Sparse Neighborhoods Research Articles

Sparse neighbor representation for classification

Recent research of sparse signal representation has aimed at learning discriminative sparse models instead of purely reconstructive ones for classification tasks, such as sparse representation based classification (SRC) which obtains state-of-the-art results in face recognition. In this paper, a new method is proposed in that direction. With the assumption of locally linear embedding, the proposed method achieves the classification goal via sparse neighbor representation, combining the reconstruction property, sparsity and discrimination power. The experiments on several data sets are performed and results show that the proposed method is acceptable for nonlinear data sets. Further, it is argued that the proposed method is well suited for the classification of low dimensional data dimensionally reduced by dimensionality reduction methods, especially the methods obtaining the low dimensional and neighborhood preserving embeddings, and it costs less time.

Phonological similarity influences word learning in adults learning Spanish as a foreign language

Neighborhood density-the number of words that sound similar to a given word (Luce & Pisoni, 1998)-influences word-learning in native English speaking children and adults (Storkel, 2004; Storkel, Armbruster, & Hogan, 2006): novel words with many similar sounding English words (i.e., dense neighborhood) are learned more quickly than novel words with few similar sounding English words (i.e., sparse neighborhood). The present study examined how neighborhood density influences word-learning in native English speaking adults learning Spanish as a foreign language. Students in their third-semester of Spanish language classes learned advanced Spanish words that sounded similar to many known Spanish words (i.e., dense neighborhood) or sounded similar to few known Spanish words (i.e., sparse neighborhood). In three word-learning tasks, performance was better for Spanish words with dense rather than sparse neighborhoods. These results suggest that a similar mechanism may be used to learn new words in a native and a foreign language.

Lexical similarity and speech production: Neighborhoods for nonwords

Lexical similarity has been shown to play a role in speech production (as in speech perception). In production, words with many phonologically similar neighbors, i.e., those that are phonologically similar to a large number of other words, are produced with more hyperarticulated vowels than words with fewer neighbors ( Wright, 1997; Munson and Solomon, 2004). The experiments presented here further investigate the role that lexical similarity, expressed in terms of phonological neighborhood density, plays in the details of phonetic implementation, looking at nasal coarticulation in particular. And the way in which the grammar might produce such effects is probed by looking for neighborhood-conditioned effects in not only real words, but also nonsense words. Eight speakers of American English produced 48 highly familiar CVN or NVC words and 48 similar CVN or NVC word-like nonsense words with either many or few phonologically similar neighbors (words differing from the target word by a single phoneme). Lexical frequency and segmental context were balanced across neighborhood conditions. Degree of nasal coarticulation was measured acoustically on each vowel (A1-P0) ( Chen, 1997), along with vowel duration and spectral hyperarticulation. Analysis showed that real words (both CVNs and NVCs) from dense neighborhoods (e.g., band, mug) were consistently and reliably produced with more coarticulation and more hyperarticulation than words from sparse neighborhoods (e.g., stem, mouth). Vowel duration was not affected. The same consistent effects of neighborhood were found for nonsense words as well (e.g., gand, mub vs. blem, maub) for approximately half of the speakers in the study. The other speakers showed consistently opposite effects for nonsense words (i.e., greater nasal coarticulation and hyperarticulation in nonsense words from sparse neighborhoods), despite patterning with the rest of the speakers for real words. These findings demonstrate that phonological similarity at the lexical level has an influence on phonetic realization: similarity to a greater number of real words is associated with increased coarticulation as well as hyperarticulation. Interestingly, the degree of coarticulation and hyperarticulation vary systematically in nonwords as well, but the nature of the systematic relationship differs across two groups of speakers: some speakers produce nonsense words just like real words with increased hyperarticulation and coarticulation in nonsense words from dense neighborhoods, while other speakers produce nonsense words with the opposite systematic relationship—with increased hyperarticulation and coarticulation in nonsense words from sparse neighborhoods. However, because nonsense words, like real words, show some kind of systematic relationship between detail in production and lexical similarity neighborhoods, the differences in phonetic implementation must not be encoded directly in the representation, as nonwords have no long-term representations. Rather, lexical similarity must be assessed online with active reference to a space in the lexicon (the neighborhood).

Repetition of words from dense and sparse phonological neighborhoods in children with hearing loss and normal hearing.

