A Spoken Dialogue System's (SDS's) dialogue strategy specifies which action it will take depending on its representation of the current dialogue context. Designing it by hand involves anticipating how users will interact with the system, and/or repeated testing and refining, and so can be a difficult, time-consuming task. Since SDSs inevitably make understanding errors, a particularly important issue is how to design ``repair strategies'', the parts of the dialogue strategy which attempt to get the dialogue ``back-on-track'' following these errors. To try to produce better dialogue strategies with less time and effort, previous researchers have modelled a dialogue strategy as a sequential decision problem called a Markov Decision Process (MDP), and then applied Reinforcement Learning (RL) algorithms to example training dialogues to generate dialogue strategies automatically. More recent research has used training dialogues conducted with simulated rather than real users and learned which action to take in all dialogue contexts, (a ``full'' as opposed to a ``partial'' dialogue strategy) - simulated users allow more training dialogues to be generated, and the exploration of new dialogue contexts not present in an original dataset. As yet however, limited insight has been provided as to which dialogue contextual features are important to include in the MDP and why. Indeed, a full dialogue strategy has not been learned from training dialogues with a realistic probabilistic user simulation derived from real user data, and then shown to work well with real users. This thesis investigates the value of adding new linguistically-motivated contextual features to the MDP when using RL to learn full dialogue strategies for SDSs. These new features are recent Dialogue Acts (DAs). DAs indicate the role or intention of an utterance in a dialogue e.g. ``provide-information'', an utterance being a complete unit of a speaker's speech, often bounded by silence. An accurate probabilistic user simulation learned from real user data is used for generating training dialogues, and the recent DAs are shown to improve performance in testing in simulation and with real users. With real users, performance is also better than other competing learned and hand-crafted strategies. Analysis of the strategies, and further simulation experiments show how the DAs improve performance through better repair strategies. The main findings are expected to apply to SDSs in general - indeed our strategies are learned and tested on real users in different domains, (flight-booking versus tourist information). Comparisons are also made to recent research which focuses on handling understanding errors in SDSs, but which does not use RL or user simulations.

Syftet med uppsatsen är att undersöka hur två reseföretag i förtroendekris bemöter kritik via Twitter samt vilka strategier inom den image reparerande diskursen de använder sig av. Studien ämnar även ta fasta på om det finns ett mönster bakom bemötandet. För att uppfylla uppsatsens syfte har två frågeställningar formulerats: · Hur bemöter SJ och SAS kritik på Twitter? · Verkar företagen vara medvetna om valda strategiers möjliga effekter? För att genomföra denna undersökning utfördes först en förundersökning där antalet dialoger mellan företagen och dess kunder sammanställdes samt dialogernas längd och antalet aktörer räknades. Till huvudundersökningen studeras 20 dialoger innehållande kritik som låg närmast i tiden. För att studera dessa använde vi oss av en kvalitativ textanalys. De viktigaste resultaten visar att båda reseföretagen tillämpar liknande strategier i situationer där de kritiseras för händelser av liknande karaktär. De mest använda strategierna var för både SJ en SAS av ignorerande eller förnekande karaktär, två strategier som i denna studie har kategoriserats som avståndstagande strategier. Den huvudsakliga slutsatsen i uppsatsen är att den valda strategin inte påverkar ett företags rykte så mycket som sättet själva strategin tillämpas på. / Aim: The purpose of this paper is to examine how two travel companies in different kinds of crises respond to criticism through social media, more specifically on Twitter. The paper also aims to identify whether crises of various kinds in the travel industry are met by specific image repair strategies. If possible one aim is also to take note if there is a certain relationship between chosen strategy and the impact on the examined companies reputation. Material and Method: The first part of the research consists of a preliminary investigations where the two companies tweets is counted and measured in order to get an overall picture of their activity on Twitter. This part of the research was based on one week ́s activity per company and the tweets were recalculated to get average numbers. The qualitative part represents the key study of this research and the material consisted of the 20 latest dialogues where customers expressed criticism and complaints directed to the companies SJ and SAS. Main Results: The main results of this research found that the two companies applied similar strategies in situations when accused for incidents of resembling nature. The most used strategies was for both SJ an SAS of ignoring or denial character, two strategies that in this research was categorized as renunciative strategies. The main conclusion this essay found is that the chosen strategy does not impact a company’s reputation as much as the application of the strategy.

Thesis (M.A.)--Indiana University, 2010. / Title from screen (viewed on July 18, 2010). Department of Communication Studies, Indiana University-Purdue University Indianapolis (IUPUI). Advisor(s): John Parrish-Sprowl, Kristina Horn Sheeler, Ronald Sandwina. Includes vitae. Includes bibliographical references (leaves 133-138).

