Search

Code-switching is the practice of mixing languages within or across sentences. Current code-switching literature has largely focused on bilingual speakers of alphabetical languages, evident in influential theories of code-switching (Poplack, 1980; Myers-Scotton, 1993b). Research emphasises experience of code-switching as a major predictor of processing costs, where habitual code-switchers do not seem to exhibit the same lexical-semantic processing costs as those who are not frequent code-switchers (Gosselin & Sabourin, 2021). Cantonese-English speakers provide an interesting population where there are much larger between-language linguistic differences. Additionally, code-switching is used commonly within this population in everyday conversation and informal written conversations (e.g., Whatsapp/social media). Further evidence is needed to understand the nature of Cantonese-English code-switching and its impact on language processing and learning. Psycholinguistic studies reporting processing costs in terms of response latencies have often used cued language-switching tasks where code-switching is forced. Meuter & Allport (1999) used a cued numeral naming task and found that response latencies were slower on switch trials than non-switch trials, and that switch costs were larger from the dominant L1 to weaker L2. Costa & Santesteban (2004) conducted a picture naming language-switching task and similarly identified asymmetrical switch costs among those with dominant L1 and weaker L2. However, they also demonstrated that highly proficient bilinguals did not show asymmetrical switching costs when tested in both of their dominant languages, and even when tested in their weaker L3. The issue with these production studies is that they are difficult to reflect natural use of language, as natural code-switching is usually not cued and delays in reaction times may simply be due to the item/numeral being more commonly used in one language or the other, i.e. its habituality. In contrast, Hui, Fong and Wang (2022) looked at Cantonese to English code-switched reading where participants read Cantonese sentences containing either an English or Cantonese target word. Using eye-tracking, they found longer reading times in non-habitual switches (i.e. using an English target word when its Cantonese equivalent is more commonly used) than in habitual switches (using an English target word that is more commonly used than its Cantonese equivalent). This raises the question of whether previously reported processing costs may be confounded by whether code-switched words are commonly or uncommonly used in each language. However, Hui et al. (2022) only investigated one switch direction (Cantonese to English) and it is therefore difficult to pull apart effects of switching from effects of target orthography. Liversedge et al. (2016) looked at sentence and word reading times in Chinese and English and results indicated that fixations on English words are significantly faster than Chinese words. The current study will therefore examine bidirectional switch costs using eye-tracking during sentence reading, with materials based on Hui et al, selecting words and sentences that are appropriate to use in either Cantonese or English.

To maintain optimal comprehension, readers and listeners must use the available sentence context to support the processing of lexically ambiguous words and access appropriate word meanings. The process of accessing appropriate word meanings is not trivial as 80% of common English words have multiple dictionary definitions. While this meaning selection process is more difficult for low-frequency word meanings, recent evidence from the word-meaning priming paradigm indicates that this difficulty can be reduced via recent encounters with ambiguous words. For instance, Rodd et al. (2016; Experiment 2) asked participants to complete a word association task across four priming conditions, where ambiguous words were unprimed or primed 1-, 20- or 40-minutes prior. At each time point, there was an increase in prime consistent responses, where rates of response consistent with the prime increased from 20% to 35%, 28%, and 28% respectively, indicating that, relative to an unprimed condition, recent encounters made a particular word-meaning more readily available even after 40-minutes. Subsequent experiments have shown that priming effects are still evident even after 24 hours (Gaskell et al., 2019). Word-meaning priming effects appear robust. There is evidence to suggest that they are not modality-specific (Gilbert et al., 2018) and can be observed when priming occurs through more naturalistic texts (Blott et al., 2022). However, the paradigm is not without limitations. It is limited by the fact that it probes word-meaning access via meta-linguistic tasks, such as word association (e.g., Rodd et al., 2013) or semantic relatedness (Gilbert et al., 2021), which are far removed from natural language processing (e.g., when reading for comprehension). Hence, it remains unclear how recent encounters support the processing of ambiguous words in