4 and 7 year-old children with normal and impaired hearing performed a listen-and-repeat task with words from dense and sparse phonological neighborhoods. Response accuracy was measured as a function of age, hearing loss, and neighborhood density. Accuracy was higher for older children, children with normal hearing, and for words from dense phonological neighborhoods. Age was evaluated cross-sectionally. Older children with normal hearing were consistently more accurate than younger children with normal hearing; children with hearing loss showed much smaller age-related differences. For children with normal hearing, accuracy for sparse neighborhoods was significantly higher for 7 year-olds than for 4 year-olds, but no such effect was observed for children with hearing loss; a similar pattern holds for words from dense neighborhoods. Relationships between accuracy, productive phonology (GFTA-II), expressive vocabulary (EVT-II), and degree of hearing loss (PTA dBHL) are also explored. These results suggest that phonological predictability, as measured by neighborhood density, is an important factor in the ability of children to perceive and reproduce words, interacting with both age and presence or absence of hearing loss. Absence of age-related performance improvement for children with hearing loss has theoretical implications for the role of auditory experience in development. [Work supported by NIH/NIDCD Grant Nos. R01DC006681 and P30DC04662.]

Lexical influences on children's perception of foreign‐accented speech.

The ability to perceive speech accurately in the face of multiple sources of variability appears to develop during early and middle childhood. However, the cognitive‐linguistic skills that underlie these improvements are unknown. Early‐middle childhood is a time of rapid development in many areas of cognition and language, including the expansion of the lexicon. Lexical development includes both changes in lexicon size and changes in lexicon structure. Research suggests that as the lexicon grows, the words within the lexicon become more highly specified moving from more holistic representations to more fine‐grained, segmental representations. This study tested how the expansion and restructuring of the lexicon are related to the ability to perceive highly variable speech signals. Specifically, the perception of foreign‐accented speech — a common, real‐world source of speech variability — was tested. The relationship between lexicon size and the perception of lexically easy (i.e., high‐frequency words from sparse neighborhoods) and lexically hard (i.e., low‐frequency words from dense neighborhoods) foreign‐accented words was investigated for 5‐ and 6‐year‐old children. [Work supported by NIDCD 1R21DC010021.]

A Short Report: Word-Level Phonological and Lexical Characteristics Interact to Influence Phoneme Awareness

In this study, we examined the influence of word-level phonological and lexical characteristics on early phoneme awareness. Typically developing children, ages 61 to 78 months, completed a phoneme-based, odd-one-out task that included consonant—vowel—consonant word sets (e.g., “chair—chain—ship”) that varied orthogonally by a phonological characteristic, sound contrast similarity (similar vs. dissimilar), and a lexical characteristic, neighborhood density (dense vs. sparse). In a subsample of the participants—those with the highest vocabularies—results were in line with a predicted interactive effect of phonological and lexical characteristics on phoneme awareness performance: word sets contrasting similar sounds were less likely to yield correct responses in words from sparse neighborhoods than words from dense neighborhoods. Word sets contrasting dissimilar sounds were most likely to yield correct responses regardless of the words’ neighborhood density. Based on these findings, theories of early phoneme awareness should consider both word-level and child-level influences on performance. Attention to these influences is predicted to result in more sensitive and specific measures of reading risk.

The influence of lexical status and neighborhood density on children's nonword repetition

ABSTRACTThis study examined effects of lexical status and neighborhood density of constituent syllables on children's nonword repetition and interactions with nonword length. Lexical status of the target syllable impacted repetition accuracy for the longest nonwords. In addition, children made more errors that changed a nonword syllable to a word syllable than the reverse. Syllables from dense versus sparse neighborhoods were repeated more accurately in three- and four-syllable nonwords, but there was no effect of density for two-syllable nonwords. The effect of neighborhood density was greater for a low versus high vocabulary group. Finally, children's error responses were from more dense neighborhoods than the target syllables. The results are congruent with models of nonword repetition that emphasize the influence of long-term lexical knowledge on children's performance.