To describe and to analyze the use of repair strategies in the syllable constituents of coda and complex onset used by children with normal phonological acquisition and phonological deviation, in order to examine the similarities as well as the differences between the studied groups. Besides this, specifically in relation to the coda constituent and based on the Non-Linear Gerative Phonology, we aimed at examining if the application of the adopted resources provides evidences of phonological knowledge. Methods: The analysis included speech data from 60 children with normal acquisition, 30 female and 30 male, in addition to 84 children with deviation, 31 female and 53 male. The age in the normal group was between 1:0 and 4:0, whereas in the deviation group it was between 3:0 and 11:0. In the analysis of the strategies used in complex onset position, the variables: age, gender, preceding and following context, obstruent of complex onset, kind of liquid of complex onset, tonicity, number of syllables and position in the word were considered. In relation to the resources used in the coda, the variables were: age, gender, preceding and following context, kind of phoneme in the coda, tonicity, number of syllables and position in the word. Then, the data were submitted to statistical analysis using the VARBRUL program. Results: The analysis of the complex onset showed that the strategies used by the children considering normal phonological development and deviation were, respectively: simplification for C1V (93%; 77%), alteration of the obstruent feature (5%; 17%), alteration of the liquid feature (1%; 5%), epenthesis (1%; 0%) and metathesis (0%; 1%). In the coda constituent, the children with normal development and deviation used the following resources, respectively: omission of target segment (71%; 71%), omission of target segment with a change in the quality of the preceding vowel (1%; 4%), omission of target syllable (6%; 1%), 14 semivocalization (11%; 14%), substitution for liquid (1%; 3%), palatalization (2%; 3%), metathesis (1%; 2%), epenthesis (1%; 1%), compensatory lengthening (5%; 0%) and other realizations (1%; 1%). In terms of the strategies used in the complex onset, the variables tonicity, preceding context and position in the word were not relevant, whereas considering the resources applied in coda, all the variables were important. Conclusion: The study of repair strategies in both constituents showed more similarities than differences in relation to the kind of repair strategies that children with normal phonological development and phonological deviation use during the acquisition process. In the complex onset constituent, there is less diversity among the resources, while the use of simplification for C1V is predominant in both studied groups. In the coda syllabic position, we found a greater variety of strategies, and the omission of the target segment was the most used by children with normal phonological acquisition and deviation. Besides this, the other strategies applied in the coda position became important due to their evidences in terms of phonological knowledge. / Descrever e analisar o uso das estratégias de reparo nos constituintes coda e onset complexo, empregados por crianças com aquisição fonológica normal e desviante, a fim de verificar as semelhanças e as diferenças existentes entre os grupos estudados. Além disso, em relação especificamente ao constituinte coda, pretendeu-se, através da Fonologia Gerativa Não-Linear, verificar se a aplicação dos recursos empregados fornece indícios de conhecimento fonológico. Método: Foram utilizados dados de fala de 60 crianças com aquisição normal, 30 do sexo feminino e 30 do sexo masculino, e 84 crianças com aquisição desviante, 31 do sexo feminino e 53 do sexo masculino. A idade do grupo normal variou de 1:0 a 4:0, enquanto que a do grupo desviante variou de 3:0 a 11:0 anos. Para a análise das estratégias empregadas na posição de onset complexo foram consideradas as variáveis: idade, sexo, contexto precedente e seguinte, obstruinte do onset complexo, tipo de líquida do onset complexo, tonicidade, número de sílabas e posição na palavra. Para os recursos utilizados em coda foram consideradas: idade, sexo, contexto precedente e seguinte, tipo de fonema em coda, tonicidade, número de sílabas e posição na palavra. Posteriormente, os dados foram submetidos à análise estatística através do programa VARBRUL. Resultados: A análise do onset complexo revelou que as estratégias que as crianças lançaram mão, considerando o desenvolvimento 12 fonológico normal e desviante, foram, respectivamente: simplificação para C1V (93%; 77%), alteração do traço da obstruinte (5%; 17%), alteração do traço da líquida (1%; 5%), epêntese (1%; 0%) e metátese (0%, 1%). No constituinte coda, as crianças com desenvolvimento normal e desviante utilizaram os seguintes recursos, respectivamente: omissão do fonema alvo (71%; 71%), omissão do fonema alvo com mudança da qualidade da vogal precedente (1%; 4%), omissão da sílaba alvo (6%; 1%), semivocalização (11%; 14%), substituição por líquida (1%; 3%), palatalização (2%; 3%), metátese (1%; 2%), epêntese (1%; 1%), alongamento compensatório (5%; 0%) e outras realizações (1%; 1%). Para as estratégias empregadas em onset complexo as variáveis tonicidade, contexto precedente e posição na palavra não foram relevantes no estudo, enquanto que, para os recursos aplicados em coda, todas as variáveis foram relevantes. Conclusão: O estudo das estratégias de reparo em ambos os constituintes revelou mais similaridades que diferenças quanto ao tipo de estratégias de reparo que as crianças com desenvolvimento fonológico normal e desviante lançam mão ao longo do percurso de aquisição. No constituinte onset complexo há menos diversidade entre os recursos, predominando o uso de simplificação para C1V para ambos os grupos estudados. Já na posição silábica de coda, constata-se uma maior variedade de estratégias, sendo o recurso de omissão do fonema alvo o preferido pelas crianças com aquisição fonológica normal e desviante. Além disso, as outras estratégias aplicadas na posição de coda se mostraram importantes pelo que evidenciaram em termos de conhecimento fonológico.

