Alternative Words Research Articles

Method of speech recognition using empirically determined word candidates

A method for performing speech recognition can include determining a recognition result for received user speech. The recognition result can include recognized text and a corresponding confidence score. The confidence score of the recognition result can correspond to a predetermined minimum threshold. If the confidence score does not exceed the predetermined minimum threshold, the user can be presented with at least one empirically determined alternate word candidate corresponding to the recognition result.

Word order in Croatian Sign Language

This paper focuses on the basic word order of Croatian Sign Language (HZJ) and factors that permit alternative word orders to occur in sentences and in context. Although they are unrelated languages, the basic word order in HZJ is the same as in spoken Croatian: SVO.One of the factors allowing alternative word orders in context is information status (old or new), which influences constituent placement, as in other languages. HZJ has a tendency to omit old, previously mentioned information, usually the Subject, and the part that is expressed is the new information (Rheme). When old information is expressed, it appears at the beginning of the sentence, preceding the Rheme.Like other languages, HZJ word order can be influenced by the nature of the arguments (Subject, Object) as well as the type of Verb. Sentences with ‘reversible’ arguments (i.e. both are animate and could be agents) tend to use the basic word order, whereas those with nonreversible arguments allow more variable word order. Basic word order also occurs more often with plain verbs (those that do not agree with their arguments). Agreeing and spatial verbs use other word orders in addition to SVO, including the tendency to position Verbs at the end of sentences.Investigation on the interaction of word order and the grammatical usage of facial expressions and head positions (nonmanual marking) indicates that nonmanual markings have pragmatic roles, and could have syntactic functions which await further research.

Lexical statistics of competition in L2 versus L1 listening

Spoken-word recognition involves multiple activation of alternative word candidates and competition between these alternatives. Phonemic confusions in L2 listening increase the number of potentially active words, thus slowing word recognition by adding competitors. This study used a 70,000-word English lexicon backed by frequency statistics from a 17,900,000-word corpus to assess the competition increase resulting from two representative phonemic confusions, one vocalic (ae/E) and one consonantal (r/l), in L2 versus L1 listening. The first analysis involved word embedding. Embedded words (cat in cattle, rib in ribbon) cause competition, which phonemic confusion can increase (cat in kettle, rib in liberty). The average increase in number of embedded words was 59.6 and 48.3 temporary ambiguity. Even when no embeddings are present, multiple alternatives are possible: para- can become parrot, paradise, etc., but also pallet, palace given /r/–/l/ confusion. Phoneme confusions (vowel or consonant) in first or second position in the word approximately doubled the number of activated candidates; confusions later in the word increased activation by on average 53 third, 42 confusions significantly increase competition for L2 compared with L1 listeners.

Self-Corrected Elaboration Effects on Incidental Memory

Subjects performed an orienting task involving 3 conditions followed by an unexpected free recall test. The conditions were designed to force 3 types of corrected elaborations: Generated Correction, Chosen Correction, and No Correction. In the Generated Correction condition the subjects were presented with a target word (e.g., Baby) and a bizarre sentence frame (e.g., "____drinks beer.") and asked to correct the target to a congruous word (e.g., Uncle) to make a common sentence. In the Chosen Correction condition, the subjects were presented with a target and its bizarre sentence frame and asked to choose one of the alternative congruous words (e.g., Uncle, Aunt) to make a common sentence. In the No Correction condition, the subjects were presented with a target and its bizarre sentence frame and asked to rate the congruity of each target to its sentence frame. Generated Correction led to a better performance than Chosen Correction and No Correction, but a difference between the last two correction types was not found. These results were interpreted as showing that, by generating correct information, self-corrected elaboration led to facilitation of incidental memory.

The Secret Life of the Dyslexic Child: How She Thinks, How He Feels, How They Can Succeed

The Secret Life of the Dyslexic Child: How She Thinks, How He Feels, How They Can Succeed

Do alternative names block young and older adults’ retrieval of proper names?

This study evaluates whether tip of the tongue experiences (TOTs) are caused by a more accessible word which blocks retrieval of the target word, especially for older adults. In a “competitor priming” paradigm, young and older adults produced the name of a famous character (e.g., Eliza Doolittle) in response to a question and subsequently named a picture of a famous actor or actress depicting this character (e.g., Audrey Hepburn as Eliza Doolittle). Older adults produced more TOTs than young adults, but prior production of a related character name did not affect TOTs, although it did reduce incorrect responses. There were no age differences in knowledge of films and TV and thus the age-related increase in TOTs is not because older adults have more relevant knowledge. The findings are compatible with models in which alternate words are a consequence not a cause of TOT.

