Search

For qualitative data analysis (QDA), researchers assign codes to text segments to arrange the information into topics or concepts. These annotations facilitate information retrieval and the identification of emerging patterns in unstructured data. However, this metadata is typically not published or reused after the research. Subsequent studies with similar research questions require a new definition of codes and do not benefit from other analysts' experience. Machine learning (ML) based classification seeded with such data remains a challenging task due to the ambiguity of code definitions and the inherent subjectivity of the exercise. Previous attempts to support QDA using ML rely on linear models and only examined individual datasets that were either smaller or coded specifically for this purpose. However, we show that modern approaches effectively capture at least part of the codes' semantics and may generalize to multiple studies. We analyze the performance of multiple classifiers across three large real-world datasets. Furthermore, we propose an ML-based approach to identify semantic relations of codes in different studies to show thematic faceting, enhance retrieval of related content, or bootstrap the coding process. These are encouraging results that suggest how analysts might benefit from prior interpretation efforts, potentially yielding new insights into qualitative data. [ABSTRACT FROM AUTHOR]

Formal pattern languages are used in industry to communicate and analyse requirements, as they are said to be both machine-readable and intuitively understandable for humans. The questions arise to what extent this intuitive understanding of a pattern language is in agreement with its formal semantics and whether this understanding can be increased systematically. We present two consecutive empirical experiments to address these questions. The formal semantics serves as an objective judge on the intuitive understanding. Our experiments confirm the practical usefulness of HanforPL insofar the intuition matches the formal semantics in most practically relevant cases. They also reveal a number of edge cases where even a prior exposure to formal logic is not a guarantee for correct understanding. We present and validate systematic adjustments to the patterns, leading to several large increases in understandability but come at the cost of new, but less impactful ambiguities. We demonstrate how an inquiry on the alignment of the intuitive and formal semantics of a pattern language can help to understand and improve the language. While results regarding the understandability of HanforPL are favourable in commonly used cases, there is potential for improvement. The systematic adaption of patterns shows that small modifications may have large effects on the alignment of formal and intuitive semantics, and that modification must be considered with caution in the context of the respective pattern to avoid unintentionally adding new ambiguities. This article is an extension of our published REFSQ paper. [ABSTRACT FROM AUTHOR]

In this paper, we investigate ambiguity in language use that is intended by the speaker. Our overall argument is that, as a general feature of language resulting from the indeterminacy of meaning at all levels of language, ambiguity can be taken advantage of in specific contexts for a host of purposes. Given that an ambiguous utterance has (at least) two possible meanings – A and B – there are two logical possibilities for a speaker in a deliberate ambiguity: to mean both A and B or to mean only A (or B). However, a closer examination of the literature and available data reveal a far more complicated picture: there are more strategies under either of these possibilities, each of which closely interacts with the context and is motivated by the speaker's purpose of communication in that context. The treatment of deliberate ambiguity in the present study is thus more coherent theoretically and more comprehensive in coverage than what is found in the sizable literature on the subject. • Ambiguity can be deliberate and advantageous for both the speaker and hearer. • Deliberate ambiguity leads to seven strategies. • Five of the seven are speaker-oriented, two are hearer-oriented. • Deliberate ambiguity strategies are motivated by specific contextual factors. [ABSTRACT FROM AUTHOR]

A natural language interface (NLI) to databases is an interface that translates a natural language question to a structured query that is executable by database management systems (DBMS). However, an NLI that is trained in the general domain is hard to apply in the spatial domain due to the idiosyncrasy and expressiveness of the spatial semantics. Moreover, there are a wide range of database servers available, and a unilingual NLI model limits its practical usage. In this article, we propose to not only address the spatial domain generalization challenge, but also support multilingual back-end, i.e., supporting different query languages, such as SQL and Prolog. For the challenge of spatial semantics, we propose a spatial comprehension model that is able to recognize the meaning of spatial entities based on the semantics of context and effectively resolve the ambiguity given the spatial semantics. The spatial semantics learned from the spatial comprehension model is then injected to the natural language question to ease the burden of capturing the spatial-specific semantics. We also propose to add a prefix symbol to support the multilingual back-end (e.g., query languages). With our spatial comprehension model and symbol injections, our NLI for the spatial domain, named SpatialNLI, is able to capture the semantic structure of the question and translate it to the corresponding syntax of an executable query accurately. We also experimentally ascertain that SpatialNLI outperforms state-of-the-art methods. [ABSTRACT FROM AUTHOR]

The article "Dynamic inquisitive semantics—looking ahead and looking back" explores the topic of dynamic inquisitive semantics (InqD) and its relationship to previous frameworks for question semantics. The authors discuss the analysis of mention-some and mention-all questions and weak/strong donkey anaphora, proposing that wh-words like "who" can be interpreted with or without the max* operator. However, a commentator presents an alternative account of donkey anaphora that eliminates this ambiguity. The article also examines the analysis of donkey anaphora using implication and the relationship between negation and implication. The authors discuss different approaches to analyzing weak and strong readings of sentences with indefinite quantifiers, highlighting the complexities and challenges of each approach. They acknowledge that further work is needed to fully explain certain phenomena. The article also discusses alternative approaches to the semantics of wh-questions and the syntax of the left periphery, as well as different logical frameworks for question semantics. The authors position InqD as a "third generation" framework that builds on earlier frameworks and emphasize the need for further research and refinement. [Extracted from the article]

Lexical ambiguity is pervasive in language, and the nature of the representations of an ambiguous word's multiple meanings is yet to be fully understood. With a special focus on Chinese characters, the present study first established that native speaker's perception about a character's number of meanings was heavily influenced by the availability of its distinct word formations, while whether these meanings would be perceived to be closely related was driven by further conceptual analysis. These notions were operationalized as two computed metrics, which assessed the degree of dispersion across individual word formations and the degree of propinquity across clusters of word formations, respectively, in a distributional semantic space. The observed correlations between the computed and the perceived metrics indicated that the utility of word formations to tap into meaning representations of Chinese characters was indeed cognitively plausible. The results have demonstrated the extent to which distributional semantics could inform about meaning representations of Chinese characters, which has theoretical implications for the representation of ambiguous words more generally. [ABSTRACT FROM AUTHOR]

