410 Linguistik
Refine
Has Fulltext
- yes (7)
Year of publication
- 2023 (7) (remove)
Document Type
- Doctoral Thesis (7) (remove)
Is part of the Bibliography
- yes (7)
Keywords
- Prosodie (2)
- Spracherwerb (2)
- Säuglinge (2)
- infants (2)
- language acquisition (2)
- prosody (2)
- sentence comprehension (2)
- Aphasie (1)
- Approximate Bayesian Computation (1)
- Cue-Gewichtung (1)
Successful sentence comprehension requires the comprehender to correctly figure out who did what to whom. For example, in the sentence John kicked the ball, the comprehender has to figure out who did the action of kicking and what was being kicked. This process of identifying and connecting the syntactically-related words in a sentence is called dependency completion. What are the cognitive constraints that determine dependency completion? A widely-accepted theory is cue-based retrieval. The theory maintains that dependency completion is driven by a content-addressable search for the co-dependents in memory. The cue-based retrieval explains a wide range of empirical data from several constructions including subject-verb agreement, subject-verb non-agreement, plausibility mismatch configurations, and negative polarity items.
However, there are two major empirical challenges to the theory: (i) Grammatical sentences’ data from subject-verb number agreement dependencies, where the theory predicts a slowdown at the verb in sentences like the key to the cabinet was rusty compared to the key to the cabinets was rusty, but the data are inconsistent with this prediction; and, (ii) Data from antecedent-reflexive dependencies, where a facilitation in reading times is predicted at the reflexive in the bodybuilder who worked with the trainers injured themselves vs. the bodybuilder who worked with the trainer injured themselves, but the data do not show a facilitatory effect.
The work presented in this dissertation is dedicated to building a more general theory of dependency completion that can account for the above two datasets without losing the original empirical coverage of the cue-based retrieval assumption. In two journal articles, I present computational modeling work that addresses the above two empirical challenges.
To explain the grammatical sentences’ data from subject-verb number agreement dependencies, I propose a new model that assumes that the cue-based retrieval operates on a probabilistically distorted representation of nouns in memory (Article I). This hybrid distortion-plus-retrieval model was compared against the existing candidate models using data from 17 studies on subject-verb number agreement in 4 languages. I find that the hybrid model outperforms the existing models of number agreement processing suggesting that the cue-based retrieval theory must incorporate a feature distortion assumption.
To account for the absence of facilitatory effect in antecedent-reflexive dependencies, I propose an individual difference model, which was built within the cue-based retrieval framework (Article II). The model assumes that individuals may differ in how strongly they weigh a syntactic cue over a number cue. The model was fitted to data from two studies on antecedent-reflexive dependencies, and the participant-level cue-weighting was estimated. We find that one-fourth of the participants, in both studies, weigh the syntactic cue higher than the number cue in processing reflexive dependencies and the remaining participants weigh the two cues equally. The result indicates that the absence of predicted facilitatory effect at the level of grouped data is driven by some, not all, participants who weigh syntactic cues higher than the number cue. More generally, the result demonstrates that the assumption of differential cue weighting is important for a theory of dependency completion processes. This differential cue weighting idea was independently supported by a modeling study on subject-verb non-agreement dependencies (Article III).
Overall, the cue-based retrieval, which is a general theory of dependency completion, needs to incorporate two new assumptions: (i) the nouns stored in memory can undergo probabilistic feature distortion, and (ii) the linguistic cues used for retrieval can be weighted differentially. This is the cumulative result of the modeling work presented in this dissertation.
The dissertation makes an important theoretical contribution: Sentence comprehension in humans is driven by a mechanism that assumes cue-based retrieval, probabilistic feature distortion, and differential cue weighting. This insight is theoretically important because there is some independent support for these three assumptions in sentence processing and the broader memory literature. The modeling work presented here is also methodologically important because for the first time, it demonstrates (i) how the complex models of sentence processing can be evaluated using data from multiple studies simultaneously, without oversimplifying the models, and (ii) how the inferences drawn from the individual-level behavior can be used in theory development.
