Semantic Error Rate Research Articles

Mapping action naming in patients with gliomas: The influence of transitivity

ObjectivePatients with left perisylvian gliomas might undergo language mapping with nTMS in preparation for awake brain surgery. Action naming is an important addition to the presurgical language mapping protocol. However, it has not yet been determined whether specific action stimuli can influence mapping outcomes in terms of number and/or localisation of induced errors. MethodsWe investigated this question by employing tractography-based nTMS language mapping of the left arcuate fasciculus (AF) with two types of verbs: transitive and intransitive. Data were collected from 22 patients with a left perisylvian glioma. ResultsOur results demonstrated that nTMS induced a higher error rate with transitive rather than intransitive verbs, specifically during stimulation of the posterior temporal terminations of the left AF (transitive error rate: 8.3%; intransitive error rate: 4.8%). The effect was absent when gliomas displaced the temporal terminations of the AF. Also, nTMS triggered a higher number of semantic errors with transitive (vs intransitive) actions during stimulation of the posterior temporal terminations of the AF (semantic error rate – transitives: 3.3%; semantic error rate – intransitives: 0%). ConclusionOur work highlights that clinical outcomes of language mapping with nTMS are affected by the choice of linguistic stimuli. Transitive verbs may be suited to achieve optimal nTMS mapping outcomes in posterior temporal areas of the left AF in this population. Displacement of white matter terminations due to the tumor can affect these results, and semantic errors may indicate core language processes that can be mapped when administering transitive verbs.

Imitation Learning-Based Implicit Semantic-Aware Communication Networks: Multi-Layer Representation and Collaborative Reasoning

Semantic communication has recently attracted significant interest from both industry and academia due to its potential to transform the existing data-focused communication architecture towards a more generally intelligent and goal-oriented semantic-aware networking system. Despite its promising potential, semantic communications and semantic-aware networking are still in their infancy. Most existing works focus on transporting and delivering the explicit semantic information, e.g., labels or features of objects, that can be directly identified from the source signal. The original definition of semantics as well as recent results in cognitive neuroscience suggest that it is the implicit semantic information, in particular the hidden relations connecting different concepts and feature items that play the fundamental role in recognizing, communicating, and delivering the real semantic meanings of messages. Motivated by this observation, we propose a novel reasoning-based implicit semantic-aware communication network architecture that allows destination users to directly learn a reasoning mechanism that can automatically generate complex implicit semantic information based on a limited clue information sent by the source users. Our proposed architecture can be implemented in a multi-tier cloud/edge computing networks in which multiple tiers of cloud data center (CDC) and edge servers can collaborate and support efficient semantic encoding, decoding, and implicit semantic interpretation for multiple end-users. We introduce a new multi-layer representation of semantic information taking into consideration both the hierarchical structure of implicit semantics as well as the personalized inference preference of individual users. We model the semantic reasoning process as a reinforcement learning process and then propose an imitation-based semantic reasoning mechanism learning (iRML) solution to learning a reasoning policy that imitates the inference behavior of the source user. A federated graph convolutional network (GCN)-based collaborative reasoning solution is proposed to allow multiple edge servers to jointly construct a shared semantic interpretation model based on decentralized semantic message samples. Extensive experiments have been conducted based on real-world datasets to evaluate the performance of our proposed architecture. Numerical results confirm that iRML offers up to 25.8 dB improvement on the semantic symbol error rate, compared to the semantic-irrelevant communication solutions.

Deep Source-Channel Coding for Sentence Semantic Transmission With HARQ

Recently, semantic communication has been brought to the forefront because of its great success in deep learning (DL), especially Transformer. Even if semantic communication has been successfully applied in the sentence transmission to reduce semantic errors, existing architecture is usually fixed in the codeword length and is inefficient and inflexible for the varying sentence length. In this paper, we exploit hybrid automatic repeat request (HARQ) to reduce semantic transmission error further. We first combine semantic coding (SC) with Reed Solomon (RS) channel coding and HARQ, called SC-RS-HARQ, which exploits the superiority of the SC and the reliability of the conventional methods successfully. Although the SC-RS-HARQ is easily applied in the existing HARQ systems, we also develop an end-to-end architecture, called SCHARQ, to pursue the performance further. Numerical results demonstrate that SCHARQ significantly reduces the required number of bits for sentence semantic transmission and sentence error rate. Finally, we attempt to replace error detection from cyclic redundancy check to a similarity detection network called Sim32 to allow the receiver to reserve the wrong sentences with similar semantic information and to save transmission resources.