When and how do listeners relate a sentence to the wider discourse? Evidence from the N400 effect

In two ERP experiments, we assessed the impact of discourse-level information on the processing of an unfolding spoken sentence. Subjects listened to sentences like Jane told her brother that he was exceptionally quick/slow, designed such that the alternative critical words were equally acceptable within the local sentence context. In Experiment 1, these sentences were embedded in a discourse that rendered one of the critical words anomalous (e.g. because Jane’s brother had in fact done something very quickly). Relative to the coherent alternative, these discourse-anomalous words elicited a standard N400 effect that started at 150–200 ms after acoustic word onset. Furthermore, when the same sentences were heard in isolation in Experiment 2, the N400 effect disappeared. The results demonstrate that our listeners related the unfolding spoken words to the wider discourse extremely rapidly, after having heard the first two or three phonemes only, and in many cases well before the end of the word. In addition, the identical nature of discourse- and sentence-dependent N400 effects suggests that from the perspective of the word-elicited comprehension process indexed by the N400, the interpretive context delineated by a single unfolding sentence and a larger discourse is functionally identical.

An investigation of the unit of activation in on-line inferences during text processing: word-unit or proposition-unit?

This study investigated the unit of activation in on-line inferences. To explore whether the activation of inferences is a word-unit or a proposition-unit, a meaningfulness-decision task was devised as a priming measure. The task required participants to decide whether a series of short sentences was meaningful or not. The result of Experiment 1 suggested that only proposition-units were activated. A priming effect was observed in targets which described inferences both at word-level and proposition-level, but not in targets which described inferences at word-level only. In Experiment 2, a meaningfulness-decision task was administered to investigate word-based priming and to make sure that the result of Experiment 1 was not caused by characteristics of the task. In Experiment 3, a priming effect was found in targets which described inferences using alternative perspectives and words. These results suggest that the unit of activation in on-line inferences is proposition-units and that these activations relate to the proposition-semantic level, not the word or single-concept level.

High-level approaches to confidence estimation in speech recognition

We describe some approaches to estimating confidence scores for the words output by a speech recognizer. By high-level we mean that the proposed measures do not rely on decoder specific side and so should find more general applicability than measures that have been developed for specific recognizers. Our main approach is to attempt to decouple the language modeling and acoustic modeling in the recognizer in order to generate independent information from these two sources that can then be used for estimation of confidence. We isolate these two information sources by using a phone recognizer working in parallel with the word recognizer. A set of techniques for estimating confidence measures using the phone recognizer output in conjunction with the word recognizer output is described. The most effective of these techniques is based on the construction of metamodels, which generate alternative word hypotheses for an utterance. An alternative approach requires no other recognizers or extra information for confidence estimation and is based on the notion that a word that is semantically distant from the other decoded words in the utterance is likely to be incorrect. We describe a method for constructing semantic similarities between words and hence estimating a confidence. Results using the UK version of the Wall Street Journal are given for each technique.

Use of cognitive interview techniques in the development of nutrition surveys and interactive nutrition messages for low-income populations

The effectiveness of dietary surveys and educational messages is dependent in part on how well the target audience's information processing needs and abilities are addressed. Use of pilot testing is helpful; however, problems with wording and language are often not revealed. Cognitive interview techniques offer 1 approach to assist dietitians in understanding how audiences process information. With this method, respondents are led through a survey or message and asked to paraphrase items; discuss thoughts, feelings, and ideas that come to mind; and suggest alternative wording. As part of a US Department of Agriculture-funded nutrition education project, 23 cognitive interviews were conducted among technical community college students in North Carolina. Interview findings informed the development of tailored computer messages and survey questions. Better understanding of respondents’ cognitive processes significantly improved the language and approach used in this intervention. Interview data indicated 4 problem areas: vague or ineffective instructions, confusing questions and response options, variable interpretation of terms, and misinterpretation of dietary recommendations. Interviews also provided insight into the meaning of diet-related stages of change. These findings concur with previous research suggesting that cognitive interview techniques are a valuable tool in the formative evaluation and development of nutrition surveys and materials. J Am Die Assoc. 2002; 102: 690-696.

Measuring party identification: an exploratory study with focus groups

Warren E. Miller is irretrievably associated with the concept of party identification. The concept was pivotal in his own research and is a recurring theme in the study of voting behaviour across the world. Throughout his career Miller stressed that analysts should take great care when defining and measuring the concept. In this paper, evidence from focus groups is used to explore the form of question most likely to enable voters to distinguish between their enduring party identification and their current political preference. The study shows that the traditional party identification question employed in the British Election Study is inadequate for the purpose. In response to Miller’s call to improve the validity of the question, an alternative question wording is suggested.

