Beyond mean RTs in visual word recognition: Extensions of a remarkably stable three-way interaction amongst word frequency, stimulus degradation, and RT distributions.

The pattern of factor effects on response time (RT) performance in multi-factor experiments provides a powerful technique for inferring the structure of underlying mental processes. The results of the present lexical decision experiment show that additive effects of stimulus quality and word frequency are observed in mean RTs, variances, and the exGaussian parameters of the RT distribution. These findings are consistent with the conclusion that word frequency and stimulus quality affect separate stages of processing. This is consistent with the conclusion that word frequency effects reflect mapping operations between stages, but, when taken in conjunction with other reports in the literature, is inconsistent with the received view in many activation models that word frequency exerts its effect within the word detector level of representation. Resume Le schema des effets du facteur sur l'execution du temps de reponse (TR) dans les experiences a multiples facteurs fournissent une technique efficace pour fixer la structure des processus mentaux sous-jacents. Les resultats de la presente experience de decision lexicale montrent que les effets additifs de la qualite du stimulus et de la frequence du mot sont observes dans les TR moyens, les variances et les parametres ex-gaussiens de la distribution du TR. Ces conclusions confirment que la frequence du mot et la qualite du stimulus influencent les phases separees du traitement. Elles confirment egalement que les effets de la frequence du mot refletent les operations de correspondance entre les phases, mais, lorsque comparees a d'autres rapports, les conclusions ne correspondent pas a la vision etablie dans plusieurs modeles d'activation qui veut que la frequence du mot exerce ses effets au niveau de representation du detecteur du mot. It is well established that a word's frequency of occurrence in printed English is a strong determinant of performance in word recognition tasks such as lexical decision. For example, high frequency words such as CAT are typically recognized more quickly than lower frequency words such as VAT (e.g., Forster & Chambers, 1973; Frederiksen & Kroll, 1976; see also Monsell, 1991, for a review). One account of this word frequency effect is given by a number of models in the Activation class. Thus, models such as Morton's (1969) logogen model and variants of McClelland and Rumelhart's (1981) interactiveactivation model (e.g., Coltheart, Curtis, Atkins, & Haller, 1993; Grainger & Jacobs, 1996) all assume a word level of representation in the form of orthographic and phonological word detectors for each word the reader knows. These word detectors accumulate activation over time, and both models assume that low frequency words take longer to reach a threshold than do high frequency words. The models differ in that the interactive-activation model assumes that high frequency words start with a higher resting level of activation than do low frequency words whereas the logogen model assumes a common resting level of activation but different recognition thresholds for high and low frequency words One difficulty with the activation explanation is that a simple main effect of word frequency in lexical decision does not allow one to determine the locus of the effect; it merely informs us that word frequency affects performance in this task. A number of investigators have therefore applied Sternberg's (1969) additive factors logic to this issue by manipulating word frequency in conjunction with other factors that affect performance in lexical decision because predictions can be derived from the models as to how the effects of these factors on response time (RT) should combine (e.g., Becker & Killion, 1977; Besner & Smith, 1992; Besner & Swan, 1982). WORD FREQUENCY, STIMULUS QUALITY, AND CONTEXT It is well documented that stimulus quality affects word recognition: Clearly presented stimuli are recognized more quickly than degraded ones. …

Stimulus quality and word frequency produce additive effects in lexical decision performance, whereas the semantic priming effect interacts with both stimulus quality and word frequency effects. This pattern places important constraints on models of visual word recognition. In Experiment 1, all three variables were investigated within a single speeded pronunciation study. The results indicated that the joint effects of stimulus quality and word frequency were dependent upon prime relatedness. In particular, an additive effect of stimulus quality and word frequency was found after related primes, and an interactive effect was found after unrelated primes. It was hypothesized that this pattern reflects an adaptive reliance on related prime information within the experimental context. In Experiment 2, related primes were eliminated from the list, and the interactive effects of stimulus quality and word frequency found following unrelated primes in Experiment 1 reverted to additive effects for the same unrelated prime conditions. The results are supportive of a flexible lexical processor that adapts to both local prime information and global list-wide context.

Across 3 different word recognition tasks, distributional analyses were used to examine the joint effects of stimulus quality and word frequency on underlying response time distributions. Consistent with the extant literature, stimulus quality and word frequency produced additive effects in lexical decision, not only in the means but also in the shape of the response time distributions, supporting an early normalization process that is separate from processes influenced by word frequency. In contrast, speeded pronunciation and semantic classification produced interactive influences of word frequency and stimulus quality, which is a fundamental prediction from interactive activation models of lexical processing. These findings suggest that stimulus normalization is specific to lexical decision and is driven by the task's emphasis on familiarity-based information.

