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 investigated pronunciation variants of word-final obstruent-liquid-schwa (OLS) clusters in casual French and the variables predicting the absence of the phonemes in these clusters. In a dataset of 291 noun tokens extracted from a corpus of casual conversations, we observed that in 80.7% of the tokens, at least one phoneme was absent and that in no less than 15.5% the whole cluster was absent (e.g., /mis/ for ministre). Importantly, the probability of a phoneme being absent was higher if the following phoneme was absent as well. These data show that reduction can affect several phonemes at once and is not restricted to just a handful of (function) words. Moreover, our results demonstrate that the absence of each single phoneme is affected by the speaker's tendency to increase ease of articulation and to adapt a word's pronunciation variant to the time available.

Abstract

The present study investigated the effect of orthography on the learning and subsequent processing of pronunciation variants in a second language. Dutch learners of French learned reduced pronunciation variants that result from schwa-zero alternation in French (e.g., reduced /ʃnij/ from chenille 'caterpillar'). Half of the participants additionally learnt the words' spellings, which correspond more closely to the full variants with schwa. On the following day, participants performed an auditory lexical decision task, in which they heard half of the words in their reduced variants, and the other half in their full variants. Participants who had exclusively learnt the auditory forms performed significantly worse on full variants than participants who had also learnt the spellings. This shows that learners integrate phonological and orthographic information to process pronunciation variants. There was no difference between both groups in their performances on reduced variants, suggesting that the exposure to spelling does not impede learners' processing of these variants.

Abstract

In an auditory lexical decision experiment, 5,541 spoken content words and pseudo-words were presented to 20 native speakers of Dutch. The words vary in phonological makeup and in number of syllables and stress pattern, and are further representative of the native Dutch vocabulary in that most are morphologically complex, comprising two stems or one stem plus derivational and inflectional suffixes, with inflections representing both regular and irregular paradigms; the pseudo-words were matched in these respects to the real words. The BALDEY data file includes response times and accuracy rates, with for each item morphological information plus phonological and acoustic information derived from automatic phonemic segmentation of the stimuli. Two initial analyses illustrate how this data set can be used. First, we discuss several measures of the point at which a word has no further neighbors, and compare the degree to which each measure predicts our lexical decision response outcomes. Second, we investigate how well four different measures of frequency of occurrence (from written corpora, spoken corpora, subtitles and frequency ratings by 70 participants) predict the same outcomes. These analyses motivate general conclusions about the auditory lexical decision task. The (publicly available) BALDEY database lends itself to many further analyses.

Abstract

This article presents two studies investigating how the situation in which speech is uttered affects the frequency with which words are reduced. Study 1 is based on the Spoken Dutch Corpus, which consists of 15 components, nearly all representing a different speech situation. This study shows that the components differ in how often ten semantically weak words are highly reduced. The differences are especially large between the components with scripted and unscripted speech. Within the component group of unscripted speech, the formality of the situation shows an effect. Study 2 investigated segment reduction in a shadowing experiment in which participants repeated Dutch carefully and casually articulated sentences. Prefixal schwa and suffixal /t/ were absent in participants' responses to both sentences types as often as in formal interviews. If a segment was absent, this appeared to be mostly due to extreme co-articulation, unlike in speech produced in less formal situations. Speakers thus adapted more to the formal situation of the experiment than to the stimuli to be shadowed. We conclude that speech situation affects the occurrence of reduced word pronunciation variants, which should be accounted for by psycholinguistic models of speech production and comprehension

Abstract

This paper investigates whether individual speakers forming a homogeneous group differ in their choice and pronunciation of words when engaged in casual conversation, and if so, how they differ. More specifically, it examines whether the Balanced Winnow classifier is able to distinguish between the twenty speakers of the Ernestus Corpus of Spontaneous Dutch, who all have the same social background. To examine differences in choice and pronunciation of words, instead of characteristics of the speech signal itself, classification was based on lexical and pronunciation features extracted from hand-made orthographic and automatically generated broad phonetic transcriptions. The lexical features consisted of words and two-word combinations. The pronunciation features represented pronunciation variations at the word and phone level that are typical for casual speech. The best classifier achieved a performance of 79.9% and was based on the lexical features and on the pronunciation features representing single phones and triphones. The speakers must thus differ from each other in these features. Inspection of the relevant features indicated that, among other things, the words relevant for classification generally do not contain much semantic content, and that speakers differ not only from each other in the use of these words but also in their pronunciation.

