Entire Word Research Articles

Compressive Sensing for Missing Data Imputation in Noise Robust Speech Recognition

An effective way to increase the noise robustness of automatic speech recognition is to label noisy speech features as either reliable or unreliable (missing), and to replace (impute) the missing ones by clean speech estimates. Conventional imputation techniques employ parametric models and impute the missing features on a frame-by-frame basis. At low signal-to-noise ratios (SNRs), these techniques fail, because too many time frames may contain few, if any, reliable features. In this paper, we introduce a novel non-parametric, exemplar-based method for reconstructing clean speech from noisy observations, based on techniques from the field of Compressive Sensing. The method, dubbed sparse imputation, can impute missing features using larger time windows such as entire words. Using an overcomplete dictionary of clean speech exemplars, the method finds the sparsest combination of exemplars that jointly approximate the reliable features of a noisy utterance. That linear combination of clean speech exemplars is used to replace the missing features. Recognition experiments on noisy isolated digits show that sparse imputation outperforms conventional imputation techniques at SNR = -5 dB when using an ideal `oracle' mask. With error-prone estimated masks sparse imputation performs slightly worse than the best conventional technique.

Does word length affect speech onset latencies when producing single words?

Most models of spoken production predict that shorter utterances should be initiated faster than longer ones. However, whether word-length effects in single word production exist is at present controversial. A series of experiments did not find evidence for such an effect. First, an experimental manipulation of word length in picture naming showed no latency differences. Second, Dutch and English speakers named 2 sets of either objects or words (monosyllabic names in Dutch and disyllabic names in English or vice versa). A length effect, which should manifest itself as an interaction between object set and response language, emerged in word naming but not in picture naming. Third, distractors consisting of the final syllable of disyllabic object names speeded up responses, but at the same time, no word-length effect was found. These results suggest that before the response is initiated, an entire word has been phonologically encoded, but only its initial syllable is placed in an articulatory buffer.

An investigation of correlations between different environmental assessments and risk assessment

The increasing deterioration of environment is affecting our entire word. Environmental management plays a crucial role in maintaining environmental sustainability. While environmental management is complicated due to the complex natural ecosystem which is poorly understood and ill-structured, and significant uncertainties. A variety of environment assessment processes, e.g., environmental impact assessment (EIA), regional strategic environmental assessment (R-SEA), the precautionary principle, sustainability assessment (SA) etc., widely used in environmental management are reviewed and their correlations with human health risk assessment (HHRA) and ecological risk assessment (ERA) are investigated. The results of this paper facilitate the decision makers selecting proper assessment framework and/or some combinations of them in environmental management.

Conjunction errors and semantic transparency

Memory conjunction errors occur when aspects of two different events are falsely recognized or recalled as having occurred as parts of the same event. One theoretical account of conjunction errors is rooted in traditional dual-process models of recognition judgments, in which responses are based on an item's familiarity or the retrieval of recollected details associated with the encoding of that item. We manipulated the familiarity of test probes by varying their semantic overlap with studied items, taking advantage of the inherent semantic transparency of compound words. Transparent compounds are those whose component parts (lexemes) are semantically related to the meaning of the entire word. In contrast, opaque compounds' lexemes do not contribute directly to the meaning of the compound. We showed that the familiarity of semantically transparent assembly lures created from their lexemes (study dog and house, test on doghouse) is greater than the familiarity of opaque assembly lures (study back and draw, test on drawback). A response-signal experiment revealed no evidence for the use of a recall-to-reject process for either semantically transparent or opaque lures.

Were we or are we? Perception of reduced function words in spontaneous conversations.

