La locution est peu étudiée en génération automatique de texte (GAT). Syntaxiquement, elle forme un syntagme, alors que sémantiquement, elle ne constitue qu’une seule unité. Le présent mémoire propose un traitement des locutions en GAT multilingue qui permet d’isoler les constituants de la locution tout en conservant le sens global de celle-ci. Pour ce faire, nous avons élaboré une solution flexible à base de patrons universels d’arbres de dépendances syntaxiques vers lesquels pointent des patrons de locutions propres au français (Pausé, 2017). Notre traitement a été effectué dans le réalisateur de texte profond multilingue GenDR à l’aide des données du Réseau lexical du français (RL-fr). Ce travail a abouti à la création de 36 règles de lexicalisation par patron (indépendantes de la langue) et à un dictionnaire lexical pour les locutions du français. Notre implémentation couvre 2 846 locutions du RL-fr (soit 97,5 %), avec une précision de 97,7 %. Le mémoire se divise en cinq chapitres, qui décrivent : 1) l’architecture classique en GAT et le traitement des locutions par différents systèmes symboliques ; 2) l’architecture de GenDR, (principalement sa grammaire, ses dictionnaires, son interface sémantique-syntaxe et ses stratégies de lexicalisations) ; 3) la place des locutions dans la phraséologie selon la théorie Sens-Texte, ainsi que le RL-fr et ses patrons syntaxiques linéarisés ; 4) notre implémentation de la lexicalisation par patron des locutions dans GenDR, et 5) notre évaluation de la couverture de la précision de notre implémentation. / Idioms are rarely studied in natural language generation (NLG). Syntactically, they form a phrase, while semantically, they correspond to a single unit. In this master’s thesis, we propose a treatment of idioms in multilingual NLG that enables us to isolate their constituents while preserving their global meaning. To do so, we developed a flexible solution based on universal templates of syntactic dependency trees, onto which we map French-specific idiom patterns (Pausé, 2017). Our work was implemented in Generic Deep Realizer (GenDR) using data from the Réseau lexical du français (RL-fr). This resulted in the creation of 36 template-based lexicalization rules (independent of language) and of a lexical dictionary for French idioms. Our implementation covers 2846 idioms of the RL-fr (i.e., 97.5%), with an accuracy of 97.7%. We divided our analysis into five chapters, which describe: 1) the classical NLG architecture and the handling of idioms by different symbolic systems; 2) the architecture of GenDR (mainly its grammar, its dictionaries, its semantic-syntactic interface, and its lexicalization strategies); 3) the place of idioms in phraseology according to Meaning-Text Theory (théorie Sens-Texte), the RL-fr and its linearized syntactic patterns; 4) our implementation of the template lexicalization of idioms in GenDR; and 5) our evaluation of the coverage and the precision of our implementation.

This thesis is about whether language affects thinking. It deals with the linguistic relativity hypothesis, which proposes that the language we speak influences the way we think. This hypothesis is investigated in the domain of caused motion (e.g., ‘The man rolled the tyre into the garage’), by looking at Spanish and Swedish, two languages that show striking differences in how motion events are encoded. The thesis consists of four studies. The first two focus on native speakers of Spanish and Swedish. Study I compares how Spanish and Swedish speakers describe the same set of caused motion events, directing the spotlight at how variable the descriptions are in each language. The results confirm earlier findings from semantic typology regarding the dominant ways of expressing the events in each language: Spanish behaves like a verb-framed language and Swedish like a satellite-framed language (Talmy, 2000). Going beyond previous findings, the study demonstrates—using the tools of entropy and Monte Carlo simulations—that there is markedly more variability in Spanish than in Swedish descriptions. Study II tests whether differences in how Spanish and Swedish speakers describe caused motion events are reflected in how they think about such events. Using a novel similarity arrangement task, it is found that Spanish and Swedish speakers partly differ in how they represent caused motion events if they can access language during the task. However, the differences disappear when the possibility to use language is momentarily blocked by an interference task. The last two studies focus on Swedish learners of Spanish as a second language (L2). Study III explores how Swedish learners (compared to native Spanish speakers) adapt their Spanish motion descriptions to recently encountered input. Using insights from the literature on structural priming, we find that Swedish learners initially expect to encounter in their L2, Spanish, those verb types that are typical in Swedish (manner verbs like ‘roll’) but that, with increasing proficiency, their expectations become increasingly attuned to the typical Spanish pattern of using path verbs (like ‘enter’).  These expectations are reflected in the way L2 learners adapt their own production to the Spanish input. Study IV asks whether recent linguistic experience in an L2 can affect how L2 learners think about motion events. It is found that encountering motion descriptions in the L2 that emphasize different types of information (path or manner) leads L2 speakers to perceive similarity along different dimensions in a subsequent similarity arrangement task. Taken together, the thesis argues that the study of the relation between language and thought affords more valuable insights when not posed as an either-or question (i.e., does language affect thought or not?). In this spirit, the thesis contributes to the wider aim of investigating the conditions under which language does or does not affect thought and explores what the different outcomes tell us about language, thought, and the intricate mechanisms that relate them. / <p>At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 1: Manuscript. Paper 3: Manuscript.</p>

