Refine
Has Fulltext
- no (47)
Year of publication
- 2021 (47) (remove)
Document Type
- Article (47) (remove)
Language
- English (47)
Is part of the Bibliography
- yes (47) (remove)
Keywords
- German (3)
- bilingualism (3)
- Cognitive development (2)
- dysphagia (2)
- exhaustivity (2)
- perception of robots (2)
- processing (2)
- syntax (2)
- variability (2)
- Action segmentation (1)
Institute
- Department Linguistik (47) (remove)
In eye-movement control during reading, advanced process-oriented models have been developed to reproduce behavioral data. So far, model complexity and large numbers of model parameters prevented rigorous statistical inference and modeling of interindividual differences. Here we propose a Bayesian approach to both problems for one representative computational model of sentence reading (SWIFT; Engbert et al., Psychological Review, 112, 2005, pp. 777-813). We used experimental data from 36 subjects who read the text in a normal and one of four manipulated text layouts (e.g., mirrored and scrambled letters). The SWIFT model was fitted to subjects and experimental conditions individually to investigate between- subject variability. Based on posterior distributions of model parameters, fixation probabilities and durations are reliably recovered from simulated data and reproduced for withheld empirical data, at both the experimental condition and subject levels. A subsequent statistical analysis of model parameters across reading conditions generates model-driven explanations for observable effects between conditions.
Advocating the inclusion of older adults in digital language learning technology and research
(2021)
Apples and oranges
(2021)
Despite scarce empirical evidence, introducing new vocabulary in semantic categories has long been standard in second language teaching. We examined the effect of learning context on encoding, immediate recall and integration of new vocabulary into semantic memory by contrasting categorically related (novel names for familiar concepts blocked by semantic category) and unrelated (mixed semantic categories) learning contexts. Two learning sessions were conducted 24 hours apart, with each participant exposed to both contexts. Subsequently, a test phase examined picture naming, translation and picture-word interference tasks. Compared to the unrelated context, the categorically related context resulted in poorer naming accuracy in the learning phase, slower response latencies at the immediate recall tasks and greater semantic interference in the picture-word interference task (picture naming in L1 with semantically related novel word distractors). We develop a theoretical account of word learning that attributes observed differences to episodic rather than semantic memory.
In successful communication, the literal meaning of linguistic utterances is often enriched by pragmatic inferences. Part of the pragmatic reasoning underlying such inferences has been successfully modeled as Bayesian goal recognition in the Rational Speech Act (RSA) framework. In this paper, we try to model the interpretation of question-answer sequences with narrow focus in the answer in the RSA framework, thereby exploring the effects of domain size and prior probabilities on interpretation. Should narrow focus exhaustivity inferences be actually based on Bayesian inference involving prior probabilities of states, RSA models should predict a dependency of exhaustivity on these factors. We present experimental data that suggest that interlocutors do not act according to the predictions of the RSA model and that exhaustivity is in fact approximately constant across different domain sizes and priors. The results constitute a conceptual challenge for Bayesian accounts of the underlying pragmatic inferences.
Argument mining on twitter
(2021)
In the last decade, the field of argument mining has grown notably. However, only relatively few studies have investigated argumentation in social media and specifically on Twitter. Here, we provide the, to our knowledge, first critical in-depth survey of the state of the art in tweet-based argument mining. We discuss approaches to modelling the structure of arguments in the context of tweet corpus annotation, and we review current progress in the task of detecting argument components and their relations in tweets. We also survey the intersection of argument mining and stance detection, before we conclude with an outlook.