Spontaneous, reduced pronunciations diverge greatly from citation forms. The quality of a single segment can change, e.g., /b/ in “about” surfacing as an approximant. But sounds, syllables, and entire words can also be deleted (e.g., “do you have time?” as [djutEm] with no acoustic trace of “have”). This work investigates the perception of reduced function words such as “he was” or “we were.” Twenty-two young American English speakers’ spontaneous conversations with close acquaintances were recorded. From these, we selected utterances containing items such as “he’s/he was, we’re/we were, got him/got them.” When hearing an entire utterance, native listeners may clearly perceive “we were,” but on hearing just the “we were” portion, they perceive an unambiguous “we’re.” The portion of the signal presented to listeners is manipulated to determine the contributions of local acoustic cues, speech rate and coarticulation, semantic and syntactic information, and overall bias toward present vs past tense. An auditory and a written task are also compared to separate the contribution of intonation from that of syntax/semantics. These results begin to elucidate the interplay of information sources listeners draw upon when parsing spontaneous speech. Future work will compare to non-native listeners’ perceptions.

Stroop effects on redemption and semantic effects on confession: simultaneous automatic activation of embedded and carrier words

The present study was aimed at assessing whether focusing attention on a task-relevant part of a word prevents processing of its meaning. Participants performed a color-naming task on a prime word followed by lexical decision on a probe. Primes were words, which could contain an embedded color word (e.g., "redemption") written in an incongruent color. Probes were either semantically related (e.g., "confession") or unrelated (e.g., "production") to the prime word. A Stroop effect emerged for color words appearing either in the initial or in the final position of the carrier word. A priming effect also emerged, with faster responses to probes semantically related to the prime. These results are evidence that focusing attention on part of a prime (i.e., the embedded color word) does not prevent the semantic processing of the entire word.

Health education prescription for Lifestyle diseases prevention

Life style are closely related to poople's health. A good life style can make a person energetic and live longer:while unhealthy lifestyle will lead to a variety of lifestyle diseases, which is a serious thent to people's health. It's an important means to prevent and control the lifestyle diseases through health education, im-provenment of people's health awareness, cultivation of a healthy and civilized way of life, and strengthening of self healthcare, which is the fundamental way to enhance health level of the entire social Key words: Life style; Life style diseases; Health education

Efficient Learning for the Poor: New Insights into Literacy Acquisition for Children

Reading depends on the speed of visual recognition and capacity of short-term memory. To understand a sentence, the mind must read it fast enough to capture it within the limits of the short-term memory. This means that children must attain a minimum speed of fairly accurate reading to understand a passage. Learning to read involves “tricking” the brain into perceiving groups of letters as coherent words. This is achieved most efficiently by pairing small units consistently with sounds rather than learning entire words. To link the letters with sounds, explicit and extensive practice is needed; the more complex the spelling of a language, the more practice is necessary. However, schools of low-income students often waste instructional time and lack reading resources, so students cannot get sufficient practice to automatize reading and may remain illiterate for years. Lack of reading fluency in the early grades creates inefficiencies that affect the entire educational system. Neurocognitive research on reading points to benchmarks and monitoring indicators. All students should attain reading speeds of 45–60 words per minute by the end of grade 2 and 120–150 words per minute for grades 6–8.

Development of an aspheric 22-diopter 50-mm diameter magnifier

To develop in an interdisciplinary approach between ophthalmology and design areas an ergonomic +22-diopter 50-mm aspheric hand magnifier for low vision. An aluminum cylinder was cut, processed using a lathe and carved to produce a ring that accommodated a 50-mm aspheric lens, with an external depression not to slide from the holder's fingers. A cylindrical steel bar was cut, processed using a lathe and carved in order to form an externally turned ring to be screwed into the internal thread of the aluminum ring, to maintain the lens in a steady position. Both rings were submitted to electrostatic painting with a dull black electrostatic Epoxi ink, except the lower border of the external ring, to indicate the correct side of the magnifier to face the material to be read. A 22-diopter 50-mm diameter aspheric lens magnifier with a black ring to be hold at its external circular depression was obtained in order to safely search the adequate reading focus with an inferior aluminum colored ring to face the object to be read and allow a less distorted reading. This is the first Brazilian high-magnification great-diameter magnifier for low vision that permits basically the focusing on an entire word, not only syllables, in order to allow a faster and more comfortable reading. By an interdisciplinary approach a 22-diopter 50-mm aspheric lens magnifier was developed with image and ergonomic characteristics such as to permit comfortable and adequate reading performance in cases of low vision.

