English Lexicon Project Research Articles

Subtlex-UK: A New and Improved Word Frequency Database for British English

We present word frequencies based on subtitles of British television programmes. We show that the SUBTLEX-UK word frequencies explain more of the variance in the lexical decision times of the British Lexicon Project than the word frequencies based on the British National Corpus and the SUBTLEX-US frequencies. In addition to the word form frequencies, we also present measures of contextual diversity part-of-speech specific word frequencies, word frequencies in children programmes, and word bigram frequencies, giving researchers of British English access to the full range of norms recently made available for other languages. Finally, we introduce a new measure of word frequency, the Zipf scale, which we hope will stop the current misunderstandings of the word frequency effect.

Keywords in the mental lexicon.

Network science draws from a number of fields to examine complex systems using nodes to represent individuals and connections to represent relationships between individuals to form a network. This approach has been used in several areas of Psychology to illustrate the influence that the structure of a network has on processing in that system. In the present study the concept of keyplayers in a network (Borgatti, 2006) was examined in the domain of Psycholinguistics. Keyplayers are nodes in a network that, when removed, result in the network fracturing into several smaller components. A set of such nodes was found in a network of phonological word-forms as was another set of foil words, comparable to the "keywords" on a number of lexical and network characteristics. In three conventional psycholinguistic tasks keywords were responded to more quickly and accurately than the foils. A similar trend was observed in an analysis of the keywords and foils (and another set of foils) in the English Lexicon Project. These results open avenues for further exploration of keywords in various areas of language processing, and demonstrate the utility of the network science approach to psycholinguistics and psychology more generally.

It takes time to prime: semantic priming in the ocular lexical decision task.

Two eye-tracking experiments were conducted in which the manual response mode typically used in lexical decision tasks (LDTs) was replaced with an eye-movement response through a sequence of 3 words. This ocular LDT combines the explicit control of task goals found in LDTs with the highly practiced ocular response used in reading text. In Experiment 1, forward saccades indicated an affirmative lexical decision (LD) on each word in the triplet. In Experiment 2, LD responses were delayed until all 3 letter strings had been read. The goal of the study was to evaluate the contribution of task goals and response mode to semantic priming. Semantic priming is very robust in tasks that involve recognition of words in isolation, such as LDT, but limited during text reading, as measured using eye movements. Gaze durations in both experiments showed robust semantic priming even though ocular response times were much shorter than manual LDs for the same words in the English Lexicon Project. Ex-Gaussian distribution fits revealed that the priming effect was concentrated in estimates of tau (τ), meaning that priming was most pronounced in the slow tail of the distribution. This pattern shows differential use of the prime information, which may be more heavily recruited in cases in which the LD is difficult, as indicated by longer response times. Compared with the manual LD responses, ocular LDs provide a more sensitive measure of this task-related influence on word recognition as measured by the LDT.

How does emotional content affect lexical processing?

Even single words in isolation can evoke emotional reactions, but the mechanisms by which emotion is involved in automatic lexical processing are unclear. Previous studies using extremely similar materials and methods have yielded apparently incompatible patterns of results. In much previous work, however, words' emotional content is entangled with other non-emotional characteristics such as frequency of occurrence, familiarity and age of acquisition, all of which have potential consequences for lexical processing themselves. In the present study, the authors compare different models of emotion using the British Lexicon Project, a large-scale freely available lexical decision database. After controlling for the potentially confounding effects of non-emotional variables, a variety of statistical approaches revealed that emotional words, whether positive or negative, are processed faster than neutral words. This effect appears to be categorical rather than graded; is not modulated by emotional arousal; and is not limited to words explicitly referring to emotions. The authors suggest that emotional connotations facilitate processing due to the grounding of words' meanings in emotional experience.

