Publications

Abstract

n this study we investigated the durational correlates of lexical stress and pitch accent at normal and fast speech rate in Dutch. Previous literature on English shows that durations of lexically unstressed vowels are reduced more than stressed vowels when speakers increase their speech rate. We found that the same holds for Dutch, irrespective of whether the unstressed vowel is schwa or a "full" vowel. In the same line, we expected that vowels in words without a pitch accent would be shortened relatively more than vowels in words with a pitch accent. This was not the case: if anything, the accented vowels were shortened relatively more than the unaccented vowels. We conclude that duration is an important cue for lexical stress, but not for pitch accent.

Abstract

Studies on fast speech have shown that word-level timing of fast speech differs from that of normal rate speech in that unstressed syllables are shortened more than stressed syllables as speech rate increases. An earlier experiment showed that the intelligibility of time-compressed speech could not be improved by making its temporal organisation closer to natural fast speech. To test the hypothesis that segmental intelligibility is more important than prosodic timing in listening to timecompressed speech, the intelligibility of bisyllabic words was tested in three time-compression conditions: either stressed and unstressed syllable were compressed to the same degree, or the stressed syllable was compressed more than the unstressed syllable, or the reverse. As was found before, imitating wordlevel timing of fast speech did not improve intelligibility over linear compression. However, the results did not confirm the hypothesis either: there was no difference in intelligibility between the three compression conditions. We conclude that segmental intelligibility plays an important role, but further research is necessary to decide between the contributions of prosody and segmental intelligibility to the word-level intelligibility of time-compressed speech.

Abstract

In Iterated Learning (IL) experiments, a participant’s learned output serves as the next participant’s learning input (Kirby et al., 2014). IL can be used to model cultural transmission and has indicated that weak biases can be amplified through repeated cultural transmission (Kirby et al., 2007). So, for example, structural language properties can emerge over time because languages come to reflect the cognitive constraints in the individuals that learn and produce the language. Similarly, we propose that languages may also reflect certain anatomical biases. Do sound systems adapt to the affordances of the articulation space induced by the vocal tract?
The human vocal tract has inherent nonlinearities which might derive from acoustics and aerodynamics (cf. quantal theory, see Stevens, 1989) or biomechanics (cf. Gick & Moisik, 2015). For instance, moving the tongue anteriorly along the hard palate to produce a fricative does not result in large changes in acoustics in most cases, but for a small range there is an abrupt change from a perceived palato-alveolar [ʃ] to alveolar [s] sound (Perkell, 2012). Nonlinearities such as these might bias all human speakers to converge on a very limited set of phonetic categories, and might even be a basis for combinatoriality or phonemic ‘universals’.
While IL typically uses discrete symbols, Verhoef et al. (2014) have used slide whistles to produce a continuous signal. We conducted an IL experiment with human subjects who communicated using a digital slide whistle for which the degree of nonlinearity is controlled. A single parameter (α) changes the mapping from slide whistle position (the ‘articulator’) to the acoustics. With α=0, the position of the slide whistle maps Bark-linearly to the acoustics. As α approaches 1, the mapping gets more double-sigmoidal, creating three plateaus where large ranges of positions map to similar frequencies. In more abstract terms, α represents the strength of a nonlinear (anatomical) bias in the vocal tract.
Six chains (138 participants) of dyads were tested, each chain with a different, fixed α. Participants had to communicate four meanings by producing a continuous signal using the slide-whistle in a ‘director-matcher’ game, alternating roles (cf. Garrod et al., 2007).
Results show that for high αs, subjects quickly converged on the plateaus. This quick convergence is indicative of a strong bias, repelling subjects away from unstable regions already within-subject. Furthermore, high αs lead to the emergence of signals that oscillate between two (out of three) plateaus. Because the sigmoidal spaces are spatially constrained, participants increasingly used the sequential/temporal dimension. As a result of this, the average duration of signals with high α was ~100ms longer than with low α. These oscillations could be an expression of a basis for phonemic combinatoriality.
We have shown that it is possible to manipulate the magnitude of an articulator-induced non-linear bias in a slide whistle IL framework. The results suggest that anatomical biases might indeed constrain the design space of language. In particular, the signaling systems in our study quickly converged (within-subject) on the use of stable regions. While these conclusions were drawn from experiments using slide whistles with a relatively strong bias, weaker biases could possibly be amplified over time by repeated cultural transmission, and likely lead to similar outcomes.

Abstract

Many factors have been proposed to explain why groups of people use different speech sounds in their language. These range from cultural, cognitive, environmental (e.g., Everett, et al., 2015) to anatomical (e.g., vocal tract (VT) morphology). How could such anatomical properties have led to the similarities and differences in speech sound distributions between human languages?

It is known that hard palate profile variation can induce different articulatory strategies in speakers (e.g., Brunner et al., 2009). That is, different hard palate profiles might induce a kind of bias on speech sound production, easing some types of sounds while impeding others. With a population of speakers (with a proportion of individuals) that share certain anatomical properties, even subtle VT biases might become expressed at a population-level (through e.g., bias amplification, Kirby et al., 2007). However, before we look into population-level effects, we should first look at within-individual anatomical factors. For that, we have developed a computer-simulated analogue for a human speaker: an agent. Our agent is designed to replicate speech sounds using a production and cognition module in a computationally tractable manner.

Previous agent models have often used more abstract (e.g., symbolic) signals. (e.g., Kirby et al., 2007). We have equipped our agent with a three-dimensional model of the VT (the production module, based on Birkholz, 2005) to which we made numerous adjustments. Specifically, we used a 4th-order Bezier curve that is able to capture hard palate variation on the mid-sagittal plane (XXX, 2015). Using an evolutionary algorithm, we were able to fit the model to human hard palate MRI tracings, yielding high accuracy fits and using as little as two parameters. Finally, we show that the samples map well-dispersed to the parameter-space, demonstrating that the model cannot generate unrealistic profiles. We can thus use this procedure to import palate measurements into our agent’s production module to investigate the effects on acoustics. We can also exaggerate/introduce novel biases.

Our agent is able to control the VT model using the cognition module.

