Meaningful semantic context has been shown to improve comprehension of spoken sentences by young and old adults, especially in difficult listening conditions. Whether older adults benefit differently than younger adults is a topic of some controversy. I asked young (14 participants, 18-25) and older adults (20 participants, 60-75) to report entire sentences which contained either a coherent or anomalous global semantic context (e.g. coherent: “Her new skirt was made of denim”, anomalous: “Her good slope was done in carrot”). Sentences were mixed with signal-correlated noise, at 10 signal-to-noise ratios (-6 to +2 dB and clear speech). Percentage scores were converted to rationalized arcsine units and subjected to a repeated-measures ANOVA; slopes from psychometric functions fitted to the transformed data were also analyzed. Cognitive and hearing threshold differences were considered as factors influencing results. Finally, individual variability in the use of context was explored. Comprehension by both groups benefited from meaningful context, without a clear difference in the overall amount of benefit obtained. Cognitive factors did not appear to influence the results, although differences in hearing thresholds likely contributed to the consistent performance decrement for older adults. Individuals varied greatly in their use of context, a possible explanation for inconsistent results in studies comparing context use by young and older people. fMRI was then used to look at neural activity associated with deriving benefit from meaningful context. Whole-brain EPI data were acquired from young (16 participants, 19-26) adults using a sparse imaging design. Participants heard coherent and anomalous sentences in the scanner, and were asked to report what they heard on half of the trials. Individual’s word-report data obtained in the scanner were used to model intelligibility in the analysis and results were compared to an analysis conducted using intelligibility estimates based on group data from another study. In addition to bilateral temporal activity associated with increasing intelligibility, I observe a large left inferior-frontal region in which BOLD signal correlated more strongly with highly intelligible anomalous compared to highly intelligible coherent prose, presumably reflecting challenged semantic integration and supporting Hagoort’s (2005) model of semantic unification. / Thesis (Master, Neuroscience Studies) -- Queen's University, 2008-09-18 23:01:40.979

Purpose: To investigate the effects of speaker age, speaker gender, semantic context, signal-to-noise ratio (SNR) and a listener’s hearing status on speech recognition and listening effort in older adults. We examined the hypothesis that older adults would recognize less speech and exert greater listening effort when listening to the speech of younger versus older adult speakers. Method: Speech stimuli were recorded from 12 adult speakers classified as “younger” (three males and three females aged 18-31 years) and “older” (three males and three females aged 69-89) respectively. A computer-based subjective rating was conducted to confirm that the speakers were representative of younger and older speakers. Listeners included 20 older adults (aged 65 years and above), who were divided into two age-matched groups with and without hearing loss. All listening and speaking participants in the study were native speakers of New Zealand English. A dual-task paradigm was used to measure speech recognition and listening effort; the primary task involved recognition of target words in sentences containing either high or low contextual cues, while the secondary task required listeners to memorise the target words for later recall, following a set number of sentences. Listening tasks were performed with a variety of listening conditions (quiet, +5 dB SNR and 0dB SNR). Results: There were no overall differences in speech recognition scores or word recall scores for the 20 older listeners, when listening to the speech of the younger versus older speakers. However, differential effects of speaker group were observed in the two semantic context conditions (high versus low context). Older male speakers were the easiest to understand when semantic context was low; however, for sentences with high semantic context, the older male group were the most difficult to understand. Word recall scores were also significantly higher in the most challenging listening condition (low semantic context, 0 dB SNR), when the speaker was an older male. Conclusion: Differential effects of speaker group were observed in the two semantic context conditions (high versus low context) suggesting that different speech cues were used by listeners, as the level of context varied. The findings provide further evidence that, in challenging listening conditions, older listeners are able to use a wide range of cues, such as prosodic features and semantic context to compensate for a degraded signal. The availability of these cues depends on characteristics of the speaker, such as rate of speech and prosody, as well as characteristics of the listener and the listening environment. .