Imageability and age of acquisition effects in disyllabic word recognition

Imageability and age of acquisition (AoA) effects, as well as key interactions between these variables and frequency and consistency, were examined via multiple regression analyses for 1,936 disyllabic words, using reaction time and accuracy measures from the English Lexicon Project. Both imageability and AoA accounted for unique variance in lexical decision and naming reaction time performance. In addition, across both tasks, AoA and imageability effects were larger for low-frequency words than high-frequency words, and imageability effects were larger for later acquired than earlier acquired words. In reading aloud, consistency effects in reaction time were larger for later acquired words than earlier acquired words, but consistency did not interact with imageability in the reaction time analysis. These results provide further evidence that multisyllabic word recognition is similar to monosyllabic word recognition and indicate that AoA and imageability are valid predictors of word recognition performance. In addition, the results indicate that meaning exerts a larger influence in the reading aloud of multisyllabic words than monosyllabic words. Finally, parallel-distributed-processing approaches provide a useful theoretical framework to explain the main effects and interactions.

Bigram Frequency, Number of Syllables and Morphemes and Their Effects on Lexical Decision and Word Naming

There has been an increasing volume of evidence supporting the role of the syllable in word processing tasks. Recently it has also been shown that orthographic redundancy, related to the pattern of bigram frequencies, could not explain the syllable number effect on lexical decision times. This was demonstrated on a large sample of words taken from the British Lexicon Project. In this study we extend this research by examining both lexical decision and word naming times taken from the English Lexicon Project. There was a syllable number effect for both tasks in the expected direction, and this effect was independent of the presence of a bigram trough. The research also examined the role of other bigram related variables and the number of morphemes on lexical decision and word naming times. The number of morphemes had a significant effect on both word processing tasks, with words with more morphemes producing faster reaction times and also fewer errors. This pattern was reversed for nonword lexical decision times. The results are discussed in the light of recent developments in models of reading.

The bigram trough hypothesis and the syllable number effect in lexical decision

There has been an increasing volume of evidence supporting the role of the syllable in various word processing tasks. It has, however, been suggested that syllable effects may be caused by orthographic redundancy. In particular, it has been proposed that the presence of bigram troughs at syllable boundaries cause what are seen as syllable effects. We investigated the bigram trough hypothesis as an explanation of the number of syllables effect for lexical decision in five-letter words and nonwords from the British Lexicon Project. The number of syllables made a significant contribution to prediction of lexical decision times along with word frequency and orthographic similarity. The presence of a bigram trough did not. For nonwords, the number of syllables made a significant contribution to prediction of lexical decision times only for nonwords with relatively long decision times. The presence of a bigram trough made no contribution. The evidence presented suggests that the bigram trough cannot be an explanation of the syllable number effect in lexical decision. A comparison of the results from words and nonwords is interpreted as providing some support for dual-route models of reading.

Dealing with zero word frequencies: A review of the existing rules of thumb and a suggestion for an evidence-based choice

In a critical review of the heuristics used to deal with zero word frequencies, we show that four are suboptimal, one is good, and one may be acceptable. The four suboptimal strategies are discarding words with zero frequencies, giving words with zero frequencies a very low frequency, adding 1 to the frequency per million, and making use of the Good-Turing algorithm. The good algorithm is the Laplace transformation, which consists of adding 1 to each frequency count and increasing the total corpus size by the number of word types observed. A strategy that may be acceptable is to guess the frequency of absent words on the basis of other corpora and then increasing the total corpus size by the estimated summed frequency of the missing words. A comparison with the lexical decision times of the English Lexicon Project and the British Lexicon Project suggests that the Laplace transformation gives the most useful estimates (in addition to being easy to calculate). Therefore, we recommend it to researchers.