Plaut and Booth’s (2006) first simulation shows that there is essentially perfect discrimination between word and nonwords sharing the same orthographic structure when the simulation is carried out in the way we suggested. We are therefore satisfied that their model can (now) be characterized as a parallel distributed processing (PDP) model of lexical processing (with the caveat that although it is beyond the scope of the present work, it is important to show at some point that the model “scales up” when it has a more realistically sized vocabulary of 40,000 words or so). We voiced concern that their model makes lexical decisions at the semantic level but there are patients with severe damage to semantics who are nonetheless as accurate at lexical decision as control patients without such damage (Coltheart, 2004). Can a model that makes lexical decisions at the semantic level simulate these data? In reply, Plaut and Booth (2006) reported new simulations showing that increasingly severe “lesioning” of their model at the semantic level impairs lexical decision accuracy in a monotonic way, but it is a remarkably small effect (see their Figure 1). They argue that “distinguishing the semantic activation of one word from that of another requires far more detailed information— and, thus, is less robust to damage—than distinguishing either from the much weaker activation produced by a nonword” (Plaut & Booth, 2006, p. 198). The results of these simulations not withstanding, a substantive difficulty for their account of lexical decision remains. Blazely, Coltheart, and Casey (in press) reported a detailed analysis of two patients (EM and PC) with semantic dementia, both of whom had significant impairments of semantic memory. EM performs slightly worse than PC on semantic tasks, but her visual lexical decision performance (two-alternative forced choice) was virtually perfect (97% correct), whereas PC’s visual lexical decision performance was significantly impaired (75% correct). It is difficult to see how this pattern can be simulated by Plaut and Booth’s (2000) model if lexical decision is carried out at the semantic level, but it is easy to understand if the decisions are carried out at the lexical level and PC’s lexical processing abilities are impaired. Blazely et al. provide converging evidence in support of this conclusion. Our third major concern was that Plaut and Booth’s (2000) PDP model’s use of the sigmoid function relating activation to reaction time (RT) would not permit the joint effect of two factors on RT to be additive when one of these factors and a third factor produce an interaction that lies within the same range of RTs. They responded to this by reporting a new simulation that purports to accomplish this. These data are reported in their Table 1. Some of these data are reproduced in Figure 1 for illustrative purposes. One point here is that this simulation is an unusual way of looking at the joint effects of word frequency and stimulus quality because it entails collapsing across the significant Context Stimulus Quality interaction in the right-hand panel. The joint effects of stimulus quality and word frequency should be examined when there is a neutral prime rather than a related or unrelated word, as argued by Neely (1991; see also Borowsky & Besner, 1993, who used nonword primes, and see Plaut & Booth, 2000, who also used a nonword baseline in their experiments and simulations but have now dropped it in their 2006 simulation). The more important question is whether the model has produced additive effects of stimulus quality and word frequency (as claimed for the data in the left hand panel). The size of the nonsignificant (underadditive) interaction in the left panel (.011 units) is almost double the size of the significant interaction in the right-hand panel (.006 units). Plaut and Booth (2006) do not comment on this. Even if Plaut and Booth were able to provide a more convincing simulation of both additive and interactive effects within the same range of RTs, we would like to know what principled reason makes it possible for the simulation to produce these results given that they now agree that the sigmoid-based explanation “only approximates the actual behavior of the model” (Plaut & Booth, 2006, p. 199). We take the view that Plaut and Booth’s (2006) new simulation work settles little beyond the fact that their model can discriminate 450 500 550 600 650 700

Additive effects of Stimulus Quality and Word Frequency on RT in the context of lexical decision when the foils are orthographically legal were first reported more than 4 decades ago, and subsequently replicated numerous times. Two accounts are considered that make different a priori predictions when the foils are orthographically illegal. Yap and Balota's (2007) Familiarity Discrimination account predicts additive effects of these two factors on mean RT and across the RT distribution because it assumes a staged normalization process that deals with the effect of low Stimulus Quality; a subsequent process produces the effect of Word Frequency. In contrast, O'Malley and Besner's (2008) context-dependent thresholding/cascading account predicts an interaction because the use of illegal foils eliminates the need for thresholding at the letter level normally used to protect against lexical capture (identifying a nonword as a word) in experiments where Stimulus Quality is a factor, and hence the system reverts to processes in cascade. Critically, the present experiment yielded an interaction in which low-frequency words were more impaired by low Stimulus Quality than were high-frequency words. These data are inconsistent with the Familiarity Discrimination account as currently constituted, but consistent with a context-specific cascaded account. Further discussion considers how the Familiarity account may be modified so as to accommodate these data. Most generally, these data add to the view that processing is highly malleable (context dependent) rather than the received view, especially in regard to computational accounts, in which interactive-activation dynamics dominate. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