Controlling the past, owning the present, and future: cholinergic modulation decreases semantic perseverations in a person with post-stroke aphasia

ABSTRACT Background Perseverations in speech production tasks represent a pervasive symptom of chronic aphasia. Semantic perseverations (SPs) are defined as repetitive and unconscious production of specific linguistic forms previously produced, heard, or seen which share semantic relatedness with the target word. Neurochemically, SPs have been attributed to weakened activation of the target word due to the depletion of neurotransmitter systems (acetylcholine and dopamine) occurring in the context of activation of semantic competitors. Aims The present pilot study seeks to evaluate the effectiveness of a cholinergic enhancing drug combined with speech language therapy in reducing SPs and other non-perseverative semantic errors in a person with chronic post-stroke aphasia. Methods & Procedures Combined therapy of the cholinesterase inhibitor donepezil and conventional speech-language therapy (SLT) (2 hours/week) over 16 weeks was administered to a woman with chronic fluent post-stroke aphasia showing high rate of semantic errors. Aphasia and SPs assessments were performed at four different time-points across the study: at baseline, after donepezil 5 mg, donepezil 10 mg, and after 4-weeks of washout. The changes induced by the treatments on the occurrence of SPs and semantic paraphasias during a picture naming task were evaluated. By using a specific statistical methodology, we performed a fine-grained analysis of the frequency of SPs and their temporal course at the different time-points to dissect changes induced by the treatment. Results At baseline, there were significantly more SPs than expected by a random distribution, whereas a marked reduction of these errors was found in the three following evaluations. A significant reduction in aphasia severity was also found with high donepezil doses and this improvement maintained after a 4-week washout period. Everyday communication improved with low doses of donepezil. Conclusion Our findings suggest that decreased activity of the cholinergic system may exert a permissive role for the production of SPs and highlight the importance of combining cholinergic agents with speech-language therapy to reduce SPs in aphasia at the time that other language deficits are also improved.

Semantic interference in speech error production in a randomised continuous naming task: evidence from aphasia

ABSTRACTNaming pictures from the same semantic category hinders subsequent naming from that category (i.e. semantic interference), irrespective of the number of intervening different-category exemplars named. Persistent semantic interference has been well documented in chronometric studies, and has been attributed to experience-driven adjustments in the strength of connections between semantic and lexical representations. However, whether parallel effects exist in speech error data remains unclear. In the current study, people with aphasia, a speaker population prone to naming errors, provided naming responses to a large picture corpus presented in random order that comprised multiple exemplars drawn from several different categories. We found persistent semantic interference in the task in semantic error rates specifically, and that semantic similarity between consecutive related exemplars modulated the effect. The results provide further evidence for the presumed lexical-semantic locus and mechanism(s) underlying semantic interference.

Regional Brain Dysfunction Associated with Semantic Errors in Comprehension.

Here we illustrate how investigation of individuals acutely after stroke, before structure/function reorganization through recovery or rehabilitation, can be helpful in answering questions about the role of specific brain regions in language functions. Although there is converging evidence from a variety of sources that the left posterior-superior temporal gyrus plays some role in spoken word comprehension, its precise role in this function has not been established. We hypothesized that this region is essential for distinguishing between semantically related words, because it is critical for linking the spoken word to the complete semantic representation. We tested this hypothesis in 127 individuals with 48 hours of acute ischemic stroke, before the opportunity for reorganization or recovery. We identified tissue dysfunction (acute infarct and/or hypoperfusion) in gray and white matter parcels of the left hemisphere, and we evaluated the association between rate of semantic errors in a word-picture verification tasks and extent of tissue dysfunction in each region. We found that after correcting for lesion volume and multiple comparisons, the rate of semantic errors correlated with the extent of tissue dysfunction in left posterior-superior temporal gyrus and retrolenticular white matter.