Abstract

This paper studies the robustness of exemplar effects in word comprehension by means of four long-term priming experiments with lexical decision tasks in Dutch. A prime and target represented the same word type and were presented with the same or different degree of reduction. In Experiment 1, participants heard only a small number of trials, a large proportion of repeated words, and stimuli produced by only one speaker. They recognized targets more quickly if these represented the same degree of reduction as their primes, which forms additional evidence for the exemplar effects reported in the literature. Similar effects were found for two speakers who differ in their pronunciations. In Experiment 2, with a smaller proportion of repeated words and more trials between prime and target, participants recognized targets preceded by primes with the same or a different degree of reduction equally quickly. Also, in Experiments 3 and 4, in which listeners were not exposed to one but two types of pronunciation variation (reduction degree and speaker voice), no exemplar effects arose. We conclude that the role of exemplars in speech comprehension during natural conversations, which typically involve several speakers and few repeated content words, may be smaller than previously assumed.

Abstract

This paper investigated the learnability and generalizability of French schwa alternation by Dutch low-proficiency second language learners. We trained 40 participants on 24 new schwa words by exposing them equally often to the reduced and full forms of these words. We then assessed participants' accuracy and reaction times to these newly learnt words as well as 24 previously encountered schwa words with an auditory lexical decision task. Our results show learning of the new words in both forms. This suggests that lack of exposure is probably the main cause of learners' difficulties with reduced forms. Nevertheless, the full forms were slightly better recognized than the reduced ones, possibly due to phonetic and phonological properties of the reduced forms. We also observed no generalization to previously encountered words, suggesting that our participants stored both of the learnt word forms and did not create a rule that applies to all schwa words.

Abstract

In speech, words are often reduced rather than fully pronounced (e.g., (/ˈsʌmri/ for /ˈsʌməri/, summary). Non-native listeners may have problems in processing these reduced forms, because they have encountered them less often. This paper addresses the question whether this also holds for highly proficient non-natives and for words with similar forms and meanings in the non-natives' mother tongue (i.e., cognates). In an English auditory lexical decision task, natives and highly proficient Dutch non-natives of English listened to cognates and non-cognates that were presented in full or without their post-stress schwa. The data show that highly proficient learners are affected by reduction as much as native speakers. Nevertheless, the two listener groups appear to process reduced forms differently, because non-natives produce more errors on reduced cognates than on non-cognates. While listening to reduced forms, non-natives appear to be hindered by the co-activated lexical representations of cognate forms in their native language.

Abstract

This paper studies the influence of adverse listening conditions on exemplar effects in priming experiments that do not instruct participants to use their episodic memories. We conducted two lexical decision experiments, in which a prime and a target represented the same word type and could be spoken by the same or a different speaker. In Experiment 1, participants listened to clear speech, and showed no exemplar effects: they recognised repetitions by the same speaker as quickly as different speaker repetitions. In Experiment 2, the stimuli contained noise, and exemplar effects did arise. Importantly, Experiment 1 elicited longer average RTs than Experiment 2, a result that contradicts the time-course hypothesis, according to which exemplars only play a role when processing is slow. Instead, our findings support the hypothesis that exemplar effects arise under adverse listening conditions, when participants are stimulated to use their episodic memories in addition to their mental lexicons.

Abstract

This paper presents DIANA, a new computational model of human speech processing. It is the first model that simulates the complete processing chain from the on-line processing of an acoustic signal to the execution of a response, including reaction times. Moreover it assumes minimal modularity. DIANA consists of three components. The activation component computes a probabilistic match between the input acoustic signal and representations in DIANA’s lexicon, resulting in a list of word hypotheses changing over time as the input unfolds. The decision component operates on this list and selects a word as soon as sufficient evidence is available. Finally, the execution component accounts for the time to execute a behavioral action. We show that DIANA well simulates the average participant in a word recognition experiment.