A cross-sectional comparison of the effects of phonotactic probability and neighborhood density on word learning by preschool children

Two experiments examined the effects of phonotactic probability and neighborhood density on word learning by 3-, 4-, and 5-year-old children. Nonwords orthogonally varying in probability and density were taught with learning and retention measured via picture naming. Experiment 1 used a within story probability/across story density exposure context. Experiment 2 used an across story probability/within story density exposure context. Results showed that probability and density interacted to create optimal learning conditions. Specifically, rare/sparse sound sequences appeared to facilitate triggering of word learning. In contrast, the optimal convergence for lexical configuration and engagement was dependent on exposure context. In particular, common sound sequences and dense neighborhoods were optimal when density was manipulated across stories, whereas rare sound sequences and sparse neighborhoods were optimal when density was manipulated within a story. Taken together, children’s phonological and lexical representations were hypothesized to be interdependent on one another resulting in a convergence of form characteristics for optimal word learning.

Phonological neighbourhood effects in French spoken-word recognition

According to activation-based models of spoken-word recognition, words with many and high-frequency phonological neighbours are processed more slowly than words with few and low-frequency phonological neighbours. Although considerable empirical support for inhibitory neighbourhood density effects has accumulated, especially in English, little or nothing is known about the effects of neighbourhood frequency and its interaction with neighbourhood density. In this study we examine both effects first separately and then simultaneously in French lexical decision experiments. As in English, we found that words in dense neighbourhoods are recognized more slowly than words in sparse neighbourhoods. Moreover, we showed that words with higher frequency neighbours are processed more slowly than words with no higher frequency neighbours, but only for words occurring in sparse neighbourhoods. Implications of these results for spoken-word recognition models are discussed.

Lexical Effects on Spoken-Word Recognition in Children with Normal Hearing

This study is the first in a series designed to develop and norm new theoretically motivated sentence tests for children. The purpose was to examine the independent contributions of word frequency (i.e., how often words occur in language) and lexical density (the number of similar sounding words or "neighbors" to a target word) to the perception of key words in the new sentence set. Twenty-four children with normal hearing aged 5 to 12 yrs served as participants; they were divided into four equal age-matched groups. The stimuli consisted of 100 semantically neutral sentences that were 5 to 7 words in length. Each sentence contained 3 key words that were controlled for word frequency and lexical density. Words with few neighbors come from sparse neighborhoods, whereas words with many neighbors come from dense neighborhoods. The key words within a sentence belonged to one of the four lexical categories: (1) high-frequency sparse, (2) low-frequency dense, (3) high-frequency dense, and (4) low-frequency sparse. Participants were administered the sentence list and the 300 key words in isolation at 65 dB SPL. Each participant group was tested in spectrally matched noise at one of the four signal-to-noise ratios (SNRs -2, 0, 2, and 4 dB). The percent of words correctly identified was calculated as a function of SNR, key word context (sentences vs. words), and key word lexical category. SNR had a significant effect on the recognition of key words in sentences and in isolation; performance improved at higher SNRs. There were significant main effects of word frequency and lexical density as well as a significant interaction between the two lexical factors. In isolation, high-frequency words were recognized more accurately than low-frequency words. In both word and sentence contexts, sparse words yielded greater accuracy than dense words, irrespective of word frequency. There was a modest but significant negative correlation between lexical density and the recognition of words in isolation and in sentences. Word frequency and lexical density seem to influence word recognition independently in children with normal hearing. This is similar to earlier results in adults with normal hearing. In addition, there seems to be an interaction between the two factors, with lexical density being more heavily weighted than word frequency. These results give us further insight into the way children organize and access words from long-term lexical memory in a relational way. Our results showed that lexical effects were most evident at poorer SNRs. This may have important implications for assessing spoken-word recognition performance in children with sensory aids because they typically receive a degraded auditory signal.

Phonological Neighborhood Density in the Trevor Corpus: Perception and Production Factors in Lexical Acquisition