Revisiting the dual-nature account of homophony: Evidence from picture naming in aphasia

Revisiting the dual-nature account of homophony: Evidence from picture naming in aphasia

Domain Transfer for Deep Natural Language Generation from Abstract Meaning Representations

Stochastic natural language generation systems that are trained from labelled datasets are often domainspecific in their annotation and in their mapping from semantic input representations to lexical-syntactic outputs. As a result, learnt models fail to generalize across domains, heavily restricting their usability beyond single applications. In this article, we focus on the problem of domain adaptation for natural language generation. We show how linguistic knowledge from a source domain, for which labelled data is available, can be adapted to a target domain by reusing training data across domains. As a key to this, we propose to employ abstract meaning representations as a common semantic representation across domains. We model natural language generation as a long short-term memory recurrent neural network encoderdecoder, in which one recurrent neural network learns a latent representation of a semantic input, and a second recurrent neural network learns to decode it to a sequence of words. We show that the learnt representations can be transferred across domains and can be leveraged effectively to improve training on new unseen domains. Experiments in three different domains and with six datasets demonstrate that the lexical-syntactic constructions learnt in one domain can be transferred to new domains and achieve up to 75-100% of the performance of in-domain training. This is based on objective metrics such as BLEU and semantic error rate and a subjective human rating study. Training a policy from prior knowledge from a different domain is consistently better than pure in-domain training by up to 10%.

Assessing Field Spectroscopy Metadata Quality

This paper presents the proposed criteria for measuring the quality and completeness of field spectroscopy metadata in a spectral archive. Definitions for metadata quality and completeness for field spectroscopy datasets are introduced. Unique methods for measuring quality and completeness of metadata to meet the requirements of field spectroscopy datasets are presented. Field spectroscopy metadata quality can be defined in terms of (but is not limited to) logical consistency, lineage, semantic and syntactic error rates, compliance with a quality standard, quality assurance by a recognized authority, and reputational authority of the data owners/data creators. Two spectral libraries are examined as case studies of operationalized metadata policies, and the degree to which they are aligned with the needs of field spectroscopy scientists. The case studies reveal that the metadata in publicly available spectral datasets are underperforming on the quality and completeness measures. This paper is part two in a series examining the issues central to a metadata standard for field spectroscopy datasets.

A fresh look at the predictors of naming accuracy and errors in Alzheimer's disease

In recent years, a considerable number of studies have tried to establish which characteristics of objects and their names predict the responses of patients with Alzheimer's disease (AD) in the picture-naming task. The frequency of use of words and their age of acquisition (AoA) have been implicated as two of the most influential variables, with naming being best preserved for objects with high-frequency, early-acquired names. The present study takes a fresh look at the predictors of naming success in Spanish and English AD patients using a range of measures of word frequency and AoA along with visual complexity, imageability, and word length as predictors. Analyses using generalized linear mixed modelling found that naming accuracy was better predicted by AoA ratings taken from older adults than conventional ratings from young adults. Older frequency measures based on written language samples predicted accuracy better than more modern measures based on the frequencies of words in film subtitles. Replacing adult frequency with an estimate of cumulative (lifespan) frequency did not reduce the impact of AoA. Semantic error rates were predicted by both written word frequency and senior AoA while null response errors were only predicted by frequency. Visual complexity, imageability, and word length did not predict naming accuracy or errors.

Effects of near and distant semantic neighbors on word production

One way to examine the dynamics of word processing is to investigate how processing is affected by the co-activation of similar words ("neighbors"). A unique prediction of attractor dynamical models is that near neighbors should exert inhibitory effects and distant neighbors should exert facilitative effects. In study 1, data from 62 unselected chronic aphasia patients revealed a higher rate of semantic errors for words with many near semantic neighbors and fewer semantic errors for words with many distant semantic neighbors. In study 2, this basic result was replicated in controls using a speeded picture-naming paradigm. Together, these two studies provide strong new evidence consistent with the attractor dynamics view of neighborhood effects. In addition, analyses of correlations between effect sizes and lesion locations, and comparisons with the existing literature on semantic deficits in aphasia and the speeded picture-naming paradigm, all provide converging evidence that the semantic error patterns found in the present studies were due to disruptions of cognitive control mechanisms.

