Publications

Abstract

The event-related potential (ERP) component named phonological mismatch negativity (PMN) arises when listeners hear an unexpected word form in a spoken sentence [1]. The PMN is thought to reflect the mismatch between expected and perceived auditory speech input. In this paper, we use the PMN to test a central premise in the predictive coding framework [2], namely that the mismatch between prior expectations and sensory input is an important mechanism of perception. We test this with natural speech materials containing approximately 50,000 word tokens. The corresponding EEG-signal was recorded while participants (n = 48) listened to these materials. Following [3], we quantify the mismatch with two word probability distributions (WPD): a WPD based on preceding context, and a WPD that is additionally updated based on the incoming audio of the current word. We use the between-WPD cross entropy for each word in the utterances and show that a higher cross entropy correlates with a more negative PMN. Our results show that listeners anticipate auditory input while processing each word in naturalistic speech. Moreover, complementing previous research, we show that predictive language processing occurs across the whole probability spectrum.

Abstract

The mismatch between top-down predicted and bottom-up perceptual input is an important mechanism of perception according to the predictive coding framework (Friston, [1]). In this paper we develop and validate a new information-theoretic measure that quantifies the mismatch between expected and observed auditory input during speech processing. We argue that such a mismatch measure is useful for the study of speech processing. To compute the mismatch measure, we use naturalistic speech materials containing approximately 50,000 word tokens. For each word token we first estimate the prior word probability distribution with the aid of statistical language modelling, and next use automatic speech recognition to update this word probability distribution based on the unfolding speech signal. We validate the mismatch measure with multiple analyses, and show that the auditory-based update improves the probability of the correct word and lowers the uncertainty of the word probability distribution. Based on these results, we argue that it is possible to explicitly estimate the mismatch between predicted and perceived speech input with the cross entropy between word expectations computed before and after an auditory update.

Abstract

Previous research has demonstrated that language use can vary depending on the context of situation. The present paper extends this finding by comparing word predictability differences between 14 speech registers ranging from highly informal conversations to read-aloud books. We trained 14 statistical language models to compute register-specific word predictability and trained a register classifier on the perplexity score vector of the language models. The classifier distinguishes perfectly between samples from all speech registers and this result generalizes to unseen materials. We show that differences in vocabulary and sentence length cannot explain the speech register classifier’s performance. The combined results show that speech registers differ in word predictability.

Abstract

Previous studies have shown that speakers align their speech to each other at multiple linguistic levels. This study investigates whether alignment is mostly the result of priming from the immediately preceding
speech materials, focussing on pitch and articulation rate (AR). Native Dutch speakers completed sentences, first by themselves (pre-test), then in alternation with Confederate 1 (Round 1), with Confederate 2 (Round 2), with Confederate 1 again
(Round 3), and lastly by themselves again (post-test). Results indicate that participants aligned to the confederates and that this alignment lasted during the post-test. The confederates’ directly preceding sentences were not good predictors for the participants’ pitch and AR. Overall, the results indicate that alignment is more of a global effect than a local priming effect.

Abstract

This paper shows that the dictation task, a well-
known testing instrument in language education, has
untapped potential as a research tool for studying
speech perception. We describe how transcriptions
can be scored on measures of lexical, orthographic,
phonological, and semantic similarity to target
phrases to provide comprehensive information about
accuracy at different processing levels. The former
three measures are automatically extractable,
increasing objectivity, and the middle two are
gradient, providing finer-grained information than
traditionally used. We evaluate the measures in an
English dictation task featuring phonetically reduced
continuous speech. Whereas the lexical and
orthographic measures emphasize listeners’ word
identification difficulties, the phonological measure
demonstrates that listeners can often still recover
phonological features, and the semantic measure
captures their ability to get the gist of the utterances.
Correlational analyses and a discussion of practical
and theoretical considerations show that combining
multiple measures improves the dictation task’s
utility as a research tool.

Abstract

Lexically guided perceptual learning has traditionally been studied with ambiguous consonant sounds to which native listeners are exposed in a purely receptive listening context. To extend previous research, we investigate whether lexically guided learning applies to a vowel shift encountered by non-native listeners in an interactive dialogue. Dutch participants played a two-player game in English in either a control condition, which contained no evidence for a vowel shift, or a lexically constraining condition, in which onscreen lexical information required them to re-interpret their interlocutor’s /ɪ/ pronunciations as representing /ε/. A phonetic categorization pre-test and post-test were used to assess whether the game shifted listeners’ phonemic boundaries such that more of the /ε/-/ɪ/ continuum came to be perceived as /ε/. Both listener groups showed an overall post-test shift toward /ɪ/, suggesting that vowel perception may be sensitive to directional biases related to properties of the speaker’s vowel space. Importantly, listeners in the lexically constraining condition made relatively more post-test /ε/ responses than the control group, thereby exhibiting an effect of lexically guided adaptation. The results thus demonstrate that non-native listeners can adjust their phonemic boundaries on the basis of lexical information to accommodate a vowel shift learned in interactive conversation.