more naturalistic scenarios. Therefore, the proposed experiment will extend the word-meaning paradigm to silent reading, where eye movements will be used to index whether recent encounters during priming can reduce the cost of encountering ambiguous words in sentences that support a subordinate meaning. Eye movement studies indicate that when prior sentence context favours the subordinate meaning of a homograph (e.g., the dog meaning of boxer), reading times are longer on homographs than on an unambiguous control word (e.g., puppy; Duffy et al., 1988; Rayner et al., 1994, 2006; Sereno et al., 2006). This effect has been termed the subordinate bias effect (c.f. Pacht & Rayner, 1993). Regarding word-meaning priming, several studies have investigated the effect of recent encounters with word meanings on the subsequent processing on ambiguous words. In Experiment 1, Rayner et al. (1994) exposed participants to biased homographs or low-ambiguity control words during a paired association task before monitoring their eye movements when reading sentence containing the same words or a set of unprimed words. Rayner et al. replicated the traditional subordinate bias effect in the unprimed condition and then reported a reduced subordinate bias effect following priming in total reading times on target words but not gaze durations. These results suggest that recent encounters with word meanings support the integration of ambiguous words into readers’ ongoing semantic representations of the text but that they do not ease initial encoding. More recently, Leinenger and Rayner (2013) had participants read homographs that were preceded by a single sentence that contained a prior instance of the homograph or a control phrase. Across both experiments, prior exposure to the subordinate meaning of a homograph induced a clear reduction in the subordinate bias effect as indexed by both early and late eye movement measures. Thus, when readers encounter a subordinate meaning of a lexically ambiguous word in the same sentence, there is clear evidence that the subordinate bias effect can be reduced. The short-term benefits of priming reported by Leinenger and Rayner are, however, very different from the benefits observed in the word-meaning priming paradigm. That is, Leinenger and Rayner’s findings do not illustrate whether priming leads to a long-lasting shift in how word-meanings are accessed and interpreted. This study looks to examine whether we can index word-meaning priming in a paradigm more akin to the word-meaning priming paradigm during silent reading after a 1-minute delay. It will act as a proof of concept before a second study which looks to examine word-meaning priming after a more substantial delay. To examine whether word-meaning priming extends to silent reading, participants will read both priming and experimental sentences containing homographs or matched low-ambiguity control words. Comparing reading times on homographs and low-ambiguity control words in the unprimed condition will serve as a baseline and provide an estimate of the subordinate bias effect. This will also provide a form of positive control - evidencing the subordinate bias effect in the unprimed condition will indicate that any failure to detect a reduction in subordinate bias effect following priming does not result from stimuli characteristics. The 1-minute delay condition will allow us to directly examine whether the subordinate bias effect can be reduced following priming.

Developmental dyslexia is a neurodevelopmental disorder characterised by impaired word recognition and slower reading speed (Peterson & Pennington, 2012). Despite these linguistic deficits, visual interventions, such as coloured overlays and backgrounds, are routinely used to ameliorate reading difficulties faced by dyslexic readers. This pre-registered experiment will examine the effectiveness of coloured backgrounds to promote reading efficiency in dyslexic and non-dyslexic readers during real-time sentence reading. Specifically, we will use eye-tracking to examine sentence and target word reading when using a yellow or white (control) background. Despite the widely held view of a central phonological deficit in dyslexia, roughly half of dyslexic individuals are also diagnosed with visual stress, also known as Meares-Irlen syndrome (Irlen, 1991, 1997). Symptoms of visual stress include fatigue, blurring of words, and headaches (Irlen, 1991; Singleton & Trotter, 2005). Coloured overlays have been used to combat visual stress, but the evidence base is far from conclusive. Early studies found favourable results (e.g., Kriss & Evans, 2005). However, it appears that these benefits may be a placebo effect as children are typically asked to select the overlay colour that they believe to be beneficial. Evidence from randomised control trials, where participants are assigned overlays, is consistent with this suggestion as there’s no benefit associated with overlays when reading jumbled sentences (e.g., Wilkins et al., 1994). In addition to the placebo effect, overlay research is limited by the fact that it often