Detecting Word Boundaries Research Articles

Jordanian EFL Students' Perception of Noncontrastive Allophonic Cues in English Speech Segmentation.

The prominent role of allophonic cues in English speech segmentation has widely been recognized by phonologists and psycholinguists. However, very meager inquiry was devoted to analysing the perception of these noncontrastive allophonic cues by Arab EFL learners. Accordingly, the present study is an attempt to examine the exploitation of allophonic cues, mainly aspiration, glottalization and approximant devoicing to English word junctures by 40 Jordanian PhD students. Moreover, it aims to find out which allophonic cues are perceived more accurately during the segmentation process and if there is any evidence for Universal Grammar markedness. The experiment is led through a forced-choice identification task adopted from Altenberg (Second Lang Res 21:325-358, 2005) and Rojczyk et al. (Res Lang 1:15-29, 2016). The results of ANOVA unveiled that there is a statistically significant difference between the three types of allophonic cues, viz. aspiration, glottalization and approximant devoicing. This implies that the participants outperformed in stimuli marked by glottalization than by aspiration and approximant devoicing. This result provided further evidence for the universality of glottalization as a boundary cue in English speech segmentation. Overall, the Jordanian PhD students failed in perceiving the allophonic cues accurately and exploiting them to detect word boundaries. The present inquiry has the potential to provide several recommendations for syllabus designers, and second/foreign language teachers and learners.

Detecting word boundaries in an undeciphered script: The Byblos syllabary

The Byblos writing system (ca. 1500 BC) comprises 15 inscriptions, with the largest one containing 461 clearly discernible signs. Some further 28 unassigned fragments found in the Levante and in Italy are tentatively assigned to the Byblos corpus too. We have created a Unicode letter for each sign variant found in these inscriptions and transcribed the corpus. Feeding the corpus into our RegEx- based deciphering tool, we established a preliminary (dynamic) syllabary laying a foundation for computer-assisted deciphering efforts of the Byblos script. A first sequence analysis allows to detect morphemes and word boundaries, so that even without knowing the sound values, we gain some insight into the syntax of the language depicted by these beautiful graphemes.

Word-based largest chunks for Agreement Groups processing: Cross-linguistic observations

The present study reports results from a series of computer experiments seeking to combine word-based Largest Chunk (LCh) segmentation and Agreement Groups (AG) sequence processing. The AG model is based on groups of similar utterances that enable combinatorial mapping of novel utterances. LCh segmentation is concerned with cognitive text segmentation, i.e. with detecting word boundaries in a sequence of linguistic symbols. Our observations are based on the text of Le petit prince (The little prince) by Antoine de Saint-Exupéry in three languages: French, English, and Hungarian. The data suggest that word-based LCh segmentation is not very efficient with respect to utterance boundaries, however, it can provide useful word combinations for AG processing. Typological differences between the languages are also reflected in the results.

Audiovisual speech segmentation in post-stroke aphasia: a pilot study

ABSTRACTBackground: Stroke may cause sentence comprehension disorders. Speech segmentation, i.e. the ability to detect word boundaries while listening to continuous speech, is an initial step allowing the successful identification of words and the accurate understanding of meaning within sentences. It has received little attention in people with post-stroke aphasia (PWA).Objectives: Our goal was to study speech segmentation in PWA and examine the potential benefit of seeing the speakers’ articulatory gestures while segmenting sentences.Methods: Fourteen PWA and twelve healthy controls participated in this pilot study. Performance was measured with a word-monitoring task. In the auditory-only modality, participants were presented with auditory-only stimuli while in the audiovisual modality, visual speech cues (i.e. speaker’s articulatory gestures) accompanied the auditory input. The proportion of correct responses was calculated for each participant and each modality. Visual enhancement was then calculated in order to estimate the potential benefit of seeing the speaker’s articulatory gestures.Results: Both in auditory-only and audiovisual modalities, PWA performed significantly less well than controls, who had 100% correct performance in both modalities. The performance of PWA was correlated with their phonological ability. Six PWA used the visual cues. Group level analysis performed on PWA did not show any reliable difference between the auditory-only and audiovisual modalities (median of visual enhancement = 7% [Q1 − Q3: −5 − 39]).Conclusion: Our findings show that speech segmentation disorder may exist in PWA. This points to the importance of assessing and training speech segmentation after stroke. Further studies should investigate the characteristics of PWA who use visual speech cues during sentence processing.