Sensory experience ratings for over 5,000 mono- and disyllabic words

Sensory experience ratings (SERs) reflect the extent to which a word evokes a sensory and/or perceptual experience in the mind of the reader. Juhasz, Yap, Dicke, Taylor, and Gullick (Quarterly Journal of Experimental Psychology 64:1683-1691, 2011) demonstrated that SERs predict a significant amount of variance in lexical-decision response times in two megastudies of lexical processing when a large number of established psycholinguistic variables are controlled for. Here we provide the SERs for the 2,857 monosyllabic words used in the Juhasz et al. study, as well as newly collected ratings on 3,000 disyllabic words. New analyses with the combined set of words confirmed that SERs predict a reliable amount of variance in the lexical-decision response times and naming times from the English Lexicon Project (Balota, Yap, Cortese, Hutchison, Kessler, Loftus, & Treiman, Behavior Research Methods 39:445-459, 2007) when a large number of surface, lexical, and semantic variables are statistically controlled for. The results suggest that the relative availability of sensory/perceptual information associated with a word contributes to lexical-semantic processing.

Age-of-acquisition ratings for 30,000 English words

We present age-of-acquisition (AoA) ratings for 30,121 English content words (nouns, verbs, and adjectives). For data collection, this megastudy used the Web-based crowdsourcing technology offered by the Amazon Mechanical Turk. Our data indicate that the ratings collected in this way are as valid and reliable as those collected in laboratory conditions (the correlation between our ratings and those collected in the lab from U.S. students reached .93 for a subsample of 2,500 monosyllabic words). We also show that our AoA ratings explain a substantial percentage of the variance in the lexical-decision data of the English Lexicon Project, over and above the effects of log frequency, word length, and similarity to other words. This is true not only for the lemmas used in our rating study, but also for their inflected forms. We further discuss the relationships of AoA with other predictors of word recognition and illustrate the utility of AoA ratings for research on vocabulary growth.

EDom: Norming software and relative meaning frequencies for 544 English homonyms

Words that are homonyms-that is, for which a single written and spoken form is associated with multiple, unrelated interpretations, such as COMPOUND, which can denote an < enclosure > or a < composite > meaning-are an invaluable class of items for studying word and discourse comprehension. When using homonyms as stimuli, it is critical to control for the relative frequencies of each interpretation, because this variable can drastically alter the empirical effects of homonymy. Currently, the standard method for estimating these frequencies is based on the classification of free associates generated for a homonym, but this approach is both assumption-laden and resource-demanding. Here, we outline an alternative norming methodology based on explicit ratings of the relative meaning frequencies of dictionary definitions. To evaluate this method, we collected and analyzed data in a norming study involving 544 English homonyms, using the eDom norming software that we developed for this purpose. Dictionary definitions were generally sufficient to exhaustively cover word meanings, and the methods converged on stable norms with fewer data and less effort on the part of the experimenter. The predictive validity of the norms was demonstrated in analyses of lexical decision data from the English Lexicon Project (Balota et al., Behavior Research Methods, 39, 445-459, 2007), and from Armstrong and Plaut (Proceedings of the 33rd Annual Meeting of the Cognitive Science Society, 2223-2228, 2011). On the basis of these results, our norming method obviates relying on the unsubstantiated assumptions involved in estimating relative meaning frequencies on the basis of classification of free associates. Additional details of the norming procedure, the meaning frequency norms, and the source code, standalone binaries, and user manual for the software are available at http://edom.cnbc.cmu.edu .

Does richness lose its luster? Effects of extensive practice on semantic richness in visual word recognition

Previous studies have reported facilitatory effects of semantic richness on word recognition (e.g., Yap et al., 2012). These effects suggest that word meaning is an important contributor to lexical decision task (LDT) performance, but what are the effects of repeated LDT practice on these semantic contributions? The current study utilized data from the British Lexicon Project (BLP) in which 78 participants made lexical decision judgments for 28,730 words over 16 h. We used linear mixed effects to detect practice-driven changes in the explanatory variance accounted for by a set of lexical predictors that included numerous indices of relative semantic richness, including imageability, the number of senses and average radius of co-occurrence (ARC). Results showed that practice was associated with decreasing effects of predictors such as word frequency and imageability. In contrast, ARC effects were only slightly diminished with repeated practice, and effects of the number of senses and the age of acquisition were unaffected by practice. We interpret our results within a framework in which variables may dynamically influence lexical processing and the post-lexical decision making mechanisms that also contribute to LDT performance.

