Patterns Of Word Use Research Articles

Using computer-scored measures of emotion and style to discriminate among disputed and undisputed Pauline and non-Pauline epistles.

The Dictionary of Affect in Language that allows measurement of Pleasantness, Activation, and Imagery in texts and a computer program that provides several additional stylistic measures were used to score samples from Disputed (n = 22 samples) and Undisputed (n = 40) Pauline epistles and from Other New Testament epistles (n= 16). All samples came from an English translation. Several significant mean differences were noted between samples from Disputed and Undisputed epistles. A discriminant function predicting Disputed or Undisputed authorship limited to five predictors was 85% successful in assigning samples to either category. The majority of samples from Other epistles were classified as Disputed, i.e., less likely to have been written by Paul. Undisputed Pauline samples were predicted to be those with lower Imagery, shorter words, less frequent words, greater repetitiveness, and greater Pleasantness. There were significant differences in patterns of word use (vocabulary) between Disputed and Undisputed samples.

Patterns, Particularism and Seeing the Similarity

I argue for a form of particularism from a reading of Wittgenstein's critique of the idea that word use is governed by rules. In place of the idea that word use is driven by rules, I show how the patterns of word use, in virtue of which we express our reasons, emerge from our ongoing practice, including our practice of seeing things as similar. I argue that the notion of seeing the similarities is primitive for Wittgenstein. The remark, ‘this and similar things are called “games”’ does not signal a form of ignorance. It signals the constitutive role that speakers, as judges, have to play in the metaphysics of the patterns of word use.

Patterns of Natural Language Use: Disclosure, Personality, and Social Integration

When people write about their deepest thoughts and feelings about an emotionally significant event, numerous benefits in many domains (e.g., health, achievement, and well-being) result. As one step in understanding how writing achieves these effects, we have developed a computer program that provides a “fingerprint” of the words people use in writing or in natural settings. Analyses of text samples indicate that particular patterns of word use predict health and also reflect personality styles. We have also discovered that language use in the laboratory writing paradigm is associated with changes in social interactions and language use in the real world. The implications for using computer-based text analysis programs in the development of psychological theory are discussed.

ENGLISH LETTER FREQUENCIES AS FOUND IN WHISSELL'S PARSIMONIOUS SAMPLING OF ENGLISH WORDS

An estimate of English letter frequencies is reported. The major part of the textual sample was 320,780 words, comprising 6.505 different words from 1998 work of Whissell as parsimoniously describing modern English patterns of word use. To this sample has been added another of 1,000 words to represent English words generally rarer than the 6,505. Comparison with estimated English letter frequencies from a sample over 160 years old suggested strong stability in letter frequencies over time. The old sample did, however, tend to overestimate the occurrence of rare letters (j, q, x, and z). An application of the findings to the teaching of letter recognition is explored.

A cache-based natural language model for speech recognition

Speech-recognition systems must often decide between competing ways of breaking up the acoustic input into strings of words. Since the possible strings may be acoustically similar, a language model is required; given a word string, the model returns its linguistic probability. Several Markov language models are discussed. A novel kind of language model which reflects short-term patterns of word use by means of a cache component (analogous to cache memory in hardware terminology) is presented. The model also contains a 3g-gram component of the traditional type. The combined model and a pure 3g-gram model were tested on samples drawn from the Lancaster-Oslo/Bergen (LOB) corpus of English text. The relative performance of the two models is examined, and suggestions for the future improvements are made. >