requires participants to read jumbled sentences containing 15 high-frequency words. This makes it difficult to infer the effects of overlays on natural reading. To tackle this issue, Henderson et al. (2013) presented 16 dyslexics and 26 non-dyslexics with two non-fictional passages of text. One passage was read with an overlay, and one passage was read without. Henderson et al. assessed the effects of overlays on both reading comprehension of the text and reading rate. From their analysis, it was clear that there was no benefit of using overlays for comprehension or reading rate; there was no significant main effect of overlay or evidence of an interaction between overlay and group for either outcome measure. In addition to reading passages of text, Henderson et al. asked the same participants to read unconnected, jumbled texts like those in previously discussed studies. When reading these jumbled texts, both dyslexics and non-dyslexics benefited from overlays, but the dyslexic participants did not benefit from overlays to a greater extent than those without dyslexia. This suggests that the benefit of overlays observed when reading jumbled, unconnected texts do not transfer over to the reading of connected texts. Despite the lack of empirical evidence supporting the use of overlays, recent studies have begun to investigate the use of colour backgrounds on reading in dyslexia. For example, Rello and Bigham (2017) investigated the benefit of 10 coloured backgrounds on reading in 341 participants (84 dyslexics), and found favourable results for warm colours, such as yellow. However, Rello and Bigham had one observation per colour per participant and they relied on a crude measure of reading speed (the time from the start to the end of the trial). In a previous study (https://osf.io/wxtvn), we looked to investigate the effect of yellow backgrounds on real-time reading using a self-paced reading paradigm. We compared sentence and target word reading times in 31 dyslexic and 30 non-dyslexic participants when reading from a white or yellow background. Importantly, reading times in both groups did not differ when reading from white or yellow backgrounds. This null effect could not be attributed to a lack of engagement with the task as scores on comprehension questions were near ceiling and we detected a frequency effect, indicating that reading time measurements captured costs associated with processing less common words. Therefore, we conclude that yellow backgrounds do not benefit sentence and target word reading times in dyslexics or non-dyslexics. One limitation of our prior study was that it made use of a self-paced reading task where words were masked and uncovered via a button press. Like studies involving the reading of high-frequency words, the experience of readers during the task is far from real sentence reading. Thus, we look to examine the benefit of coloured backgrounds in a more naturalistic manner (i.e., free reading) using eye-tracking. We retain the 2 (group: dyslexic, non-dyslexic), x 2 (background: yellow, white) x 2 (frequency: high, low) design used in our previous study (https://osf.io/wxtvn) and all experimental items remain the same.

In recent years there has been a proliferation in research on return-sweep saccades (Adedeji et al., 2022; Parker et al., 2017; Parker, Nikolova et al., 2019; Parker, Slattery et al., 2019; Parker & Slattery, 2019, 2021; Parker et al., 2013; Slattery & Parker, 2019; Slattery & Vasilev, 2019; Vasilev et al., 2021), i.e., the eye movement that brings a reader’s gaze from the end of one line to the start of the next. One of the most consistently reported findings is that fixations preceding a return-sweep (i.e., line-final fixations) are shorter than intra-line fixations (e.g., Parker, Nikolova et al., 2019). Several explanations for shorter line-final fixations exist. For example, they may present general oculomotor programming in preparation for return-sweep execution, reduced skipping costs at line boundaries, or reduced lateral masking (Parker et al., 2023). Another robust finding is that readers' fixations following a return-sweep also differ in duration from intra-line fixations (e.g., Parker et al., 2019). Fixations following a return-sweep can be categorized as accurate line-initial fixations and under-sweep fixations. What differentiates these two populations of fixations is the direction of the following saccade with accurate line-initial fixations followed by a rightwards progressive saccade. In contrast, under-sweeps are followed by a leftwards regressive saccade that is thought to correct return-sweep undershoot errors. Accurate line-initial fixations tend to be longer than intra-line reading fixations and have been argued to reflect a reliance on foveal encoding and an absence of parafoveal preview (Parker et al., 2017; Parker & Slattery, 2019). In contrast, under-sweep fixations are much shorter than intra-line fixations and this is thought to be driven by retinal feedback and return-sweep error (Schebilske, 1976). Current arguments about the differences in fixation