Among various types of nominal phrases, the appositive construction has long been a relatively under-researched subject in the literature of Chinese linguistics. This paper centers on the use of the appositive construction [P(ronoun)-Num(eral)-Cl(assifier)-Nominal Phrase (NP)] in Mandarin Chinese. Upon revealing a series of asymmetries in syntax, semantics, and pragmatics, this paper proposes an attributional versus identificational distinction for Chinese appositives. Specifically, for the attributional case, the apposition (i.e. [Num-Cl-NP]) is property-denoting and serves to rationalize the speaker's evaluation about the referent denoted by the anchor (i.e. the P); for the identificational case, the apposition is individual-denoting and serves to facilitate referent identification of the anchor by picking out an identifiable quantified set of discourse referents from the given context. To formally capture this distinction, this paper develops a dichotomous analysis for the syntax of Chinese appositives. The non-unified treatment not only offers an effective explanation for the asymmetries exhibited by the appositives under different cases, but also contributes to a better understanding of Chinese appositives in general. [ABSTRACT FROM AUTHOR]

In this paper, we derive a notion of word meaning in context that characterizes meaning as both intensional and conceptual. We introduce a framework for specifying local as well as global constraints on word meaning in context, together with their interactions, thus modelling a wide range of lexical shifts and ambiguities observed in utterance interpretation. We represent sentence meaning as a situation description system , a probabilistic model which takes utterance understanding to be the mental process of describing to oneself one or more situations that would account for an observed utterance. We show how the system can be implemented in practice, and apply it to examples containing various contextualisation phenomena. [ABSTRACT FROM AUTHOR]

Most words have multiple meanings, but there are foundationally distinct accounts for this. Categorical theories posit that humans maintain discrete entries for distinct word meanings, as in a dictionary. Continuous ones eschew discrete sense representations, arguing that word meanings are best characterized as trajectories through a continuous state space. Both kinds of approach face empirical challenges. In response, we introduce two novel "hybrid" theories, which reconcile discrete sense representations with a continuous view of word meaning. We then report on two behavioral experiments, pairing them with an analytical approach relying on neural language models to test these competing accounts. The experimental results are best explained by one of the novel hybrid accounts, which posits both distinct sense representations and a continuous meaning space. This hybrid account accommodates both the dynamic, context-dependent nature of word meaning, as well as the behavioral evidence for category-like structure in human lexical knowledge. We further develop and quantify the predictive power of several computational implementations of this hybrid account. These results raise questions for future research on lexical ambiguity, such as why and when discrete sense representations might emerge in the first place. They also connect to more general questions about the role of discrete versus gradient representations in cognitive processes and suggest that at least in this case, the best explanation is one that integrates both factors: Word meaning is both categorical and continuous. [ABSTRACT FROM AUTHOR]

Recent experimental studies have claimed to find evidence for the view that natural kind terms such as "water" are ambiguous: that they have two extensions, one determined by superficial properties, the other by underlying essence. In an online experiment, we presented to 600 participants scenarios describing discoveries of novel samples that differ in deep structure from samples of a familiar kind but are superficially identical, such as a water‐like substance that is not composed of H2O. We used three different types of question sets to probe whether the participants considered the sample as a member of the kind or not. Our results did not confirm the predictions of the ambiguity view. They were, rather, consistent with views that take underlying essences to be the sole criterion for membership in a natural kind. [ABSTRACT FROM AUTHOR]

In her engaging book, The referential mechanism of proper names, Li presents empirical studies involving American and Chinese laypeople. Li interprets her results as supporting an epistemic‐perspective reading of the variability in referential intuitions on proper names. Building upon this thesis, Li defends the ambiguity view, claiming that names are ambiguous between a descriptivist and a causal‐historical meaning. I argue that either Li's data do not enable a comparison of the two theories of reference, or support for the ambiguity view is limited to the Chinese sample at most and does not rely upon the inference that Li employs. [ABSTRACT FROM AUTHOR]

Aim: We aimed to apply ontological techniques to address semantic ambiguities in protected area and conservation informatics. By doing so, we aimed to create a coherent, machine‐actionable semantic representation of the biogeographic areas (which often overlap protected areas) to support more efficient and standardized informatics, supporting research and decision‐making. We present BIOREALM, the first informatic ontology for comparative biogeography. Location: Global. Taxon: Any taxon can be integrated in BIOREALM. Methods: We convert a cladogram of biogeographic areas—generated by a process known as bioregionalization—into a series of ontological classes. Areas of endemism are treated as formal objects related by hierarchical relationships and constrained by a condition of monophyly. We use semantic web approaches to extend the Environment Ontology (ENVO) with classes for (often semantically confounded) biogeographic entities, including biogeographic areas, areas of endemism and endemic areas. We applied this approach to a bioregionalization of Australia as a case study. In all, 20 subregions which are part of the Austral Bioregionalisation Atlas have been selected for the study and integrated in BIOREALM. Results: We have created an ontology—formatted in the Web Ontology Language and adhering to the practices of the Open Biomedical and Biological Ontology Foundry—which provides a rigorous, extensible and machine‐actionable framework that can improve biogeographic analyses and interoperability between systems. One main class and 20 individuals per class were implemented. Main Conclusions: BIOREALM encodes a model‐theoretic view of endemism using semantic web approaches, offering new avenues to express and analyse biogeographic units. This approach offers a means to identify monophyletic biogeographic areas for conservation, based on specific combinations of monophyletic endemic taxa. Such an ontology provides knowledge representation solutions which supports interoperability along the FAIR (Findable, Accessible, Interoperable, Reusable) principles, thus fostering more consistent ecological informatics. [ABSTRACT FROM AUTHOR]

Word sense disambiguation (WSD) is a process used to determine the most appropriate meaning of a word in a given contextual framework, particularly when the word is ambiguous. While WSD has been extensively studied for English, it remains a challenging problem for resource-scarce languages such as Hindi. Therefore, it is crucial to address ambiguity in Hindi to effectively and efficiently utilize it on the web for various applications such as machine translation, information retrieval, etc. The rich linguistic structure of Hindi, characterized by complex morphological variations and syntactic nuances, presents unique challenges in accurately determining the intended sense of a word within a given context. This review paper presents an overview of different approaches employed to resolve the ambiguity of Hindi words, including supervised, unsupervised, and knowledge-based methods. Additionally, the paper discusses applications, identifies open problems, presents conclusions, and suggests future research directions. [ABSTRACT FROM AUTHOR]