PALILALIA IN SIGN LANGUAGE

Sign languages, which use the hands and arms as articulators, are the natural languages of deaf people. Research indicates that sign, like speech, can break down at the level of motor control, in the context of a movement disorder such as Parkinson disease (PD).1 This article describes the first known case of a deaf sign language user with progressive supranuclear palsy (PSP). PSP affects the rostral brainstem and its projections to the basal ganglia, cerebellum, and cerebral cortex.2 Dysarthria typically emerges early in the course of the disease and disrupts several aspects of speech.3 The characteristic features of PSP dysarthria are articulatory incoordination and palilalia—the repetition of entire words at decreasing amplitudes without pause.2 This descriptive case study suggests that those same features are prominent in the sign production of a deaf man with PSP. ### Methods. #### Participants. The individual with PSP was a 79-year-old right-handed man, who was born deaf. He learned British Sign Language (BSL) at school at age 7, and it became his primary language. He developed left-sided weakness and dysphagia at age …

The impact of semantic impairment on verbal short-term memory in stroke aphasia and semantic dementia: A comparative study

This study presents the first direct comparison of immediate serial recall in semantic dementia (SD) and transcortical sensory aphasia (TSA). Previous studies of the effect of semantic impairment on verbal short-term memory (STM) have led to important theoretical advances. However, different conclusions have been drawn from these two groups. This research aimed to explain these inconsistencies. We observed (a) qualitative differences between SD and TSA in the nature of the verbal STM impairment and (b) considerable variation within the TSA group. The SD and TSA patients all had poor semantic processing and good phonology. Reflecting this, both groups remained sensitive to phonological similarity and showed a reduced effect of lexicality in immediate serial recall. The SD patients showed normal serial position effects; in contrast, the TSA patients had poor recall of the initial list items and exhibited large recency effects on longer lists. The error patterns of the two groups differed: the SD patients made numerous phoneme migration errors whereas the TSA group were more likely to produce entire words in the wrong order, often initiating recall with terminal list items. The SD cases also showed somewhat larger effects of word frequency and imageability. We propose that these contrasting performance patterns are explicable in terms of the nature of the underlying semantic impairment. SD is associated with anterior lobe atrophy and produces degradation of semantic knowledge—this is more striking for less frequent/imageable items, accentuating the effects of these lexical/semantic variables in STM. SD patients frequently recombine the phonemes of different list items due to the reduced semantic constraint upon phonology (semantic binding: Patterson, K., Graham, N., & Hodges, J. R. (1994). The impact of semantic memory loss on phonological representations. Journal of Cognitive Neuroscience, 6, 57–69). In contrast, the semantic impairment in TSA follows frontal or temporoparietal lesions and is associated with poor executive control of semantic processing (deregulated semantic cognition: Jefferies, E., & Lambon Ralph, M. A. (2006). Semantic impairment in stroke aphasia vs. semantic dementia: a case-series comparison. Brain, 129, 2132–2147), explaining why these patients are liable to recall entire words out of serial order.

An examination of semantic radical combinability effects with lateralized cues in Chinese character recognition

Auclair and Siéroff examined lateralized cuing effects in the identification of centrally presented letter strings and reported no cuing effects for short word stimuli. They argued for a redistribution of attention over the entire word for short familiar words. We explored cuing effects with Chinese phonetic compounds, which can be considered extreme examples of short words, in a character-level semantic judgment task. When the semantic radical position was placed on the left of the characters, strong radical combinability and semantic transparency effects were observed. There was also a significant interaction between cue position (left vs. right) and radical combinability: A left cue facilitated semantic judgment of characters with small radical combinability more than did a right cue. This behavior reflects the information profile of Chinese phonetic compounds. Semantic radicals with small combinability are more informative than those with large combinability in determining the meaning of the whole character; they therefore benefit more from a left than a right cue. A mechanism redistributing attention over the whole of the character was not in evidence at the level of semantic processing.