Improved language modelling through better language model evaluation measures

This paper explores the interaction between a language model’s perplexity and its effect on the word error rate of a speech recognition system. Much recent research has indicated that these two measures are not as well correlated as was once thought, and many examples exist of models which have a much lower perplexity than the equivalent N -gram model, yet lead to no improvement in recognition accuracy. This paper investigates the reasons for this apparent discrepancy. Perplexity’s calculation is based solely on the probabilities of words contained within the test text; it disregards the probabilities of alternative words which will be competing with the correct word within the decoder. It is shown that by considering the probabilities of the alternative words it is possible to derive measures of language model quality which are better correlated with word error rate than perplexity is. Furthermore, optimizing language model parameters with respect to these new measures leads to a significant reduction in the word error rate.

A multinomial model for short-term priming in word identification.

A simple multinomial model for short-term priming in perceptual word identification is presented. In the experiments to which the model is applied, prime words are presented just prior to a flashed target word, and subjects must decide which of 2 alternative words matches the target. The model assumes that on some proportion of trials, confusion among the words leads to the decision being based on 1 of the prime words instead of the target. In addition, it is assumed that subjects sometimes discount a prime that matches 1 of the test alternatives and so choose the alternative that does not match. With these assumptions, the model fits the data from 5 experiments (including 4 used to develop the model known as ROUSE [responding optimally with unknown sources of evidence]; D. E. Huber, R. Shiffrin, K. Lyle, & K. Ruys, 2001). The multinomial model fits the data about as well as the ROUSE model and so should lead to further development and critical testing of both models.

Understanding assurance services reports: A user perspective

In 1997 the International Auditing Practices Committee (IAPC) issued an exposure draft (ED) outlining a framework for providing assurance services. It was recommended that the level of assurance provided be expressed on a continuum from absolute to low. The user was to infer this assurance from, among other things, the nature of the engagement and the description of the work undertaken. Using a 2x2 between‐subjects research design this study examined whether assurance report users would differ in their identified level of assurance as a result of the description of the nature of the engagement (historical versus prospective) and the amount of work performed (higher level versus lower level). It was found that shareholders desired such assurance reports, but many did not find the format suggested by the IAPC useful for their decision‐making. A higher level of assurance was found to be attached to historical compared with prospective reports, but no difference was found for reports attempting to communicate higher versus lower levels of assurance as a result of the description of work performed. These results suggest that the IAPC should consider the use of alternative wording in order to accurately communicate the level of work performed in assurance engagements.

Early detection of developmental and behavioral problems.

1. Frances Page Glascoe, PhD* 1. 2. *Associate Professor of Pediatrics, Division of Child Development, Vanderbilt University School of Medicine, Nashville, TN. After completing this article, readers should be able to: 1. List the percentage of children who drop out of school and have undetected disabilities or known environmental risk factors. 2. Describe the ways in which early intervention is effective. 3. Delineate methods of detecting disabilities and development delays. 4. Describe the percentage of children in whom assessment tools can detect disabilities correctly. 5. Determine how often children should undergo developmental testing. 6. Describe the role of parents in detecting and addressing developmental and behavioral problems. Approximately 15% to 18% of children in the United States have developmental or behavioral disabilities. An additional 7% to 10% experience substantive school failure and drop out before completing high school. Overall, one in four children has serious psychosocial problems. To ensure that these children are detected early and their difficulties addressed, the American Academy of Pediatrics’ Committee on Children with Disabilities recommends that pediatricians use validated screening tools at each health supervision visit. Many pediatricians find it difficult to comply with this recommendation because of minimal reimbursement, young patients’ limited compliance with requests to stack blocks or answer questions, time constraints, and concerns about the accuracy and length of well-known screening tools. Finally, children who are at environmental risk for developmental delays and subsequent school failure due to poverty, limited parental education, and similar risk factors do not always receive health supervision visits. Accordingly, they are unavailable at times when pediatricians typically are most vigilant in their search for developmental problems. ### UNDERDETECTION Most physicians depend on clinical judgment rather than screening tools. Unfortunately, research shows that clinical judgment detects fewer than 30% of children who have mental retardation, learning disabilities, language impairments, and other developmental disabilities. Clinical judgment also identifies fewer than 50% of children who have serious emotional and behavioral disturbances. Use of improved classification systems, such as the Primary …