Previous research has focused on detailed neurocomputation (e.g., Kröger et al., 2014) that highlights e.g., neurobiological principles or speech recognition performance. However, the brain is not the focus of our current study. Furthermore, present-day computing throughput likely does not allow for large-scale deployment of these architectures, as required by the population model we are developing. Thus, the question whether a very simple cognition module is able to replicate sounds in a computationally tractable manner, and even generalize over novel stimuli, is one worthy of attention in its own right.

Our agent’s cognition module is based on running an evolutionary algorithm on a large population of feed-forward neural networks (NNs). As such, (anatomical) bias strength can be thought of as an attractor basin area within the parameter-space the agent has to explore. The NN we used consists of a triple-layered (fully-connected), directed graph. The input layer (three neurons) receives the formants frequencies of a target-sound. The output layer (12 neurons) projects to the articulators in the production module. A hidden layer (seven neurons) enables the network to deal with nonlinear dependencies. The Euclidean distance (first three formants) between target and replication is used as fitness measure. Results show that sound replication is indeed possible, with Euclidean distance quickly approaching a close-to-zero asymptote.

Statistical analysis should reveal if the agent can also: a) Generalize: Can it replicate sounds not exposed to during learning? b) Replicate consistently: Do different, isolated agents always converge on the same sounds? c) Deal with consolidation: Can it still learn new sounds after an extended learning phase (‘infancy’) has been terminated? Finally, a comparison with more complex models will be used to demonstrate robustness.

Abstract

Mutations of the epidermal growth factor receptor (EGFR) gene are found at a relatively high frequency in glioma, with the most common being the de2-7 EGFR (or EGFRvIII). This mutation arises from an in-frame deletion of exons 2-7, which removes 267 amino acids from the extracellular domain of the receptor. Despite being unable to bind ligand, the de2-7 EGFR is constitutively active at a low level. Transfection of human glioma cells with the de2-7 EGFR has little effect in vitro, but when grown as tumor xenografts this mutated receptor imparts a dramatic growth advantage. We mapped the phosphorylation pattern of de2-7 EGFR, both in vivo and in vitro, using a panel of antibodies specific for different phosphorylated tyrosine residues. Phosphorylation of de2-7 EGFR was detected constitutively at all tyrosine sites surveyed in vitro and in vivo, including tyrosine 845, a known target in the wild-type EGFR for src kinase. There was a substantial upregulation of phosphorylation at every yrosine residue of the de2-7 EGFR when cells were grown in vivo compared to the receptor isolated from cells cultured in vitro. Upregulation of phosphorylation at tyrosine 845 could be stimulated in vitro by the addition of specific components of the ECM via an integrindependent mechanism. These observations may partially explain why the growth enhancement mediated by de2-7 EGFR is largely restricted to the in vivo environment

Abstract

This study investigates effects of prosodic factors (prominence, boundary) on coarticulatory Vnasalization in Australian English (AusE) in CVN and NVC in comparison to those in American English
(AmE). As in AmE, prominence was found to
lengthen N, but to reduce V-nasalization, enhancing N’s nasality and V’s orality, respectively (paradigmatic contrast enhancement). But the prominence effect in CVN was more robust than that in AmE. Again similar to findings in AmE, boundary
induced a reduction of N-duration and V-nasalization phrase-initially (syntagmatic contrast enhancement), and increased the nasality of both C and V phrasefinally.
But AusE showed some differences in terms
of the magnitude of V nasalization and N duration. The results suggest that the linguistic contrast enhancements underlie prosodic-structure modulation of coarticulatory V-nasalization in
comparable ways across dialects, while the fine phonetic detail indicates that the phonetics-prosody interplay is internalized in the individual dialect’s phonetic grammar.

Abstract

It has been suggested that some individuals may benefit more from social robots than do others. Using second
language (L2) as an example, the present study examined how individual differences in attitudes toward robots and personality
traits may be related to learning outcomes. Preliminary results with 24 Turkish-speaking adults suggest that negative attitudes
toward robots, more specifically thoughts and anxiety about the negative social impact that robots may have on the society,
predicted how well adults learned L2 words from a social robot. The possible implications of the findings as well as future directions are also discussed

Abstract

In Korean, focus is expressed in accentual phrasing. To ascertain whether words focused in this manner enjoy a processing advantage analogous to that conferred by focus as expressed in, e.g, English and Dutch, we devised sentences with target words in one of four conditions: prosodic focus, syntactic focus, prosodic + syntactic focus, and no focus as a control. 32 native speakers of Korean listened to blocks of 10 sentences, then were presented visually with words and asked whether or not they had heard them. Overall, words with focus were recognised significantly faster and more accurately than unfocused words. In addition, words with syntactic focus or syntactic + prosodic focus were recognised faster than words with prosodic focus alone. As for other languages, Korean focus confers processing advantage on the words carrying it. While prosodic focus does provide an advantage, however, syntactic focus appears to provide the greater beneficial effect for recognition memory

Abstract

Human speakers often produce sentences incrementally. They can start speaking having in mind only a fragmentary idea of what they want to say, and while saying this they refine the contents underlying subsequent parts of the utterance. This capability imposes a number of constraints on the design of a syntactic processor. This paper explores these constraints and evaluates some recent computational sentence generators from the perspective of incremental production.

Abstract

The previous literature has suggested that the hand movement in co-speech gestures and signs consists of a series of phases with qualitatively different dynamic characteristics. In this paper, we propose a syntagmatic rule system for movement phases that applies to both co-speech gestures and signs. Descriptive criteria for the rule system were developed for the analysis video-recorded continuous production of signs and gesture. It involves segmenting a stream of body movement into phases and identifying different phase types. Two human coders used the criteria to analyze signs and cospeech gestures that are produced in natural discourse. It was found that the criteria yielded good inter-coder reliability. These criteria can be used for the technology of automatic recognition of signs and co-speech gestures in order to segment continuous production and identify the potentially meaningbearing phase.

Abstract

Two types of phonological variation in Dutch, resulting from optional rules, are schwa epenthesis and schwa deletion. In a lexical decision experiment it was investigated whether the phonological variants were processed similarly to the standard forms. It was found that the two types of variation patterned differently. Words with schwa epenthesis were processed faster and more accurately than the standard forms, whereas words with schwa deletion led to less fast and less accurate responses. The results are discussed in relation to the role of consonant-vowel alternations in speech processing and the perceptual integrity of onset clusters.