Mirror writing and reversing single letters in stroke patients and normal elderly

Mirror writing (MW) refers to the production of individual letters or whole word strings in reversed direction. When held to a mirror, these letters or words can be read normally. We observed MW in a considerable number of stroke patients. Of the 86 patients screened 15 (17.5%) showed at least one instance of mirror writing in any of the tasks. Both right (14% of 36 patients) and left (20% of 50 patients) hemisphere damaged patients produced reversed letters only when writing with their left hand, respectively the contralesional and ipsilesional hand. The dissociated performance between the two hands in brain damaged patients is relevant to the interpretation of MW because, unlike all other peripheral dysgraphias, MW affects the non-dominant hand only. Importantly, healthy elderly also showed MW solely when writing with their left hand (6.9% of 86 participants). MW in controls was less frequent but qualitatively similar to that observed in brain damaged patients. This finding is consistent with the motor interpretation of MW that assumes an inability to transform the stored letter forming programmes for left hand writing. However, several cases have been reported in the literature of a more pervasive form of MW whereby patients mirror reverse entire words or sentences. This pattern has been observed in children learning to write but it has never been observed in healthy adult volunteers. We propose that the diagnosis of MW should be limited to the reversal of whole words, multi-digit numbers and full sentences, which reveal a disorder in coding the correct direction of writing rather than an inability to accomplish the correct spatial orientation of single letters.

Variable Perception of White Noise in Ambiguous Phonetic Contexts: The Case of /p/ and /f/

The roles of spectro-temporal coherence, lexical status, and word position in the perception of speech in acoustic signals containing a mixture of speech and nonspeech sounds were investigated. Stimuli consisted of nine (non)words in which either white noise was inserted only into the silent interval preceding and/or following the onset of vocalic transitions ambiguous between /p/ and /f/, or in which white noise overlaid the entire utterance. Ten listeners perceived 85% /f/s when noise was inserted only into the silent interval signaling a stop closure, 47% /f/s when noise overlaid the entire (non)words, and 1% in the control condition that contained no noise. Effects of spectro-temporal coherence seemed to have dominated perceptual outcomes, although the lexical status and position of the critical phoneme also appeared to affect responses. The results are explained more adequately by the theory of Auditory Scene Analysis than by the Motor Theory of Speech Perception.

Auditory word identification in dyslexic and normally achieving readers

The integrity of phonological representation/processing in dyslexic children was explored with a gating task in which children listened to successively longer segments (gates) of a word. At each gate, the task was to decide what the entire word was. Responses were scored for overall accuracy as well as the children’s sensitivity to coarticulation from the final consonant. As a group, dyslexic children were less able than normally achieving readers to detect coarticulation present in the vowel portion of the word, particularly on the most difficult items, namely those ending in a nasal sound. Hierarchical regression and path analyses indicated that phonological awareness mediated the relation of gating and general language ability to word and pseudoword reading ability.

The Dynamics of Lexical Competition During Spoken Word Recognition

The sounds that make up spoken words are heard in a series and must be mapped rapidly onto words in memory because their elements, unlike those of visual words, cannot simultaneously exist or persist in time. Although theories agree that the dynamics of spoken word recognition are important, they differ in how they treat the nature of the competitor set-precisely which words are activated as an auditory word form unfolds in real time. This study used eye tracking to measure the impact over time of word frequency and 2 partially overlapping competitor set definitions: onset density and neighborhood density. Time course measures revealed early and continuous effects of frequency (facilitatory) and on set based similarity (inhibitory). Neighborhood density appears to have early facilitatory effects and late inhibitory effects. The late inhibitory effects are due to differences in the temporal distribution of similarity within neighborhoods. The early facilitatory effects are due to subphonemic cues that inform the listener about word length before the entire word is heard. The results support a new conception of lexical competition neighborhoods in which recognition occurs against a background of activated competitors that changes over time based on fine-grained goodness-of-fit and competition dynamics.