Baseball's Steroid Era put many different high-profile athletes under pressure to explain steroid allegations that were made against them. This thesis used textual analysis of news reports and media portrayals of the athletes, along with analysis of their image repair strategies to combat those allegations, to determine how successful the athletes were in changing public opinion as evidenced through the media. The contexts, media reports, and strategies of Jason Giambi, Mark McGwire, Andy Pettitte, and Roger Clemens were analyzed and revealed important implications involving effective use of image repair strategies. They provided a deeper framework for the success of mortification strategies. An authentic, sincere mortification strategy has more power to change the media's reporting and portrayal of the athlete, while stunted or incentivized mortification strategies provide diminishing results. The four different situations of the players and the different combinations of strategies used provide insight into how much a public persona matters in confronting allegations. They show how ineffective the strategy of minimization is against allegations that involve on-field performance. The situations reveal how the promise of future on-field actions, along with actual on-field success can help repair an athlete's image without a solid rhetorical strategy. They show the amount of information offered, along with the strategies used, influences the amount of persuasion that occurs. The different situations also showed how a complete image repair strategy is successful in ending news coverage of the allegations and not just changing the media portrayal.

<p>Speech intervention for children with language impairment aims to both strengthen specific language skills and to reinforce the ability to participate in interactional contexts. Children with language impairment often have problems in interaction; therefore the contributions of the conversational partner are of great importance. Thus, the demands are high on the speech and language pathologist (SLP) for sense and cohesion in intervention. The aim of the present study was to examine how SLPs and children with language impairment interact in language intervention. An approach influenced by Conversation Analysis was used in order to identify various phenomena that occurred, and their effect on the structure of the interaction was analyzed.</p><p>Four SLPs, two men and two women, were enrolled in the study.  They participated in intervention with two children with language impairment each, i.e. eight conversations. The analysis showed that different forms of feedback and different repair strategies had an impact on the interaction. In addition, it was shown that the interactions contained several situations related to social face work.  Both differences and similarities were found between the participants in the analyzed material. Different interactional patterns of the SLPs appeared to affect the outcome of the intervention in different ways.</p> / <p>Logopedisk intervention för barn med språkstörning har för avsikt att både stärka de specifika språkförmågor som brister och att göra barnet mer redo för kommunikativa sammanhang. Språkstörning hos barn innebär ofta svårigheter i interaktionen med andra, därför är samtalspartnerns agerande av stor betydelse. Således är kraven på logopeden stora för att skapa mening och sammanhållning i interventionen. Syftet med föreliggande studie var att undersöka hur logopeder och barn med språkstörning interagerar vid behandling av språkstörning. Ett arbetssätt influerat av Conversation Analysis användes för att belysa olika fenomen som förekom i interaktionen och vilken inverkan dessa hade på interaktionens struktur.</p><p>I studien deltog fyra logopeder, två män och två kvinnor med två behandlingsbesök av barn med språkstörning vardera, således åtta samtal. Intressanta fenomen som förekom var olika typer av återkopplingar, reparationer och ansiktshotande situationer. Det förelåg både skillnader och likheter mellan deltagarna i det analyserade materialet. Logopedernas olika interaktionsmönster förefaller påverka interventionen på olika sätt.</p>