Abstract

A recurrently connected attractor neural network with a Hebbian learning rule is currently our best ANN analogy for a piece cortex. Functionally biological memory operates on a spectrum of time scales with regard to induction and retention, and it is modulated in complex ways by sub-cortical neuromodulatory systems. Moreover, biological memory networks are commonly believed to be highly distributed and engage many co-operating cortical areas. Here we focus on the temporal aspects of induction and retention of memory in a connectionist type attractor memory model of a piece of cortex. A continuous time, forgetful Bayesian-Hebbian learning rule is described and compared to the characteristics of LTP and LTD seen experimentally. More generally, an attractor network implementing this learning rule can operate as a long-term, intermediate-term, or short-term memory. Modulation of the print-now signal of the learning rule replicates some experimental memory phenomena, like e.g. the von Restorff effect.

Abstract

Citations play a crucial role in the scientific discourse, in information retrieval, and in bibliometrics. Many initiatives are currently promoting the idea of having free and open citation data. Creation of citation data, however, is not part of the cataloging workflow in libraries nowadays.
In this paper, we present our project Linked Open Citation Database, in which we design distributed processes and a system infrastructure based on linked data technology. The goal is to show that efficiently cataloging citations in libraries using a semi-automatic approach is possible. We specifically describe the current state of the workflow and its implementation. We show that we could significantly improve the automatic reference extraction that is crucial for the subsequent data curation. We further give insights on the curation and linking process and provide evaluation results that not only direct the further development of the project, but also allow us to discuss its overall feasibility.

Abstract

It is widely held that smells and flavors are impossible to put into words. In this paper we test this claim by seeking predictive patterns in wine reviews, which ostensibly aim to provide guides to perceptual content. Wine reviews have previously been critiqued as random and meaningless. We collected an English corpus of wine reviews with their structured metadata, and applied machine learning techniques to automatically predict the wine's color, grape variety, and country of origin. To train the three supervised classifiers, three different information sources were incorporated: lexical bag-of-words features, domain-specific terminology features, and semantic word embedding features. In addition, using regression analysis we investigated basic review properties, i.e., review length, average word length, and their relationship to the scalar values of price and review score. Our results show that wine experts do share a common vocabulary to describe wines and they use this in a consistent way, which makes it possible to automatically predict wine characteristics based on the review text alone. This means that odors and flavors may be more expressible in language than typically acknowledged.

Abstract

Recent work has investigated the emergence of structure in speech using experiments which use artificial continuous signals. Some experiments have had no limit on the duration which signals can have (e.g. Verhoef et al., 2014), and others have had time limitations (e.g. Verhoef et al., 2015). However, the effect of time constraints on the structure in signals has never been experimentally investigated.

Abstract

Structure of language is not only caused by cognitive processes, but also by physical aspects of the signalling modality. We test the assumptions surrounding the role which the physical aspects of the signal space will have on the emergence of structure in speech. Here, we use a signal creation task to test whether a signal space and a meaning space having similar dimensionalities will generate an iconic system with signal-meaning mapping and whether, when the topologies differ, the emergence of non-iconic structure is facilitated. In our experiments, signals are created using infrared sensors which use hand position to create audio signals. We find that people take advantage of signal-meaning mappings where possible. Further, we use trajectory probabilities and measures of variance to show that when there is a dimensionality mismatch, more structural strategies are used.

Abstract

Verb learning is important for young children. While most
previous research has focused on linguistic and conceptual
challenges in early verb learning (e.g. Gentner, 1982, 2006),
the present paper examined early verb learning at the
attentional level and quantified the input for early verb learning
by measuring verb-action co-occurrence statistics in parent-
child interaction from the learner’s perspective. To do so, we
used head-mounted eye tracking to record fine-grained
multimodal behaviors during parent-infant joint play, and
analyzed parent speech, parent and infant action, and infant
attention at the moments when parents produced verb labels.
Our results show great variability across different action verbs,
in terms of frequency of verb utterances, frequency of
corresponding actions related to verb meanings, and infants’
attention to verbs and actions, which provide new insights on
why some verbs are harder to learn than others.

Abstract

Studies of sound symbolism have shown that people can associate sound and meaning in consistent ways when presented with maximally contrastive stimulus pairs of nonwords such as bouba/kiki (rounded/sharp) or mil/mal (small/big). Recent work has shown the effect extends to antonymic words from natural languages and has proposed a role for shared cross-modal correspondences in biasing form-to-meaning associations. An important open question is how the associations work, and particularly what the role is of sound-symbolic matches versus mismatches. We report on a learning task designed to distinguish between three existing theories by using a spectrum of sound-symbolically matching, mismatching, and neutral (neither matching nor mismatching) stimuli. Synthesized stimuli allow us to control for prosody, and the inclusion of a neutral condition allows a direct test of competing accounts. We find evidence for a sound-symbolic match boost, but not for a mismatch difficulty compared to the neutral condition.

Abstract

We present the Narrative Brain Dataset, an fMRI dataset that was collected during spoken presentation of short excerpts of three
stories in Dutch. Together with the brain imaging data, the dataset contains the written versions of the stimulation texts. The texts are
accompanied with stochastic (perplexity and entropy) and semantic computational linguistic measures. The richness and unconstrained
nature of the data allows the study of language processing in the brain in a more naturalistic setting than is common for fMRI studies.
We hope that by making NBD available we serve the double purpose of providing useful neural data to researchers interested in natural
language processing in the brain and to further stimulate data sharing in the field of neuroscience of language.