Abstract

DIANA is an end-to-end computational model of speech processing, which takes as input the speech signal, and provides as output the orthographic transcription of the stimulus, a word/non-word judgment and the associated estimated reaction time. So far, the model has only been tested for Dutch. In this paper, we extend DIANA such that it can also process North American English. The model is tested by having it simulate human participants in a large scale North American English lexical decision experiment. The simulations show that DIANA can adequately approximate the reaction times of an average participant (r = 0.45). In addition, they indicate that DIANA does not yet adequately model the cognitive processes that take place after stimulus offset.

Abstract

This study examines whether the production of words with two phonological variants involves single or multiple lexical phonological representations. Three production experiments investigated the roles of the relative frequencies of the two pronunciation variants of French words with schwa: the schwa variant (e.g., Image ) and the reduced variant (e.g., Image ). In two naming tasks and in a symbol–word association learning task, variants with higher relative frequencies were produced faster. This suggests that the production lexicon keeps a frequency count for each variant and hence that schwa words are represented in the production lexicon with two different lexemes. In addition, the advantage for schwa variants over reduced variants in the naming tasks but not in the learning task and the absence of a variant relative frequency effect for schwa variants produced in isolation support the hypothesis that context affects the variants’ lexical activation and modulates the effect of variant relative frequency.

Abstract

This corpus-based study shows that the presence and duration of schwa in Dutch word-initial syllables are affected by a word’s predictability and its morphological structure. Schwa is less reduced in words that are more predictable given the following word. In addition, schwa may be longer if the syllable forms a prefix, and in prefixes the duration of schwa is positively correlated with the frequency of the word relative to its stem. Our results suggest that the conditions which favor reduced realizations are more complex than one would expect on the basis of the current literature.

Abstract

Despite using different algorithms, most unsupervised automatic phone segmentation methods achieve similar performance in terms of percentage correct boundary detection. Nevertheless, unsupervised segmentation algorithms are not able to perfectly reproduce manually obtained reference transcriptions. This paper investigates fundamental problems for unsupervised segmentation algorithms by comparing a phone segmentation obtained using only the acoustic information present in the signal with a reference segmentation created by human transcribers. The analyses of the output of an unsupervised speech segmentation method that uses acoustic change to hypothesize boundaries showed that acoustic change is a fairly good indicator of segment boundaries: over two-thirds of the hypothesized boundaries coincide with segment boundaries. Statistical analyses showed that the errors are related to segment duration, sequences of similar segments, and inherently dynamic phones. In order to improve unsupervised automatic speech segmentation, current one-stage bottom-up segmentation methods should be expanded into two-stage segmentation methods that are able to use a mix of bottom-up information extracted from the speech signal and automatically derived top-down information. In this way, unsupervised methods can be improved while remaining flexible and language-independent.

Abstract

This paper presents a study on the acoustic sub-segmental properties of word-final /t/ in conversational standard Dutch and how these properties contribute to whether humans and an ASR system classify the /t/ as acoustically present or absent. In general, humans and the ASR system use the same cues (presence of a constriction, a burst, and alveolar frication), but the ASR system is also less sensitive to fine cues (weak bursts, smoothly starting friction) than human listeners and misled by the presence of glottal vibration. These data inform the further development of models of human and automatic speech processing.

Abstract

Spontal-N is a corpus of spontaneous, interactional Norwegian. To our knowledge, it is the first corpus of Norwegian in which the majority of speakers have spent significant parts of their lives in Sweden, and in which the recorded speech displays varying degrees of interference from Swedish. The corpus consists of studio quality audio- and video-recordings of four 30-minute free conversations between acquaintances, and a manual orthographic transcription of the entire material. On basis of the orthographic transcriptions, we automatically annotated approximately 50 percent of the material on the phoneme level, by means of a forced alignment between the acoustic signal and pronunciations listed in a dictionary. Approximately seven percent of the automatic transcription was manually corrected. Taking the manual correction as a gold standard, we evaluated several sources of pronunciation variants for the automatic transcription. Spontal-N is intended as a general purpose speech resource that is also suitable for investigating phonetic detail.