Abstract

An important dimension of linguistic variation is formality. This study investigates the role of social distance between interlocutors. Twenty-five native Dutch speakers retold eight short films to confederates, who acted either formally or informally. Speakers were familiarized with the informal confederates, whereas the formal confederates remained strangers. Results show that the two types of interlocutors elicited different versions of the same stories. Formal interlocutors (large social distance) elicited lower articulation rates, and more nouns and prepositions, both indicators of explicit information. Speakers addressing interlocutors to whom social distance was small, however, provided more explicit information with an involved character (i.e. adjectives with subjective meanings). They also used the word and more often as a gap filler or as a way to keep the floor. Furthermore, they were more likely to laugh and to use more interjections, first-person pronouns and direct speech, which are all indicators of involvement, empathy and subjectivity.

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

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

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

Many psycholinguistic models of speech sequence planning make claims about the onset and offset times of planning units, such as words, syllables, and phonemes. These predictions typically go untested, however, since psycholinguists have assumed that the temporal dynamics of the speech signal is a poor index of the temporal dynamics of the underlying speech planning process. This article argues that this problem is tractable, and presents and validates two simple metrics that derive planning unit onset and offset times from the acoustic signal and articulatographic data.

Abstract

This study addresses the extent to which the location of primary stress in Dutch, German, and English monomorphemic words is affected by the syllables preceding the three final syllables. We present analyses of the monomorphemic words in the CELEX lexical database, which showed that penultimate primary stress is less frequent in Dutch and English trisyllabic than quadrisyllabic words. In addition, we discuss paper-and-pencil experiments in which native speakers assigned primary stress to pseudowords. These experiments provided evidence that in all three languages penultimate stress is more likely in quadrisyllabic than in trisyllabic words. We explain this length effect with the preferences in these languages for word-initial stress and for alternating patterns of stressed and unstressed syllables. The experimental data also showed important intra- and interspeaker variation, and they thus form a challenging test case for theories of language variation.

Abstract

This paper explores the relationship between the acoustic duration of phonemic sequences and their frequencies of occurrence. The data were obtained from large (sub)corpora of spontaneous speech in Dutch, English, German, and Italian. Acoustic duration of an n-phone is shown to codetermine the n-phone's frequency of use, such that languages preferentially use diphones and triphones that are neither very long nor very short. The observed distributions are well approximated by a theoretical function that quantifies the concurrent action of the self-regulatory processes of minimization of articulatory effort and minimization of perception effort

Abstract

This study reports a shadowing experiment, in which one has to repeat a speech stimulus as fast as possible. We tested claims about a direct link between perception and production based on speech gestures, and obtained two types of counterevidence. First, shadowing is not slowed down by a gestural mismatch between stimulus and response. Second, phonetic detail is more likely to be imitated in a shadowing task if it is phonologically relevant. This is consistent with the idea that speech perception and speech production are only loosely coupled, on an abstract phonological level.

Abstract

In four experiments, we investigated how listeners compensate for reduced /t/ in Dutch. Mitterer and Ernestus [Mitterer,H., & Ernestus, M. (2006). Listeners recover /t/s that speakers lenite: evidence from /t/-lenition in Dutch. Journal of Phonetics, 34, 73–103] showed that listeners are biased to perceive a /t/ more easily after /s/ than after /n/, compensating for the tendency of speakers to reduce word-final /t/ after /s/ in spontaneous conversations. We tested the robustness of this phonological context effect in perception with three very different experimental tasks: an identification task, a discrimination task with native listeners and with non-native listeners who do not have any experience with /t/-reduction,and a passive listening task (using electrophysiological dependent measures). The context effect was generally robust against these experimental manipulations, although we also observed some deviations from the overall pattern. Our combined results show that the context effect in compensation for reduced /t/ results from a complex process involving auditory constraints, phonological learning, and lexical constraints.

Abstract

This study investigates the roles of phonetic analogy and lexical frequency in an ongoing sound change, the devoicing of fricatives in Dutch, which occurs mainly in the Netherlands and to a lesser degree in Flanders. In the experiment, Dutch and Flemish students read two variants of 98 words: the standard and a nonstandard form with the incorrect voice value of the fricative. Dutch students chose the non-standard forms with devoiced fricatives more often than Flemish students. Moreover, devoicing, though a gradual process, appeared lexically diffused, affecting first the words that are low in frequency and phonetically similar to words with voiceless fricatives.