durations, at least for accurate line-initial fixations, are underpinned by the assumption that lexical processing is responsible for these differences in fixation duration. However, these differences may also be a consequence of low-level oculomotor control. One approach to tease these two explanations apart is to compare multiline text reading to the scanning of multiline z-strings. The scanning of z-strings has been conceptualised as an oculomotor control condition to normal reading as it approximates reading without lexical and post-lexical processing (Kuperman et al., 2010). So, eye movements during paragraph reading and multiline z-string scanning should be influenced similarly by visual and oculomotor factors. Therefore, if the line-final and line-initial fixation duration differences are not linguistic in nature, then we would assume similar reductions during paragraph reading and multiline z-string scanning. If, however, these differences are an artefact of linguistic processing, we might expect less of a reduction in durations for line-final fixations (relative to intra-line fixations) and less of an increase for line-initial fixations (relative to intra-line fixations). Previously, Kuperman et al. (2010), in a reanalysis of single-line reading and z-string data from Nuthmann et al. (2007) and Nuthmann and Engbert (2009), showed that shorter initial fixation during single-line text reading (representing “start-up” effects) and longer line0final fixations during single-line text reading (representing “wrap-up” effects) were present during sentence reading but z-string scanning. Based on these findings, Kuperman et al. reasoned that both start-up and wrap-up effects require mindful reading (i.e., lexical and post-lexical processing) and are not simply a result of a generic oculomotor programme. Following Kuperman et al.’s comparison of single sentence reading and z-string scanning, we may expect that reading time differences at line boundaries may not be present during z-string scanning; indicating that lexical and post-lexical processing are responsible for shorter line-final and longer line-initial paragraph reading fixations.

Stage 1 IPA at Cortex

Laterality for language can be assessed behaviourally using both visual tasks and auditory tasks. Visual half-field paradigms typically present stimuli in the left (LVF) or right visual fields (RVF). More rapid and more accurate recognition of stimuli in the RVF is generally considered to be indicative of left hemisphere dominance for language (e.g., Hellige, 1990). Similar, to dichotic listening tasks, where stimuli are presented to the left (LEA) or right ear (REA), better performance for stimuli presented in the right ear is considered to indicate left hemisphere dominance for language (see Kimura, 2011). These findings reflect the fact that these tasks rely on contralateral projects of information in the brain such that information in the right field of vision/presented to the right ear project initially to the language dominant left hemisphere and receive more efficient processing while stimuli presented in the left visual field/ear require additional colossal transfer for processing. When examining hemispheric dominance for language, researchers will often display words or images for processing in visual tasks and consonant-vowel (CV) strings in auditory tasks. To date, only a handful of studies have examined the performance of lateralised auditory and visual tasks that make use of the same stimuli. One such study was conducted by Voyer and Boudreau (2003), who reported opposite brain laterality for CV strings in analogous auditory and visual tasks; a REA and LVF advantage. Oltedal and Hugdahl (2017) have since replicated this behavioural observation and confirmed, using fMRI, that the opposite pattern of lateralisation is present at the brain level. As such, the data lend support to the conclusion that auditory and visual lateralised presentation of CV strings measure different cognitive processes– tentatively linguistic and spatial processing respectively. Oltedal and Hugdahl argue that because CVs were shown repeatedly under rapid presentation, these units were processed spatially. In the proposed study, we will revisit the topic of opposite brain laterality for analogous auditory and visual tasks. First, in a conceptual replication of Oltedal and Hugdahl (2017), we will present participants with analogous auditory and visual CV tasks (i.e., a dichotic listening task and a lateralised visual half-field task). Second, we will present participants with analogous auditory and visual word recognition tasks where there is a small pool of stimuli that could potentially be learned and processed spatially. Together, these tasks will be able to disentangle previous findings. If we observe opposite perceptual biases between tasks for CV string and words, then we may conclude that the presentation method allows spatial learning of the stimuli. If we only observe opposite perceptual biases for the CV tasks, then this would indicate that CVs are processed differently from word units.