Abstract

Linguistic traits are often seen as reflecting cognitive biases and constraints (e.g. Christiansen & Chater, 2008). However, language must also adapt to properties of the channel through which communication between individuals occurs. Perhaps the most basic aspect of any communication channel is noise. Communicative signals can be blocked, degraded or distorted by other sources in the environment. This poses a fundamental problem for communication. On average, channel disruption accompanies problems in conversation every 3 minutes (27% of cases of other-initiated repair, Dingemanse et al., 2015). Linguistic signals must adapt to this harsh environment. While modern language structures are robust to noise (e.g. Piantadosi et al., 2011), we investigate how noise might have shaped the early emergence of structure in language. The obvious adaptation to noise is redundancy. Signals which are maximally different from competitors are harder to render ambiguous by noise. Redundancy can be increased by adding differentiating segments to each signal (increasing the diversity of segments). However, this makes each signal more complex and harder to learn. Under this strategy, holistic languages may emerge. Another strategy is reduplication - repeating parts of the signal so that noise is less likely to disrupt all of the crucial information. This strategy does not increase the difficulty of learning the language - there is only one extra rule which applies to all signals. Therefore, under pressures for learnability, expressivity and redundancy, reduplicated signals are expected to emerge. However, reduplication is not a pervasive feature of words (though it does occur in limited domains like plurals or iconic meanings). We suggest that this is due to the pressure for redundancy being lifted by conversational infrastructure for repair. Receivers can request that senders repeat signals only after a problem occurs. That is, robustness is achieved by repeating the signal across conversational turns (when needed) instead of within single utterances. As a proof of concept, we ran two iterated learning chains with pairs of individuals in generations learning and using an artificial language (e.g. Kirby et al., 2015). The meaning space was a structured collection of unfamiliar images (3 shapes x 2 textures x 2 outline types). The initial language for each chain was the same written, unstructured, fully expressive language. Signals produced in each generation formed the training language for the next generation. Within each generation, pairs played an interactive communication game. The director was given a target meaning to describe, and typed a word for the matcher, who guessed the target meaning from a set. With a 50% probability, a contiguous section of 3-5 characters in the typed word was replaced by ‘noise’ characters (#). In one chain, the matcher could initiate repair by requesting that the director type and send another signal. Parallel generations across chains were matched for the number of signals sent (if repair was initiated for a meaning, then it was presented twice in the parallel generation where repair was not possible) and noise (a signal for a given meaning which was affected by noise in one generation was affected by the same amount of noise in the parallel generation). For the final set of signals produced in each generation we measured the signal redundancy (the zip compressibility of the signals), the character diversity (entropy of the characters of the signals) and systematic structure (z-score of the correlation between signal edit distance and meaning hamming distance). In the condition without repair, redundancy increased with each generation (r=0.97, p=0.01), and the character diversity decreased (r=-0.99,p=0.001) which is consistent with reduplication, as shown below (part of the initial and the final language): Linear regressions revealed that generations with repair had higher overall systematic structure (main effect of condition, t = 2.5, p < 0.05), increasing character diversity (interaction between condition and generation, t = 3.9, p = 0.01) and redundancy increased at a slower rate (interaction between condition and generation, t = -2.5, p < 0.05). That is, the ability to repair counteracts the pressure from noise, and facilitates the emergence of compositional structure. Therefore, just as systems to repair damage to DNA replication are vital for the evolution of biological species (O’Brien, 2006), conversational repair may regulate replication of linguistic forms in the cultural evolution of language. Future studies should further investigate how evolving linguistic structure is shaped by interaction pressures, drawing on experimental methods and naturalistic studies of emerging languages, both spoken (e.g Botha, 2006; Roberge, 2008) and signed (e.g Senghas, Kita, & Ozyurek, 2004; Sandler et al., 2005).

Abstract

Continuous Bag of Words (CBOW) is a powerful text embedding method. Due to its strong capabilities to encode word content, CBOW embeddings perform well on a wide range of downstream tasks while being efficient to compute. However, CBOW is not capable of capturing the word order. The reason is that the computation of CBOW's word embeddings is commutative, i.e., embeddings of XYZ and ZYX are the same. In order to address this shortcoming, we propose a
learning algorithm for the Continuous Matrix Space Model, which we call Continual Multiplication of Words (CMOW). Our algorithm is an adaptation of word2vec, so that it can be trained on large quantities of unlabeled text. We empirically show that CMOW better captures linguistic properties, but it is inferior to CBOW in memorizing word content. Motivated by these findings, we propose a hybrid model that combines the strengths of CBOW and CMOW. Our results show that the hybrid CBOW-CMOW-model retains CBOW's strong ability to memorize word content while at the same time substantially improving its ability to encode other linguistic information by 8%. As a result, the hybrid also performs better on 8 out of 11 supervised downstream tasks with an average improvement of 1.2%.

Abstract

For (semi-)automated subject indexing systems in digital libraries, it is often more practical to use metadata such as the title of a publication instead of the full-text or the abstract. Therefore, it is desirable to have good text mining and text classification algorithms that operate well already on the title of a publication. So far, the classification performance on titles is not competitive with the performance on the full-texts if the same number of training samples is used for training. However, it is much easier to obtain title data in large quantities and to use it for training than full-text data. In this paper, we investigate the question how models obtained from training on increasing amounts of title training data compare to models from training on a constant number of full-texts. We evaluate this question on a large-scale dataset from the medical domain (PubMed) and from economics (EconBiz). In these datasets, the titles and annotations of millions of publications are available, and they outnumber the available full-texts by a factor of 20 and 15, respectively. To exploit these large amounts of data to their full potential, we develop three strong deep learning classifiers and evaluate their performance on the two datasets. The results are promising. On the EconBiz dataset, all three classifiers outperform their full-text counterparts by a large margin. The best title-based classifier outperforms the best full-text method by 9.4%. On the PubMed dataset, the best title-based method almost reaches the performance of the best full-text classifier, with a difference of only 2.9%.

Abstract

The human body is unique: it is both an object of perception and the source of human experience. Its universality makes it a perfect resource for asking questions about how cognition, brain and typology relate to one another. For example, we can ask how speakers of different languages segment and categorize the human body. A dominant view is that body parts are “given” by visual perceptual discontinuities, and that words are merely labels for these visually determined parts (e.g., Andersen, 1978; Brown, 1976; Lakoff, 1987). However, there are problems with this view. First it ignores other perceptual information, such as somatosensory and motoric representations. By looking at the neural representations of sesnsory representations, we can test how much of the categorization of the human body can be done through perception alone. Second, we can look at language typology to see how much universality and variation there is in body-part categories. A comparison of a range of typologically, genetically and areally diverse languages shows that the perceptual view has only limited applicability (Majid, Enfield & van Staden, in press). For example, using a “coloring-in” task, where speakers of seven different languages were given a line drawing of a human body and asked to color in various body parts, Majid & van Staden (in prep) show that languages vary substantially in body part segmentation. For example, Jahai (Mon-Khmer) makes a lexical distinction between upper arm, lower arm, and hand, but Lavukaleve (Papuan Isolate) has just one word to refer to arm, hand, and leg. This shows that body part categorization is not a straightforward mapping of words to visually determined perceptual parts.

