Function words are known to be shorter than content words. I investigate the function words BE and HAVE (with its content word homonym) and show that more reduction, operationalized as word shortening or contraction, is found in some grammaticalized meanings of these words. The difference between the words’ uses cannot be attributed to differences in frequency or semantic weight. Instead I argue that these words are often shortened and reduced when they occur in constructions in which they are highly predictable. This suggests that particular grammaticalized uses of a word are stored with their own exemplar clouds of context-specific phonetic realizations. The phonetics of any instance of a word are then jointly determined by the exemplar cloud for that word and the particular context. A given instance of an auxiliary can be reduced either because it is predictable in the current context or because that use of the auxiliary usually occurs in predictable contexts. The effects cannot be attributed to frequency or semantic weight. The present study compares function word production in the speech of school-aged children and their caregivers and in inter-adult speech. The effects of predictability in context and average predictability across contexts are replicated across the datasets. However, I find that as children get older their function words shorten relative to content words, even when controlling for increasing speech rate, showing that as their language experience increases they spend less time where it is not needed for comprehensibility. Caregivers spend less time on function words with older children than younger children, suggesting that they expect function words to be more difficult for younger interlocutors to decode than for older interlocutors. Additionally, while adults use either word shortening or contraction to increase the efficiency of speech, children tend to either use contraction and word shortening or neither until age seven, where they start to use one strategy or the other like adults. Young children with better vocabulary employ an adult-like strategy earlier, suggesting earlier onset of efficient yet effective speech behavior, namely allocating less signal to function words when they are especially easy for the listener to decode.

This thesis examined the factors of L1, input frequency and emergent productivity in child L2 acquisition. This thesis is the first study to look at the interplay of L1 and usage factors in children learning a L2. The focus of the thesis was an investigation of these factors in the acquisition of article and auxiliary systems of English, which have been proven to be problematic areas for both L1 and L2 learners. While accounts of L1 transfer in L2 are better developed in generative theory, the roles of input frequency and emergent productivity are better developed in constructivist theory. The thesis assessed these two approaches against the data from L2 children from various L1 backgrounds. The children’s accuracy and error patterns with articles and auxiliaries were investigated. The main findings were as follows. L1 typology facilitated the acquisition of the structure of the NP and VP, but it only extended as far as the awareness of the presence of the functional morpheme (article or auxiliary). L1 transfer effects were observed only in the first 1.5 years of acquisition, which could be due to the unstable L1 knowledge in child L2 learners. The use of articles and auxiliaries was also influenced by their input frequencies and distribution, as more frequent forms were supplied more accurately and were substituted for less frequent forms. Different forms of articles and auxiliaries emerged separately and followed different paths of development. It was argued that they were acquired piecemeal and that productivity with these forms emerged gradually. It was concluded that constructionist theories were better supported by the data, since the findings on input frequency and productivity were not compatible with the generative approach, and L1 transfer was incorporated into the constructionist approach to account for the findings. It was argued that by the onset of acquisition, child L2 learners had established constructions in their L1 that were abstract enough to be transferred to L2 and did not rely on lexically specific information. As all children learned specific morphological forms of L2 piecemeal, in doing so they demonstrated input effects that held across all L1 backgrounds.

This study investigates the nature of semantic compatibility between constructions and lexical items that occur in them in relation with language use, and the related concept, coercion, based on a usage-based approach to language, in which linguistic knowledge (grammar) is grounded in language use. This study shows that semantic compatibility between linguistic elements is a gradient phenomenon, and that speakers’ knowledge about the degree of semantic compatibility is intimately correlated with language use. To show this, I investigate two constructions of English: the sentential complement construction and the ditransitive construction. I observe speakers’ knowledge of the semantic compatibility between the constructions and lexical items and compared it with empirical data obtained from linguistic corpora and experiments on sentence processing and acceptability judgments. My findings specifically show that the relative semantic compatibility of the lexical items and the construction is significantly correlated with the frequency of use of their co-occurrences and the processing effort and speakers’ acceptability judgments for the co-occurrences. The empirical data show that a lexical item and a construction which are less than fully compatible can be actually used together when the incompatibility is resolved. The resolution of the semantic incompatibility between the lexical item and its host construction has been called coercion. Coercion has been invoked as a theoretical concept without being examined in depth, particularly without regard to language use. By correlating degree of semantic compatibility with empirical data of language use, this study highlights that coercion is an actual psychological process which occurs during the composition of linguistic elements. Moreover, by examining in detail how the semantics of a lexical item and a construction interact in order to reconcile the incompatibility, this study reveals that coercion is semantic integration that involves not only dynamic interaction of linguistic components but also non-linguistic contexts. Investigating semantic compatibility and coercion in detail with empirical data tells about the processes by which speakers compose linguistic elements into larger units. It also supports the assumption of the usage-based model that grammar and usage are not independent, and ultimately sheds light on the dynamic aspect of our linguistic system.