In the first part of this project (Parker et al., 2022), we aimed to extend the word-meaning paradigm to silent reading. This pre-registration covers a second experiment that is an extension of our first study. For brevity’s sake, a fuller project description can be found at: https://osf.io/5gxst. The word-meaning priming paradigm, which typically employs meta-linguistic tasks, has shown that recent encounters with the subordinate meaning of a word have relatively long-lasting effects on how ambiguous words are interpreted. For example, Rodd et al. (2016; Experiment 2) asked participants to complete a word association task across four priming conditions, where ambiguous words were unprimed or primed 1-, 20- or 40-minutes prior. At each time point, there was an increase in prime consistent responses, where rates of response consistent with the prime increased from 20% to 35%, 28%, and 28% respectively, indicating that, relative to an unprimed condition, recent encounters made a particular word-meaning more readily available, even after a substantial delay. Regarding the processing of ambiguous words during silent reading, studies have examined how the size of the subordinate bias effect changes following a recent encounter with a particular meaning (cf. Pacht & Rayner, 1993). The subordinate bias effect refers to the observation that when prior sentence context favours the subordinate meaning of a homograph (e.g., the dog meaning of boxer), reading times are longer on homographs than on an unambiguous control word (e.g., puppy; Duffy et al., 1988; Rayner et al., 1994, 2006; Sereno et al., 2006). The subordinate bias effect is thought to reflect the increased difficultly of accessing a subordinate meaning of a word relative to a control word. With respect to improving processing of subordinate word meanings, Leinenger and Rayner (2013) reported promising results. Participants read homographs that were preceded by a single sentence that contained a prior instance of the homograph or a control phrase. In two experiments, prior exposure to the subordinate meaning of a homograph within the same sentence induced a clear reduction in the subordinate bias effect as indexed by both early and late eye movement measures. Thus, when readers encounter a subordinate meaning of a lexically ambiguous word in the same sentence, there is clear evidence that the subordinate bias effect can be reduced. The short-term benefits of priming reported by Leinenger and Rayner are, however, very different from the benefits observed in the word-meaning priming paradigm. In particular, they do not illustrate whether priming leads to a long-lasting shift in how word meanings are accessed and interpreted. In a previous experiment (Parker et al., 2022) we investigated whether word-meaning priming extends to silent reading and if the subordinate bias effect can be reduced when priming occurred 1-minute prior. Specifically, we examined readers’ eye movements as they read experimental items containing high-ambiguity or low-ambiguity (control) target words where the target word was either primed (where participants previously encountered the same word within a different sentence contex) 1-minute prior or unprimed. Thus, the experiment employed a 2 (ambiguity) x 2 (priming) design. In both early and late measures of processing (i.e., gaze durations, go-past times, and total reading times) we found clear evidence of word meaning priming where the cost of processing an ambiguous word was reduced following priming that occurred 1-minute prior. Furthermore, this reduction in reading times was stronger for high- than low-ambiguity control words, which indicates that this reduction of a subordinate bias effect is over and above a simple benefit of repetition. One drawback of this study is that it did not provide insight into whether the reduction of the subordinate bias effect occurs at a more substantial delay. Therefore, the aim of the current study is to investigate whether there is a reduction in the subordinate bias effect when priming occurs 30-minutes prior, using an experimental design that is almost identical to our prior study.

Studies of cerebral lateralization often involve participants completing a series of perceptual tasks under laboratory conditions. This has constrained the number of participants recruited in such studies. Online testing can allow for much larger sample sizes but limits the amount of experimental control that is feasible. Here we considered whether online testing could give valid and reliable results on four tasks: a rhyme decision visual half-field task, a dichotic listening task, a chimeric faces task, and a finger tapping task. We recruited 392 participants, oversampling left-handers, who completed the battery twice. Three of the tasks showed evidence of both validity and reliability, insofar as they showed hemispheric advantages in the expected direction and test-retest reliability of at least r = .75. The reliability of the rhyme decision task was less satisfactory (r = .62). We also confirmed a prediction that extreme left-handers were more likely to depart from typical lateralization. Lateralization across the two language tasks (dichotic listening and rhyme judgement) was weakly correlated, but unrelated to lateralization on the chimeric faces task. We conclude that three of the tasks, dichotic listening, chimeric faces and finger tapping, show considerable promise for online evaluation of cerebral lateralization.