Brett Kessler

The English Lexicon Project.

The English Lexicon Project is a multiuniversity effort to provide a standardized behavioral and descriptive data set for 40,481 words and 40,481 nonwords. It is available via the Internet at elexicon.wustl.edu. Data from 816 participants across six universities were collected in a lexical decision task (approximately 3400 responses per participant), and data from 444 participants were collected in a speeded naming task (approximately 2500 responses per participant). The present paper describes the motivation for this project, the methods used to collect the data, and the search engine that affords access to the behavioral measures and descriptive lexical statistics for these stimuli.

Automatic Detection of Text Genre

As the text databases available to users become larger and more heterogeneous, genre becomes increasingly important for computational linguistics as a complement to topical and structural principles of classification. We propose a theory of genres as bundles of facets, which correlate with various surface cues, and argue that genre detection based on surface cues is as successful as detection based on deeper structural properties.

Computational dialectology in Irish Gaelic

Dialect groupings can be discovered objectively and automatically by cluster analysis of phonetic transcriptions such as those found in a linguistic atlas. The first step in the analysis, the computation of linguistic distance between each pair of sites, can be computed as Levenshtein distance between phonetic strings. This correlates closely with the much more laborious technique of determining and counting isoglosses, and is more accurate than the more familiar metric of computing Hamming distance based on whether vocabulary entries match. In the actual clustering step, traditional agglomerative clustering works better than the top-down technique of partitioning around medoids. When agglomerative clustering of phonetic string comparison distances is applied to Gaelic, reasonable dialect boundaries are obtained, corresponding to national and (within Ireland) provincial boundaries.

Spelling as statistical learning : Using consonantal context to spell vowels

Although English lacks 1-to-1 relationships between sounds and spellings, considering the context in which a phoneme occurs can often aid in selecting a spelling. For example, /a/ is typically spelled as a when it follows /w/, as in wand, but as o when it follows other consonants, as in pond. In 2 experiments, the authors asked whether childrens spellings of vowels in nonwords were affected by the following (Experiment 1) and preceding (Experiment 2) consonants. The participants in both experiments had spelling levels that ranged from kindergarten and 1st grade through high school. Children with higher levels of spelling skill took more advantage of context, and use of preceding context generally emerged earlier than use of following context. The results are interpreted within the framework of a statistical learning view of spelling and spelling development.

Statistical patterns in children's early writing

Many theories of spelling development claim that before children begin to spell phonologically, their spellings are random strings of letters. We evaluated this idea by testing young children (mean age=4 years 9 months) in Brazil and the United States and selecting a group of prephonological spellers. The spellings of this prephonological group showed a number of patterns that reflected things such as the frequencies of letters and bigrams in childrens language. The prephonological spellers in the two countries produced spellings that differed in some respects, consistent with their exposure to different written languages. We found no evidence for reportedly universal patterns in early spelling such as the idea that children write one letter for each syllable. Overall, our results reveal that early spellings that are not phonological are by no means random or universal and preserve certain patterns in the writing to which children have been exposed.

Use of morphology in spelling by children with dyslexia and typically developing children

In English and some other languages, spelling problems that arise at a phonological level can sometimes be solved through consideration of morphology. For example, children could infer that tuned should contain ann and thatfighting should contain a t because their stems include these letters. Children could thus avoid misspellings that might otherwise occur, such as “tudrd and “fiding.” We used a spelling-level match design to examine the extent to which children with dyslexia and younger typical children use morphology in this way. Both groups of children benefited from morphology to some extent, but not as much as they could have given their knowledge of the stems. Our results suggest that the spellings produced by older children with dyslexia are similar to those of younger normal children in their morphological characteristics, as well as in other ways.