Anterior temporal involvement in semantic word retrieval: voxel-based lesion-symptom mapping evidence from aphasia

Analysis of error types provides useful information about the stages and processes involved in normal and aphasic word production. In picture naming, semantic errors (horse for goat) generally result from something having gone awry in lexical access such that the right concept was mapped to the wrong word. This study used the new lesion analysis technique known as voxel-based lesion-symptom mapping to investigate the locus of lesions that give rise to semantic naming errors. Semantic errors were obtained from 64 individuals with post-stroke aphasia, who also underwent high-resolution structural brain scans. Whole brain voxel-based lesion-symptom mapping was carried out to determine where lesion status predicted semantic error rate. The strongest associations were found in the left anterior to mid middle temporal gyrus. This area also showed strong and significant effects in further analyses that statistically controlled for deficits in pre-lexical, conceptualization processes that might have contributed to semantic error production. This study is the first to demonstrate a specific and necessary role for the left anterior temporal lobe in mapping concepts to words in production. We hypothesize that this role consists in the conveyance of fine-grained semantic distinctions to the lexical system. Our results line up with evidence from semantic dementia, the convergence zone framework and meta-analyses of neuroimaging studies on word production. At the same time, they cast doubt on the classical linkage of semantic error production to lesions in and around Wernicke's area.

“L” is for tiger: Effects of phonological (mis)cueing on picture naming in semantic aphasia

Semantic aphasia (SA) refers to a subset of aphasic patients who exhibit multimodal semantic deficits (Jefferies, E., & Lambon Ralph, M. (2006). Semantic impairment in stroke aphasia vs semantic dementia: a case-series comparison. Brain, 129(8), 2132–2147). Consistent with their underlying semantic control deficit, SA picture naming accuracy can be improved considerably with a correct phonological cue. The performance of normal individuals in the tempo picture naming paradigm reveals an increased impact of both correct and incorrect phonological cueing, and it has been suggested that this technique reduces resources available for semantic control in neurologically-intact participants (Hodgson, C., & Lambon Ralph, M. (2008). Mimicking aphasic semantic naming errors in normal speech production: Evidence from a novel experimental paradigm. Brain and Language, 104(1), 89–101.). We tested this hypothesis by considering the impact of both correct and incorrect phonological cues on picture naming in a case series of SA patients, using exactly the same items as those presented to normal participants for tempo naming. The results confirmed the positive effect of correct cues and revealed for the first time the negative effects of category co-ordinate miscues amongst these patients in both overall accuracy and semantic error rates. The implications of our results for current speech production models are considered.

The many places of frequency: Evidence for a novel locus of the lexical frequency effect in word production

The effect of lexical frequency on language-processing tasks is exceptionally reliable. For example, pictures with higher frequency names are named faster and more accurately than those with lower frequency names. Experiments with normal participants and patients strongly suggest that this production effect arises at the level of lexical access. Further work has suggested that within lexical access this effect arises at the level of lexical representations. Here we present patient E.C. who shows an effect of lexical frequency on his nonword error rate. The best explanation of his performance is that there is an additional locus of frequency at the interface of lexical and segmental representational levels. We confirm this hypothesis by showing that only computational models with frequency at this new locus can produce a similar error pattern to that of patient E.C. Finally, in an analysis of a large group of Italian patients, we show that there exist patients who replicate E.C.'s pattern of results and others who show the complementary pattern of frequency effects on semantic error rates. Our results combined with previous findings suggest that frequency plays a role throughout the process of lexical access.

An Integrative and Discriminative Technique for Spoken Utterance Classification

Traditional methods of spoken utterance classification (SUC) adopt two independently trained phases. In the first phase, an automatic speech recognition (ASR) module returns the most likely sentence for the observed acoustic signal. In the second phase, a semantic classifier transforms the resulting sentence into the most likely semantic class. Since the two phases are isolated from each other, such traditional SUC systems are suboptimal. In this paper, we present a novel integrative and discriminative learning technique for SUC to alleviate this problem, and thereby, reduce the semantic classification error rate (CER). Our approach revolves around the effective use of the N-best lists generated by the ASR module to reduce semantic classification errors. The N-best list sentences are first rescored using all the available knowledge sources. Then, the sentence that is most likely to helps reduce the CER are extracted from the N-best lists as well as those sentences that are most likely to increase the CER. These sentences are used to discriminatively train the language and semantic-classifier models to minimize the overall semantic CER. Our experiments resulted in a reduction of CER from its initial value of 4.92% to 4.04% in the standard ATIS task.