The following chapter offers an introduction and overview to the special issue, Why antiracism and critical whiteness now? It highlights how racial concerns are—at their core—fundamentally relational, emblematizing not only antiracism's greatest hope but also its most tenacious, enduring challenge. [ABSTRACT FROM AUTHOR]

We propose a new modal logic endowed with a simple deductive system to interpret Aristotle's theory of the modal syllogism. While being inspired by standard propositional modal logic, it is also a logic of terms that admits a (sound) extensional semantics involving possible states-of-affairs in a given world. Applied to the analysis of Aristotle's modal syllogistic as found in the Prior Analytics A8-22, it sheds light on various fine-grained distinctions which when made allow us to clarify some ambiguities and obtain a completely consistent system and prove all of the modal syllogisms considered valid by Aristotle. This logic allows us also to make a connection with the axioms of modern propositional modal logic and to perceive to what extent these are implicit in Aristotle's reasoning. Further work will involve addressing the question of the completeness of this logic (or variants thereof) together with an extension of the logic to include a calculus of relations (for instance the relational syllogistic treated in Galen's Introduction to Logic) which Slomkowsky has argued is already found in the Topics. [ABSTRACT FROM AUTHOR]

Some words have more than one translation across languages. Such translation-ambiguous words are harder to learn, recognize, and produce for individuals across the language learning spectrum. Past research demonstrates that learning both translations of translation-ambiguous words on consecutive trials confers an accuracy advantage relative to learning them on separate sessions. We tested the proposal that presenting the two translations of translation-ambiguous words simultaneously on the screen could facilitate the formation of a more integrated mapping, because this would enable learners to make direct comparisons between them, whether implicitly or explicitly. We predicted that this facilitation would especially hold for translation-ambiguous words with related translations. Fifty native English speakers learned 48 German words with one or two translations that varied in the meaning similarity of their translations. Paired associate training took place on a Monday, and a first language (L1) to second language (L2) translation production test took place on Wednesday and Friday. Generally, higher translation similarity facilitated translation speed. In accuracy, training condition interacted with the similarity of the translations; translation accuracy was more affected by translation similarity in the simultaneous condition and went up as similarity increased. Overall, the consecutive condition demonstrated higher accuracy and faster reaction times than the simultaneous training condition, suggesting that learners may have been unable to successfully divide their study time between multiple words on the screen without explicit instruction. [ABSTRACT FROM AUTHOR]

The present study investigates native (L1) and second language (L2) processing of scope ambiguities in English sentences containing the universal quantifier every in subject NP and negation. Previous studies in L1 and L2 processing of scope ambiguities have found speakers to generally employ a 'minimal effort' principle that highly prefers the surface scope reading regardless of contextual support because accessing the inverse scope reading incurs significant processing cost. The present study compared L1 and L2 scope judgments and processing strategies of sentences such as Every horse didn't jump over the fence and examined whether the two groups differ in their speed and manner of analysis. Thirty native English speakers and 42 Korean learners of English participated in a self-paced reading/interpretation task that has context (ambiguous vs. unambiguous) and scope reading (surface vs. inverse) as variables. The results revealed significant differences in scope endorsement rates with L2 learners arriving at the surface scope as the dominant reading and L1 learners' judgments being highly dependent on contextual ambiguity. Moreover, L1 vs. L2 differences in processing strategies were found: L2 learners exhibited a strong tendency to arrive at the most economical interpretation while L1 speakers consulted detailed syntactic and semantic rules of computation. [ABSTRACT FROM AUTHOR]

In this paper, we demonstrate the effects of Information Theory measures on the processing of polysemous nouns and reveal that the sensitivity to multiple related senses can be learned from the linguistic context. We collected large-scale data and applied a correlation design to show that an increase in sense uncertainty (or sense diversity) is followed by a faster visual lexical decision. The facilitatory effect of sense uncertainty was revealed by the predictive power of entropy, followed by the additional analysis, which revealed that both the number of senses and the balance of sense probabilities affected processing. For the first time, the balance of sense probabilities was described via redundancy to demonstrate the effect of the numerical description of the balance of sense probabilities. Finally, we crossed distribution semantics and discrimination learning to show that polysemy effects can arise as a consequence of the principles of error-driven learning. [ABSTRACT FROM AUTHOR]

The article discusses the concept of copredication, which is the use of multiple predicates to describe a single subject. It argues that copredication poses a challenge for traditional truth-conditional semantics and suggests that it should be seen as an illusion, where the linguistic content may be coherent but not linguistically determined. The text explores the relationship between copredication and polysemy, and discusses different models of lexical items and their role in determining truth conditions. The author proposes that copredication should be understood as a form of illusion, requiring a nuanced approach to semantics. [Extracted from the article]

One of the main features of José Saramago's fictional prose is the reflection on the use of language and, in particular, the lexicon. No other recent Portuguese author has or has had such an evident and reiterated concern. In Saramago's works, we find at each step an author that is concerned to explain to the reader the use of a certain word, reflecting on its origin and semantic peculiarities. Many of these reflections may be considered part of what in linguistics is known as implicit lexicography (VERDELHO, 1995), which refers to the act of recording, in the text itself, lexicographic or dictionarystyle definitions. Implicit lexicography occurs whenever an author presents a definition of a word or expression, with the aim of clarifying its meaning or avoiding ambiguity. The objective of this study is to identify in José Saramago's fictional works the passages in which the author presents lexicographic definitions. Such definitions are well-marked in the text and are easily distinguished from the rest of the discourse, although they are also part of it. The second term of the definition is often introduced by the verbal form é ('is'), as in «um machuco é um chaparro novo», although the author makes use of other expressions: quer dizer ('that is to say'), isto quer dizer ('this means'), isto é ('that is'), significa ('it means'), maneira de dizer ('another way of saying'), para quem não saiba ('for those who don't know'), etc. [ABSTRACT FROM AUTHOR]

