure out why. Preprint, data and analysis script here. WARNING: I CANNOT REPRODUCE ALL OF THE ANALYSES REPORTED IN THE FIRST EXPERIMENT DESCRIBED IN THE PREPRINT.

Crowding, letter-position and letter-identity coding: How are they related?

There is a series of studies showing that increasing letter spacing improves reading ability. It is unclear why this is the case. With Teresa Schubert, Sachiko Kinoshita and Dennis Norris, we wanted to see if increasing letter spacing improves letter-position and/or letter-identity coding, using a masked priming same-different decision task. The answer is: probably not, but we need follow-up experiments to draw stronger conclusions. We haven't written it up, but you can find the preregistration report, materials, data and analysis script here