Early sensitivity to prosodic phrase boundary cues: Behavioral evidence from German-learning infants
(2023)
This dissertation seeks to shed light on the relation of phrasal prosody and developmental speech perception in German-learning infants. Three independent empirical studies explore the role of acoustic correlates of major prosodic boundaries, specifically pitch change, final lengthening, and pause, in infant boundary perception. Moreover, it was examined whether the sensitivity to prosodic phrase boundary markings changes during the first year of life as a result of perceptual attunement to the ambient language (Aslin & Pisoni, 1980).
Using the headturn preference procedure six- and eight-month-old monolingual German-learning infants were tested on their discrimination of two different prosodic groupings of the same list of coordinated names either with or without an internal IPB after the second name, that is, [Moni und Lilli] [und Manu] or [Moni und Lilli und Manu]. The boundary marking was systematically varied with respect to single prosodic cues or specific cue combinations.
Results revealed that six- and eight-month-old German-learning infants successfully detect the internal prosodic boundary when it is signaled by all the three main boundary cues pitch change, final lengthening, and pause. For eight-, but not for six-month-olds, the combination of pitch change and final lengthening, without the occurrence of a pause, is sufficient. This mirrors an adult-like perception by eight-months (Holzgrefe-Lang et al., 2016). Six-month-olds detect a prosodic phrase boundary signaled by final lengthening and pause. The findings suggest a developmental change in German prosodic boundary cue perception from a strong reliance on the pause cue at six months to a differentiated sensitivity to the more subtle cues pitch change and final lengthening at eight months. Neither for six- nor for eight-month-olds the occurrence of pitch change or final lengthening as single cues is sufficient, similar to what has been observed for adult speakers of German (Holzgrefe-Lang et al., 2016).
The present dissertation provides new scientific knowledge on infants’ sensitivity to individual prosodic phrase boundary cues in the first year of life. Methodologically, the studies are pathbreaking since they used exactly the same stimulus materials – phonologically thoroughly controlled lists of names – that have also been used with adults (Holzgrefe-Lang et al., 2016) and with infants in a neurophysiological paradigm (Holzgrefe-Lang, Wellmann, Höhle, & Wartenburger, 2018), allowing for comparisons across age (six/ eight months and adults) and method (behavioral vs. neurophysiological methods). Moreover, materials are suited to be transferred to other languages allowing for a crosslinguistic comparison. Taken together with a study with similar French materials (van Ommen et al., 2020) the observed change in sensitivity in German-learning infants can be interpreted as a language-specific one, from an initial language-general processing mechanism that primarily focuses on the presence of pauses to a language-specific processing that takes into account prosodic properties available in the ambient language. The developmental pattern is discussed as an interplay of acoustic salience, prosodic typology (prosodic regularity) and cue reliability.
Die vorgelegte Dissertation befasst sich mit der frühen Wortsegmentierung im monolingualen und bilingualen Spracherwerb. Die Wortsegmentierung stellt eine der wesentlichen Herausforderungen für Säuglinge im Spracherwerb dar, da gesprochene Sprache kontinuierlich ist und Wortgrenzen nicht zuverlässig durch akustische Pausen markiert werden. Zahlreiche Studien konnten für mehrere Sprachen zeigen, dass sich Segmentierungsfähigkeiten von monolingualen Säuglingen zwischen dem 6. und 12. Lebensmonat herausbilden (z. B. Englisch: Jusczyk, Houston & Newsome, 1999; Französisch: Nazzi, Mersad, Sundara, Iakimova & Polka, 2014; Deutsch: Höhle & Weissenborn, 2003; Bartels, Darcy & Höhle, 2009). Frühe Wortsegmentierungsfähigkeiten sind sprachspezifisch (Polka & Sundara, 2012). Crosslinguistische Studien zeigten, dass eine sprachübergreifende Segmentierung für einsprachig aufwachsende Säuglinge nur erfolgreich bewältigt wird, wenn die nicht-native Sprache rhythmische Eigenschaften mit ihrer Muttersprache teilt (Houston, Jusczyk, Kuijpers, Coolen & Cutler, 2000; Höhle, 2002; Polka & Sundara, 2012).