Abstract

This paper presents the Kachna corpus of spontaneous speech, in which ten Czech and ten Norwegian speakers were recorded both in their native language and in English. The dialogues are elicited using a picture replication task that requires active cooperation and interaction of speakers by asking them to produce a drawing as close to the original as possible. The corpus is appropriate for the study of interactional features and speech reduction phenomena across native and second languages. The combination of productions in non-native English and in speakers’ native language is advantageous for investigation of L2 issues while providing a L1 behaviour reference from all the speakers. The corpus consists of 20 dialogues comprising 12 hours 53 minutes of recording, and was collected in 2008. Preparation of the transcriptions, including a manual orthographic transcription and an automatically generated phonetic transcription, is currently in progress. The phonetic transcriptions are automatically generated by aligning acoustic models with the speech signal on the basis of the orthographic transcriptions and a dictionary of pronunciation variants compiled for the relevant language. Upon completion the corpus will be made available via the European Language Resources Association (ELRA).

Abstract

This study investigates phrase-medial vowel devoicing in European French (e.g. /ty po/ [typo] 'you can'). Our spontaneous speech data confirm that French phrase-medial devoicing is a frequent phenomenon affecting high vowels preceded by voiceless consonants. We also found that devoicing is more frequent in temporally reduced and coarticulated vowels. Complete and partial devoicing were conditioned by the same variables (speech rate, consonant type and distance from the end of the AP). Given these results, we propose that phrase-medial vowel devoicing in French arises mainly from the temporal compression of vocalic gestures and the aerodynamic conditions imposed by high vowels.

Abstract

This article describes the preparation, recording and orthographic transcription of a new speech corpus, the Nijmegen Corpus of Casual French (NCCFr). The corpus contains a total of over 36 h of recordings of 46 French speakers engaged in conversations with friends. Casual speech was elicited during three different parts, which together provided around 90 min of speech from every pair of speakers. While Parts 1 and 2 did not require participants to perform any specific task, in Part 3 participants negotiated a common answer to general questions about society. Comparisons with the ESTER corpus of journalistic speech show that the two corpora contain speech of considerably different registers. A number of indicators of casualness, including swear words, casual words, verlan, disfluencies and word repetitions, are more frequent in the NCCFr than in the ESTER corpus, while the use of double negation, an indicator of formal speech, is less frequent. In general, these estimates of casualness are constant through the three parts of the recording sessions and across speakers. Based on these facts, we conclude that our corpus is a rich resource of highly casual speech, and that it can be effectively exploited by researchers in language science and technology.

Abstract

This article describes the preparation, recording and orthographic transcription of a new speech corpus, the Nijmegen Corpus of Casual Spanish (NCCSp). The corpus contains around 30 hours of recordings of 52 Madrid Spanish speakers engaged in conversations with friends. Casual speech was elicited during three different parts, which together provided around ninety minutes of speech from every group of speakers. While Parts 1 and 2 did not require participants to perform any specific task, in Part 3 participants negotiated a common answer to general questions about society. Information about how to obtain a copy of the corpus can be found online at http://mirjamernestus.ruhosting.nl/Ernestus/NCCSp

Abstract

Previous research suggests that bilinguals presented with low and high predictability sentences benefit from semantics in clear but not in conversational speech [1]. In everyday speech, however, many words are not highly predictable. Previous research has shown that native listeners can use also more subtle semantic contextual information [2]. The present study reports two auditory lexical decision experiments investigating to what extent late Asian-English bilinguals benefit from subtle semantic cues in their processing of English unreduced and reduced speech. Our results indicate that these bilinguals are less sensitive to semantic cues than native listeners for both speech registers.