Information Weights of Nucleotides in DNA Sequences

The protein sequence is coded with the help of the triplets of nucleotides, each corresponding to one amino acid in a protein se- quence. The triplet code of the coding sequences possesses some informative redundancy. Some triplets are more probable than others. The analogous redundancies appear in all natural languages. The non-equal frequency of the characters in plain text makes possible that entire words can be predicted given the context of the plain text. This is typical problem in cryptanalysis that a plain text is compressed before en- crypting it in order to reduce the language redundancies. The nucleotides represent the natural units to discuss the redundancies in the coding sequences of natural genomes. The mutation pressure and selection pressure are the main factors responsible for the observed redundancies. Then, the nucleotide frequency in DNA seems to be the natural information weight. We show, that the probability of a nucleotide to stay non- mutated becomes another, the more efficient information weight. It has smaller redundancy although it is correlated with the nucleotide fre- quency. We have found the values of probability for nucleotide to stay non-mutated in the particular case of the Borrellia burgdorferi ge- nome. In order to examine the usefulness of the new frequencies we used them in a problem of bit-string packing in a channel with a given capacity. We performed a computer experiment, in which we have generated all possible oligomers consisting of k nucleotides and we have shown, that if the number of bits of the information carried out by the oligomers does not exceed a given threshold value, the same as calcu- lated for genes of the Borrelia burgdorferi genome, then the distribution of the generated oligomers resembles the one used by these genes.

Biomedical information retrieval across languages

This work presents a new dictionary-based approach to biomedical cross-language information retrieval (CLIR) that addresses many of the general and domain-specific challenges in current CLIR research. Our method is based on a multilingual lexicon that was generated partly manually and partly automatically, and currently covers six European languages. It contains morphologically meaningful word fragments, termed subwords. Using subwords instead of entire words significantly reduces the number of lexical entries necessary to sufficiently cover a specific language and domain. Mediation between queries and documents is based on these subwords as well as on lists of word-n-grams that are generated from large monolingual corpora and constitute possible translation units. The translations are then sent to a standard Internet search engine. This process makes our approach an effective tool for searching the biomedical content of the World Wide Web in different languages. We evaluate this approach using the OHSUMED corpus, a large medical document collection, within a cross-language retrieval setting.

Method for generating a statistic for phone lengths and method for determining the length of individual phones for speech synthesis

A statistic for phone lengths is generated by determining the length of individual phones for speech synthesis. A primary statistic is based on primary clusters (for example triphones), and a secondary statistic is based on secondary clusters (for example phonemes of entire words). Both statistics include average phone lengths and, for example, the standard variation of the average phone lengths. During the determination of phone lengths, it is firstly attempted to determine the average phone lengths and standard variation of the average phone lengths by reference to the secondary statistic which is more language-specific. If this is not the case, the primary statistic, which can always be applied, is resorted to. By this two stage method, a phone length is determined which corresponds significantly better to a natural language than has been possible with the conventional single stage method.

Nonuniform Error Correction Using Low-Density Parity-Check Codes

This correspondence introduces a framework to design and analyze low-density parity-check (LDPC) codes over nonuniform channels. We study LDPC codes for channels with nonuniform noise distributions, rate-adaptive coding, and unequal error protection. First, we propose a technique to design LDPC codes for volume holographic memory (VHM) systems for which the noise distribution is nonuniform. We show that the proposed coding scheme has an easy design procedure and results in efficient codes for holographic memories. An important property of the proposed technique is the design of the codes that have a low error floor and low variable node degrees, while maintaining performance close to the Shannon limit. We then show that punctured LDPC codes can be studied as a special case of our design methodology for nonuniform channels. Finally, we propose a method to generate LDPC codes that can provide unequal error protection in addition to having a good overall performance. Moreover, the highly protected bits can be decoded without requiring the entire word to be decoded.