Speech intervention for children with language impairment aims to both strengthen specific language skills and to reinforce the ability to participate in interactional contexts. Children with language impairment often have problems in interaction; therefore the contributions of the conversational partner are of great importance. Thus, the demands are high on the speech and language pathologist (SLP) for sense and cohesion in intervention. The aim of the present study was to examine how SLPs and children with language impairment interact in language intervention. An approach influenced by Conversation Analysis was used in order to identify various phenomena that occurred, and their effect on the structure of the interaction was analyzed. Four SLPs, two men and two women, were enrolled in the study.  They participated in intervention with two children with language impairment each, i.e. eight conversations. The analysis showed that different forms of feedback and different repair strategies had an impact on the interaction. In addition, it was shown that the interactions contained several situations related to social face work.  Both differences and similarities were found between the participants in the analyzed material. Different interactional patterns of the SLPs appeared to affect the outcome of the intervention in different ways. / Logopedisk intervention för barn med språkstörning har för avsikt att både stärka de specifika språkförmågor som brister och att göra barnet mer redo för kommunikativa sammanhang. Språkstörning hos barn innebär ofta svårigheter i interaktionen med andra, därför är samtalspartnerns agerande av stor betydelse. Således är kraven på logopeden stora för att skapa mening och sammanhållning i interventionen. Syftet med föreliggande studie var att undersöka hur logopeder och barn med språkstörning interagerar vid behandling av språkstörning. Ett arbetssätt influerat av Conversation Analysis användes för att belysa olika fenomen som förekom i interaktionen och vilken inverkan dessa hade på interaktionens struktur. I studien deltog fyra logopeder, två män och två kvinnor med två behandlingsbesök av barn med språkstörning vardera, således åtta samtal. Intressanta fenomen som förekom var olika typer av återkopplingar, reparationer och ansiktshotande situationer. Det förelåg både skillnader och likheter mellan deltagarna i det analyserade materialet. Logopedernas olika interaktionsmönster förefaller påverka interventionen på olika sätt.

Indiana University-Purdue University Indianapolis (IUPUI) / The importance of image management has created the need to for organizations to continually work in order to improve their image or defend it against perceived threats. Since organizations engage in a constant struggle to preserve their reputation, it is important to understand the persuasive discourse associated with image repair strategies. In addition, a successful rhetor must also acknowledge the importance of perception of the appropriateness and effectiveness of the apologetic discourse from the perspective of an audience. Focus groups were conducted and analyzed in order to better understand the perceived appropriateness and effectiveness of the image repair strategies employed by the Big Three Automakers as perceived by the audience. The findings of this study complement the original findings of Benoit and Drew’s quantitative study assessing the appropriateness and effectiveness of image repair strategies in an interpersonal setting. However, there were observable differences between the studies in terms of the perception of both the appropriateness and effectiveness of bolstering and the effectiveness of differentiation. The implications of these differences can be important in developing a better understanding of the utilization of image repair strategies in the apologetic discourse of organizations. Specifically, the results demonstrate how the audience determines the appropriateness and effectiveness of the strategies and how rhetors are able to successfully use different strategies based in context.

Samtalsanalys har tidigare använts som ett komplement till traditionellt använda standardiserade tester för att undersöka mer funktionella aspekter av kommunikation (Wells &amp; Local, 1992; Yont, Hewitt &amp; Miccio, 2002; Samuelsson 2009). Syftet med föreliggande studie är att utifrån sytematiska analyser av vardagliga samtal och logopedisk intervention utforska relationen mellan kommunikativa behov i vardagen och mål med intervention för barn med språkstörning. Tre barn med språkstörning filmades i vardagliga samtal hemma och på förskolan samt vid ett behandlingsbesök hos logopeden. Materialet transkriberades och analyserades enligt samtalsanalytiska principer. Intervjuer och retrospektioner genomfördes med föräldrar, förskolepersonal samt behandlande logoped för att ytterligare belysa barnens kommunikativa situation. I föreliggande studie framkommer att de kommunikativa behov barnen har relaterar till de mål som sätts upp i logopedisk behandling. På grund av deras fonologiska svårigheter påverkas barnens kommunikation då det ofta uppstår problemkällor, deras yttranden ignoreras och deras talutrymme inskränks. Reparationsstrategier används av både vuxna och barn för att reda ut de situationer där missförstånd uppstår. Dessa strategier kan, med hjälp av retrospektioner, uppmärksammas och diskuteras i logopedisk intervention i syfte att undvika framtida situationer där barnens svårförståeliga yttranden inte reds ut. / Conversation analysis has been used as a complement to traditional standardized testing to evaluate the functional aspect of communication (Wells &amp; Local, 1992; Yont, Hewitt &amp; Miccio, 2002; Samuelsson 2009). The purpose of the present study is to explore the relationship between everyday communicative needs and objectives set in speech and language intervention through systematic analysis of everyday conversations and speech and language intervention. Three children with language impairment were video recorded during everyday interaction at home, at the preschool and during one speech and language therapy session. The material was transcribed and analyzed according to principles of conversation analysis. Interviews and retrospections were conducted with parents, preschool teachers and the treating speech and language pathologists in order to further illustrate the participating children’s communicational situation. The results in the present study show that the communicational needs does relate to the objectives set up in speech and language intervention for the children in the current study. The children’s communication is primarily afflicted by their phonological difficulties in the way that it fuels for sources of communicational breakdowns, the children’s utterances become ignored and it restrains their participation in conversation. Repair strategies are used by both adults and children to solve sequences where mutual understanding is not achieved. These strategies may, through retrospections, be identified and discussed in speech and language intervention in order to avoid future situations where unintelligible utterances are not solved.