Scientific literature, as the major medium that carries knowledge between scientists, exhibits explosive growth in the last century. Despite the frequent use of many tangible measures, to quantify the influence of literature from different perspectives, it remains unclear how knowledge is embodied and measured among tremendous scientific productivity, as knowledge underlying scientific literature is abstract and difficult to concretize. In this regard, there has laid a vacancy in the theoretical embodiment of knowledge for their evaluation and excavation. Here, for the first time, we quantify the knowledge from the perspective of information structurization and define a new measure of knowledge quantification index (KQI) that leverages the extent of disorder difference caused by hierarchical structure in the citation network to represent knowledge production in the literature. Built upon 214 million articles, published from 1800 to 2021, KQI is demonstrated for mining influential classics and laureates that are omitted by traditional metrics, thanks to in-depth utilization of structure. Due to the additivity of entropy and the interconnectivity of the network, KQI assembles numerous scientific impact metrics into one and gains interpretability and resistance to manipulation. In addition, KQI explores a new perspective regarding knowledge measurement through entropy and structure, utilizing structure rather than semantics to avoid ambiguity and attain applicability. [ABSTRACT FROM AUTHOR]

Translating experimental tasks that were designed to investigate differences between conditions at the group-level into valid and reliable instruments to measure individual differences in cognitive skills is challenging (Hedge et al., 2018; Rouder et al., 2019; Rouder & Haaf, 2019). For psycholinguists, the additional complexities associated with selecting or constructing language stimuli, and the need for appropriate well-matched baseline conditions make this endeavour particularly complex. In a typical experiment, a process-of-interest (e.g. ambiguity resolution) is targeted by contrasting performance in an experimental condition with performance in a wellmatched control condition. In many cases, careful between-condition matching precludes the same participant from encountering all stimulus items. Unfortunately, solutions that work for group-level research (e.g. constructing counterbalanced experiment versions) are inappropriate for individual-differences designs. As a case study, we report an ambiguity resolution experiment that illustrates the steps that researchers can take to address this issue and assess whether their measurement instrument is both valid and reliable. On the basis of our findings, we caution against the widespread approach of using datasets from group-level studies to also answer important questions about individual differences. [ABSTRACT FROM AUTHOR]

The memory effects of semantic attributes (e.g., concreteness, familiarity, valence) have long been studied by manipulating their average perceived intensities, as quantified in word rating norms. The semantic ambiguity hypothesis specifies that the uncertainty as well as the intensity of semantic attributes is processed when words are encoded. Testing that hypothesis requires a normed measure of ambiguity, so that ambiguity and intensity can be manipulated independently. The standard deviation (SD) of intensity ratings has been used for that purpose, which has produced three characteristic ambiguity effects. Owing to the recency of such research, fundamental questions remain about the validity of this method of measuring ambiguity and about its process-level effects on memory. In a validity experiment, we found that the rating SDs of six semantic attributes (arousal, concreteness, familiarity, meaningfulness, negative valence, positive valence) passed tests of concurrent and predictive validity. In three memory experiments, we found that manipulating rating SDs had a specific effect on retrieval: It influenced subjects' ability to use reconstructive retrieval to recall words. That pattern was predicted by the current theoretical explanation of how ambiguity benefits memory. [ABSTRACT FROM AUTHOR]

Purpose: This study investigated the influence of L1 co-activation and meaning frequency of homonyms on L2 lexical access of Brazilian Portuguese–English bilinguals during an L2 meaning decision task. Design: Eighty-four university students completed a meaning decision task—with cognate homonyms—a meaning recognition task, a Language History Questionnaire, and an L2 proficiency test. The meaning decision task was composed of 128 prime words followed by one out of three possible targets: related to the dominant meaning, related to the subordinate one, or unrelated. Independent variables were cognate status (64 cognate × 64 non-cognate words), homonym status (64 homonym × 64 non-homonyms), prime-sharedness (16 dominant meaning shared words × 16 subordinate meaning shared words), and target-relatedness (64 related to dominant meaning × 64 related to subordinate meaning × 64 unrelated words). Dependent variables were reaction times and accuracy. Data and analysis: Data were analyzed using linear mixed effect models in R. Findings: Results showed that, in this sample of bilinguals, processing of target words related to the meaning of the prime was confounded when primes were homonyms. In general, this interference was not decreased by the co-activation of cognate words across languages. Accuracy scores benefited from cognate status only when no ambiguity was present. Word processing was faster and more accurate when target words were related to the dominant meaning of the homonym prime, which shared this same meaning across languages. Originality: L1 co-activation via cognate words was not enough to decrease homonym interference effect in L2 word processing. Implications: This study lends support for a non-selective bilingual lexical access view and for frequency approaches during L2 word recognition. It also indicates limitations of the cognate facilitation effect. [ABSTRACT FROM AUTHOR]

Over the last couple of years, few-shot learning (FSL) has attracted significant attention towards minimizing the dependency on labeled training examples. An inherent difficulty in FSL is handling ambiguities resulting from having too few training samples per class. To tackle this fundamental challenge in FSL, we aim to train meta-learner models that can leverage prior semantic knowledge about novel classes to guide the classifier synthesis process. In particular, we propose semantically-conditioned feature attention and sample attention mechanisms that estimate the importance of representation dimensions and training instances. We also study the problem of sample noise in FSL, towards utilizing meta-learners in more realistic and imperfect settings. Our experimental results demonstrate the effectiveness of the proposed semantic FSL model with and without sample noise. [ABSTRACT FROM AUTHOR]

This article investigates whether human masculine plural noun phrases (NPs) in Spanish, which can be interpreted with an exclusively masculine or a mixed-gender meaning, are a case of balanced or unbalanced ambiguity. The results of an experiment using a sentence continuation task with oral stimuli are consistent with the claim that masculine grammatical gender biases listeners toward an exclusively masculine interpretation. The acceptance rate of continuations with the pronoun uno/una referring to a masculine plural antecedent showed that the exclusively masculine meaning of the NP is accessed more frequently and involves a lower cognitive cost than the mixed-gender interpretation. Further, this effect interacts with the stereotypicality of the noun: nouns independently established to carry a masculine stereotype are less likely to be associated with a mixed-gender interpretation. The study also found that the speakers' attitudes toward nonsexist language predict their acceptance of the mixed-gender interpretation of masculine NPs. [ABSTRACT FROM AUTHOR]

Digital texts in many languages have examples of missing or misused diacritics which makes it hard for natural language processing applications to disambiguate the meaning of words. Therefore, diacritics restoration is a crucial step in natural language processing applications for many languages. In this study we approach this problem as bidirectional transformation of diacritical letters and their ASCII counterparts, rather than unidirectional diacritic restoration. We propose a context-aware character-level sequence to sequence model for this transformation. The model is language independent in the sense that no language-specific feature extraction is necessary other than the utilization of word embeddings and is directly applicable to other languages. We trained the model for Turkish diacritics correction task and for the assessment we used Turkish tweets benchmark dataset. Our best setting for the proposed model improves the state-of-the-art results in terms of F1 score by 4.7% on ambiguous words and 1.24% over all cases. [ABSTRACT FROM AUTHOR]