This study investigates the nature of semantic compatibility between constructions and lexical items that occur in them in relation with language use, and the related concept, coercion, based on a usage-based approach to language, in which linguistic knowledge (grammar) is grounded in language use. This study shows that semantic compatibility between linguistic elements is a gradient phenomenon, and that speakers’ knowledge about the degree of semantic compatibility is intimately correlated with language use. To show this, I investigate two constructions of English: the sentential complement construction and the ditransitive construction. I observe speakers’ knowledge of the semantic compatibility between the constructions and lexical items and compared it with empirical data obtained from linguistic corpora and experiments on sentence processing and acceptability judgments. My findings specifically show that the relative semantic compatibility of the lexical items and the construction is significantly correlated with the frequency of use of their co-occurrences and the processing effort and speakers’ acceptability judgments for the co-occurrences. The empirical data show that a lexical item and a construction which are less than fully compatible can be actually used together when the incompatibility is resolved. The resolution of the semantic incompatibility between the lexical item and its host construction has been called coercion. Coercion has been invoked as a theoretical concept without being examined in depth, particularly without regard to language use. By correlating degree of semantic compatibility with empirical data of language use, this study highlights that coercion is an actual psychological process which occurs during the composition of linguistic elements. Moreover, by examining in detail how the semantics of a lexical item and a construction interact in order to reconcile the incompatibility, this study reveals that coercion is semantic integration that involves not only dynamic interaction of linguistic components but also non-linguistic contexts. Investigating semantic compatibility and coercion in detail with empirical data tells about the processes by which speakers compose linguistic elements into larger units. It also supports the assumption of the usage-based model that grammar and usage are not independent, and ultimately sheds light on the dynamic aspect of our linguistic system.

No description available.

Användarbaserade försäkringar är förhållandevis nya produkter på den svenska försäkringsmarknaden. Till skillnad från ”traditionella” försäkringar bygger användarbaserade försäkringar i större utsträckning på personlig data om försäkringstagaren vilket ger förutsättningar för mer individanpassade premier och villkor. Med hjälp av ”smarta” produktersom exempelvis smartphones i kombination med verktyg som automatiskt analyserar datan har det blivit möjligt att frångå riskbedömningar som grundar sig på generell statistik och istället fokusera på den individuella risken. Tekniken kan användas i hela försäkringskedjan men i denna uppsats är det riskurvalet och premiesättningen som är i fokus. Trots ett brett användningsområde tillämpas användarbaserade försäkringar främst inom tre områden: hem, fordon och hälsa. Fördelarna med dessa produkter är bland annat att försäkringstagaren själv kan påverka sin premie genom att justera sitt risktagande. Det kan i sin tur ge ringar på vattnet och ge positiva effekter även i andra delar av samhället. Användarbaserad fordonsförsäkring utgör ett bra exempel på detta. Genom att uppmuntra en säkrare körstil minskar risken för olyckor även för andra trafikanter som inte valt denna försäkringslösning. Färre olyckor innebär även en lägre belastning på sjukvården, att färre personer sjukskrivs till följd av trafikolyckor minskar sannolikt också företagens kostnader.  Att behöva dela personlig data med försäkringsgivaren kan dock ses som ett intrång i den egna integriteteten. Det finns viss risk att användarbaserade försäkringar leder till att försäkrings-barheten minskar i samhället och att vissa grupper riskerar att stängas ute från den så viktiga försäkringsmarknaden. Ett annat problem är att försäkringstagarnas data analyseras automatiskt med hjälp av artificiell intelligens (AI). Det kan därför bli svårt att avgöra om ett negativt beslut för en person beror på att hänsyn tagits till otillåtna faktorer som exempelvis etnicitet eller om risken faktiskt varit högre. Det gör det också svårare för konsumenter att jämföra olika försäkringsprodukter och välja den som erbjuder det bästa priset.  Min slutsats är att det inte finns någon lagstiftning som förbjuder användarbaserade försäkringar i dagsläget. Dock inskränker viss lagstiftning försäkringsbolagens möjligheter att fritt utforma dessa försäkringsprodukter. Min bedömning är att det inte föreligger någon akut risk för att försäkringsbarheten skulle minska på grund av användarbaserade försäkringar eller att vissa grupper missgynnas, men det är ett tänkbart scenario i framtiden – vilket innebär att ny lagstiftning på området kan bli nödvändig. Tekniken kan också användas i andra delar av försäkringskedjan och därav ge ytterligare skäl till att överväga ny lagstiftning.