Abstract

Many studies in cognitive science address how people categorize objects, but there has been comparatively little research on event categorization. This study investigated the categorization of events involving material destruction, such as “cutting” and “breaking”. Speakers of 28 typologically, genetically, and areally diverse languages described events shown in a set of video-clips. There was considerable cross-linguistic agreement in the dimensions along which the events were distinguished, but there was variation in the number of categories and the placement of their boundaries.

Abstract

Studies have claimed that blind people’s spatial representations are different from sighted people, and blind people display superior auditory processing. Due to the nature of auditory and haptic information, it has been proposed that blind people have spatial representations that are more sequential than sighted people. Even the temporary loss of sight—such as through blindfolding—can affect spatial representations, but not much research has been done on this topic. We compared blindfolded and sighted people’s linguistic spatial expressions and non-linguistic localization accuracy to test how blindfolding affects the representation of path in auditory motion events. We found that blindfolded people were as good as sighted people when localizing simple sounds, but they outperformed sighted people when localizing auditory motion events. Blindfolded people’s path related speech also included more sequential, and less holistic elements. Our results indicate that even temporary loss of sight influences spatial representations of auditory motion events

Abstract

Lombard speech, speech produced in noise, is acoustically different from speech produced in quiet (plain speech) in several ways, including having a higher and wider F0 range (pitch). Extensive research on native Lombard speech does not consider that non-natives experience a higher cognitive load while producing
speech and that the native language may influence the non-native speech. We investigated pitch range in plain and Lombard speech in native and non-natives.
Dutch and American-English speakers read contrastive question-answer pairs in quiet and in noise in English, while the Dutch also read Dutch sentence pairs. We found that Lombard speech is characterized by a wider pitch range than plain speech, for all speakers (native English, non-native English, and native Dutch).
This shows that non-natives also widen their pitch range in Lombard speech. In sentences with early-focus, we see the same increase in pitch range when going from plain to Lombard speech in native and non-native English, but a smaller increase in native Dutch. In sentences with late-focus, we see the biggest increase for the native English, followed by non-native English and then native Dutch. Together these results indicate an effect of the native language on non-native Lombard speech.

Abstract

Lombard speech, speech produced in noise, is
typically produced with a higher fundamental
frequency (F0, pitch) compared to speech in quiet. This paper examined the potential differences in native and non-native Lombard speech by analyzing median pitch in sentences with early- or late-focus produced in quiet and noise. We found an increase in pitch in late-focus sentences in noise for Dutch speakers in both English and Dutch, and for American-English speakers in English. These results
show that non-native speakers produce Lombard speech, despite their higher cognitive load. For the early-focus sentences, we found a difference between the Dutch and the American-English speakers. Whereas the Dutch showed an increased F0 in noise
in English and Dutch, the American-English speakers did not in English. Together, these results suggest that some acoustic characteristics of Lombard speech, such as pitch, may be language-specific, potentially
resulting in the native language influencing the non-native Lombard speech.

Abstract

The results of a word-spotting experiment are presented in which Dutch listeners tried to spot different types of bisyllabic Dutch words embedded in different types of nonsense contexts. Embedded verbs were not reliably harder to spot than embedded nouns; this suggests that nouns and verbs are recognised via the same basic processes. Iambic words were no harder to spot than trochaic words, suggesting that trochaic words are not in principle easier to recognise than iambic words. Words were harder to spot in consonantal contexts (i.e., contexts which themselves could not be words) than in longer contexts which contained at least one vowel (i.e., contexts which, though not words, were possible words of Dutch). A control experiment showed that this difference was not due to acoustic differences between the words in each context. The results support the claim that spoken-word recognition is sensitive to the viability of sound sequences as possible words.

Abstract

Lexical knowledge influences how human listeners make decisions about speech sounds. Positive lexical effects (faster responses to target sounds in words than in nonwords) are robust across several laboratory tasks, while negative effects (slower responses to targets in more word-like nonwords than in less word-like nonwords) have been found in phonetic decision tasks but not phoneme monitoring tasks. The present experiments tested whether negative lexical effects are therefore a task-specific consequence of the forced choice required in phonetic decision. We compared phoneme monitoring and phonetic decision performance using the same Dutch materials in each task. In both experiments there were positive lexical effects, but no negative lexical effects. We observe that in all studies showing negative lexical effects, the materials were made by cross-splicing, which meant that they contained perceptual evidence supporting the lexically-consistent phonemes. Lexical knowledge seems to influence phonemic decision-making only when there is evidence for the lexically-consistent phoneme in the speech signal.

Abstract

We briefly describe the Merge model of phonemic decision-making, and, in the light of general arguments about the possible role of feedback in spoken-word recognition, defend Merge's feedforward structure. Merge not only accounts adequately for the data, without invoking feedback connections, but does so in a parsimonious manner.

Abstract

This paper presents lexical statistics on the pattern of occurrence of words embedded in other words. We report the results of an analysis of 25000 words, varying in length from two to six syllables, extracted from a phonetically-coded English dictionary (The Longman Dictionary of Contemporary English). Each syllable, and each string of syllables within each word was checked against the dictionary. Two analyses are presented: the first used a complete list of polysyllables, with look-up on the entire dictionary; the second used a sublist of content words, counting only embedded words which were themselves content words. The results have important implications for models of human speech recognition. The efficiency of these models depends, in different ways, on the number and location of words within words.

Abstract

Humans learn language by interaction with their environment and listening to other humans. It should also be possible for computational models to learn language directly from speech but so far most approaches require text. We improve on existing neural network approaches to create visually grounded embeddings for spoken utterances. Using a combination of a multi-layer GRU, importance sampling, cyclic learning rates, ensembling and vectorial self-attention our results show a remarkable increase in image-caption retrieval performance over previous work. Furthermore, we investigate which layers in the model learn to recognise words in the input. We find that deeper network layers are better at encoding word presence, although the final layer has slightly lower performance. This shows that our visually grounded sentence encoder learns to recognise words from the input even though it is not explicitly trained for word recognition.