Pronunciation modeling by sharing Gaussian densities across phonetic models

Conversational speech exhibits considerable pronunciation variability, which has been shown to have a detrimental effect on the accuracy of automatic speech recognition. There have been many attempts to model pronunciation variation, including the use of decision trees to generate alternate word pronunciations from phonemic baseforms. Use of pronunciation models during recognition is known to improve accuracy. This paper describes the incorporation of pronunciation models into acoustic model training in addition to recognition. Subtle difficulties in the straightforward use of alternatives to canonical pronunciations are first illustrated: it is shown that simply improving the accuracy of the phonetic transcription used for acoustic model training is of little benefit. Acoustic models trained on the most accurate phonetic transcriptions result in worse recognition than acoustic models trained on canonical baseforms. Analysis of this counterintuitive result leads to a new method of accommodating nonstandard pronunciations: rather than allowing a phoneme in the canonical pronunciation to be realized as one of a few distinct alternate phones, the hidden Markov model (HMM) states of the phoneme’s model are instead allowed to share Gaussian mixture components with the HMM states of the model(s) of the alternate realization(s). Qualitatively, this amounts to making a soft decision about which surface form is realized. Quantitatively, experiments show that this method is particularly well suited for acoustic model training for spontaneous speech: a 1.7 %(absolute) improvement in recognition accuracy on the Switchboard corpus is presented.

Method and system for buffering recognized words during speech recognition

A method and system for editing words that have been misrecognized. The system allows a speaker to specify a number of alternative words to be displayed in a correction window by resizing the correction window. The system also displays the words in the correction window in alphabetical order. A preferred system eliminates the possibility, when a misrecognized word is respoken, that the respoken utterance will be again recognized as the same misrecognized word. The system, when operating with a word processor, allows the speaker to specify the amount of speech that is buffered before transferring to the word processor.

Myths of Neuropsychology: Further Considerations

The publication of “Myths of Neuropsychology’” (Dodrill, 1997) has stimulated three responses, each of which takes issue with points in the original publication. These three responses especially address Myth #4 (“Above-average performances on neuropsychological tests are expected when intellectual abilities are above average’”) and to a lesser degree Myth #1 (“We have a good knowledge of the constructs that our tests measure’”). The present paper reviews the thoughts and new data which these colleagues have provided in response to the original paper, and it also presents additional data bearing on the issues at hand. Based on the composite of information now available, it is conceded that Myth #4 was not accurately worded in the original paper. This was a mistake, and an alternate wording for Myth #4 is offered. Other changes in the original publication do not appear to be required. In order to resolve many of the issues at hand and thereby advance our profession, research in a number of areas is needed, many of which are specifically identified in the present paper.

COOP Charts in French: translation and preliminary data on instrument properties.

This paper describes the procedure used to translate the COOP Charts into French and provides preliminary information on the instrument's acceptability, reliability and validity. The charts were translated in several steps: seven initial translations were combined into a first pilot version, which was then tested for acceptability, clarity and alternative wordings in two convenience samples taken from the general population (n = 53). The modified version was then reviewed by a lay panel and another translator and submitted by mail to 209 congress participants to test several construct validity hypotheses through known-groups comparisons. A panel of public health professionals discussed the content validity of the charts. Finally, test-retest reliability and concurrent validity with SF-36 Health Survey scores were examined among 65 patients with end-stage renal disease. The translation process identified a wide variability in translation options for several items. The acceptability of the charts was excellent. The test-retest correlations ranged from 0.60 to 0.87. Content validity appeared to be appropriate, except for the chart on 'social support', which combines the questions of need and availability of social support. The utility of illustrations was questioned by some respondents: many claimed not to have used the illustrations in selecting their response, while others found them to be not expressive enough. Most preliminary tests of construct validity were consistent with theory. This French translation of the COOP Charts appears to be ready for more extensive testing in the intended target population of ambulatory patients.

391 Neural interactions during synactic and/or semantic processing of compound sentences: An ERP study

A lexical decision experiment investigated hemisphere asymmetries in resolving lexical ambiguity within a sentence context. Sentences that biased a single meaning (either dominant or subordinate) of sentence-final ambiguous words were followed by a lateralized target related to the sentence-congruent or -incongruent meaning of the ambiguous word, or an unrelated word. In the RVF sentence-congruent targets were facilitated, while incongruent targets were not primed. In contrast, related targets were facilitated in the LVF, regardless of sentence context. This suggests that selecting the contextually appropriate word meaning requires the left hemisphere, and supports a right hemisphere role in maintaining alternate word senses.