In vier Studien dieser Dissertation wurden mit behavioralen (Headturn Preference Paradigma) und elektrophysiologischen Untersuchungen (Elektroenzephalografie) monolingual Deutsch aufwachsende und bilingual Deutsch-Französisch aufwachsende Säuglinge im Alter von 9 Monaten untersucht. Dabei wurde der Frage nachgegangen, ob monolingual Deutsch aufwachsende Säuglinge im Alter von 9 Monaten in der Lage sind, ihre Muttersprache Deutsch und die rhythmisch unähnliche Sprache Französisch zu segmentieren. Mit anderen Worten: Können monolinguale Säuglinge im Alter von 9 Monaten ihre Segmentierungsprozeduren modifizieren bzw. von ihrer bevorzugten Segmentierung abweichen, um auch nicht-muttersprachlichen Input erfolgreich zu segmentieren?
Bezogen auf die bilingualen Sprachlerner wurde der Frage nachgegangen, ob zweisprachig aufwachsende Säuglinge vergleichbare Segmentierungsfähigkeiten wie monolingual aufwachsende Säuglinge aufweisen und ob sich zudem ein Einfluss der Sprachdominanz auf die Entwicklung der Wortsegmentierungsfähigkeiten in einer bilingualen Population zeigt.
Durch die gewählten Methoden konnten sowohl Verhaltenskorrelate als auch elektrophysiologische Korrelate zur Beantwortung der Fragestellungen herangezogen werden. Darüber hinaus ermöglichte das EEG durch ereigniskorrelierte Potenziale (EKPs) einen Einblick in Lern- und Verarbeitungsprozesse, die mit Verhaltensmethoden nicht erfassbar waren.
Die Ergebnisse zeigen, dass monolingual Deutsch aufwachsende Säuglinge im Alter von 9 Monaten sowohl ihre Muttersprache als auch die nicht-native Sprache Französisch erfolgreich segmentieren. Die Fähigkeit zur Segmentierung der nicht-nativen Sprache Französisch wird jedoch beeinflusst von der Muttersprache: monolinguale Säuglinge, die mit Französisch zuerst getestet wurden, segmentierten sowohl das Französische als auch das im Anschluss präsentierte deutsche Sprachmaterial. Monolinguale Säuglinge die zuerst mit Deutsch und anschließend mit Französisch getestet wurden, segmentierten die deutschen Stimuli, jedoch nicht das französische Sprachmaterial.
Bilingual Deutsch-Französisch aufwachsende Säuglinge segmentieren im Alter von 9 Monaten beide Muttersprachen erfolgreich. Die Ergebnisse deuten zudem auf einen Einfluss der Sprachdominanz auf die Wortsegmentierungsfähigkeiten von zweisprachig aufwachsenden Säuglingen. Die balancierten Bilingualen segmentieren beide Muttersprachen erfolgreich, die unbalancierten Bilingualen zeigen nur für die jeweils dominante Sprache eine erfolgreiche Segmentierung.
Zusammenfassend liefert diese Arbeit erstmals Evidenz für eine erfolgreiche sprachübergreifende Segmentierung in prosodisch differenten Sprachen unterschiedlicher Rhythmusklassen in einer monolingualen Population. Darüber hinaus liefern die Studien dieser Arbeit Evidenz dafür, dass bilingual aufwachsende Säuglinge bezogen auf die Wortsegmentierungsfähigkeiten eine vergleichbare Entwicklung wie einsprachig aufwachsende Sprachlerner zeigen. Dieses Ergebnis erweitert die Datenlage bisheriger Studien, die für verschiedene Entwicklungsschritte im Spracherwerb keine Verzögerung, sondern eine zu monolingual aufwachsenden Säuglingen vergleichbare Entwicklung innerhalb einer bilingualen Population nachweisen konnten (Sprachdiskrimination: Byers-Heinlein, Burns & Werker, 2010; Bosch & Sebastian-Galles, 1997; Phonemdiskrimination: Albareda-Castellot, Pons & Sebastián-Gallés, 2011; Wahrnehmung rhythmischer Eigenschaften: Bijeljac-Babic, Höhle & Nazzi, 2016).