Abstract

The ultimate goal of our research is to improve an existing speech-based computational model of human speech recognition on the task of simulating the role of fine-grained phonetic information in human speech processing. As part of this work we are investigating articulatory feature classifiers that are able to create reliable and accurate transcriptions of the articulatory behaviour encoded in the acoustic speech signal. Articulatory feature (AF) modelling of speech has received a considerable amount of attention in automatic speech recognition research. Different approaches have been used to build AF classifiers, most notably multi-layer perceptrons. Recently, deep neural networks have been applied to the task of AF classification. This paper aims to improve AF classification by investigating two different approaches: 1) investigating the usefulness of a deep Convolutional neural network (CNN) for AF classification; 2) integrating the Mel filtering operation into the CNN architecture. The results showed a remarkable improvement in classification accuracy of the CNNs over state-of-the-art AF classification results for Dutch, most notably in the minority classes. Integrating the Mel filtering operation into the CNN architecture did not further improve classification performance.

Abstract

This study demonstrates how interaction – specifically negotiation and repair – facilitates the emergence, evolution, and conventionalization of a silent gesture communication system. In a modified iterated learning paradigm, partners communicated noun-verb meanings using only silent gesture. The need to disambiguate similar noun-verb pairs drove these "new" language users to develop a morphology that allowed for quicker processing, easier transmission, and improved accuracy. The specific morphological system that emerged came about through a process of negotiation within the dyad, namely by means of repair. By applying a discourse analytic approach to the use of repair in an experimental methodology for language evolution, we are able to determine not only if interaction facilitates the emergence and learnability of a new communication system, but also how interaction affects such a system

Abstract

The anterior lordosis of the cervical spine is thought
to contribute to pitch (fo) production by influencing
cricoid rotation as a function of larynx height. This
study examines the matter of inter-individual
variation in cervical spine shape and whether this has
an influence on how fo is produced along increasing
or decreasing scales, using the ArtiVarK dataset,
which contains real-time MRI pitch production data.
We find that the cervical spine actively participates in
fo production, but the amount of displacement
depends on individual shape. In general, anterior
spine motion (tending toward cervical lordosis)
occurs for low fo, while posterior movement (tending
towards cervical kyphosis) occurs for high fo.

Abstract

Analyzing EEG signals recorded while participants are listening to continuous speech with the purpose of testing linguistic hypotheses is complicated by the fact that the signals simultaneously reflect exogenous acoustic excitation and endogenous linguistic processing. This makes it difficult to trace subtle differences that occur in mid-sentence position. We apply an analysis based on multivariate temporal response functions to uncover subtle mid-sentence effects. This approach is based on a per-stimulus estimate of the response of the neural system to speech input. Analyzing EEG signals predicted on the basis of the response functions might then bring to light conditionspecific differences in the filtered signals. We validate this approach by means of an analysis of EEG signals recorded with isolated word stimuli. Then, we apply the validated method to the analysis of the responses to the same words in the middle of meaningful sentences.

Abstract

In order to study the cognitive processes underlying speech comprehension, neuro-physiological measures (e.g., EEG and MEG), or behavioural measures (e.g., reaction times and response accuracy) can be applied. Compared to behavioural measures, EEG signals can provide a more fine-grained and complementary view of the processes that take place during the unfolding of an auditory stimulus.

EEG signals are often analysed after having chosen specific time windows, which are usually based on the temporal structure of ERP components expected to be sensitive to the experimental manipulation. However, as the timing of ERP components may vary between experiments, trials, and participants, such a-priori defined analysis time windows may significantly hamper the exploratory power of the analysis of components of interest. In this paper, we explore a wide-window analysis method applied to EEG signals collected in an auditory repetition priming experiment.

This approach is based on a bank of temporal filters arranged along the time axis in combination with linear mixed effects modelling. Crucially, it permits a temporal decomposition of effects in a single comprehensive statistical model which captures the entire EEG trace.

Abstract

Two word-spotting experiments are reported that examine whether the Possible-Word Constraint (PWC) [1] is a language-specific or language-universal strategy for the segmentation of continuous speech. The PWC disfavours parses which leave an impossible residue between the end of a candidate word and a known boundary. The experiments examined cases where the residue was either a CV syllable with a lax vowel, or a CVC syllable with a schwa. Although neither syllable context is a possible word in English, word-spotting in both contexts was easier than with a context consisting of a single consonant. The PWC appears to be language-universal rather than language-specific.

Abstract

We report two experiments in which vowels and steady-state consonants served as targets in a speeded detection task. In the first experiment, two vowels were compared with one voiced and once unvoiced fricative. Response times (RTs) to the vowels were longer than to the fricatives. The error rate was higher for the consonants. Consonants in word-final position produced the shortest RTs, For the vowels, RT correlated negatively with target duration. In the second experiment, the same two vowel targets were compared with two nasals. This time there was no significant difference in RTs, but the error rate was still significantly higher for the consonants. Error rate and length correlated negatively for the vowels only. We conclude that RT differences between phonemes are independent of vocalic or consonantal status. Instead, we argue that the process of phoneme detection reflects more finely grained differences in acoustic/articulatory structure within the phonemic repertoire.

Abstract

Simulations explored the inability of the TRACE model of spoken-word recognition to model the effects on human listening of subcategorical mismatch in word forms. The source of TRACE's failure lay not in interactive connectivity, not in the presence of inter-word competition, and not in the use of phonemic representations, but in the need for continuously optimised interpretation of the input. When an analogue of TRACE was allowed to cycle to asymptote on every slice of input, an acceptable simulation of the subcategorical mismatch data was achieved. Even then, however, the simulation was not as close as that produced by the Merge model, which has inter-word competition, phonemic representations and continuous optimisation (but no interactive connectivity).