Context sensitivity in the spelling of English vowels

Abstract Statistical analyses of English sound-to-spelling correspondences ( Kessler & Treiman, 2001 ) show that vowel spellings become more predictable, in some cases, when the preceding and following consonants are taken into account. In four experiments, we asked whether adult spellers are sensitive to such associations. We found evidence for sensitivity to associations involving both preceding and following consonants when examining adults’ spellings of vowels in nonwords (Experiments 1 and 2) and their substitution errors on vowels in real words (Experiment 3). The results show that phoneme-to-grapheme mapping is sensitive to a broader array of context than just rime context. Additional findings suggest that the context must be within the same syllable to be influential (Experiment 4). To the extent that rimes play a special role in spelling, this role may derive from the fact that associations between vowels and codas are more common in English than associations between vowels and onsets, not from spellers’ greater sensitivity to within-rime associations.

Learning about the letter name subset of the vocabulary: Evidence from US and Brazilian preschoolers

To examine the factors that affect the learning of letter names, an important foundation for literacy, we asked 318 US and 369 Brazilian preschoolers to identify each uppercase letter. Similarity of letter shape was the major determinant of confusion errors in both countries, and children were especially likely to interchange letters that were similar in shape as well as name. Errors were also affected by letter frequency, both general frequency and occurrence of letters in children’s own names. Differences in letter names and letter frequencies between English and Portuguese led to certain differences in the patterns of performance for children in the two countries. Other differences appeared to reflect US children’s greater familiarity with the conventional order of the alphabet. Boys were overrepresented at the low end of the continuum of letter name knowledge, suggesting that some boys begin formal reading instruction lacking important foundational skills. A child’s ability to identify the letters of the alphabet by name is one of the best predictors of how readily he or she will learn to read. Kindergarten letter identification accounts for nearly one-third of the variance in reading ability in Grades 1 to 3, and it is almost as successful at predicting later reading skill as an entire reading readiness test (Snow, Burns, & Griffin, 1998). Knowledge of letter names aids would-be readers and spellers in several ways (see Foulin, 2005). It helps them make some sense of printed words such as jail, where the entire name of one or more of the letters is heard in the spoken word. In addition, letter name knowledge helps children learn about the sound-symbolizing function of letters, because the phoneme that a letter represents is usually heard in the letter’s name. Effects of letter name knowledge on reading, spelling, and letter sound knowledge have been documented in languages as diverse as English (McBride-Chang, 1999; Treiman, Tincoff, Rodriguez, Mouzaki, & Francis, 1998), Portuguese (Abreu & Cardoso-Martins, 1998), and Hebrew (Levin, Patel, Margalit, & Barad, 2002). Given the foundational role of alphabet knowledge in literacy development, it is important to understand the processes involved in letter name learning. In the present study, we explore the hypothesis that these are the same processes

IS ENGLISH SPELLING CHAOTIC? MISCONCEPTIONS CONCERNING ITS IRREGULARITY

An overview of the goals of English orthography counters the misconception that its spelling is chaotic and unprincipled. Direct representation of the speaker’s phonemes is not its only goal. But even the sound-to-letter correspondences are not as inconsistent as widely believed. A survey of first-grade text vocabulary shows that spelling consistency is increased significantly if one takes into account the position of the phoneme within the syllable and the identity of the phonemes in the environment. Environmental influences within the rime are especially important. Understanding these patterns may reduce the complexity of spelling for learners and those with spelling problems.

Vowel Processing During Silent Reading: Evidence from Eye Movements.

Two eye movement experiments examined whether skilled readers include vowels in the early phonological representations used in word recognition during silent reading. Target words were presented in sentences preceded by parafoveal previews in which the vowel phoneme was concordant or discordant with the vowel phoneme in the target word. In Experiment 1, the orthographic vowel differed from the target in both the concordant and discordant preview conditions. In Experiment 2, the vowel letters in the preview were identical to those in the target word. The phonological vowel was ambiguous, however, and the final consonants of the previews biased the vowel phoneme either toward or away from the targets vowel phoneme. In both experiments, shorter reading times were observed for targets preceded by concordant previews than by discordant previews. Implications for models of word recognition are discussed.