UC Berkeley Phonology Lab Annual Report (2010) Phonological Neighborhood Density in the Trevor Corpus: Perception and Production Factors in Lexical Acquisition Melinda Woodley University of California, Berkeley Since at least the 1980's, phonological neighborhood density (henceforth PND) has been recognized as an important factor in speech processing (e.g. Goldinger, Luce, & Pisoni, 1989). One of the reasons it has provoked such interest is that in contrast to many other factors relevant for speech processing (e.g. word frequency, imageability, age of acquisition, etc.) PND is known to have opposite effects on speech perception and production; words in dense neighborhoods are more difficult to perceive, but easier to produce than words in sparse neighborhoods (see e.g. Dell & Gordon, 2003 for a review). More recently, phonological neighborhood density has been invoked as a possible driving factor in the development of segment-based lexical representations. Researchers in phonological development have argued that the lower overall density of young children's lexicons allows them to initially store and access words in a more holistic manner (Charles-Luce & Luce, 1990, 1995; Edwards, Beckman, & Munson, 2004; Logan, 1992; Vihman & Croft, 2007, among others). As words are added to the lexicon, however, the potential for confusion between similar sounding words increases, forcing language learners to pay attention to finer grained phonetic details in order to make the necessary contrasts between lexical items. One potential shortcoming in most studies of developing phonological neighborhoods is that phonological development is considered in the aggregate; in many studies (e.g. Charles-Luce & Luce, 1990, 1995; Logan, 1992) the estimated time course for adding words to the lexicon is based on large corpora of many children. While this approach may gain something in its generalizability, it may also be missing some of the more interesting details that emerge from examining the development of individual children's phonological systems; since each child approaches the learning task based on his or her own experience with language, and his or her own inductive processes, cases studies of individual children's phonological development can shed light on the process by which early phonological generalizations are made and how they eventually develop into a full-fledged phonological system.

Word learning by children with phonological delays: Differentiating effects of phonotactic probability and neighborhood density

This study examined the ability of 20 preschool children with functional phonological delays and 34 age- and vocabulary-matched typical children to learn words differing in phonotactic probability (i.e., the likelihood of occurrence of a sound sequence) and neighborhood density (i.e., the number of words that differ from a target by one phoneme). Children were exposed to nonwords paired with novel objects in a story and learning was measured by a picture naming task. Results showed that both groups created lexical representations for rare sound sequences from sparse neighborhoods. However, only children with typical development appeared to build on this initial lexical representation to create a full representation of the word (i.e., lexical–semantic connection and semantic representation). It was hypothesized that creating a lexical representation may be too resource demanding for children with phonological delays, leaving few resources available to create a lexical–semantic connection and/or a semantic representation. Learning outcomes: The reader will be able to (1) define phonotactic probability; (2) define neighborhood density; (3) identify how these variables impact the word learning process in general; (4) identify potential areas of deficit in the word learning process for children with functional phonological delays.

Associative and orthographic neighborhood density effects in normal aging and Alzheimer’s disease.

A group of patients with Alzheimer's disease and a group of healthy elderly controls were tested with a lexical-decision task that included words with dense or sparse orthographic and associative neighborhoods to investigate whether there is automatic orthographic and semantic activation of related representations in these populations similar to that found with younger samples. Although some studies support the idea of deteriorated connections in semantic networks in Alzheimer's disease, other studies propose that the automatic spread of activation at lexicosemantic levels remains intact and that intergroup differences are a consequence of impaired retrieval or attentional deficits. In this study, participants responded to words with dense orthographic and associative neighborhoods faster than they did to words with sparse neighborhoods, providing evidence in favor of a preserved automatic spread of activation through intact orthographic and semantic representations. Furthermore, no differences were found between the two groups regarding the magnitude of the effects, although control participants responded significantly faster than patients.

Neighborhood effects in spelling in adults

The participants were asked to spell aloud words for which there were either many orthographically similar words (a dense neighborhood) or few orthographically similar words (a sparse neighborhood). Words with a dense neighborhood were spelled faster and more accurately than were words with a sparse neighborhood. This finding is consistent with the hypothesis of Rapp, Epstein, and Tainturier (2002), that the cognitive spelling system has an interactive architecture incorporating feedback between individual graphemes and orthographic lexeme representations.

Children's spoken word recognition and contributions to phonological awareness and nonword repetition: A 1-year follow-up

ABSTRACTThis study examined effects of lexical factors on children's spoken word recognition across a 1-year time span, and contributions to phonological awareness and nonword repetition. Across the year, children identified words based on less input on a speech-gating task. For word repetition, older children improved for the most familiar words. There was a competition effect for the word repetition task, but this effect was present only for the most familiar words on the gating task. Recognition for words from sparse neighborhoods predicted phonological awareness 1 year later, and children poorer at recognizing these words in Year 1 scored lower on word reading in Year 2. Spoken word recognition also accounted for unique variance in nonword repetition across the 1-year time span. Findings are discussed in terms of understanding the effects of vocabulary growth on spoken word recognition, phonological awareness, and nonword repetition.