Semantic segmentation of parts of objects is a marginally explored and challenging task in which multiple instances of objects and multiple parts within those objects must be recognized in an image. We introduce a novel approach (GMENet) for this task combining object-level context conditioning, part-level spatial relationships, and shape contour information. The first target is achieved by introducing a class-conditioning module that enforces class-level semantics when learning the part-level ones. Thus, intermediate-level features carry object-level prior to the decoding stage. To tackle part-level ambiguity and spatial relationships among parts we exploit an adjacency graph-based module that aims at matching the spatial relationships between parts in the ground truth and predicted maps. Last, we introduce an additional module to further leverage edges localization. Besides testing our framework on the already used Pascal-Part-58 and Pascal-Person-Part benchmarks, we further introduce two novel benchmarks for large-scale part parsing, i.e., a more challenging version of Pascal-Part with 108 classes and the ADE20K-Part benchmark with 544 parts. GMENet achieves state-of-the-art results in all the considered tasks and furthermore allows to improve object-level segmentation accuracy. [ABSTRACT FROM AUTHOR]

The paper discusses the hypothesis of parameters for the expression of comparison and degree constructions discussing data from Brazilian Portuguese (BP). Beck et alii (2009) and Beck, Oda and Sugisaki (2004) propose three parameters: i) the language has degree predicates (DSP -- Degree Structure Parameter); ii) the language has or does not have binding of degree variables in the syntax (DAP -Degree Abstraction Parameter); and iii) the language has a syntactic degree argument filled (DegPP -- Degree Phrase Parameter). Languages with negative setting in the DSP do not show any degree construction at all, especially comparative constructions with an expression like "more". The DAP explains structures with binding of degree variables in the syntax, like comparative clauses with scope ambiguities. At last, there is the impossibility of filling the syntactic degree argument of gradable adjectives with measure phrases (cf. *2 metros alto/2 meters tall). This fact is explained by a negative setting in the DegPP. It is shown that BP receives positive setting in the first ones but negative setting in the last. It is also discussed degree interrogatives and measure phrases like 2 metros de altura "2 meters of height". The paper concludes that these structures should be better discussed to understand their place in the degree parameter hypothesis. [ABSTRACT FROM AUTHOR]

This paper explores a taxonomy of uses of 'because' from the linguistics literature. It traces the apparent semantic differences between content, epistemic, and act 'because' to differences in attachment height. But it argues that the fact that these uses of 'because' never occur in the same environments is evidence of an underlying semantic unity. Arguments from such a distribution to underlying unity are familiar from phonology and morphology, and they are implicit in Quine's comments on ambiguity in Word and Object, but they have too rarely been deployed in semantics. Here, such an argument points the way towards an underexplored conception of polysemy, while also making progress on the rather more limited topic of the lexical semantics of 'because'. [ABSTRACT FROM AUTHOR]

In this paper, we propose an ambiguousness-aware state evolution (AASE) method which represents the uncertainty of the input sequence and evolves the subsequent skeletons to generate a reasonable full-length sequence for action prediction. Unlike most existing methods that enforce partial sequences with the labels of full-length videos and ignore the semantic information of the subsequent action, we develop an evolution method by predicting the instructional actions and generating the reasonable candidate subsequent actions, so that the ambiguity of the full sequence’s label supervising for the partial actions can be effectively alleviated. Our method generates the rational subsequent actions under the instructional action class to complement the partially observed action sequence. We design two criteria for a rational generation: 1) the instruction of subsequent action keeps the semantic consistency with the observed sequence; 2) the generation sequence is satisfied with the distribution of the sequence of real data. Moreover, we design an uncertainty module to decide the instructional action class for the generation network. AASE predicts instructional actions with uncertainty learning and evolves different instructional actions by generating the subsequent skeletons, which find the most probable action to represent the partially observed action by learning the way of perceiving the tendency of the ongoing action. We conduct experiments on seven widely used action datasets: NTU-60, NTU-120, UCF101, UT-Interaction, BIT, PKU-MMD and HMDB51, and our experimental results clearly demonstrate that our method achieves very competitive performance with state-of-the-art. [ABSTRACT FROM AUTHOR]

Due to the miscellaneous ambiguity of semantics in open-domain conversation, current deep dialogue models disregard to detect potential emotional and action response features in the latent space, which leads to the general tendency to produce inaccurate and irrelevant sentences. To address this problem, we propose a semantic-aware conditional variational autoencoder that discriminates the sentiment and action responses features in the latent space for one-to-many open-domain dialogue generation. Specifically, explicit controllable variables are leveraged from the proposed module to create diverse conversational texts. This controllable variable can constrain the distribution of the latent space, disentangling the latent space features during training. Furthermore, the feature disentanglement improves the dialogue generation in terms of deep learning interpretability and text quality, which also reveals the latent features of different emotions on the logic of text generation. [ABSTRACT FROM AUTHOR]

Spoken language is interpreted incrementally, with listeners considering multiple candidate meanings as words unfold over time. Due to incremental interpretation, when a speaker refers to something in the world, there is often temporary ambiguity regarding which of several candidate items in the referential context the speaker is referring to. Subsequent tests of recognition memory show that listeners have good memory for referenced items, but that listeners also sometimes recognize non-referenced items from the referential context that share features with items that were mentioned. Predicted or inferred (but not experienced) interpretations of what was said are also sometimes retained in memory. While these findings indicate that multiple items from the referential context may be encoded in memory, the mechanisms supporting memory for the context of language use remain poorly understood. This paper tests the hypothesis that a consequence of temporary ambiguity in spoken language is enhanced memory for the items in the referential context. Two experiments demonstrate that periods of temporary referential ambiguity boost memory for non-referenced items in the referential context. Items that temporarily matched the unfolding referring expression were better remembered than those that did not. The longer the period of ambiguity, the stronger the memory boost, particularly for items activated early in the expression. In sum, the fact that spoken language unfolds over time creates momentary ambiguity about the speaker's intention; this ambiguity, in turn, allows the listener to later remember not only what the speaker did say, but also what they could have, but did not. [ABSTRACT FROM AUTHOR]