The British Lexicon Project: Lexical decision data for 28,730 monosyllabic and disyllabic English words

We present a new database of lexical decision times for English words and nonwords, for which two groups of British participants each responded to 14,365 monosyllabic and disyllabic words and the same number of nonwords for a total duration of 16 h (divided over multiple sessions). This database, called the British Lexicon Project (BLP), fills an important gap between the Dutch Lexicon Project (DLP; Keuleers, Diependaele, & Brysbaert, Frontiers in Language Sciences. Psychology, 1, 174, 2010) and the English Lexicon Project (ELP; Balota et al., 2007), because it applies the repeated measures design of the DLP to the English language. The high correlation between the BLP and ELP data indicates that a high percentage of variance in lexical decision data sets is systematic variance, rather than noise, and that the results of megastudies are rather robust with respect to the selection and presentation of the stimuli. Because of its design, the BLP makes the same analyses possible as the DLP, offering researchers with a new interesting data set of word-processing times for mixed effects analyses and mathematical modeling. The BLP data are available at http://crr.ugent.be/blp and as Electronic Supplementary Materials.Electronic supplementary materialThe online version of this article (doi:10.3758/s13428-011-0118-4) contains supplementary material, which is available to authorized users.

Imageability and body–object interaction ratings for 599 multisyllabic nouns

We collected imageability and body-object interaction (BOI) ratings for 599 multisyllabic nouns. We then examined the effects of these variables on a subset of these items in picture-naming, word-naming, lexical decision, and semantic categorization. Picture-naming latencies were taken from the International Picture-Naming Project database (Szekely, Jacobsen, D'Amico, Devescovi, Andonova, Herron, et al. Journal of Memory and Language, 51, 247-250, 2004), word-naming and lexical decision latencies were taken from the English Lexicon Project database (Balota, Yap, Cortese, Hutchison, Kessler, Loftis, et al. Behavior Research Methods, 39, 445-459, 2007), and we collected semantic categorization latencies. Results from hierarchical multiple regression analyses showed that imageability and BOI separately accounted for unique latency variability in each task, even with several other predictor variables (e.g., print frequency, number of syllables and morphemes, age of acquisition) entered first in the analyses. These ratings should be useful to researchers interested in manipulating or controlling for the effects of imageability and BOI for multisyllabic stimuli in lexical and semantic tasks.

Rapid communication: Semantic size does not matter: “Bigger” words are not recognized faster

Sereno, O'Donnell, and Sereno (2009) reported that words are recognized faster in a lexical decision task when their referents are physically large than when they are small, suggesting that "semantic size" might be an important variable that should be considered in visual word recognition research and modelling. We sought to replicate their size effect, but failed to find a significant latency advantage in lexical decision for "big" words (cf. "small" words), even though we used the same word stimuli as Sereno et al. and had almost three times as many subjects. We also examined existing data from visual word recognition megastudies (e.g., English Lexicon Project) and found that semantic size is not a significant predictor of lexical decision performance after controlling for the standard lexical variables. In summary, the null results from our lab experiment--despite a much larger subject sample size than Sereno et al.--converged with our analysis of megastudy lexical decision performance, leading us to conclude that semantic size does not matter for word recognition. Discussion focuses on why semantic size (unlike some other semantic variables) is unlikely to play a role in lexical decision.

Location, location, location: How it affects the neighborhood (effect)

Reaction times in lexical decision are more sensitive to a words’ length and orthographic-neighborhood density when the stimulus is presented to the left visual field (LVF) than to the right visual field (RVF). We claim that the length effect is equivalent to the neighborhood effect, and propose a novel explanation of why the LVF, but not the RVF, is sensitive to density, based on different firing rates of abstract-letter representations encoding letters falling in the LVF versus RVF. We support this proposal with a large-scale implemented model of lexical decision utilizing spiking units, which provides a reasonable fit to the data from the English Lexicon Project under simulated central presentation, while replicating the observed hemifield asymmetries under simulated lateralized presentation.