Der Sprechakt teilt sich in zwei Vorgänge: Zunächst muss das richtige Wort aus dem mentalen Lexikon abgerufen werden und anschließend wird es in der Kommunikation verwendet. Zur Erforschung des ersteren Vorgangs werden oft Ein-Personen-Studien verwendet, in denen durch Beobachten der Reaktion auf Stimuli (z. B. Bilder) die Mikrostruktur des lexikalisch- semantischen Systems beleuchtet wird. Für die Anwendung von Sprache in der Kommunikation hingegen nutzt man Partnerexperimente, um die Koordination zwischen den Gesprächspartnern zu beobachten und zu ergründen, wie sich gegenseitiges Verstehen und biografisches Wissen darauf auswirken. Wenig erforscht ist aber, wie ein von einem Gesprächspartner eingebrachter Bedeutungskontext die traditionell in Ein-Personen-Studien untersuchten lexikalisch-semantischen Effekte beeinflusst. Im Rahmen meiner Dissertation möchte ich die Lücke zwischen den beiden Forschungsansätzen schließen, indem ich einen kommunikativen Kontext in etablierte Paradigmen der Bildbenennung integriere. Hierzu betrachte ich zunächst klassische semantische Kontexteffekte, die durch nähere oder entferntere kategorische Relationen zwischen Begriffen hervorgerufen werden (Studie 1), um anschließend lose thematische Beziehungen zu untersuchen, die mit alltäglichen Ereignissen verbunden sind (Studie 2 & Studie 3). Um die hochgradig verflochtenen Ebenen der lexikalischen und semantischen Verarbeitung voneinander zu trennen, habe ich ereigniskorrelierte Hirnpotentiale (ERPs) eingesetzt, um die elektrophysiologischen Signaturen des konzeptuellen Primings und der lexikalischen Auswahl zu verfolgen. Die vorliegende Arbeit liefert sowohl theoretische als auch praktische Beiträge. Erstens stützen unsere Ergebnisse die theoretischen Annahmen, dass sich semantisches Priming und lexikalische Interferenz vorübergehend überschneiden und gemeinsam das Benennungsverhalten in einem Trade-off beeinflussen. Auch die Gegenwart eines Kommunikationspartners kann Auswirkungen auf dieses Zusammenspiel haben. Zweitens ergänzen diese Ergebnisse die aktuelle Literatur zu verschiedenen Arten von semantischen Beziehungen, wie z. B. Nulleffekte für entfernte Beziehungen und Kontexteffekte, die systematisch mit der Stärke der Verwandtschaft zunehmen. Und schließlich bietet unser neuartiges Design eines kommunikativen Kontextes ein praktisches Instrument, um die Lücke zwischen Ein-Personen-Studien und Kommunikationsstudien zu schließen. Alles in allem tragen diese Ergebnisse zu einem besseren Verständnis der neuronalen Mechanismen unseres Sprachproduktionssystems bei, das in der Lage ist, sich flexibel sowohl an sprachliche als auch an soziale Kontexte anzupassen. / Speaking could be divided into two processes: first, the correct word must be retrieved from the mental lexicon, and then it is used in communication. To study the former process, single-person studies are often used, in which the microstructure of the lexical-semantic system is illuminated by observing reaction times to name stimuli (e.g., pictures). For the language use in communication, on the other hand, partner experiments are used to observe coordination between interlocutors and to explore how mutual understanding and biographical knowledge affect it. However, how a meaningful context brought by an interlocutor influences the established lexical-semantic effects from single-person studies remains underexplored. Within the scope of my dissertation, I aim to bridge the gap between these two research approaches by integrating a communicative context into well-established picture naming paradigms. To this end, I first investigate classic semantic context effects induced by close or distant categorical relations (Study 1), and then examine loose thematic relations associated with everyday events (Study 2 & Study 3). To separate the highly intertwined strata of lexical and semantic processing, I used event-related brain potentials (ERPs) to track the electrophysiological signatures of conceptual priming and lexical selection. The present work makes both theoretical and practical contributions. First, our results support the theoretical assumptions that semantic priming and lexical interference temporarily overlap, and jointly modulate naming behavior in a trade-off. Such interplay may be greatly influenced by the presence of a communicating partner. Second, these findings add to the current literature on different types of semantic relations, such as null effects for distant relations and context effects that systematically increase with the strength of relatedness. Finally, our novel design of a communicative context provides a practical tool to bridge the gap between single-person studies and communication studies. All in all, these findings advance our understanding of the neural mechanisms of our speech production system, which is capable of flexibly adapting to both linguistic and social contexts.

Research on semantic processing focused mainly on isolated units in language, which does not reflect the complexity of language. In order to understand how semantic information is processed in a wider context, the first goal of this thesis was to determine whether Swedish pre-school children are able to comprehend semantic context and if that context is semantically built up over time. The second goal was to investigate how the brain distributes attentional resources by means of brain activation amplitude and processing type. Swedish preschool children were tested in a dichotic listening task with longer children’s narratives. The development of event-related potential N400 component and its amplitude were used to investigate both goals. The decrease of the N400 in the attended and unattended channel indicated semantic comprehension and that semantic context was built up over time. The attended stimulus received more resources, processed the stimuli in more of a top-down manner and displayed prominent N400 amplitude in contrast to the unattended stimulus. The N400 and the late positivity were more complex than expected since endings of utterances longer than nine words were not accounted for. More research on wider linguistic context is needed in order to understand how the human brain comprehends natural language. / Tidigare forskning på semantisk processning har fokuserad på isolerade språkliga enheter vilket inte reflekterar språkets komplexitet. För att kunna förstår hur semantisk information processeras i en större kontext, var studiens första syfte att undersöka om svenska förskolebarn om svenska förskolebarn förmår att förstå semantisk kontext och om denna kontext byggs upp över tid. Det andra syftet var att undersöka hur hjärnan fördelar uppmärksamhetsbaserade resurser i avseende på hjärnaktiveringsamplitud och olika processesningstyper. För detta testades svenska förskolebarn i ett dikotiskt lyssningstest med olika barnsagor. Utvecklingen av N400-komponenten, en händelse-relaterad potential, användes för detta. Nedgången av N400 komponenten och den sena positiviteten visades i både de uppmärksammade och ouppmärksammade kanalerna och detta indikerar semantisk förståelse och att semantisk kontext byggdes upp över tid. Därutöver kunde en större N400-amplitud observeras i den uppmärksammade kanalen, vilket indikerar att den fick mer hjärnresurser och använde sig av top-down-bearbetning i större utsträckning än bottom-up-processer. N400-komponenten och sena positiviteten visade sig vara mer komplex än förväntat. Det kan bero för att de sista orden i ett yttrande som var längre än nio ord exkluderades från analysen. Det finns ett behov av forskning som använder sig av längre lingvistiska kontexter och deras effekter i människohjärnan.