Usage-based theories assume that all aspects of language processing are shaped by the distributional properties of the language. The frequency not only of words but also of larger chunks plays a major role in language processing. These theories predict that the frequency of phrases influences the time needed to prepare these phrases for production and their acoustic duration. By contrast, dominant psycholinguistic models of utterance production predict no such effects. In these models, the system keeps track of the frequency of individual words but not of co-occurrences. This study investigates the extent to which the frequency of phrases impacts naming latencies and acoustic duration with a balanced design, where the same words are recombined to build high- and low-frequency phrases. The brain signal of participants is recorded so as to obtain information on the electrophysiological bases and functional locus of frequency effects. Forty-seven participants named pictures using high- and low-frequency adjective-noun phrases. Naming latencies were shorter for high-frequency than low-frequency phrases. There was no evidence that phrase frequency impacted acoustic duration. The electrophysiological signal differed between high- and low-frequency phrases in time windows that do not overlap with conceptualization or articulation processes. These findings suggest that phrase frequency influences the preparation of phrases for production, irrespective of the lexical properties of the constituents, and that this effect originates at least partly when speakers access and encode linguistic representations. Moreover, this study provides information on how the brain signal recorded during the preparation of utterances changes with the frequency of word combinations.
Perceptual narrowing in the domain of face perception typically begins to reduce infants' sensitivity to differences distinguishing other-race faces from approximately 6 months of age. The present study investigated whether it is possible to re-sensitize Caucasian 12-month-old infants to other-race Asian faces through statistical learning by familiarizing them with different statistical distributions of these faces. The familiarization faces were created by generating a morphed continuum from one Asian face identity to another. In the unimodal condition, infants were familiarized with a frequency distribution wherein they saw the midpoint face of the morphed continuum the most frequently. In the bimodal condition, infants were familiarized with a frequency distribution wherein they saw faces closer to the endpoints of the morphed continuum the most frequently. After familiarization, infants were tested on their discrimination of the two original Asian faces. The infants' looking times during the test indicated that infants in the bimodal condition could discriminate between the two faces, while infants in the unimodal condition could not. These findings therefore suggest that 12-month-old Caucasian infants could be re-sensitized to Asian faces by familiarizing them with a bimodal frequency distribution of such faces.
Comparisons of equality with German so ... wie, and the relationship between degrees and properties
(2021)
We present a compositionally transparent, unified semantic analysis of two kinds of so ... wie-equative constructions in German, namely degree equatives and property equatives in the domain of individuals or events. Unlike in English and many other European languages (Haspelmath & Buchholz 1998, Rett 2013), both equative types in German feature the parameter marker so, suggesting a unified analysis. We show that the parallel formal expression of German degree and property equatives is accompanied by a parallel syntactic distribution (in predicative, attributive, and adverbial position), and by identical semantic properties: Both equative types allow for scope ambiguities, show negative island effects out of context, and license the negative polarity item uberhaupt 'at all' in the complement clause. As the same properties are also shared by German comparatives, we adopt the influential quantificational analysis of comparatives in von Stechow (1984ab), Heim (1985, 2001, 2007), and Beck (2011), and treat both German equative types in a uniform manner as expressing universal quantification over sets of degrees or over sets of properties (of individuals or events). Conceptually, the uniform marking of degree-related and property-related meanings is expected given that the abstract semantic category degree (type ) can be reconstructed in terms of equivalence classes, i.e., ontologically simpler sets of individuals (type ) or events (type ). These are found in any language, showing that whether or not a language makes explicit reference to degrees (by means of gradable adjectives, degree question words, degree-only equatives) does not follow on general conceptual or semantic grounds, but is determined by the grammar of that language.
Coordinated subjects often show variable number agreement with the finite verb, but linguistic approaches to this phenomenon have rarely been informed by systematically collected data. We report the results from three experiments investigating German speakers' agreement preferences with complex subjects joined by the correlative conjunctions sowohl horizontal ellipsis als auch ('both horizontal ellipsis and'), weder horizontal ellipsis noch ('neither horizontal ellipsis nor') or entweder horizontal ellipsis oder ('either horizontal ellipsis or'). We examine to what extent conjunction type and a conjunct's relative proximity to the verb affect the acceptability and processibility of singular vs. plural agreement. Experiment 1 was an untimed acceptability rating task, Experiment 2 a timed sentence completion task, and Experiment 3 was a self-paced reading task. Taken together, our results show that number agreement with correlative coordination in German is primarily determined by a default constraint triggering plural agreement, which interacts with linear order and semantic factors. Semantic differences between conjunctions only affected speakers' agreement preferences in the absence of processing pressure but not their initial agreement computation. The combined results from our offline and online experimental measures of German speakers' agreement preferences suggest that the constraints under investigation do not only differ in their relative weighting but also in their relative timing during agreement computation.