• The findings highlight the semantic relation and grammatical factor in the brain's representation of lexical ambiguities. • Compared to unambiguous words, homonyms and polysemous activate the inferior frontal gyri and posterior temporal region. • Compared to polysemous words, homonyms elicit increased activity in dorsal prefrontal and posterior parietal cortices. • Meanings of homonyms with different syntactic categories are represented distinctively in frontal and temporal cortices. In human languages, it is a common phenomenon for a single word to have multiple meanings. This study used fMRI to investigate how the brain processed different types of lexical ambiguity, and how it differentiated the meanings of ambiguous words. We focused on homonyms and polysemy that differed in the relatedness among multiple meanings. Participants (N = 35) performed a prime-target semantic relatedness task, where a specific meaning of an ambiguous word was primed. Results showed that homonyms elicited greater activation in bilateral dorsal prefrontal and posterior parietal cortices than polysemous words, suggesting that these regions may be more engaged in cognitive control when the meanings of ambiguous words are unrelated. Multivariate pattern analysis further revealed that meanings of homonyms with different syntactic categories were represented differently in the frontal and temporal cortices. The findings highlighted the importance of semantic relations and grammatical factors in the brain's representation of lexical ambiguities. [ABSTRACT FROM AUTHOR]

In real-world vision, people prioritise the most informative scene regions via eye-movements. According to the cognitive guidance theory of visual attention, viewers allocate visual attention to those parts of the scene that are expected to be the most informative. The expected information of a scene region is coded in the semantic distribution of that scene. Meaning maps have been proposed to capture the spatial distribution of local scene semantics in order to test cognitive guidance theories of attention. Notwithstanding the success of meaning maps, the reason for their success has been contested. This has led to at least two possible explanations for the success of meaning maps in predicting visual attention. On the one hand, meaning maps might measure scene semantics. On the other hand, meaning maps might measure scene features, overlapping with, but distinct from, scene semantics. This study aims to disentangle these two sources of information by considering both conceptual information and non-semantic scene entropy simultaneously. We found that both semantic and non-semantic information is captured by meaning maps, but scene entropy accounted for more unique variance in the success of meaning maps than conceptual information. Additionally, some explained variance was unaccounted for by either source of information. Thus, although meaning maps may index some aspect of semantic information, their success seems to be better explained by non-semantic information. We conclude that meaning maps may not yet be a good tool to test cognitive guidance theories of attention in general, since they capture non-semantic aspects of local semantic density and only a small portion of conceptual information. Rather, we suggest that researchers should better define the exact aspect of cognitive guidance theories they wish to test and then use the tool that best captures that desired semantic information. As it stands, the semantic information contained in meaning maps seems too ambiguous to draw strong conclusions about how and when semantic information guides visual attention. [ABSTRACT FROM AUTHOR]

In communication, textual data are a vital attribute. In all languages, ambiguous or polysemous words' meaning changes depending on the context in which they are used. The ability to determine the ambiguous word's correct meaning is a Know‐distill challenging task in natural language processing (NLP). Word sense disambiguation (WSD) is an NLP process to analyze and determine the correct meaning of polysemous words in a text. WSD is a computational linguistics task that automatically identifies the polysemous word's set of senses. Based on the context some word comes into view, WSD recognizes and tags the word to its correct priori known meaning. Semitic languages like Arabic have even more significant challenges than other languages since Arabic lacks diacritics, standardization, and a massive shortage of available resources. Recently, many approaches and techniques have been suggested to solve word ambiguity dilemmas in many different ways and several languages. In this review paper, an extensive survey of research works is presented, seeking to solve Arabic word sense disambiguation with the existing AWSD datasets. This article is categorized under:Algorithmic Development > Text MiningTechnologies > Machine Learning [ABSTRACT FROM AUTHOR]

Some generic sentences seem to be true despite the fact that almost all the members of the relevant kind are exceptions. It's controversial whether generics of this type express relatively weak generalizations or relatively strong ones. If the latter, then we're systematically mistaken about their truth, but they make no trouble for our semantic theorizing. In this brief note, I present several arguments for the former: sentences of the relevant type are weak generics. [ABSTRACT FROM AUTHOR]

In the process of semantic capture, traditional sentence representation methods tend to lose a lot of global and contextual semantics and ignore the internal structure information of words in sentences. To address these limitations, we propose a sentence representation method for character-assisted construction-Bert (CharAs-CBert) to improve the accuracy of sentiment text classification. First, based on the construction, a more effective construction vector is generated to distinguish the basic morphology of the sentence and reduce the ambiguity of the same word in different sentences. At the same time, it aims to strengthen the representation of salient words and effectively capture contextual semantics. Second, character feature vectors are introduced to explore the internal structure information of sentences and improve the representation ability of local and global semantics. Then, to make the sentence representation have better stability and robustness, character information, word information, and construction vectors are combined and used together for sentence representation. Finally, the evaluation and verification are carried out on various open-source baseline data such as ACL-14 and SemEval 2014 to demonstrate the validity and reliability of sentence representation, namely, the F1 and ACC are 87.54% and 92.88% on ACL14, respectively. [ABSTRACT FROM AUTHOR]

Numerous deep-learning methods have been successfully applied to semantic segmentation (SS) and height estimation (EH) of remote-sensing imagery. It has also been proved that such a framework can be reusable for multiple tasks to reduce computational resource overhead. However, there are still some technical limitations due to the semantic inconsistency between 3-D and 2-D features and the strong interference of different objects with similar spectral–spatial properties. Previous works have sought to address these issues through hard (HPS) or soft parameter sharing (SPS) schemes. But due to unintentional integration, the specific information transmitted between multiple tasks is not clear or in a lot of redundancy. Furthermore, tuning the weights by hand between classification and regression loss functions is challenging. In this article, a novel multitask learning (MTL) method, termed associatively segmenting semantics and estimating height (ASSEH), is proposed to associatively segment semantics and estimate height from monocular remote-sensing imagery. First, considering semantic inconsistency across tasks, we design a task-specific distillation (TSD) module containing a set of task-specific gating units (TSGUs) for each task at the cost of fewer parameters. The module allows for task-specific features to be tailored from the backbone while allowing for task-shared features to be transmitted. Second, we leverage the proposed cross-task propagation (CTP) module to construct and diffuse the local pattern graphlets at the common positions across tasks. Such a high-order recursive method can bridge two tasks explicitly to effectively settle semantic ambiguities caused by similar spectral characteristics with less computational burden and memory requirements. Third, a dynamic weighted geometric mean (DWGeoMean) strategy is introduced to dynamically learn the weights of each task and be more robust to the magnitude of the loss function. Finally, the results of the ISPRS Vaihingen dataset and the urban semantic 3-D (US3D) dataset well demonstrate that our ASSEH achieves state-of-the-art performance. [ABSTRACT FROM AUTHOR]