The current thesis is based on four individual studies which aim to account for the expression of motion events (ME) in Spanish and Swedish as first languages (L1), in Swedish as a second language (L2), and in Spanish as a heritage language (SHL). The data, resulting from audio-recordings of different sorts of stimuli, have been analyzed with special focus on (1) the most common structures used for referring to various types of ME, (2) the types and amount of information provided by the participants, in particular as regards the semantic components Path, Manner and Ground, and (3) grammatical aspect and types of syntactic structures resorted to, including the correlation between the two latter factors and speakers’ discursive preferences.      Study 1 sets out to explore how Spanish and Swedish native speakers convey information about motion. The results show that the Swedish L1 speakers produced a wider range of descriptions concerning Manner and Path than the Spanish L1 speakers; furthermore, both groups delivered detailed Ground descriptions, although the Swedish native speakers expressed final destinations (endpoints) of ME to a greater extent.      Study 2 aims to investigate to what extent Swedish L1 patterns for motion encoding are still at play in the acquisition of Spanish L2 even at advanced stages of L2 acquisition. The results show that the learner group used a larger amount of Path particles and Ground adjuncts (in particular those referring to endpoints) than did the Spanish natives; this finding supports the claim that L2 learners rely on the lexicalization patterns of their L1 when describing ME in an L2. As for Manner, the L2 speakers were found to express this component mainly outside the verb, and to deliver more information about Manner than the Spanish natives.      Study 3 addresses the construal of ME in Swedish speakers of L2 Spanish, in particular concerning the encoding of motion endpoints and Manner of motion. The results show that the Swedish learners of Spanish exhibited the same, high frequencies of endpoint marking as did their monolingual Swedish peers, thus deviating from the Spanish native pattern. Moreover, the L2 speakers used the same amount of Manner verbs as did the Spanish natives but tended consistently to provide additional Manner information in periphrastic constructions.      Finally, Study 4 sets out to analyze the ways in which L1 Spanish/L2 Swedish early and late bilinguals express ME in SHL. The aim is to show in which ways and to what extent the typological patterns for motion encoding in the L2 may impact on motion encoding in the L1 with regard to three parameters: (1) age of onset (AO) of the acquisition of L2, (2) length of residence (LoR) in the L2 environment and (3) contact level with the L1 (CL). The focus data, consisting of oral re-tellings produced by the bilinguals, were compared to analogous data produced by two control groups (native speakers of Spanish and Swedish) in order to analyze conflation patterns regarding Manner, Path and Ground information. The analysis points to the conclusion that both the individuals’ AO of L2 acquisition and their LoR in the L2 environment have affected their L1 conceptualization patterns while their CL plays a subordinate role.      In summary, the findings lend support to the idea that the habitual conceptualization of events in the L1 influences L2 acquisition; conversely, the conceptual patterns of the L2 have an impact on L1 usage in bilinguals, especially in combination with an early AO and a long LoR. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: In press.</p>

The current thesis is based on four individual studies which aim to account for the expression of motion events (ME) in Spanish and Swedish as first languages (L1), in Swedish as a second language (L2), and in Spanish as a heritage language (SHL). The data, resulting from audio-recordings of different sorts of stimuli, have been analyzed with special focus on (1) the most common structures used for referring to various types of ME, (2) the types and amount of information provided by the participants, in particular as regards the semantic components Path, Manner and Ground, and (3) grammatical aspect and types of syntactic structures resorted to, including the correlation between the two latter factors and speakers’ discursive preferences.      Study 1 sets out to explore how Spanish and Swedish native speakers convey information about motion. The results show that the Swedish L1 speakers produced a wider range of descriptions concerning Manner and Path than the Spanish L1 speakers; furthermore, both groups delivered detailed Ground descriptions, although the Swedish native speakers expressed final destinations (endpoints) of ME to a greater extent.      Study 2 aims to investigate to what extent Swedish L1 patterns for motion encoding are still at play in the acquisition of Spanish L2 even at advanced stages of L2 acquisition. The results show that the learner group used a larger amount of Path particles and Ground adjuncts (in particular those referring to endpoints) than did the Spanish natives; this finding supports the claim that L2 learners rely on the lexicalization patterns of their L1 when describing ME in an L2. As for Manner, the L2 speakers were found to express this component mainly outside the verb, and to deliver more information about Manner than the Spanish natives.      Study 3 addresses the construal of ME in Swedish speakers of L2 Spanish, in particular concerning the encoding of motion endpoints and Manner of motion. The results show that the Swedish learners of Spanish exhibited the same, high frequencies of endpoint marking as did their monolingual Swedish peers, thus deviating from the Spanish native pattern. Moreover, the L2 speakers used the same amount of Manner verbs as did the Spanish natives but tended consistently to provide additional Manner information in periphrastic constructions.      Finally, Study 4 sets out to analyze the ways in which L1 Spanish/L2 Swedish early and late bilinguals express ME in SHL. The aim is to show in which ways and to what extent the typological patterns for motion encoding in the L2 may impact on motion encoding in the L1 with regard to three parameters: (1) age of onset (AO) of the acquisition of L2, (2) length of residence (LoR) in the L2 environment and (3) contact level with the L1 (CL). The focus data, consisting of oral re-tellings produced by the bilinguals, were compared to analogous data produced by two control groups (native speakers of Spanish and Swedish) in order to analyze conflation patterns regarding Manner, Path and Ground information. The analysis points to the conclusion that both the individuals’ AO of L2 acquisition and their LoR in the L2 environment have affected their L1 conceptualization patterns while their CL plays a subordinate role.      In summary, the findings lend support to the idea that the habitual conceptualization of events in the L1 influences L2 acquisition; conversely, the conceptual patterns of the L2 have an impact on L1 usage in bilinguals, especially in combination with an early AO and a long LoR. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: In press.</p>