There are multiple reports, in the context of the time taken to read aloud, that the joint effects of stimulus quality and word frequency (a) interact when only words appear in the list but (b) are additive when nonwords are intermixed with words (O'Malley & Besner, 2008). This triple interaction has been explained in terms of the idea that different processing modes are in play in these different contexts. Processing is cascaded when only words appear in the list, allowing the effect of stimulus quality to influence the downstream process(es) affected by word frequency. In contrast, when nonwords appear in the list an early process affected by stimulus quality, but not word frequency, is staged (thresholded) so as to reduce the probability of lexicalizations (reading a nonword as a word) when stimulus quality is low. The present experiment addresses the issue of whether such thresholding in the presence of nonwords is driven by the orthography or phonology of the nonwords included in the stimulus set. Participants read words aloud that varied in word frequency and were randomly intermixed with nonwords that all sounded identical to words (e.g., BRANE for BRAIN). Stimulus quality and word frequency had additive effects on the time to read aloud in this context, consistent with the view that it is the orthography of the nonwords that matters. Other aspects of the results suggest that between level feed-back is in play when this particular kind of nonword is used. (PsycINFO Database Record

Reductions in stimulus quality may disrupt the reading performance of older adults more when compared with young adults because of sensory declines that begin early in middle age. However, few studies have investigated adult age differences in the effects of stimulus quality on reading, and none have examined how this affects lexical processing and eye movement control. Accordingly, we report two experiments that examine the effects of reduced stimulus quality on the eye movements of young (18–24 years), middle-aged (41–51 years), and older (65+ years) adult readers. In Experiment 1, participants read sentences that contained a high- or low-frequency critical word and that were presented normally or with contrast reduced so that words appeared faint. Experiment 2 further investigated effects of reduced stimulus quality using a gaze-contingent technique to present upcoming text normally or with contrast reduced. Typical patterns of age-related reading difficulty (e.g., slower reading, more regressions) were observed in both experiments. In addition, eye movements were disrupted more for older than younger adults when all text (Experiment 1) or just upcoming text (Experiment 2) appeared faint. Moreover, there was an interaction between stimulus quality and word frequency (Experiment 1), such that readers fixated faint low-frequency words for disproportionately longer. Crucially, this effect was similar across all age groups. Thus, although older readers suffer more from reduced stimulus quality, this additional difficulty primarily affects their visual processing of text. These findings have important implications for understanding the role of stimulus quality on reading behavior across the lifespan.

In the lexical decision task, the additive effects of stimulus quality and word frequency have been used to infer the presence of independent processing stages in visual word recognition. Importantly, this pattern can be moderated by semantic priming, suggesting the presence of a retrospective prime retrieval mechanism that is selectively engaged based on task context (i.e., utility of the primes). We examine the sensitivity of this mechanism in two lexical decision experiments that manipulate stimulus quality, word frequency, and semantic priming. Critically, we studied these joint effects when the proportion of related primes was set at .50 or .25. Results indicated that with a .50 relatedness proportion, a three-way interaction was obtained such that additivity between frequency and stimulus quality was found following related semantic primes, but an overadditive pattern was exhibited following unrelated primes. When the relatedness proportion was reduced to .25, this interaction was eliminated. Furthermore, relatedness proportion affected the magnitude of the stimulus quality by priming interaction but not the frequency by priming interaction. These results are interpreted within the context of a flexible lexical processer that adaptively engages processes in response to task context.

Is it possible to identify when lexical access in visual word recognition is based on an orthographic code and when it is based on a phonological code? Some researchers have argued that the joint effects of semantic context and stimulus quality in lexical decision are diagnostic. Their argument is that when phonological recoding is employed, it serves to keep the effects of stimulus quality and context from interacting. Semantic context and stimulus quality are therefore predicted to have additive effects on RT. In contrast, when lexical access is not mediated by a phonological recoding stage, then the effect of stimulus quality interacts with context, as commonly seen in the literature. A strong test of these claims was devised in which participants were forced to use phonological recoding for the purpose of lexical access. An interaction between context and stimulus quality was observed. This finding is taken as evidence that the joint effects of semantic context and stimulus quality are not diagnostic with respect to the nature of the lexical access code (i.e., orthographic versus phonological) for readers of English.