It has been shown that, in language comprehension, listeners model certain attributes of their interlocutor (e.g., dialectic background, age, gender) and interpret speech against that model; for example, they understand cross-dialectally ambiguous words such as flat and gas for their American English (AE) meanings more often when listening to an AE interlocutor than a British English (BE) interlocutor. This study further investigated whether listeners construct concurrent interlocutor models when communicating with interleaved interlocutors of different dialectic backgrounds, and, if they do, how they choose between concurrent models to interpret words. In two experiments, participants heard a word (e.g., flat) spoken by a BE or AE interlocutor and provided a word associate (indicating which meaning of the word was accessed). When different interlocutors were encountered in separate blocks, participants accessed more AE meanings when listening to an AE rather than a BE interlocutor, and the accent effect was not larger for words pronounced more differently in BE and AE (e.g., fall sounds more distinctly British vs. American than flat does). These results suggest that participants constructed an interlocutor model (e.g., of a BE or an AE speaker) and used it (instead of accent details in a word) to guide word meaning access. When interlocutors were interleaved in the same block, we observed a comparable accent effect, which increased as a function of between-accent differences in pronunciation. These results suggest that participants constructed concurrent interlocutor models and used accent details in a word to select the appropriate interlocutor model. We also observed that the accent effect was comparable for two interleaved interlocutors of the same gender (e.g., a female BE interlocutor and a female AE interlocutor) and for two interleaved interlocutors of different genders (e.g., a female BE interlocutor and a male AE interlocutor). These results suggest that participants did not use gender-related voice details for model selection when accent details were sufficient for interlocutor model selection. [ABSTRACT FROM AUTHOR]

Brain systems dealing with multiple meanings of ambiguous stimuli are relatively well studied, while the processing of non-selected meanings is less investigated in the neurophysiological literature and provokes controversy between existing theories. It is debated whether these meanings are actively suppressed and, if yes, whether suppression characterizes any task that involves alternative solutions or only those tasks that emphasize semantic processing or the existence of alternatives. The current functional MRI event-related study used a modified version of the word fragment completion task to reveal brain mechanisms involved in implicit processing of the non-selected solutions of ambiguous fragments. The stimuli were pairs of fragmented adjectives and nouns. Noun fragments could have one or two solutions (resulting in two words with unrelated meanings). Adjective fragments had one solution and created contexts strongly suggesting one solution for ambiguous noun fragments. All fragmented nouns were presented twice during the experiment (with two different adjectives). We revealed that ambiguity resolution was associated with a reduced BOLD signal within several regions related to language processing, including the anterior hippocampi and amygdala and posterior lateral temporal cortex. Obtained findings were interpreted as resulting from brain activity inhibition, which underlies a hypothesized mechanism of suppression of non-selected solutions. [ABSTRACT FROM AUTHOR]

Point cloud semantic segmentation is an ingredient in understanding real-world scenes. Most existing approaches perform poorly on scene boundaries and struggle with recognizing objects of different scales. In this paper, we propose a novel framework that incorporates Transformer into the U-Net architecture for inferring pointwise semantics. Specifically, the Transformer-based cross-feature fusion module is designed first to employ geometric and semantic information to learn feature offsets to overcome the border ambiguity of segmentation results, and then it utilizes the Transformer to learn cross-feature enhanced and fused encoder features. Additionally, to facilitate the overall network's structure-to-detail perception capabilities, the adaptive perception module is designed, which employs cross-attention to adaptively allocate weights to encoder features at varying resolutions, establishing long-range contextual dependencies. Ablation studies validate the individual contributions of our module design choices. Compared with the existing competitive methods, our approach achieves state-of-the-art performance and exhibits superior results on benchmarks. Code is available at https://github.com/xiluo-cug/TCFAP-Net. • A novel framework that incorporates Transformer into the U-Net architecture. • An encoder leverages Transformer to facilitate cross-feature fusion. • An adaptive perception module captures multiscale and contextual information. • Achieving outstanding performance on diverse large-scale point cloud datasets. [ABSTRACT FROM AUTHOR]

This article presents a cognitive model of event structure that can be used to explain several features of the semantics of verbs. The model consists of four basic components: agent, patient, force vector and result vector. Each component is described in terms of the theory of conceptual spaces. The force vector is the cause of the result vector. Unlike other event models both the cause and the effect are included in the representation of a single event. The model is used for two central topics. Firstly, to provide a force dynamic representation of causation. Secondly, to give a unified analysis of Aktionsart in terms of different forms of vectors, using force diagrams that are extensions of those used by Croft and others. It is then shown that the event model can be used to derive a variety of semantic features of verbs. In particular, I analyze manner-result complementarity, the ambiguity of the passive participle, and the role of goals (telicity). • A cognitively grounded model of events is proposed. • The model can account for many different types of causation. • The model provides a unified analysis of many forms of Aktionsart. • The model explains several central semantics features of verbs. [ABSTRACT FROM AUTHOR]

Modality translation, which aims to translate images from a source modality to a target one, has attracted a growing interest in the field of remote sensing recently. Compared to translation problems in multimedia applications, modality translation in remote sensing often suffers from inherent ambiguities, i.e., a single input image could correspond to multiple possible outputs, and the results may not be valid in the following image interpretation tasks, such as classification and change detection. To address these issues, we make the attempt to utilizing time-series data to resolve the ambiguities. We propose a novel multimodality image translation framework, which exploits temporal information from two aspects: 1) by introducing a guidance image from given temporally neighboring images in the target modality, we employ a feature mask module and transfer semantic information from temporal images to the output without requiring the use of any semantic labels and 2) while incorporating multiple pairs of images in time series, a temporal constraint is formulated during the learning process in order to guarantee the uniqueness of the prediction result. We also build a multimodal and multitemporal dataset that contains synthetic aperture radar (SAR), visible, and short-wave length infrared band (SWIR) image time series of the same scene to encourage and promote research on modality translation in remote sensing. Experiments are conducted on the dataset for two cross-modality translation tasks (SAR to visible and visible to SWIR). Both qualitative and quantitative results demonstrate the effectiveness and superiority of the proposed model. [ABSTRACT FROM AUTHOR]

