Word Overlap Research Articles

The Lexical Restructuring Hypothesis and Graph Theoretic Analyses of Networks Based on Random Lexicons

The mental lexicon of words used for spoken word recognition has been modeled as a complex network or graph. Do the characteristics of that graph reflect processes involved in its growth (M. S. Vitevitch, 2008) or simply the phonetic overlap between similar-sounding words? Three pseudolexicons were generated by randomly selecting phonological segments from a fixed set. Each lexicon was then modeled as a graph, linking words differing by one segment. The properties of those graphs were compared with those of a graph based on real English words. The properties of the graphs built from the pseudolexicons matched the properties of the graph based on English words. Each graph consisted of a single large island and a number of smaller islands and hermits. The degree distribution of each graph was better fit by an exponential than by a power function. Each graph showed short path lengths, large clustering coefficients, and positive assortative mixing. The results suggest that there is no need to appeal to processes of growth or language acquisition to explain the formal properties of the network structure of the mental lexicon. These properties emerged because the network was built based on the phonetic overlap of words.

Promoting Generalization of Reading: A Comparison of Two Fluency-Based Interventions for Improving General Education Student’s Oral Reading Rate

The generalization effects of empirically-supported fluency-based reading interventions have been largely ignored. The purpose of this study was to evaluate the effects of two fluency-based reading interventions, Repeated Readings and Multiple Exemplars, on elementary-aged children’s immediate and generalized oral reading fluency rate. Using a within-subjects group design, a total of 42 second (n = 25) and fourth (n = 17) grade general education students were administered both interventions and their generalized responding to passages containing high word and medium word overlap was assessed. Results indicated that children’s oral fluency on intervention passages was significantly greater during the Repeated Readings intervention. However, children’s oral reading fluency on generalization passages containing medium word overlap was significantly greater following the Multiple Exemplars intervention. This difference was however possibly due to the Multiple Exemplar medium word overlap passage being easier for students to read. No significant differences between the two interventions were observed in children’s oral reading fluency on generalization passages containing high word overlap. Implications and limitations of the study are discussed in relation to improving students’ reading fluency on generalization passages.

Metrics for the scope of a collection

AbstractSome collections cover many topics, while others are narrowly focused on a limited number of topics. We introduce the concept of the “scope” of a collection of documents and we compare two ways of measuring it. These measures are based on the distances between documents. The first uses the overlap of words between pairs of documents. The second measure uses a novel method that calculates the semantic relatedness to pairs of words from the documents. Those values are combined to obtain an overall distance between the documents. The main validation for the measures compared Web pages categorized by Yahoo. Sets of pages sampled from broad categories were determined to have a higher scope than sets derived from subcategories. The measure was significant and confirmed the expected difference in scope. Finally, we discuss other measures related to scope.

Classification of Text Documents

The exponential growth of the internet has led to a great deal of interest in developing useful and efficient tools and software to assist users in searching the Web. Document retrieval, categorization, routing and filtering can all be formulated as classification problems. However, the complexity of natural languages and the extremely high dimensionality of the feature space of documents have made this classification problem very difficult. We investigate four different methods for document classification: the naive Bayes classifier, the nearest neighbour classifier, decision trees and a subspace method. These were applied to seven-class Yahoo news groups (business, entertainment, health, international, politics, sports and technology) individually and in combination. We studied three classifier combination approaches: simple voting, dynamic classifier selection and adaptive classifier combination. Our experimental results indicate that the naive Bayes classifier and the subspace method outperform the other two classifiers on our data sets. Combinations of multiple classifiers did not always improve the classification accuracy compared to the best individual classifier. Among the three different combination approaches, our adaptive classifier combination method introduced here performed the best. The best classification accuracy that we are able to achieve on this seven-class problem is approximately 83%, which is comparable to the performance of other similar studies. However, the classification problem considered here is more difficult because the pattern classes used in our experiments have a large overlap of words in their corresponding documents.

The Effectiveness of the Keyword Method for Receptive and Productive Foreign Vocabulary Learning in the Elderly

This paper reports on an experiment in which a group of elderly healthy active female subjects were required to learn 20 Spanish vocabulary items using the keyword method of foreign language learning. This involves using a mediator to link an English word to its Spanish target. As West (1995) has recently pointed out, despite considerable evidence that the keyword method is effective for normal populations, there is little work studying its effectiveness in elderly populations. The present study shows that the keyword method significantly enhances recall of the English word given its Spanish equivalent (receptive learning) and significantly enhances the learning of Spanish equivalents of English words (productive learning) using a “soft” criterion of correctness, compared to a control group given no instruction on how to learn. This latter finding fits in with previous studies showing the effectiveness of the keyword method in productive learning to the extent that the keyword and the target word overlap (Pressley, Levin, Hall, Hiller, & Berry, 1980). The methodological problems of investigating the keyword method where images as well as the keyword are provided are discussed, as are the practical implications of the study.

How Text Difficulty and Reader Skill Interact to Produce Differential Reliance on Word and Content Overlap in Reading Transfer

Four experiments examined the factors that influence across-text transfer for children. Transfer was indicated by increases in the reading speed and accuracy of a second text following reading of different first texts. The first texts were related to the second by overlap in words only, in content only, in words and content, or in neither words nor content. Results indicated that the extent to which readers benefited from word or content overlap depended on the interaction of reader skill with text difficulty. Children who read texts that were easy for them showed transfer only when a pair of stories shared content. However, when children read stories that were difficult for them, they also showed transfer when words alone were shared by a pair of stories. The results are discussed in terms of how easy and difficult texts are represented in memory and retrieved to produce transfer.