Abstract

There is a long-standing assumption that gestural forms are geared by a set of modes of representation (acting, representing, drawing, moulding) with each technique expressing speakers’ focus of attention on specific aspects of referents (Müller, 2013). Beyond different taxonomies describing the modes of representation, it remains unclear what factors motivate certain depicting techniques over others. Results from a pantomime generation task show that pantomimes are not entirely idiosyncratic but rather follow generalisable patterns constrained by their semantic category. We show that a) specific modes of representations are preferred for certain objects (acting for manipulable objects and drawing for non-manipulable objects); and b) that use and ordering of deictics and modes of representation operate in tandem to distinguish between semantically related concepts (e.g., “to drink” vs “mug”). This study provides yet more evidence that our ability to communicate through silent gesture reveals systematic ways to describe events and objects around us

Abstract

A database is presented of relative familiarity ratings for 24 sets of Japanese words, each set comprising words overlapping in the initial portions. These ratings are useful for the generation of material sets for research in the recognition of spoken words.

Abstract

-* indicates shared first authorship -
The present study investigated the effects of selective attention on the processing of morphosyntactic errors in unattended parts of speech. Two groups of German native (L1) speakers participated in the present study. Participants listened to sentences in which irregular verbs were manipulated in three different conditions (correct, incorrect but attested ablaut pattern, incorrect and crosslinguistically unattested ablaut pattern). In order to track fast dynamic neural reactions to the stimuli, electroencephalography was used. After each sentence, participants in Experiment 1 performed a semantic judgement task, which deliberately distracted the participants from the syntactic manipulations and directed their attention to the semantic content of the sentence. In Experiment 2, participants carried out a syntactic judgement task, which put their attention on the critical stimuli. The use of two different attentional tasks allowed for investigating the impact of selective attention on speech processing and whether morphosyntactic processing steps are performed automatically. In Experiment 2, the incorrect attested condition elicited a larger N400 component compared to the correct condition, whereas in Experiment 1 no differences between conditions were found. These results suggest that the processing of morphosyntactic violations in irregular verbs is not entirely automatic but seems to be strongly affected by selective attention.

Abstract

Iconicity is a fundamental feature of human language. However its processing consequences at the behavioral and neural level in spoken word comprehension are not well understood. The current paper presents the behavioral and electrophysiological outcome of an auditory lexical decision task in which native speakers of Dutch listened to onomatopoeic words and matched control words while their electroencephalogram was recorded. Behaviorally, onomatopoeic words were processed as quickly and accurately as words with an arbitrary mapping between form and meaning. Event-related potentials time-locked to word onset revealed a significant decrease in negative amplitude in the N2 and N400 components and a late positivity for onomatopoeic words in comparison to the control words. These findings advance our understanding of the temporal dynamics of iconic form-meaning mapping in spoken word comprehension and suggest interplay between the neural representations of real-world sounds and spoken words.

Abstract

Spontaneously occurring speech is often seamlessly accompanied by hand gestures. Detailed
observations of video data suggest that speech and gesture are tightly synchronized in time,
consistent with a dynamic interplay between body and mind. However, spontaneous gesturespeech
synchrony has rarely been objectively quantified beyond analyses of video data, which
do not allow for identification of kinematic properties of gestures. Consequently, the point in
gesture which is held to couple with speech, the so-called moment of “maximum effort”, has
been variably equated with the peak velocity, peak acceleration, peak deceleration, or the onset
of the gesture. In the current exploratory report, we provide novel evidence from motiontracking
and acoustic data that peak velocity is closely aligned, and shortly leads, the peak pitch
(F0) of speech

Abstract

Word count estimation (WCE) from audio recordings has a number of applications, including quantifying the amount of speech that language-learning infants hear in their natural environments, as captured by daylong recordings made with devices worn by infants. To be applicable in a wide range of scenarios and also low-resource domains, WCE tools should be extremely robust against varying signal conditions and require minimal access to labeled training data in the target domain. For this purpose, earlier work has used automatic syllabification of speech, followed by a least-squares-mapping of syllables to word counts. This paper compares a number of previously proposed syllabifiers in the WCE task, including a supervised bi-directional long short-term memory (BLSTM) network that is trained on a language for which high quality syllable annotations are available (a “high resource language”), and reports how the alternative methods compare on different languages and signal conditions. We also explore additive noise and varying-channel data augmentation strategies for BLSTM training, and show how they improve performance in both matching and mismatching languages. Intriguingly, we also find that even though the BLSTM works on languages beyond its training data, the unsupervised algorithms can still outperform it in challenging signal conditions on novel languages.

Abstract

Infants, children and adults are capable of implicitly extracting regularities from their environment through statistical learning (SL). SL is present from early infancy and found across tasks and modalities, raising questions about the domain generality of SL. However, little is known about its’ developmental trajectory: Is SL fully developed capacity in infancy, or does it improve with age, like other cognitive skills? While SL is well established in infants and adults, only few studies have looked at SL across development with conflicting results: some find age-related improvements while others do not. Importantly, despite its postulated role in language learning, no study has examined the developmental trajectory of auditory SL throughout childhood. Here, we conduct a large-scale study of children's auditory SL across a wide age-range (5-12y, N=115). Results show that auditory SL does not change much across development. We discuss implications for modality-based differences in SL and for its role in language acquisition.

Abstract

Recent work suggests that cultural transmission can lead to the emergence of linguistic structure as speakers’ weak individual biases become amplified through iterated learning. However, to date, no published study has demonstrated a similar emergence of linguistic structure in children. This gap is problematic given that languages are mainly learned by children and that adults may bring existing linguistic biases to the task. Here, we conduct a large-scale study of iterated language learning in both children and adults, using a novel, child-friendly paradigm. The results show that while children make more mistakes overall, their languages become more learnable and show learnability biases similar to those of adults. Child languages did not show a significant increase in linguistic structure over time, but consistent mappings between meanings and signals did emerge on many occasions, as found with adults. This provides the first demonstration that cultural transmission affects the languages children and adults produce similarly.

Abstract

Thematic roles such as Agent and Instrument have a long-standing place in theories of event representation. Nonetheless, the structure of these categories has been difficult to determine. We investigated how instrumental events, such as someone slicing bread with a knife, are categorized in English. Speakers described a variety of typical and atypical instrumental events, and we determined the similarity structure of their descriptions using correspondence analysis. We found that events where the instrument is an extension of an intentional agent were most likely to elicit similar language, highlighting the importance of agency in structuring instrumental categories.