Background: Research on how patients with Broca's aphasia and concomitant agrammatism process dependency structures online offers a window into the understanding of how language is organized in the mind. Previous studies have shown that patients evince differential performances on dependency structures generated at different levels of linguistic representation. The present study extends this line of research to investigate patients' online processing of referential dependencies involving ambiguous pronominal interpretations. Aims: Mandarin, unlike English, allows the object to be missing (null) in some contexts, rendering these null pronominal forms as ambiguous between two interpretations (strict vs. sloppy), which are generated in different levels of linguistic representation. This study examined Mandarin-speaking healthy and agrammatic aphasic participants' processing of the two interpretations of the null pronominal form such as Zhangsan qile ta de ma, Lisi ye qile e "literally: Zhangsan rode his horse and Lisi rode, too", where Lisi rode either Zhangsan's horse (strict) or Lisi's horse (sloppy). Methods & Procedures: Two experiments, employing a Cross-Modal Picture Priming paradigm, manipulated three probe picture conditions (strict, sloppy, control) and two probe positions (at the verb phrase offset and 400 ms afterwards). Experiment 1 tested neurologically unimpaired controls in order to establish the time course of processing as a standard of comparison. Experiment 2 tested patients with agrammatic aphasia. Outcomes and Results: Immediately at the verb phrase offset where the null object is structurally licensed, unimpaired individuals primed both the strict and the sloppy interpretations, with no significant difference in reaction times between the two conditions. At the position 400 ms after the verb phrase offset; however, unimpaired participants did not show priming of either of the two interpretations. Patients showed the opposite pattern: at verb phrase offset, they did not show threshold-level priming of either of the two interpretations; 400 ms afterwards, they showed priming of both the strict and sloppy interpretations, with the strict interpretation eliciting significantly longer reaction times than the sloppy interpretation. Conclusions: Patients' delay in priming both the strict and the sloppy interpretations shows that patients evince a temporal delay when processing dependency structures, which may arise from patients' deficit with lexical integration. Additionally, patients spent significantly longer reaction times on the strict interpretation relative to the sloppy interpretation, which confirms the prediction that that the former, which is generated in the level of discourse, is computationally more costly than the latter, which is generated in the level of semantics. [ABSTRACT FROM AUTHOR]

This corpus-based study investigates the inter-relation between discourse markers (DMs) and other contextual signals that contribute to the interpretation of coherence relations. The objectives are three-fold: (i) to provide a comprehensive and systematic portrait of the syntax and semantics of a set of coherence relations in English; (ii) to draw a distinction between mere tendencies of co-occurrence and strong predictive signals; (iii) to identify factors that account for the variation of these signals, focusing on relation complexity, DM strength and genre preferences. The methodology combines systematic coding (description) and multivariate statistical modelling (prediction). While the effect of genre and relation complexity was found to be null or moderate, the presence of discourse signals systematically varies with the ambiguity of the DM in the relation: signals co-occur more with ambiguous DMs than with more informative ones. [ABSTRACT FROM AUTHOR]

China's period of disunion in the tenth century posed moral challenges to individuals navigating the complex and unstable political situation. Later historians exploited the ambiguities of these challenges in giving their own assessments of those times, resulting in a complex array of assertions on matters pertaining to the Southern Tang. Appraisal of the character of the official Xu Xuan includes criticism around his work on the history of the Southern Tang and his meetings with the Song emperor Taizu. Accounts of Xu Xuan's conduct evolved in meaning over time as existing material found its way into new compilations and contexts. Tracing the course of these changes reveals how the history of the Southern Tang served the varying purposes of Song writers. Résumé: La période de désunion de la Chine au Xe siècle a posé des défis moraux aux individus qui évoluaient dans une situation politique complexe et instable. Les historiens ultérieurs ont exploité les ambiguïtés de ces défis pour formuler leurs propres jugements sur cette époque, ce qui a produit un éventail complexe de positions sur les questions relatives aux Tang du Sud. Les jugements sur le caractère de Xu Xuan et sa carrière officielle comprennent des critiques autour de son travail sur l' histoire des Tang du Sud et de ses rencontres avec l' empereur Song Taizu. Les récits de la conduite de Xu Xuan ont évolué au fil du temps, à mesure que les sources existantes trouvaient leur place dans de nouvelles compilations et de nouveaux contextes. Retracer l' évolution de ces changements révèle comment l' histoire des Tang du Sud a servi les différents objectifs des auteurs Song. 摘要: 十世紀中國的大分裂對需要應對這種複雜且動蕩政局的個人構成了一種道德上的挑戰。隨後的史家就這種挑戰的不明確性紛紛提出了自己的判斷，從而引申出有關南唐事宜的一些列繁複主張。對仕官徐鉉之品格的評價涉及了對其有關南唐歷史的文章的批評以及他與宋太祖的朝會。隨著相關材料不斷被納入新的文集和語境，有關徐鉉作為的敘述也歷經了意義上的變化。對這些變化的追溯因此也能揭示出南唐歷史如何服務於宋代文人不同的意圖。 [ABSTRACT FROM AUTHOR]

Previous studies have reported that temporarily ambiguous sentences sometimes cause reading disruption (garden-path effects). These studies have interpreted their finding as indicating that the human sentence processing device (the processor) initially assigns incorrect structures and subsequently attempts revision. That is a logical interpretation. However, no previous studies have demonstrated evidence of a causal relationship between garden-path effects and initial misanalysis. Besides, there is currently limited evidence regarding whether the processor conducts revision. The present study reports two self-paced reading experiments that investigated these fundamental issues about garden-path effects. Experiment 1 demonstrated that the processor initially misanalyses temporarily ambiguous sentences and consequently encounters garden-path effects and persists with initial misinterpretations. Experiment 2 similarly observed garden-path effects. Additionally, there was evidence that the processor constructs globally correct structures during revision. These findings provide evidence that garden-path effects result from initial misanalysis, and the processor conducts revision upon disambiguation. [ABSTRACT FROM AUTHOR]