Summation Priming and Coarse Semantic Coding in the Right Hemisphere

Abstract There are now numerous observations of subtle right hemisphere (RH) contributions to language comprehension. It has been suggested that these contributions reflect coarse semantic coding in the RH. That is, the RH weakly activates large semantic fields-including concepts distantly related to the input word-whereas the left hemisphere (LH) strongly activates small semantic fields-limited to concepts closely related to the input (Beeman, 1993a,b). This makes the RH less effective at interpreting single words, but more sensitive to semantic overlap of multiple words. To test this theory, subjects read target words preceded by either "Summation" primes (three words each weakly related to the target) or Unrelated primes (three unrelated words), and target exposure duration was manipulated so that subjects correctly named about half the target words in each hemifield. In Experiment 1, subjects benefited more from Summation primes when naming target words presented to the left visual field-RH (Ivf-RH) than when naming target words presented to the right visual field-LH (rvf-LH), suggesting a RH advantage in coarse semantic coding. In Experiment 2, with a low proportion of related prime-target trials, subjects benefited more from "Direct" primes (one strong associate flanked by two unrelated words) than from Summation primes for rvf-LH target words, indicating that the LH activates closely related information much more strongly than distantly related information. Subjects benefited equally from both prime types for Ivf-RH target words, indicating that the RH activates closely related information only slightly more strongly, at best, than distantly related information. This suggests that the RH processes words with relatively coarser coding than the LH, a conclusion consistent with a recent suggestion that the RH coarsely codes visual input (Kosslyn, Chabris, Mar-solek, & Koenig, 1992).

First and second moment of counts of words in random texts generated by Markov chains.

An exact expression for the variance of random frequency that a given word has in text generated by a Markov chain is presented. The result is applied to periodic Markov chains, which describe the protein-coding DNA sequences better than simple Markov chains. A new solution to the problem of word overlap is proposed. It was found that the expected frequency and overlapping properties determine most of the variance. The expectation and variance of counts for triplets are compared with experimental counts in Escherichia coli coding sequences.

Curriculum bias and reading achievement test performance

The overlap of words specifically taught in reading textbooks with the contents of standardized reading achievement tests may be a source of bias that is frequently overlooked in psychoeducational assessments. This study compares the standardized achievement test performance of 62 second graders receiving instruction in two different reading curricula (Open Court and Houghton-Mifflin) to determine whether either curriculum generates different quantitative estimates of reading achievement. Reading subtest scores derived from the Kaufman Test of Educational Achievement-Brief Form (K-TEA), the Wide Range Achievement Test-Revised (WRAT-R), and the Reading Recognition and Reading Comprehension subtests from the Peabody Individual Achievement Test (PIAT) were examined. Grade level equivalents and scaled scores from the California Achievement Test (CAT) were also examined. Three Curriculum × Test repeated measures ANOVAs were conducted using grade level scores (2×7), standard scores (2×4), and CAT scaled scores (2 × 5) as dependent measures. A significant Curriculum × Test interaction was identified, suggesting differences among tests in estimates of reading ability as a function of the reading program.

Repeated Reading and Reading Fluency in Learning Disabled Children

THIS STUDY investigated whether improved fluency and comprehension across different stories in repeated reading depend on the degree of word overlap among passages and whether repeated reading is more effective than an equivalent amount of nonrepetitive reading. Nonfluent, learning disabled students read passages presented and timed by a computer under three different conditions. In Conditions 1 and 2 (repeated reading) the same passage was read four times before proceeding to a new story in the next session. Stories in Condition 2 contained three times as many overlapping words as stories in Condition 1. In Condition 3 (nonrepetitive reading), each of the four passages in a session was different. Results suggest that over short periods of time, increases in reading speed with the repeated reading method depend on the amount of shared words among stories, and that if stories have few shared words, repeated reading is not more effective for improving speed than an equivalent amount of nonrepetitive reading.

Profile of a dictionary compiled from scanning over one million words of surgical pathology narrative text

An anatomic pathology natural language dictionary (LEXICON) has evolved over a nine-year period, a result of scanning over one million words of narrative text from tissue examination request forms and surgical pathology reports. The text is parsed into individual words which are looked up in LEXICON and flagged by action codes which determine usage in constructing a KWIC index file and an on-line database retrievable by keywords. The LEXICON now resides on an IBM 370 168 system and has survived several transfers between computer systems. An update program is used after each batch of narrative text is scanned to modify LEXICON. LEXICON now contains 24,228 medical and nonmedical terms, 24.8% are errors (misspellings), 45.9% are keywords retrievable on and off line, 52.2% of the words are cross-referenced to a supplementary word. A preliminary study shows that many of the “nonmedical” terms in LEXICON carry significant medical information, and that there is considerable overlap of medical words among LEXICON, SNOMED, and ICDA-8. Our LEXICON appears to be an intermediate step in the process of evolving an algorithm capable of “understanding” medical narrative text.