L’objectif général de la présente thèse est d’évaluer dans quelle mesure la grammaire des verbes, aussi appelée complémentation verbale, peut être basée sur l’usage de la langue. La linguistique de l’usage (usage-based approach) constitue un récent changement de paradigme dans les sciences du langage, qui défend l’idée que la grammaire est un inventaire dynamique d’unités symboliques qui émergent et sont constamment redéfinies par l’usage de la langue. En adoptant une approche constructionnelle de la complémentation verbale et sur la base de données de l’anglais, nous traitons la question de la relation entre la grammaire des verbes et leur usage à trois niveaux d’organisation.Au niveau des verbes, nous comparons des résultats expérimentaux à des données de corpus, et trouvons que les plus fréquentes valences d’un verbe sont traitées plus facilement, ce qui montre que la valence est basée sur l’usage. Au niveau des constructions, nous montrons que, dans le cas de la construction conative, il est possible de formuler le sens d’une construction sur la base du sens des verbes en se plaçant au niveau de classes sémantiques, mais moins facilement au niveau le plus abstrait. Nous considérons ceci comme une preuve supplémentaire de l’importance des schémas de bas niveau sur les généralisations abstraites. Au niveau des alternances, nous suggérons que la productivité verbale peut être basée sur des relations d’alternances. Nous montrons que l’alternance dative présente une asymétrie en productivité, et que cette asymétrie peut être expliquée par des différences correspondantes en termes du nombre de verbes utilisés dans chaque construction. / The general goal of this thesis is to investigate to what extent the grammar of verbs, also called argument realization, can be based on linguistic usage. The usage-based approach is a recent paradigm shift in linguistics which takes the view that grammar is a dynamic inventory of symbolic conventions that emerges through, and is likewise shaped by, actual language use. Adopting a constructional approach to argument structure and on the basis of English data, we address the question of the usage basis of argument realization at three levels of organization.At the level of verbs, we compare experimental results to usage data, and find that more frequent valency patterns of a verb are processed more easily. These findings provide evidence for the usage basis of valency. At the level of constructions, we show that, in the case of the conative construction, it is possible to formulate constructional generalizations on the basis of verbal meaning at the level of semantically defined verb classes, but not so easily at the most abstract level. We take this as further evidence of the importance of lower-level schemas over broad generalizations. At the level of alternations, we present usage-based evidence that productivity can be based on alternation relations. We report that the dative alternation displays a productivity asymmetry, and we show that these differences can be explained by corresponding asymmetries in type frequencies.

Usage-based approaches to Second Language Acquisition (SLA) hold that grammatical development can at least in part be explained by a trajectory from initially entrenched formulaic chunks of linguistic material through partially abstract low-scope patterns to abstract constructions. Although earlier empirical research found little to support such a trajectory, more recent research suggests that when the methodological focus is on the breakdown of tokens not on the acquisition of abstract morphology formulaic chunks do seem to develop into at least low-scope patterns. What is not clear from these later studies, however, is why users select tokens from the environment, how these selections are repurposed to meet communicative needs and what grammatical development may be the result of such repurposing. Drawing on insights into usage-based approaches to SLA in general, and specifically predictions that the analogical processing of similar tokens can explain in part a trajectory from tokens in the linguistic environment to abstract constructions, this study investigates whether intervention in the form of task demands and written task productions can drive the selection and repurposing of task-relevant input, and whether such repurposing may influence the development of past-counterfactual constructions in instructed SLA. The study uses a classroom-based pre-test/post-test quasi-experimental design with pre-sessional university students in intact classes (n=92). Out of the three groups in the study, one group were exposed to instances of past counterfactuals which were identical to the forms needed for task completion in terms of function and lexical items (Literal Group). A second group were exposed to instances which called for the same function but different lexical items for successful task completion (Analogous Group), while a third group were not exposed to input. Results show that the analogous processing of task-relevant tokens led to positive and significant gains on most acceptability judgment and production test measures. Regression models further show that the selection of tokens explained very little variance, while analogical processing, operationalised as the repurposing of tokens to fit task demands, explained a significant amount of the variance in the same models. These findings highlight the importance of analogical processing in SLA and the significant implications such processing has for cognitive accounts of SLA and for second language teaching.