Cross-document word matching for segmentation and retrieval of Ottoman divans

Motivated by the need for the automatic indexing and analysis of huge number of documents in Ottoman divan poetry, and for discovering new knowledge to preserve and make alive this heritage, in this study we propose a novel method for segmenting and retrieving words in Ottoman divans. Documents in Ottoman are difficult to segment into words without a prior knowledge of the word. In this study, using the idea that divans have multiple copies (versions) by different writers in different writing styles, and word segmentation in some of those versions may be relatively easier to achieve than in other versions, segmentation of the versions (which are difficult, if not impossible, with traditional techniques) is performed using information carried from the simpler version. One version of a document is used as the source dataset and the other version of the same document is used as the target dataset. Words in the source dataset are automatically extracted and used as queries to be spotted in the target dataset for detecting word boundaries. We present the idea of cross-document word matching for a novel task of segmenting historical documents into words. We propose a matching scheme based on possible combinations of sequence of sub-words. We improve the performance of simple features through considering the words in a context. The method is applied on two versions of Layla and Majnun divan by Fuzuli. The results show that, the proposed word-matching-based segmentation method is promising in finding the word boundaries and in retrieving the words across documents.

Improving recognition of proper nouns in ASR through generating and filtering phonetic transcriptions

Accurate phonetic transcription of proper nouns can be an important resource for commercial applications that embed speech technologies, such as audio indexing and vocal phone directory lookup. However, an accurate phonetic transcription is more difficult to obtain for proper nouns than for regular words. Indeed, phonetic transcription of a proper noun depends on both the origin of the speaker pronouncing it and the origin of the proper noun itself.This work proposes a method that allows the extraction of phonetic transcriptions of proper nouns using actual utterances of those proper nouns, thus yielding transcriptions based on practical use instead of mere pronunciation rules.The proposed method consists in a process that first extracts phonetic transcriptions, and then iteratively filters them. In order to initialize the process, an alignment dictionary is used to detect word boundaries. A rule-based grapheme-to-phoneme generator (LIA_PHON), a knowledge-based approach (JSM), and a Statistical Machine Translation based system were evaluated for this alignment. As a result, compared to our reference dictionary (BDLEX supplemented by LIA_PHON for missing words) on the ESTER 1 French broadcast news corpus, we were able to significantly decrease the Word Error Rate (WER) on segments of speech with proper nouns, without negatively affecting the WER on the rest of the corpus.

Pause detection in continuous sign language

Sign language segmentation breaks a continuous sentence into its basic lexical units by detecting word boundaries. For robust recognition, the majority of direct segmentation approaches exploit these inter-sign pauses in a stream of hand gestures to demarcate word boundaries. Recent attempts to segment a continuous discourse exploit the constancy or directional variations of sign parameters (mainly spatial parameters). The delayed absolute difference (DAD) signature of hand positions provides means for analysing the segmentation features like pauses, repetitions and directional variations in a unique tool. In this paper, a DAD-based pause detection algorithm has been described. The performance of this deterministic algorithm is compared with three segmentation approaches. All the experiments and comparisons are done using the subjective annotation by 15 native New Zealand Sign Language (NZSL) signers. The proposed algorithm correctly and consistently detected the various lengths of pauses as compared to the existing segmentation approaches.

DYNAMIC THRESHOLDING ON SPEECH SEGMENTATION

Word is the preferred and natural unit of speech, b ecause word units have well defined acoustic repres entation. This paper presents several dynamic thresholding approaches for segment ing continuous Bangla speech sentences into words/s ub-words. We have proposed three efficient methods for speech segment ation: two of them are usually used in pattern clas sification (i.e., k-means and FCM clustering) and one of them is used in image se gmentation (i.e., Otsu’s thresholding method). We a lso used new approaches blocking black area and boundary detection techniqu es to properly detect word boundaries in continuous speech and label the entire speech sentence into a sequence of words/sub-words. K-Means and FCM clustering methods produce better segmentation results than that of Otsu’s Method. All the algorithms and metho ds used in this research are implemented in MATLAB and the proposed system achieved the average segmentation accuracy of 94% a

Implicit Segmentation of a Stream of Syllables Based on Transitional Probabilities: An MEG Study

Statistical segmentation of continuous speech, i.e., the ability to utilise transitional probabilities between syllables in order to detect word boundaries, is reflected in the brain's auditory event-related potentials (ERPs). The N1 and N400 ERP components are typically enhanced for word onsets compared to random syllables during active listening. We used magnetoencephalography (MEG) to record event-related fields (ERFs) simultaneously with ERPs to syllables in a continuous sequence consisting of ten repeating tri-syllabic pseudowords and unexpected syllables presented between these pseudowords. We found the responses to differ between the syllables within the pseudowords and between the expected and unexpected syllables, reflecting an implicit process extracting the statistical characteristics of the sequence and monitoring for unexpected syllables.