In syntactic dependency trees, when arcs are drawn from syntactic heads to dependents, they rarely cross. Constraints on these crossing dependencies are critical for determining the syntactic properties of human language, because they define the position of natural language in formal language hierarchies. We study whether the apparent constraints on crossing syntactic dependencies in natural language might be explained by constraints on dependency lengths (the linear distance between heads and dependents). We compare real dependency trees from treebanks of 52 languages against baselines of random trees which are matched with the real trees in terms of their dependency lengths. We find that these baseline trees have many more crossing dependencies than real trees, indicating that a constraint on dependency lengths alone cannot explain the empirical rarity of crossing dependencies. However, we find evidence that a combined constraint on dependency length and the rate of crossing dependencies might be able to explain two of the most-studied formal restrictions on dependency trees: gap degree and well-nestedness.
Background
Simple water-swallowing screening tools are not predictive of aspiration and dysphagia in patients with Parkinson's Disease (PD). We investigated the diagnostic accuracy of a multi-texture screening tool, the Gugging Swallowing Screen (GUSS) to identify aspiration and dysphagia/penetration in PD patients compared to flexible endoscopic evaluation of swallowing (FEES).
Methods
Swallowing function was evaluated in 51 PD participants in clinical 'on-medication' state with the GUSS and a FEES examination according to standardized protocols. Inter-rater reliability and convergent validity were determined and GUSS- and FEES-based diet recommendations were compared.
Key Results
Inter-rater reliability of GUSS ratings was high (r(s) = 0.8; p < 0.001). Aspiration was identified by the GUSS with a sensitivity of 50%, and specificity of 51.35% (PPV 28%, NPV 73%, LR+ 1.03, LR- 0.97), dysphagia/penetration was identified with 72.97% sensitivity and 35.71% specificity (PPV 75%, NPV 33.33%, LR+ 1.14, LR- 0.76). Agreement between GUSS- and FEES-based diet recommendations was low (r(s) = 0.12, p = 0.42) with consistent NPO (Nil per Os) allocation by GUSS and FEES in only one participant.
Conclusions and Inferences
The multi-texture screening tool GUSS in its current form, although applicable with good inter-rater reliability, does not detect aspiration in PD patients with acceptable accuracy. Modifications of the GUSS parameters "coughing," "voice change" and "delayed swallowing" might enhance validity. The GUSS' diet recommendations overestimate the need for oral intake restriction in PD patients and should be verified by instrumental swallowing examination.
Examining group differences in between-participant variability in non-native speech sound learning
(2021)
Many studies on non-native speech sound learning report a large amount of between-participant variability. This variability allows us to ask interesting questions about non-native speech sound learning, such as whether certain training paradigms give rise to more or less between-participant variability. This study presents a reanalysis of Fuhrmeister and Myers (Attention, Perception, and Psychophysics, 82(4), 2049-2065, 2020) and tests whether different types of phonetic training lead to group differences in between-participant variability. The original study trained participants on a non-native speech sound contrast in two different phonological (vowel) contexts and tested for differences in means between a group that received blocked training (one vowel context at a time) and interleaved training (vowel contexts were randomized). No statistically significant differences in means were found between the two groups in the original study on a discrimination test (a same-different judgment). However, the current reanalysis tested group differences in between-participant variability and found greater variability in the blocked training group immediately after training because this group had a larger proportion of participants with higher-than-average scores. After a period of offline consolidation, this group difference in variability decreased substantially. This suggests that the type and difficulty of phonetic training (blocked vs. interleaved) may initially give rise to differences in between-participant variability, but offline consolidation may attenuate that variability and have an equalizing effect across participants. This reanalysis supports the view that examining between-participant variability in addition to means when analyzing data can give us a more complete picture of the effects being tested.
