京都大学 / 新制・課程博士 / 博士(人間・環境学) / 甲第24685号 / 人博第1058号 / 新制||人||248(附属図書館) / 2022||人博||1058(吉田南総合図書館) / 京都大学大学院人間・環境学研究科共生人間学専攻 / (主査)教授 藤田 耕司, 教授 谷口 一美, 准教授 守田 貴弘, 教授 田中 伸一 / 学位規則第4条第1項該当 / Doctor of Human and Environmental Studies / Kyoto University / DFAM

This dissertation is an empirical and theoretical study of sentence-level prosody in Conamara (Connemara) Irish. It addresses the architecture of the syntax-phonology interface and the relation between syntactic constituent structure and prosodic structure formation. It argues for a fully interactional view of the interface, in which the phonological form may be influenced by a number of competing factors, including constraints governing syntax-prosody correspondence, linearization, and prosodic well-formedness. The specific proposal is set within the framework of Match Theory (Selkirk 2009, 2011), an indirect-reference theory of the syntax-prosody interface in which correspondence between syntactic and prosodic constituents is governed by a family of violable Match constraints. These constraints call for a one-to-one correspondence between syntactic and prosodic structure, to the extent that prosodic structure may be recursive under pressure from the recursive nature of syntactic phrases. However, this direct correspondence can be overruled by other interacting constraints, including prosodic markedness constraints and, as proposed here, other correspondence relations, as on the linearization of hierarchical syntactic structures. This dissertation argues that the distribution of pitch accents in Conamara Irish provides direct evidence for Match Theory. It is proposed that two phrasal pitch accents, L-H and H-L, demarcate the edges of phonological phrases, where L-H accents specifically target only those phrases which are recursive. Using the distribution of these pitch accents as indicators for the presence of prosodic boundaries, the dissertation investigates a variety of syntactic structures in both the clausal and nominal domain. It is argued that there is a close correspondence between syntactic and prosodic structure in default cases, but that this direct correspondence may be subverted in favour of a structure which better satisfies higher-ranked prosodic markedness constraints. Finally, this dissertation addresses pronoun postposing, a process pervasive in Irish dialects in word order appears to be sensitive to prosodic structure. This dissertation proposes to account for this phenomenon using the theoretical framework developed in the dissertation, in which the main patterns are accounted for through the interaction of Match constraints, prosodic markedness constraints, and a proposed violable constraint on the linearization of syntactic structure.

This thesis examines the effects of prosody on serial recall. Serial recall is an experimental task commonly used to evaluate the capacity of short-term memory. The Working Memory model by Baddeley and Hitch is a theoretical framework that describes the inner operation of short-term memory. Its hierarchies are supported by empirical evidence, but details of the core mechanisms remain unclear. In an attempt to refine the framework, this thesis investigated prosody as a factor in serial recall accuracy. Two behavioural experiments were conducted on native speakers of Canadian English. In the first experiment, the explicit awareness of word stress was examined. Results showed a main effect of word stress type, where iambic words received higher stress identification scores compared to trochaic words. In the second experiment, an immediate serial recall task was used to examine serial recall of word lists. The lists consisted of disyllabic words from Canadian English sources. The lists had mixed or uniform stress patterns. A main effect of list stress patterns was found, where mixed lists elicited better recall of the order of list items compared to uniform lists. Overall, the present thesis offers a new interpretation on how word stress is represented in the short-term memory. It adds support to the proposed interaction between short-term and long-term memory. / Thesis / Master of Science (MSc) / Many daily experiences require us to remember a list of items, such as drafting a grocery list and reciting phone numbers. How well we can remember a list can be influenced by various characteristics of the list. This thesis looks at a seldom studied characteristic: prosody. Prosody in the English language can be realized via word stress, which is the amount of emphasis we place on certain parts of a word. This thesis explores the impact of word stress on our ability to remember a list of common English words. It reports data from an online survey and a laboratory experiment. Results show that mixed word stress patterns in a list lead to better memory for said list. Overall, this thesis offers new suggestions on the role of prosody in memory. As well, it offers a novel set of data that supports current theories in memory.

No description available.

No description available.

No description available.

Timing, and ergo rhythm, are intrinsic features of language that help facilitate real-time speech comprehension. However, work exploring how variable timing is processed in speech is limited. This dissertation addresses this gap in literature by exploring the tenets of how temporal variability is cognitively processed, particularly in the context of real-time stimulus processing. This research is one of the first works to examine temporal variations in linguistic and other acoustically complex contexts. Using electroencephalography (EEG) and behavioural methods, participants were tested on their perception of temporal variations within a continuous stream of either simple tones, complex waves, or syllables. Two timing deviants were presented that occurred early or late compared to other stimuli in the sequence. Event-related potentials (ERPs) were recorded for each stimulus type across three experiments. A fourth experiment tested participant recognition memory for syllable order. Results showed differential processing between the two timing deviants. Unexpectedly earlier tokens elicited larger pre-attentive responses compared to late, suggesting a saliency for the earlier tokens that was not present for the delayed ones. This pattern was observed across all three levels of acoustic and linguistic complexity. Compared to sequences with no timing deviants or an early timing deviant, unexpectedly late tokens were more detrimental to memory, suggesting a negative impact of delays on verbal recognition. Thus, not only were early and late timing variations processed differently, but delays in continuous sequences were also more cognitively taxing for working memory. The results reported in this dissertation contribute to existing knowledge by enriching our understanding of the fundamentals of how aspects of prosodic timing may affect attention and memory. Additionally, it provides new insights into how speech synthesis can be used in neurolinguistic research by tracking how neurophysiological responses change with increasing acoustic complexity and linguistic familiarity. / Dissertation / Doctor of Philosophy (PhD) / This thesis examines how unexpectedly shorter or longer pauses in speech may affect speech comprehension. Specifically, the research reported here examined how stimuli that are presented unexpectedly early or unexpectedly late in a steady-rate sequence impact a listener’s attention and memory. Although the speed at which we speak has been studied previously, work on unexpected changes in its timing has been limited. This research begins to explore this aspect of speech processing. It contributes to our understanding of how speech timing is processed in two important ways. First, we found that unexpected delays in both non-linguistic sounds and syllables were less noticeable than unexpectedly early presentations, when participants were not required to pay attention to them. Second, we found that unexpected delays made recognition memory for stimulus order worse. Overall, the results of these studies indicate that unexpected delays in the rhythm of speech make it more difficult to understand.