This thesis is concerned with the phenomenon of quantifier scope ambiguities. This phenomenon has been researched extensively, both from a theoretical and from an empirical point of view. Nevertheless, there are still a number of under-researched topics in the field of quantifier scope, which will be the main focus of this thesis. I will take a closer look at three languages, English, German, and the Asante Twi dialect of Akan (Kwa, Niger-Kongo). The goal is a better understanding of the phenomenon of quantifier scope both within each language, as well as from a cross-linguistic perspective. First, this thesis will provide a series of experiments that allow a direct cross-linguistic comparison between English and German – two languages about which specific claims have been made in the literature. I will also provide exploratory research in the case of Asante Twi, where so far, no work has been dedicated specifically to the study of quantifier scope. The work on Asante Twi will go beyond quantifier scope and also target the quantifier and determiner system in general. The question is not only if particular scope readings are possible or not, but also which factors contribute to an increase or decrease of scope availability, and if there are factors that block certain scope readings altogether. While some of the results confirm and thereby strengthen previous claims, other results contradict general assumptions in the literature. This is particularly the case for inverse readings in German and inverse readings across clause-boundaries.
This dissertation focuses on the handling of time in dialogue. Specifically, it investigates how humans bridge time, or “buy time”, when they are expected to convey information that is not yet available to them (e.g. a travel agent searching for a flight in a long list while the customer is on the line, waiting). It also explores the feasibility of modeling such time-bridging behavior in spoken dialogue systems, and it examines
how endowing such systems with more human-like time-bridging capabilities may affect humans’ perception of them.
The relevance of time-bridging in human-human dialogue seems to stem largely from a need to avoid lengthy pauses, as these may cause both confusion and discomfort among the participants of a conversation (Levinson, 1983; Lundholm Fors, 2015). However, this avoidance of prolonged silence is at odds with the incremental nature of speech production in dialogue (Schlangen and Skantze, 2011): Speakers often start to verbalize their contribution before it is fully formulated, and sometimes even before they possess the information they need to provide, which may result in them running out of content mid-turn.
In this work, we elicit conversational data from humans, to learn how they avoid being silent while they search for information to convey to their interlocutor. We identify commonalities in the types of resources employed by different speakers, and we propose a classification scheme. We explore ways of modeling human time-buying behavior computationally, and we evaluate the effect on human listeners of embedding this behavior in a spoken dialogue system.
Our results suggest that a system using conversational speech to bridge time while searching for information to convey (as humans do) can provide a better experience in several respects than one which remains silent for a long period of time. However, not all speech serves this purpose equally: Our experiments also show that a system whose time-buying behavior is more varied (i.e. which exploits several categories from the classification scheme we developed and samples them based on information from human data) can prevent overestimation of waiting time when compared, for example, with a system that repeatedly asks the interlocutor to wait (even if these requests for waiting are phrased differently each time). Finally, this research shows that it is possible to model human time-buying behavior on a relatively small corpus, and that a system using such a model can be preferred by participants over one employing a simpler strategy, such as randomly choosing utterances to produce during the wait —even when the utterances used by both strategies are the same.
Individuals with aphasia vary in the speed and accuracy they perform sentence comprehension tasks. Previous results indicate that the performance patterns of individuals with aphasia vary between tasks (e.g., Caplan, DeDe, & Michaud, 2006; Caplan, Michaud, & Hufford, 2013a). Similarly, it has been found that the comprehension performance of individuals with aphasia varies between homogeneous test sentences within and between sessions (e.g., McNeil, Hageman, & Matthews, 2005). These studies ascribed the variability in the performance of individuals with aphasia to random noise. This conclusion would be in line with an influential theory on sentence comprehension in aphasia, the resource reduction hypothesis (Caplan, 2012). However, previous studies did not directly compare variability in language-impaired and language-unimpaired adults. Thus, it is still unclear how the variability in sentence comprehension differs between individuals with and without aphasia. Furthermore, the previous studies were exclusively carried out in English. Therefore, the findings on variability in sentence processing in English still need to be replicated in a different language.
This dissertation aims to give a systematic overview of the patterns of variability in sentence comprehension performance in aphasia in German and, based on this overview, to put the resource reduction hypothesis to the test. In order to reach the first aim, variability was considered on three different dimensions (persons, measures, and occasions) following the classification by Hultsch, Strauss, Hunter, and MacDonald (2011). At the dimension of persons, the thesis compared the performance of individuals with aphasia and language-unimpaired adults. At the dimension of measures, this work explored the performance across different sentence comprehension tasks (object manipulation, sentence-picture matching). Finally, at the dimension of occasions, this work compared the performance in each task between two test sessions. Several methods were combined to study variability to gain a large and diverse database. In addition to the offline comprehension tasks, the self-paced-listening paradigm and the visual world eye-tracking paradigm were used in this work.