This eye-tracking study establishes basic benchmarks of eye movements during reading in heritage language (HL) by Russian-speaking adults and adolescents of high (n = 21) and low proficiency (n = 27). Heritage speakers (HSs) read sentences in Cyrillic, and their eye movements were compared to those of Russian monolingual skilled adult readers, 8-year-old children and L2 learners. Reading patterns of HSs revealed longer mean fixation durations, lower skipping probabilities, and higher regressive saccade rates than in monolingual adults. High-proficient HSs were more similar to monolingual children, while low-proficient HSs performed on par with L2 learners. Low-proficient HSs differed from high-proficient HSs in exhibiting lower skipping probabilities, higher fixation counts, and larger frequency effects. Taken together, our findings are consistent with the weaker links account of bilingual language processing as well as the divergent attainment theory of HL.
Fitts' law, perhaps the most celebrated law of human motor control, expresses a relation between the kinematic property of speed and the non-kinematic, task-specific property of accuracy. We aimed to assess whether speech movements obey this law using a metronome-driven speech elicitation paradigm with a systematic speech rate control. Specifically, using the paradigm of repetitive speech, we recorded via electromagnetic articulometry speech movement data in sequences of the form /CV.../ from 6 adult speakers. These sequences were spoken at 8 distinct rates ranging from extremely slow to extremely fast. Our results demonstrate, first, that the present paradigm of extensive metronome-driven manipulations satisfies the crucial prerequisites for evaluating Fitts' law in a subset of our elicited rates. Second, we uncover for the first time in speech evidence for Fitts' law at the faster rates and specifically beyond a participant-specific critical rate. We find no evidence for Fitts' law at the slowest metronome rates. Finally, we discuss implications of these results for models of speech.
How to embrace variation and accept uncertainty in linguistic and psycholinguistic data analysis
(2021)
The use of statistical inference in linguistics and related areas like psychology typically involves a binary decision: either reject or accept some null hypothesis using statistical significance testing. When statistical power is low, this frequentist data-analytic approach breaks down: null results are uninformative, and effect size estimates associated with significant results are overestimated. Using an example from psycholinguistics, several alternative approaches are demonstrated for reporting inconsistencies between the data and a theoretical prediction. The key here is to focus on committing to a falsifiable prediction, on quantifying uncertainty statistically, and learning to accept the fact that - in almost all practical data analysis situations - we can only draw uncertain conclusions from data, regardless of whether we manage to obtain statistical significance or not. A focus on uncertainty quantification is likely to lead to fewer excessively bold claims that, on closer investigation, may turn out to be not supported by the data.
I can see it in your eyes
(2021)
Over the past years, extensive research has been dedicated to developing robust platforms and data-driven dialog models to support long-term human-robot interactions. However, little is known about how people's perception of robots and engagement with them develop over time and how these can be accurately assessed through implicit and continuous measurement techniques. In this paper, we explore this by involving participants in three interaction sessions with multiple days of zero exposure in between. Each session consists of a joint task with a robot as well as two short social chats with it before and after the task. We measure participants' gaze patterns with a wearable eye-tracker and gauge their perception of the robot and engagement with it and the joint task using questionnaires. Results disclose that aversion of gaze in a social chat is an indicator of a robot's uncanniness and that the more people gaze at the robot in a joint task, the worse they perform. In contrast with most HRI literature, our results show that gaze toward an object of shared attention, rather than gaze toward a robotic partner, is the most meaningful predictor of engagement in a joint task. Furthermore, the analyses of gaze patterns in repeated interactions disclose that people's mutual gaze in a social chat develops congruently with their perceptions of the robot over time. These are key findings for the HRI community as they entail that gaze behavior can be used as an implicit measure of people's perception of robots in a social chat and of their engagement and task performance in a joint task.