<p> Emotion perception in speech has been shown to facilitate greater understanding and learning in adults as well as children. There is an atypical RH specialization for speech perception that exists in individuals with Down Syndrome (DS). Individuals with DS have a typical left hemisphere specialization for motor control and speech production, similar to those individuals from the general population which may cause a functional dissociation between speech perception and motor control for these individuals (Heath et al., 2000). What remains unknown is how this atypical lateralization may influence speech perception when emotional intonation is included with verbal stimuli. Using a free recall dichotic listening paradigm, it was found that individuals with DS process verbal stimuli similarly to mental-age matched peers (individuals with a developmental delay, and individuals for the general population.) To investigate this further, a directed attention paradigm was employed. Participants listened to a particular ear for either a particular word or emotion. It was found that individuals from the general population were more accurate than individuals from either of the other two groups for perception of the word. Also, an effect for Ear was found with the right ear being significantly more accurately perceived than the left for all three groups. When emotion was attended to specifically, the left ear was more accurately perceived than the right. These results are somewhat consistent with previous findings (e.g., Bulman-Fleming & Bryden, 1994) for participants from the general population however the expected lateralization in DS group was not evident. This lack of atypical RHA in individuals with DS may be related to the task itself.</p> / Thesis / Master of Science (MSc)

The speech commonly addressed to infants (infant-directed speech or IDS) is believed to have multiple functions, including communication of emotion and highlighting linguistic aspects of speech. However, these two functions are most often studied separately so that the influence of emotional prosody (the changes in intonation and vocal quality that relate to emotion) on linguistic processing in infants has rarely been addressed. Given that language learning during infancy occurs in the context of natural infant-caretaker exchanges that most certainly include emotion communication and co-regulation, it is important to integrate the concepts of emotional communication and linguistic communication in studying language learning. This study examined the influence of both positive (happy) and negative (sad) face+voice contexts on word recognition in 11-to-13-month-old infants. It was hypothesized that infants would learn and subsequently recognize words when they were delivered in a happy context, but will experience more difficulty in learning and/or recognition of the same words when delivered in a sad context. The general pattern of results confirmed these predictions in that after habituating to sentences containing a specific target nonsense word, infants in the Happy Condition recovered their attention to the same sentences with a novel target word. In contrast, infants in the Sad Condition showed no significant recovery to a change in target words. These results contribute to our understanding of how emotional tone can facilitate and/or attenuate attention in older infants as they engage in language learning with their caretakers. / Ph. D.

Statistical parametric speech synthesis (SPSS) has seen improvements over recent years, especially in terms of intelligibility. Synthetic speech is often clear and understandable, but it can also be bland and monotonous. Proper generation of natural speech prosody is still a largely unsolved problem. This is relevant especially in the context of expressive audiobook speech synthesis, where speech is expected to be fluid and captivating. In general, prosody can be seen as a layer that is superimposed on the segmental (phone) sequence. Listeners can perceive the same melody or rhythm in different utterances, and the same segmental sequence can be uttered with a different prosodic layer to convey a different message. For this reason, prosody is commonly accepted to be inherently suprasegmental. It is governed by longer units within the utterance (e.g. syllables, words, phrases) and beyond the utterance (e.g. discourse). However, common techniques for the modeling of speech prosody - and speech in general - operate mainly on very short intervals, either at the state or frame level, in both hidden Markov model (HMM) and deep neural network (DNN) based speech synthesis. This thesis presents contributions supporting the claim that stronger representations of suprasegmental variation are essential for the natural generation of fundamental frequency for statistical parametric speech synthesis. We conceptualize the problem by dividing it into three sub-problems: (1) representations of acoustic signals, (2) representations of linguistic contexts, and (3) the mapping of one representation to another. The contributions of this thesis provide novel methods and insights relating to these three sub-problems. In terms of sub-problem 1, we propose a multi-level representation of f0 using the continuous wavelet transform and the discrete cosine transform, as well as a wavelet-based decomposition strategy that is linguistically and perceptually motivated. In terms of sub-problem 2, we investigate additional linguistic features such as text-derived word embeddings and syllable bag-of-phones and we propose a novel method for learning word vector representations based on acoustic counts. Finally, considering sub-problem 3, insights are given regarding hierarchical models such as parallel and cascaded deep neural networks.