Assessing the Usefulness of Google Books’ Word Frequencies for Psycholinguistic Research on Word Processing

In this Perspective Article we assess the usefulness of Google's new word frequencies for word recognition research (lexical decision and word naming). We find that, despite the massive corpus on which the Google estimates are based (131 billion words from books published in the United States alone), the Google American English frequencies explain 11% less of the variance in the lexical decision times from the English Lexicon Project (Balota et al., 2007) than the SUBTLEX-US word frequencies, based on a corpus of 51 million words from film and television subtitles. Further analyses indicate that word frequencies derived from recent books (published after 2000) are better predictors of word processing times than frequencies based on the full corpus, and that word frequencies based on fiction books predict word processing times better than word frequencies based on the full corpus. The most predictive word frequencies from Google still do not explain more of the variance in word recognition times of undergraduate students and old adults than the subtitle-based word frequencies.

Beyond single syllables: Large-scale modeling of reading aloud with the Connectionist Dual Process (CDP++) model

Most words in English have more than one syllable, yet the most influential computational models of reading aloud are restricted to processing monosyllabic words. Here, we present CDP++, a new version of the Connectionist Dual Process model ( Perry, Ziegler, & Zorzi, 2007). CDP++ is able to simulate the reading aloud of mono- and disyllabic words and nonwords, and learns to assign stress in exactly the same way as it learns to associate graphemes with phonemes. CDP++ is able to simulate the monosyllabic benchmark effects its predecessor could, and therefore shows full backwards compatibility. CDP++ also accounts for a number of novel effects specific to disyllabic words, including the effects of stress regularity and syllable number. In terms of database performance, CDP++ accounts for over 49% of the reaction time variance on items selected from the English Lexicon Project, a very large database of several thousand of words. With its lexicon of over 32,000 words, CDP++ is therefore a notable example of the successful scaling-up of a connectionist model to a size that more realistically approximates the human lexical system.

The French Lexicon Project: Lexical decision data for 38,840 French words and 38,840 pseudowords

The French Lexicon Project involved the collection of lexical decision data for 38,840 French words and the same number of nonwords. It was directly inspired by the English Lexicon Project (Balota et al., 2007) and produced very comparable frequency and word length effects. The present article describes the methods used to collect the data, reports analyses on the word frequency and the word length effects, and describes the Excel files that make the data freely available for research purposes. The word and pseudoword data from this article may be downloaded from http://brm.psychonomic-journals.org/content/supplemental.

Practice Effects in Large-Scale Visual Word Recognition Studies: A Lexical Decision Study on 14,000 Dutch Mono- and Disyllabic Words and Nonwords

In recent years, psycholinguistics has seen a remarkable growth of research based on the analysis of data from large-scale studies of word recognition, in particular lexical decision and word naming. We present the data of the Dutch Lexicon Project (DLP) in which a group of 39 participants made lexical decisions to 14,000 words and the same number of nonwords. To examine whether the extensive practice precludes comparison with the traditional short experiments, we look at the differences between the first and the last session, compare the results with the English Lexicon Project (ELP) and the French Lexicon Project (FLP), and examine to what extent established findings in Dutch psycholinguistics can be replicated in virtual experiments. Our results show that when good nonwords are used, practice effects are minimal in lexical decision experiments and do not invalidate the behavioral data. For instance, the word frequency curve is the same in DLP as in ELP and FLP. Also, the Dutch–English cognate effect is the same in DLP as in a previously published factorial experiment. This means that large-scale word recognition studies can make use of psychophysical and psychometrical approaches. In addition, our data represent an important collection of very long series of individual reaction times that may be of interest to researchers in other areas.