Morphological structure in the lexical representation of prefixed words: Evidence from speech errors

The present study aims to determine whether semantic relatedness plays a role in the production of speech errors involving derivational morphemes. A word order competition technique was used to induce morpheme and syllable exchange errors. Semantic relatedness was manipulated by contrasting error rates for prefixed words derived from free stems to those derived from bound roots. Significantly more morpheme errors were elicited in the two prefixed conditions compared with the control condition, replicating prior findings in French. Crucially, the two prefixed conditions elicited equal numbers of morpheme errors and there was no correlation between semantic relatedness and error rates. Taken together, the results strongly support a model of speech production in which derivational morphemes are represented at the form level and are not influenced by the degree of semantic relatedness within a morphological family.

Deep dyslexia in the two languages of an Arabic/French bilingual patient

We present a single case study of an Arabic/French bilingual patient, ZT, who, at the age of 32, suffered a cerebral vascular accident that resulted in a massive infarct in the left peri-sylvian region. ZT's reading displays the characteristics of the deep dyslexia syndrome in both languages, that is, production of semantic, visual, and morphological errors, and concreteness effect in reading aloud and impossibility of reading nonwords. In the first part of this paper, using a three-route model of reading, we account for the patient's performance by positing functional lesions, which affect the non-lexical, the semantic lexical and the non-semantic lexical routes of reading. Phonological priming observed in a cross-language visual lexical decision task indicates that implicit assembled phonological recoding is possible. The above lesions and implicit nonword reading characterize the output form of deep dyslexia. However, error distribution reveals dissociations across languages (e.g. the semantic error rate is higher in French whereas translations are more frequent in the Arabic testing) that cannot be accounted for within a three-route model. In the second part, extensions to Plaut and Shallice's connectionist model ( Cognitive Neuropsychology, 10 (5) (1993) 377) are proposed to account for the translinguistic errors observed. ZT's error distribution is compared to that obtained by Plaut and Shallice after lesions had been applied at different locations through the 40–60 network. The overall syndrome of deep dyslexia found in both languages is explained as resulting from lesions along the direct (O→I) and output (S→Ip, Ip→P) pathways of reading. Lesions along the output pathway mostly affecting S→Ip connections in French and Ip→P connections in Arabic account for discrepancies in ZT's error pattern across tasks and languages. This case study demonstrates the superiority of a connectionist approach for predicting the error pattern in deep dyslexia.

Are Semantic Errors Actually Semantic?: Evidence from Alzheimer's Disease

Patients with Alzheimer's disease (AD) produce a high rate of semantic errors when naming to confrontation. This is considered to be one of the many consequences of their semantic memory deficit. However, it has been shown, in aphasic patients with focal lesions, that semantic errors could arise from impairment to any one of the levels in the naming process. To check this hypothesis in AD, we assessed in 15 patients the capacity to name and access semantic knowledge (by multiple-choice probe questions) about 14 objects presented successively in the visual, tactile, auditory, and verbal modalities. In the visual naming task, 33 errors were recorded: 26 (78.8%) were semantic and 7 (21.2%) were unrelated errors. Of the 26 semantic errors, 8 were related to a deficit of the semantic knowledge related to the item and 17 to a deficit in the retrieval of the phonological form of the word. One was associated with a deficit of access to semantic knowledge in the visual modality. The 7 unrelated errors were associated with a loss of semantic knowledge for 4 and deficit of access to the phonological form for 3. In conclusion, this study shows that semantic errors do not systematically reflect a deficit of semantic knowledge in Alzheimer's disease. It also seems that unrelated errors are more frequently related to semantic deficits than semantic errors in this population.

On knowing the meaning of words we are unable to report--confirmation of a guessing explanation.

This experiment investigated Allport's (1977) claim that normal subjects can be induced to make semantic errors of the type characterised by “deep dyslexia”. The procedure followed that used by Allport except that a subject based control for the chance level of semantic errors was included. The results replicated the original finding by producing a semantic error rate of 12·7%. It was shown, however, that this rate was not significantly higher than the calculated chance rate of 11·5%. It was concluded, therefore, that the semantic error rates reported by Allport can be accounted for in terms of a guessing artefact and thus they bear no relevance to the problem of deep dyslexia.