Abstract

This paper presents the steps needed to make a corpus of Dutch spontaneous dialogues accessible for automatic phonetic research aimed at increasing our understanding of reduction phenomena and the role of fine phonetic detail. Since the corpus was not created with automatic processing in mind, it needed to be reshaped. The first part of this paper describes the actions needed for this reshaping in some detail. The second part reports the results of a preliminary analysis of the reduction phenomena in the corpus. For this purpose a phonemic transcription of the corpus was created by means of a forced alignment, first with a lexicon of canonical pronunciations and then with multiple pronunciation variants per word. In this study pronunciation variants were generated by applying a large set of phonetic processes that have been implicated in reduction to the canonical pronunciations of the words. This relatively straightforward procedure allows us to produce plausible pronunciation variants and to verify and extend the results of previous reduction studies reported in the literature.

Abstract

This study reports general and language-specific patterns in phoneme identification. In a series of phoneme monitoring experiments, Castilian Spanish, Catalan, Dutch, English, and Polish listeners identified vowel, fricative, and stop consonant targets that are phonemic in all these languages, embedded in nonsense words. Fricatives were generally identified more slowly than vowels, while the speed of identification for stop consonants was highly dependent on the onset of the measurements. Moreover, listeners' response latencies and accuracy in detecting a phoneme correlated with the number of categories within that phoneme's class in the listener's native phoneme repertoire: more native categories slowed listeners down and decreased their accuracy. We excluded the possibility that this effect stems from differences in the frequencies of occurrence of the phonemes in the different languages. Rather, the effect of the number of categories can be explained by general properties of the perception system, which cause language-specific patterns in speech processing.

Abstract

This paper discusses four experiments on Dutch which show that distinctive phonological features differ in their relevance for word recognition. The relevance of a feature for word recognition depends on its phonological stability, that is, the extent to which that feature is generally realized in accordance with its lexical specification in the relevant word position. If one feature value is uninformative, all values of that feature are less relevant for word recognition, with the least informative feature being the least relevant. Features differ in their relevance both in spoken and written word recognition, though the differences are more pronounced in auditory lexical decision than in self-paced reading.

Abstract

This study addresses the question to what extent the production of regular past tense forms in Dutch is a¤ected by analogical processes. We report an experiment in which native speakers of Dutch listened to existing regular verbs over headphones, and had to indicate which of the past tense allomorphs, te or de, was appropriate for these verbs. According to generative analyses, the choice between the two su‰xes is completely regular and governed by the underlying [voice]-specification of the stem-final segment. In this approach, no analogical e¤ects are expected. In connectionist and analogical approaches, by contrast, the phonological similarity structure in the lexicon is expected to a¤ect lexical processing. Our experimental results support the latter approach: all participants created more nonstandard past tense forms, produced more inconsistency errors, and responded more slowly for verbs with stronger analogical support for the nonstandard form.

Abstract

Listeners cannot recognize highly reduced word forms in isolation, but they can do so when these forms are presented in context (Ernestus, Baayen, & Schreuder, 2002). This suggests that not all possible surface forms of words have equal status in the mental lexicon. The present study shows that the reduced forms are linked to the canonical representations in the mental lexicon, and that these latter representations induce reconstruction processes. Listeners restore suffixes that are partly or completely missing in reduced word forms. A series of phoneme-monitoring experiments reveals the nature of this restoration: the basis for suffix restoration is mainly phonological in nature, but orthography has an influence as well.

Abstract

In this paper, we show that both token and type-based effects in lexical processing can result from a single, token-based, system, and therefore, do not necessarily reflect different levels of processing. We report three Simple Recurrent Networks modeling Dutch past-tense formation. These networks show token-based frequency effects and type-based analogical effects closely matching the behavior of human participants when producing past-tense forms for both existing verbs and pseudo-verbs. The third network covers the full vocabulary of Dutch, without imposing predefined linguistic structure on the input or output words.

Abstract

This paper discusses some pitfalls in corpus research and suggests solutions on the basis of examples and computer simulations. We first address reliability problems in language transcriptions, agreement between transcribers, and how disagreements can be dealt with. We then show that the frequencies of occurrence obtained from a corpus cannot always be analyzed with the traditional X2 test, as corpus data are often not sequentially independent and unit independent. Next, we stress the relevance of the power of statistical tests, and the sizes of statistically significant effects. Finally, we point out that a t-test based on log odds often provides a better alternative to a X2 analysis based on frequency counts.