The neural basis of speech parsing in children and adults

Word segmentation, detecting word boundaries in continuous speech, is a fundamental aspect of language learning that can occur solely by the computation of statistical and speech cues. Fifty-four children underwent functional magnetic resonance imaging (fMRI) while listening to three streams of concatenated syllables that contained either high statistical regularities, high statistical regularities and speech cues, or no easily detectable cues. Significant signal increases over time in temporal cortices suggest that children utilized the cues to implicitly segment the speech streams. This was confirmed by the findings of a second fMRI run, in which children displayed reliably greater activity in the left inferior frontal gyrus when listening to 'words' that had occurred more frequently in the streams of speech they had just heard. Finally, comparisons between activity observed in these children and that in previously studied adults indicate significant developmental changes in the neural substrate of speech parsing.

Statistical language learning in neonates revealed by event-related brain potentials

BackgroundStatistical learning is a candidate for one of the basic prerequisites underlying the expeditious acquisition of spoken language. Infants from 8 months of age exhibit this form of learning to segment fluent speech into distinct words. To test the statistical learning skills at birth, we recorded event-related brain responses of sleeping neonates while they were listening to a stream of syllables containing statistical cues to word boundaries.ResultsWe found evidence that sleeping neonates are able to automatically extract statistical properties of the speech input and thus detect the word boundaries in a continuous stream of syllables containing no morphological cues. Syllable-specific event-related brain responses found in two separate studies demonstrated that the neonatal brain treated the syllables differently according to their position within pseudowords.ConclusionThese results demonstrate that neonates can efficiently learn transitional probabilities or frequencies of co-occurrence between different syllables, enabling them to detect word boundaries and in this way isolate single words out of fluent natural speech. The ability to adopt statistical structures from speech may play a fundamental role as one of the earliest prerequisites of language acquisition.

Word boundary detection based on suprasegmental features: A case study on Bangla speech

This paper proposes a method for detecting word boundaries in continuous speech signal for Standard Colloquial Bengali (SCB), commonly referred to as Bangla. Bangla is a bound stress language with stress on the first syllable. Stress introduces its signature on the supra-segmental parameters of the speech signal, which may help to detect the word boundary in the continuous speech signal. The parameters used in this present study are: (1) Difference of the nucleus vowel duration across the syllable boundary, (2) Difference of the normalized nucleus vowel power across the syllable boundary, (3) Normalized F0 difference across the syllable boundary, (4) Difference of the average normalized F0 across the syllable boundary, (5) Difference of the normalized maximum periodic power of nucleus vowels across the syllable boundary, (6) Onset duration of the nucleus vowel. Altogether 225 sentences spoken by five native Bangla informants of both the sexes, in the age group of 20–50 years in normal laboratory environment are used in this study. These sentences contain 2734 syllables and 1103 words, sentence terminal words being excluded. A recognition score of 87.8% with a classifier, based on a distance function, weighted by inverse of variance is reported. Both speaker dependent as well as speaker independent studies are included.

Cracking the Language Code: Neural Mechanisms Underlying Speech Parsing

Word segmentation, detecting word boundaries in continuous speech, is a critical aspect of language learning. Previous research in infants and adults demonstrated that a stream of speech can be readily segmented based solely on the statistical and speech cues afforded by the input. Using functional magnetic resonance imaging (fMRI), the neural substrate of word segmentation was examined on-line as participants listened to three streams of concatenated syllables, containing either statistical regularities alone, statistical regularities and speech cues, or no cues. Despite the participants' inability to explicitly detect differences between the speech streams, neural activity differed significantly across conditions, with left-lateralized signal increases in temporal cortices observed only when participants listened to streams containing statistical regularities, particularly the stream containing speech cues. In a second fMRI study, designed to verify that word segmentation had implicitly taken place, participants listened to trisyllabic combinations that occurred with different frequencies in the streams of speech they just heard ("words," 45 times; "partwords," 15 times; "nonwords," once). Reliably greater activity in left inferior and middle frontal gyri was observed when comparing words with partwords and, to a lesser extent, when comparing partwords with nonwords. Activity in these regions, taken to index the implicit detection of word boundaries, was positively correlated with participants' rapid auditory processing skills. These findings provide a neural signature of on-line word segmentation in the mature brain and an initial model with which to study developmental changes in the neural architecture involved in processing speech cues during language learning.