Human infants can segment action sequences into their constituent actions already during the first year of life. However, work to date has almost exclusively examined the role of infants' conceptual knowledge of actions and their outcomes in driving this segmentation. The present study examined electrophysiological correlates of infants' processing of lower-level perceptual cues that signal a boundary between two actions of an action sequence. Specifically, we tested the effect of kinematic boundary cues (pre-boundary lengthening and pause) on 12-month-old infants' (N = 27) processing of a sequence of three arbitrary actions, performed by an animated figure. Using the Event-Related Potential (ERP) approach, evidence of a positivity following the onset of the boundary cues was found, in line with previous work that has found an ERP positivity (Closure Positive Shift, CPS) related to boundary processing in auditory stimuli and action sequences in adults. Moreover, an ERP negativity (Negative Central, Nc) indicated that infants' encoding of the post-boundary action was modulated by the presence or absence of prior boundary cues. We therefore conclude that 12-month-old infants are sensitive to lower-level perceptual kinematic boundary cues, which can support segmentation of a continuous stream of movement into individual action units.
People perceive sentences more favourably after hearing or reading them many times. A prominent approach in linguistic theory argues that these types of exposure effects (satiation effects) show direct evidence of a generative approach to linguistic knowledge: only some sentences improve under repeated exposure, and which sentences do improve can be predicted by a model of linguistic competence that yields natural syntactic classes. However, replications of the original findings have been inconsistent, and it remains unclear whether satiation effects can be reliably induced in an experimental setting at all. Here we report four findings regarding satiation effects in wh-questions across German and English. First, the effects pertain to zone of well-formedness rather than syntactic class: all intermediate ratings, including calibrated fillers, increase at the beginning of the experimental session regardless of syntactic construction. Second, though there is satiation, ratings asymptote below maximum acceptability. Third, these effects are consistent across judgments of superiority effects in English and German. Fourth, wh-questions appear to show similar profiles in English and German, despite these languages being traditionally considered to differ strongly in whether they show effects on movement: violations of the superiority condition can be modulated to a similar degree in both languages by manipulating subject-object initiality and animacy congruency of the wh-phrase. We improve on classic satiation methods by distinguishing between two crucial tests, namely whether exposure selectively targets certain grammatical constructions or whether there is a general repeated exposure effect. We conclude that exposure effects can be reliably induced in rating experiments but exposure does not appear to selectively target certain grammatical constructions. Instead, they appear to be a phenomenon of intermediate gradient judgments.
We present computational modeling results based on a self-paced reading study investigating number attraction effects in Eastern Armenian. We implement three novel computational models of agreement attraction in a Bayesian framework and compare their predictive fit to the data using k-fold cross-validation. We find that our data are better accounted for by an encoding-based model of agreement attraction, compared to a retrieval-based model. A novel methodological contribution of our study is the use of comprehension questions with open-ended responses, so that both misinterpretation of the number feature of the subject phrase and misassignment of the thematic subject role of the verb can be investigated at the same time. We find evidence for both types of misinterpretation in our study, sometimes in the same trial. However, the specific error patterns in our data are not fully consistent with any previously proposed model.
Morphological variability in bilingual language production is widely attested. Producing inflected words has been found to be less reliable and consistent in bilinguals than in first-language (functionally monolingual) L1 speakers, even for bilingual speakers at advanced proficiency levels. The sources for these differences are not well understood. The current study presents a detailed investigation of morphological generalization processes in bilingual speakers' language production. We examined past participle formation of German using an elicited-production experiment containing nonce verbs with varying degrees of similarity to existing verbs testing a large group of bilingual Turkish/German speakers relative to L1 German speakers. We compared similarity-based lexical extensions with generalizations of morphological rules. The results show that rule-based generalizations are used less often and more variably within the bilingual group than within the L1 group. Our results also show a selective effect of age of acquisition on the bilingual speakers' morphological generalizations.