Prior research with adult participants reported a rightward bias in the reading version of the flankers task. Here, we investigated how this bias evolves as a function of reading expertise. We tested two groups of French primary school children from Cycle 2 (grades 1 and 2) and Cycle 3 (grades 4 and 5) and one group of adult participants. In the related flanker conditions, the central target word was flanked by the same word either on the left (park park ####), the right (#### park park), or on both sides (park park park). In the unrelated conditions, the repeated flanker words were replaced by a different unrelated word. In the analysis of standardized reaction times (RTs), there was a three-way interaction between the three groups of participants and the impact of flanker relatedness as a function of the position of the related flankers. This three-way interaction reflected the significantly greater increase in effects of flanker relatedness between Cycle 2 and Cycle 3 for the bilateral flanker and the right flanker conditions compared with the left flanker condition. This suggests that the rightward bias is driven by attentional asymmetries that develop during the process of learning to read.

To investigate whether similar control processes are used during single and dual language production, we compared register switching (formal and informal speech in the same language) vs. language switching (French and English). The results across two experiments showed a positive correlation of overall register- and language-switch costs and similar formal French switch costs across the two switching tasks. However, whereas increasing the cue-to-stimulus interval resulted in a reduction of language-switch costs, register-switch costs were unaffected by the interval manipulation. This difference in switch-cost pattern indicates that control processes are not entirely identical during single and dual language production.

One of the cool aspects of the original implementation of the BIA model (van Heuven, Dijkstra & Grainger, 1998) was the discovery that inhibitory connections between language nodes and lexical representations was a necessary feature for the model to be able to simulate the target data set at that time. This demonstrates the importance of computational modeling, a key point of the present target article, since inhibitory connections were postulated to occur only between representations at the same level in the conceptual model (Grainger & Dijkstra, 1992). Top-down inhibition was subsequently dropped in the BIA+ model (Dijkstra & van Heuven, 2002), and the Multilink model of the present target article (Dijkstra, Wahl, Buytenhuijs, van Halem, Al-jibouri, de Korte & Rekké, 2018) goes one step further by removing all kinds of inhibitory connections, both between and within levels. Instead, the authors of the model propose that bilingual language processing relies on bidirectional excitatory connections between representations at different levels. This is curious given that even more evidence has accumulated in favor of inhibition since the original implementation of the BIA model, both between neighboring lexical representations (i.e., lateral inhibition) and from language membership representations (e.g., language nodes and tags) down to lexical representations. In this commentary, we focus on whether the exclusion of these two inhibitory processes is warranted, and how the inclusion of these processes might benefit future developments of the model.

In the current study, we investigated bilingual error detection by measuring the repair rate of language intrusions (i.e., involuntary production of nontarget language words) that arose while bilinguals produced sentences in a language switching context. This allowed us to compare two prominent accounts of error detection in a bilingual setting. According to the conflict monitoring account, error detection is initiated by interference. Since language switching increases bilingual language interference, error detection should be better in switch relative to repetition trials. According to the perceptual loop theory, error detection is based on language comprehension. Since language switching is known to impair language comprehension, it follows that error detection should be worse in switch relative to repetition trials. The results showed that the repair rate of language intrusions was higher in switch than repetition trials, thus providing evidence that bilingual language interference instigates error detection, in line with the conflict monitoring account.

Frost proposes a new agenda for reading research, whereby cross-linguistic experiments would uncover linguistic universals to be integrated within a universal theory of reading. We reveal the dangers of following such a call, and demonstrate the superiority of the very approach that Frost condemns.

A central component in the E-Z Reader model is a two-stage word processing mechanism made responsible for both the triggering of eye movements and sequential shifts of attention. We point to problems with both the verbal description of this mechanism and its computational implementation in the simulation. As an alternative, we consider the use of a connectionist processing module in combination with a more indirect form of cognitive eye-movement control.

The model of phoneme monitoring proposed by Norris et al. is implausible when implemented in a localist connectionist network. Lexical representations mysteriously inform phoneme decision nodes as to the presence or absence of a target phoneme.

The experiments presented here used a visual version of the syllable monitoring technique (Mehler, Dommergues, Frauenfelder, & Segui, 1981) to investigate the role of syllabic units in beginning and adult readers. Participants responded whenever a visually presented target syllable (e.g., BA) appeared at the beginning of a subsequently presented printed word (e.g., BALANCE). The target was either a consonant–vowel (CV) or consonant–vowel–consonant (CVC) structure and either did or did not correspond to the initial syllable of the target-bearing word. Skilled adult readers showed significant effects of syllable compatibility (faster detection times when the targets corresponded to the initial syllable of target-bearing words than when they did not), but this occurred only when the carrier words had low printed frequencies. First grade readers did not show a syllable compatibility effect when tested in February of the first year of schooling; only target length influenced detection times. When tested four months later (June), however, the children with the highest scores on a reading ability test did show a syllable compatibility effect. These results suggest that reading instruction rapidly allows syllable-sized units to be accessed from print, and that this type of coding continues to influence how adult readers process low frequency words.

Levelt et al. attempt to “model their theory” with weaver++. Modeling theories requires a model theory. The time is ripe for a methodology for building, testing, and evaluating computational models. We propose a tentative, five-step framework for tackling this problem, within which we discuss the potential strengths and weaknesses of Levelt et al.'s modeling approach.