Abstract

The question of whether intonation events have a categorical mental representation has long been a puzzle in prosodic research, and one that experiments testing production and perception across category boundaries have failed to definitively resolve. This paper takes the alternative approach of looking for evidence of structure within a postulated category by testing for a Perceptual Magnet Effect (PME). PME has been found in boundary tones but has not previously been conclusively found in pitch accents. In this investigation, perceived goodness and discriminability of re-synthesised Dutch nuclear rise contours (L*H H%) were evaluated by naive native speakers of Dutch. The variation between these stimuli was quantified using a polynomial-parametric modelling approach (i.e. the SOCoPaSul model) in place of the traditional approach whereby excursion size, peak alignment and pitch register are used independently of each other to quantify variation between pitch accents. Using this approach to calculate the acoustic-perceptual distance between different stimuli, PME was detected: (1) rated goodness, decreased as acoustic-perceptual distance relative to the prototype increased, and (2) equally spaced items far from the prototype were less frequently generalised than equally spaced items in the neighbourhood of the prototype. These results support the concept of categorically distinct intonation events.

Abstract

Many previous studies have shown that both infants and adults are skilled statistical learners. Because statistical learning is affected by attention, learners' ability to manage their attention can play a large role in what they learn. However, it is still unclear how learners allocate their attention in order to gain information in a visual environment containing multiple objects, especially how prior visual experience (i.e., familiarly of objects) influences where people look. To answer these questions, we collected eye movement data from adults exploring multiple novel objects while manipulating object familiarity with global (frequencies) and local (repetitions) regularities. We found that participants are sensitive to both global and local statistics embedded in their visual environment and they dynamically shift their attention to prioritize some objects over others as they gain knowledge of the objects and their distributions within the task.

Abstract

For almost two decades, the poor performance observed with the so-called Director task has been interpreted as evidence of limited use of Theory of Mind in communication. Here we propose a probabilistic model of common ground in referential communication that derives three inferences from an utterance: what the speaker is talking about in a visual context, what she knows about the context, and what referential expressions she prefers. We tested our model by comparing its inferences with those made by human participants and found that it closely mirrors their judgments, whereas an alternative model compromising the hearer’s expectations of cooperativeness and efficiency reveals a worse fit to the human data. Rather than assuming that common ground is fixed in a given exchange and may or may not constrain reference resolution, we show how common ground can be inferred as part of the process of reference assignment.

Abstract

In this paper, I will argue that although the study of multimodal interaction offers exciting new prospects for Human Computer Interaction and human-human communication research, language is the primary form of communication, even in multimodal systems. I will support this claim with theoretical and empirical arguments, mainly drawn from human-human communication research, and will discuss the implications for multimodal communication research and Human-Computer Interaction.

Abstract

While there are many studies on information retrieval models using full-text, there are presently no comparison studies of full-text retrieval vs. retrieval only over the titles of documents. On the one hand, the full-text of documents like scientific papers is not always available due to, e.g., copyright policies of academic publishers. On the other hand, conducting a search based on titles alone has strong limitations. Titles are short and therefore may not contain enough information to yield satisfactory search results. In this paper, we compare different retrieval models regarding their search performance on the full-text vs. only titles of documents. We use different datasets, including the three digital library datasets: EconBiz, IREON, and PubMed. The results show that it is possible to build effective title-based retrieval models that provide competitive results comparable to full-text retrieval. The difference between the average evaluation results of the best title-based retrieval models is only 3% less than those of the best full-text-based retrieval models.

Abstract

Most of the study of basic emotion expressions has focused on facial expressions and little work has been done to specifically investigate happiness, the only positive of the basic emotions (Ekman & Friesen, 1971). However, a theoretical suggestion has been made that happiness could be broken down into discrete positive emotions, which each fulfil the criteria of basic emotions, and that these would be expressed vocally (Ekman, 1992). To empirically test this hypothesis, 20 participants categorised 80 paralinguistic sounds using the labels achievement, amusement, contentment, pleasure and relief. The results suggest that achievement, amusement and relief are perceived as distinct categories, which subjects accurately identify. In contrast, the categories of contentment and pleasure were systematically confused with other responses, although performance was still well above chance levels. These findings are initial evidence that the positive emotions engage distinct vocal expressions and may be considered to be distinct emotion categories.

Abstract

Fine-Tracker is a speech-based model of human speech
recognition. While previous work has shown that Fine-Tracker
is successful at modelling aspects of human spoken-word
recognition, its speech recognition performance is not
comparable to that of human performance, possibly due to
suboptimal intermediate articulatory feature (AF)
representations. This study investigates the effect of improved
AF representations, obtained using a state-of-the-art deep
convolutional network, on Fine-Tracker’s simulation and
recognition performance: Although the improved AF quality
resulted in improved speech recognition; it, surprisingly, did
not lead to an improvement in Fine-Tracker’s simulation power.

Abstract

This work focuses on efficient use of the training material by selecting the optimal set of model topologies. We do this by training multiple word models of each word class, based on a subclassification according to a priori knowledge of the training material. We will examine classification criteria with respect to duration of the word, gender of the speaker, position of the word in the utterance, pauses in the vicinity of the word, and combinations of these. Comparative experiments were carried out on a corpus consisting of Dutch spoken connected digit strings and isolated digits, which are recorded in a wide variety of acoustic conditions. The results show, that classification based on gender of the speaker, position of the digit in the string, pauses in the vicinity of the training tokens, and models based on a combination of these criteria perform significantly better than the set with single models per digit.

Abstract

In this paper, we investigate the ability of SpeM, our recognition system based on the combination of an automatic phone recogniser and a wordsearch module, to determine as early as possible during the word recognition process whether a word is likely to be recognised correctly (this we refer to as ‘on-line’ early word recognition). We present two measures that can be used to predict whether a word is correctly recognised: the Bayesian word activation and the amount of available (acoustic) information for a word. SpeM was tested on 1,463 polysyllabic words in 885 continuous speech utterances. The investigated predictors indicated that a word activation that is 1) high (but not too high) and 2) based on more phones is more reliable to predict the correctness of a word than a similarly high value based on a small number of phones or a lower value of the word activation.