Aujourd'hui, un utilisateur peut interagir avec des assistants virtuels, comme Alexa, Siri ou Cortana, pour accomplir des tâches dans un environnement numérique. Dans ces systèmes, les liens entre des ordres exprimés en langage naturel et leurs réalisations concrètes sont précisées lors de la phase de conception. Une approche plus adaptative consisterait à laisser l'utilisateur donner des instructions en langage naturel ou des démonstrations lorsqu'une tâche est inconnue de l'assistant. Une solution adaptative devrait ainsi permettre à l'assistant d'agir sur un environnement numérique plus vaste composé de multiples domaines d'application et de mieux répondre aux besoins des utilisateurs. Des systèmes robotiques, inspirés par des études portant sur le développement du langage chez l'humain, ont déjà été développés pour fournir de telles capacités d'adaptation. Ici, nous étendons cette approche à l'interaction humaine avec un assistant virtuel qui peut, premièrement, apprendre le lien entre des commandes verbales et la réalisation d'actions basiques d'un domaine applicatif spécifique. Ensuite, il peut apprendre des liens plus complexes en combinant ses connaissances procédurales précédemment acquises en interaction avec l'utilisateur. La flexibilité du système est démontrée par sa forte adaptabilité au langage naturel, sa capacité à apprendre des actions dans de nouveaux domaines (Email, Wikipedia,...), et à former des connaissances procédurales hybrides en utilisant plusieurs services numériques, par exemple, en combinant une recherche Wikipédia avec un service de courrier électronique / Today users can interact with popular virtual assistants such as Siri to accomplish their tasks on a digital environment. In these systems, links between natural language requests and their concrete realizations are specified at the conception phase. A more adaptive approach would be to allow the user to provide natural language instructions or demonstrations when a task is unknown by the assistant. An adaptive solution should allow the virtual assistant to operate a much larger digital environment composed of multiple application domains and providers and better match user needs. We have previously developed robotic systems, inspired by human language developmental studies, that provide such a usage-based adaptive capacity. Here we extend this approach to human interaction with a virtual assistant that can first learn the mapping between verbal commands and basic action semantics of a specific domain. Then, it can learn higher level mapping by combining previously learned procedural knowledge in interaction with the user. The flexibility of the system is demonstrated as the virtual assistant can learn actions in a new domains (Email, Wikipedia,...), and can then learn how email and Wikipedia basic procedures can be combined to form hybrid procedural knowledge

This study investigates the acquisition of Chinese connectives by second language learners under the theoretical framework of usage-based theory. Language is not a random set of words and phrases, but rather a coherent and cohesive set of utterances. As such, learning a second language (L2) entails, among other processes, learners’ development of employing cohesive devices to construct a coherent discourse in their target language. One type of cohesive devices frequently used by L2 learners is connectives. In Chinese, connectives are utilized to denote various semantic relationships between the clauses in a compound sentence. Due to their flexibility and complexity in nature, Chinese connectives present a huge challenge to L2 learners’ learning. However, to date no study has been set up to explore the learners’ development of Chinese connectives within L2 Chinese research community. This study aims to fill this gap in the literature and build an L2 acquisitional model of Chinese connectives under the theoretical framework of Constructionist Usage-based Theory. Constructionist Usage-based Theory maintains that the basic unit of language is constructions and that the syntactic and lexical form of constructions and its corresponding semantic and discourse functions are conventionalized in language usage. According to these notions, language learning is believed to be driven by the factors grounded in the form and function of constructions in language usage. This study specifically examines how the factors of frequency, form, function, contingency (interaction of form and function), and L1-tuned attention affect L2 Chinese learners’ development of Chinese connectives. Furthermore, the study investigates the learners’ knowledge about the distribution of Chinese connectives across different proficiency levels. Specifically, this study aims to address four research questions: (1) what is the relationship between L2 learners’ proficiency level and language background and the acquisition of Chinese connectives?; (2) do L2 learners overuse or underuse Chinese connectives in constructing responses when the other in a pair is given and what errors do L2 learners make when using Chinese connectives?; (3) how can 12 target pairs of Chinese connectives be categorized into (hierarchical) groups based on L2 Chinese learners’ performance?; and (4) how do theoretically-motivated models represent the factorial structure underlying L2 acquisition of Chinese connectives? To address the four research questions, this study elicited L2 Chinese learners’ performance in two tests: a mini-discourse completion test and a form-function association test. In the mini-discourse completion test, learners were required to supply a missing clause to complete a three-clause discourse in which one of paired connectives was embedded; in the form-function association test, learners were asked to choose options of paired connectives to link two given clauses where connectives were omitted. Results showed that the development of all Chinese paired connectives was positively correlated to L2 learners’ L2 proficiency level. Learners with heritage language background seemed to have an advantage over less frequent and less prototypical connectives. Predominantly, L2 learners underused Chinese connectives, resulting from the cross-linguistic influence of disparity between English and Chinese connectives at the structural level in particular and between English and Chinese textual cohesion at the discourse level in general. Based on L2 learners’ performances in the two tests, the 12 pairs of Chinese connectives were classifier into four hierarchical groups. Confirmatory factor analyses showed that the usage-based factors (i.e., frequency, co-occurrence strength, formulaicity, prototypicality, contingency, and L1-tuned attention) jointly determined the L2 acquisition and development of Chinese connectives in a complex, adaptive, dynamic manner. Summarizing these findings, this study proposed a usage-based acquisitional model of L2 Chinese connectives, providing theoretical contributions to the usage-based theory and pedagogical implications for Chinese connectives.