Word Length and Lexical Competition: Longer is the Same as Shorter

Neighborhood density refers to the number of words that sound similar to a given word. Previous studies have found that neighborhood density influences the recognition of spoken words (Luce & Pisoni, 1998); however, this work has focused almost exclusively on monosyllabic words in English. To investigate the effects of neighborhood density on longer words, bisyllabic words varying in neighborhood density were presented auditorily to participants in a perceptual identification task and a lexical decision task. In the perceptual identification task, words with sparse neighborhoods were more accurately identified than words with dense neighborhoods. In the lexical decision task, words with sparse neighborhoods were responded to more quickly and more accurately than words with dense neighborhoods. These results are similar to those found in studies examining the influence of neighborhood density on the recognition of monosyllabic words in English. In order to better understand lexical processing, models of spoken word recognition must account for the processing of words of all types.

NoA’s ark: Influence of the number of associates in visual word recognition

The main aim of this study was to explore the extent to which the number of associates of a word (NoA) influences lexical access, in four tasks that focus on different processes of visual word recognition: lexical decision, reading aloud, progressive demasking, and online sentence reading. Results consistently showed that words with a dense associative neighborhood (high-NoA words) were processed faster than words with a sparse neighborhood (low-NoA words), extending previous findings from English lexical decision and categorization experiments. These results are interpreted in terms of the higher degree of semantic richness of high-NoA words as compared with low-NoA words.

Long-term priming of neighbours biases the word recognition process: Evidence from a lexical decision task.

The role of orthographically similar words (i.e., neighbours) in the word recognition process has been studied extensively using short-term priming paradigms (e.g., Colombo, 1986). Here we demonstrate that long-term effects of neighbour priming can also be obtained. Experiment 1 showed that prior study of a neighbour (e.g., TANGO) increased later lexical decision performance for similar words (e.g., MANGO), but decreased performance for similar pseudowords (e.g., LANGO). Experiment 2 replicated this bias effect and showed that the increase in lexical decision performance due to neighbour priming is selectively due to words from a relatively sparse neighbourhood. Explanations of the bias effect in terms of lexical activation and episodic memory retrieval are discussed.

Lexical restructuring in the absence of literacy

Vocabulary growth was suggested to prompt the implementation of increasingly finer-grained lexical representations of spoken words in children (e.g., [Metsala, J. L., & Walley, A. C. (1998). Spoken vocabulary growth and the segmental restructuring of lexical representations: precursors to phonemic awareness and early reading ability. In J. L. Metsala & L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 89–120). Hillsdale, NJ: Erlbaum.]). Although literacy was not explicitly mentioned in this lexical restructuring hypothesis, the process of learning to read and spell might also have a significant impact on the specification of lexical representations (e.g., [Carroll, J. M., & Snowling, M. J. (2001). The effects of global similarity between stimuli on children’s judgments of rime and alliteration. Applied Psycholinguistics, 22, 327–342.]; [Goswami, U. (2000). Phonological representations, reading development and dyslexia: Towards a cross-linguistic theoretical framework. Dyslexia, 6, 133–151.]). This is what we checked in the present study. We manipulated word frequency and neighborhood density in a gating task (Experiment 1) and a word-identification-in-noise task (Experiment 2) presented to Portuguese literate and illiterate adults. Ex-illiterates were also tested in Experiment 2 in order to disentangle the effects of vocabulary size and literacy. There was an interaction between word frequency and neighborhood density, which was similar in the three groups. These did not differ even for the words that are supposed to undergo lexical restructuring the latest (low frequency words from sparse neighborhoods). Thus, segmental lexical representations seem to develop independently of literacy. While segmental restructuring is not affected by literacy, it constrains the development of phoneme awareness as shown by the fact that, in Experiment 3, neighborhood density modulated the phoneme deletion performance of both illiterates and ex-illiterates.

The curious case of competition in Spanish speech production

In previous studies in English examining the influence of phonological neighbourhood density in spoken word production, words with many similar sounding words, or a dense neighbourhood, were produced more quickly and accurately than words with few similar sounding words, or a sparse neighbourhood. The influence of phonological neighbourhood density on the process of spoken word production in Spanish was examined with a picture-naming task. The results showed that pictures with Spanish names from sparse neighbourhoods were named more quickly than pictures with Spanish names from dense neighbourhoods. The present pattern of results is the opposite of what has been previously found in speech production in English. We hypothesise that differences in the morphology of Spanish and English and/or the location in the word where phonological neighbours tend to occur may contribute to the processing differences observed in the two languages.