S. O'Malley and D. Besner (2008) showed that additive effects of stimulus degradation and word frequency in reading aloud occur in the presence of nonwords but not in pure word lists. They argued that this dissociation presents a major challenge to interactive computational models of reading aloud and claimed that no currently implemented model is able to simulate additive effects in these conditions. In the current article, it is shown that the connectionist dual process model (CDP+) can simulate these effects because its nonlexical route is thresholded. The authors present a series of simulations showing that CDP+ can not only simulate the precise dissociation observed by O'Malley and Besner but more generally can produce additive effects for a wide range of parameter combinations and different sets of items. The nonlexical route of CDP+ was not modified post hoc to deal with the effects of stimulus quality, but it had been thresholded for principled reasons before it was known that these effects existed. Together, the effects of stimulus quality on word frequency do not challenge CDP+ but rather provide unexpected support for its architecture and processing dynamics.

There have been multiple reports over the last 3 decades that stimulus quality and word frequency have additive effects on the time to make a lexical decision. However, it is surprising that there is only 1 published report to date that has investigated the joint effects of these two factors in the context of reading aloud, and the outcome of that study is ambiguous. The present study shows that these factors interact in the context of reading aloud and at the same time replicate the standard pattern reported for lexical decision. The main implication of these results is that lexical activation, at least as indexed by the effect of word frequency, does not unfold in a uniform way in the contexts reported here. The observed dissociation also implies, contrary to J. A. Fodor's (1983) view, that the mental lexicon is penetrable rather than encapsulated. The distinction between cascaded and thresholded processing offers one way to understand these and related results. A direction for further research is briefly noted.

The joint effects of stimulus quality and word frequency in lexical decision were examined in 4 experiments as a function of nonword type (legal nonwords, e.g., BRONE, vs. pseudohomophones, e.g., BRANE). When familiarity was a viable dimension for word-nonword discrimination, as when legal nonwords were used, additive effects of stimulus quality and word frequency were observed in both means and distributional characteristics of the response-time distributions. In contrast, when the utility of familiarity was undermined by using pseudohomophones, additivity was observed in the means but not in distributional characteristics. Specifically, opposing interactive effects in the underlying distribution were observed, producing apparent additivity in means. These findings are consistent with the suggestion that, when familiarity is deemphasized in lexical decision, cascaded processing between letter and word levels is in play, whereas, when familiarity is a viable dimension for word-nonword discrimination, processing is discrete.

Important insights in visual word recognition have been provided by studies examining the combined influence of multiple factors on participants' mean response times to English words in the lexical decision task. However, to make progress towards a complete understanding of how meaning is activated by print, researchers need to conduct more detailed analyses of behavioral patterns beyond mean response latencies and accuracies, particularly how variables influence different components of response time distributions. Moreover, it is critical to extend patterns found in English to the diverse scripts encountered by readers across the world. The present study is the first to explore the theoretically important effects of stimulus quality and word frequency on lexical decisions involving two-character Mandarin Chinese and Cantonese Chinese words, using participants from Singapore and Hong Kong, respectively. Despite the profound differences between the English and Chinese writing systems, we observed remarkably similar trade-offs in the stimulus quality × word frequency interaction across different portions of the response time distribution for both orthographies, indicating that the optimization of lexical processing by leveraging available codes in response to task demands extends across multiple and highly diverse writing systems.

Virtually all theories of visual word recognition assume (typically implicitly) that when a pathway is used, processing within that pathway always unfolds in the same way. This view is challenged by the observation that simple variations in list composition are associated with qualitative changes in performance. The present experiments demonstrate that when reading aloud, the joint effects of stimulus quality and word frequency on response time are driven by the presence/absence of nonwords in the list. Interacting effects of these factors are seen when only words appear in the experiment, whereas additive effects are seen when words and nonwords are randomly intermixed. One way to explain these and other data appeals to the distinction between cascaded processing (or interactive activation) on the one hand versus a thresholded mode of processing on the other, with contextual factors determining which mode of processing dominates.

Eye movements of Chinese readers were recorded for sentences in which high- and low-frequency target words were presented normally or with reduced stimulus quality in two experiments. We found stimulus quality and word frequency produced strong additive effects on fixation durations for target words. The results demonstrate that stimulus quality and word frequency affect different stages of processing (e.g., visual processing and lexical processing). These results are consistent with the findings of previous single-word lexical decision studies, which showed that stimulus quality manipulation primarily affects the early preattentive stage of visual processing, whereas word frequency affects lexical processes. We discuss these findings in terms of the role of stimulus quality in word recognition and in relation to the E-Z Reader model of eye movement control.