The findings are in line with the previous results. As in the previous studies, variability in sentence comprehension in individuals with aphasia emerged between test sessions and between tasks. Additionally, it was possible to characterize the variability further using hierarchical Bayesian models. For individuals with aphasia, it was shown that both between-task and between-session variability are unsystematic. In contrast to that, language-unimpaired individuals exhibited systematic differences between measures and between sessions. However, these systematic differences occurred only in the offline tasks. Hence, variability in sentence comprehension differed between language-impaired and language-unimpaired adults, and this difference could be narrowed down to the offline measures.
Based on this overview of the patterns of variability, the resource reduction hypothesis was evaluated. According to the hypothesis, the variability in the performance of individuals with aphasia can be ascribed to random fluctuations in the resources available for sentence processing. Given that the performance of the individuals with aphasia varied unsystematically, the results support the resource reduction hypothesis. Furthermore, the thesis proposes that the differences in variability between language-impaired and language-unimpaired adults can also be explained by the resource reduction hypothesis. More specifically, it is suggested that the systematic changes in the performance of language-unimpaired adults are due to decreasing fluctuations in available processing resources. In parallel, the unsystematic variability in the performance of individuals with aphasia could be due to constant fluctuations in available processing resources. In conclusion, the systematic investigation of variability contributes to a better understanding of language processing in aphasia and thus enriches aphasia research.
This thesis explores the variation in coreference patterns across language modes (i.e., spoken and written) and text genres. The significance of research on variation in language use has been emphasized in a number of linguistic studies. For instance, Biber and Conrad [2009] state that “register/genre variation is a fundamental aspect of human language” and “Given the ubiquity of register/genre variation, an understanding of how linguistic features are used in patterned ways across text varieties is of central importance for both the description of particular languages and the development of cross-linguistic theories of language use.”[p.23]
We examine the variation across genres with the primary goal of contributing to the body of knowledge on the description of language use in English. On the computational side, we believe that incorporating linguistic knowledge into learning-based systems can boost the performance of automatic natural language processing systems, particularly for non-standard texts. Therefore, in addition to their descriptive value, the linguistic findings we provide in this study may prove to be helpful for improving the performance of automatic coreference resolution, which is essential for a good text understanding and beneficial for several downstream NLP applications, including machine translation and text summarization.
In particular, we study a genre of texts that is formed of conversational interactions on the well-known social media platform Twitter. Two factors motivate us: First, Twitter conversations are realized in written form but resemble spoken communication [Scheffler, 2017], and therefore they form an atypical genre for the written mode. Second, while Twitter texts are a complicated genre for automatic coreference resolution, due to their widespread use in the digital sphere, at the same time they are highly relevant for applications that seek to extract information or sentiments from users’ messages. Thus, we are interested in discovering more about the linguistic and computational aspects of coreference in Twitter conversations. We first created a corpus of such conversations for this purpose and annotated it for coreference. We are interested in not only the coreference patterns but the overall discourse behavior of Twitter conversations. To address this, in addition to the coreference relations, we also annotated the coherence relations on the corpus we compiled. The corpus is available online in a newly developed form that allows for separating the tweets from their annotations.
This study consists of three empirical analyses where we independently apply corpus-based, psycholinguistic and computational approaches for the investigation of variation in coreference patterns in a complementary manner. (1) We first make a descriptive analysis of variation across genres through a corpus-based study. We investigate the linguistic aspects of nominal coreference in Twitter conversations and we determine how this genre relates to other text genres in spoken and written modes. In addition to the variation across genres, studying the differences in spoken-written modes is also in focus of linguistic research since from Woolbert [1922]. (2) In order to investigate whether the language mode alone has any effect on coreference patterns, we carry out a crowdsourced experiment and analyze the patterns in the same genre for both spoken and written modes. (3) Finally, we explore the potentials of domain adaptation of automatic coreference resolution (ACR) for the conversational Twitter data. In order to answer the question of how the genre of Twitter conversations relates to other genres in spoken and written modes with respect to coreference patterns, we employ a state-of-the-art neural ACR model [Lee et al., 2018] to examine whether ACR on Twitter conversations will benefit from mode-based separation in out-of-domain training data.