Call number: LD2668 .T4 1966 H873 / Master of Science

published_or_final_version / abstract / Linguistics / Master / Master of Philosophy

published_or_final_version / abstract / Linguistics / Master / Master of Arts

The Interface Hypothesis (Sorace and Filiaci, 2006) conjectures that adult second language learners (L2 learners) who have reached near-native levels of proficiency in their second language exhibit difficulties at the interface between syntax and other cognitive domains, most notably at the syntax-discourse interface. However, research in this area was limited, in that the data were offline, and thus unable to provide evidence for the nature of the deficit shown by L2 learners. This thesis presents online data which address the question of the underlying nature of the difficulties observed in L2 learners at the syntaxdiscourse interface. This thesis has extended work on the syntax-discourse interface in L2 learners by investigating the acquisition of two phenomena at the syntax-discourse interface in German: the role of word order and pronominalization with respect to information structure (Experiments 1-3), and the antecedent preferences of anaphoric demonstrative (the der, die, das series homophonous with the definite article) and personal pronouns (the er, sie, es series) (Experiments 4- 8). Crucially, this work has used an on-line methodology, the visual-world paradigm, which allows an insight into the incremental interpretation of interface phenomena in real-time processing. The data from these experiments show that L2 learners have difficulty integrating different sources of information in real-time comprehension efficiently, supporting the Interface Hypothesis. However, the nature of the processing difficulties which L2 learners demonstrate in on-line processing was not determined by these studies, resulting in the question: are L2 learners’ difficulties a result of a limitation of processing resources, or the inability to deploy those resources effectively? A novel dualtask experiment (Experiment 9), in which native speakers of German were placed under processing load simulated the results previously obtained for L2 learners. It is concluded that syntactic dependencies were constrained by resource limitation, whereas discourse based dependencies were constrained by processing resource allocation.

Natural languages display a great variety of different word orders, and one of the major challenges facing statistical machine translation is in modelling these differences. This thesis is motivated by a survey of 110 different language pairs drawn from the Europarl project, which shows that word order differences account for more variation in translation performance than any other factor. This wide ranging analysis provides compelling evidence for the importance of research into reordering. There has already been a great deal of research into improving the quality of the word order in machine translation output. However, there has been very little analysis of how best to evaluate this research. Current machine translation metrics are largely focused on evaluating the words used in translations, and their ability to measure the quality of word order has not been demonstrated. In this thesis we introduce novel metrics for quantitatively evaluating reordering. Our approach isolates the word order in translations by using word alignments. We reduce alignment information to permutations and apply standard distance metrics to compare the word order in the reference to that of the translation. We show that our metrics correlate more strongly with human judgements of word order quality than current machine translation metrics. We also show that a combined lexical and reordering metric, the LRscore, is useful for training translation model parameters. Humans prefer the output of models trained using the LRscore as the objective function, over those trained with the de facto standard translation metric, the BLEU score. The LRscore thus provides researchers with a reliable metric for evaluating the impact of their research on the quality of word order.

It is widely accepted that Old English personal pronouns often turn up in ‘special’ positions, i.e. positions in which functionally equivalent nominals rarely, if ever, appear. Leading theories of Old English syntax (e.g. van Kemenade 1987, Pintzuk 1991, 1996, Hulk & van Kemenade 1997, Kroch & Taylor 1997) account for the syntax of specially placed pronouns in different ways, but all treat special placement as a freely available option. Focusing on pronominal objects of prepositions in particular, this thesis shows, firstly, that current theories fail to account for the variety of special positions in which these pronouns appear and argues that at least three special positions must be recognised. The central concern of this thesis, however, is whether special placement is the freely available option that leading theories assume. Drawing on evidence from a number of descriptive studies of the syntax of pronominal objects of prepositions (e.g. Wende 1915, Taylor 2008, Alcorn 2009), statistical evidence is presented to show that, in a number of contexts, the probability of special placement is either too high or else too low to be plausibly ascribed to free variation. The thesis explores the linguistic basis of each of the statistically significant parameters identified, finding answers in some cases and intriguing puzzles in others.

The purpose of this paper was to do an investigation of adverbials in fiction and non-fiction texts translated into Swedish and English. Adverbials are more flexible regarding position in sentences than other constituents. It has been of interest to find out if there are any remarkable differences in mean-ing due to repositioned adverbials in translation, and the focus has been on adverbials in initial, medial and final position. The results showed that most adverbials retained their position, and also their meaning in translation. There were no noteworthy differences in how adverbials were translated in fiction compared to non-fiction. The preferred position of adverbials was the end position for most types of adverbials in English and Swedish.

In this thesis I examine the placement of the dependent genitive relative to its head noun in Classical Latin prose. The corpus is drawn from the works of four first-century B.C. authors: Caesar, Cicero, Sallust and Varro. The thesis itself is split into two main sections, a qualitative analysis and a quantitative analysis. The qualitative analysis discusses a number of factors that may influence genitive position, drawn from literature on the subject as well as my pilot studies. These factors are information structure, the genitive’s grammatical function, discontinuity, lexical category, animacy, prepositions governing the head noun, reported speech, idioms, lexical items, and grammatical number of the genitive. This analysis examines individual instances of genitive position in context, providing examples and counter-examples of the ordering patterns found with each potential factor. The qualitative analysis suggests that a number of these factors have an effect on genitive position, particularly information structure. These results are tested by the quantitative analysis. By performing a multivariate statistical analysis using the programme GoldVarb, the combined effects of multiple factors are determined and the statistically significant factors ranked in order of importance and strength of effect. The statistics show that information structure is the most important of the factors. Other significant influences are the presence of prepositions, the function of the genitive and its lexical category. By combining the two types of analysis, qualitative and quantitative, this thesis shows in detail how the factors combine to influence word order, which of them are independent, which interact, which are important and which have little to no effect at all, resulting in a better understanding of the data and the way that the contextual factors work together to produce the variant orders of the dependent genitive.

This thesis aims to find out how animacy affects sentence structure choice in Turkish learners of L2 English. The study compares three different L2 English proficiency levels with each other as well to L1 English and L1 Turkish. In this way the effect of English, a rigid word order language, and Turkish, a free word order language on sentence structure choice have been compared. A picture description task was applied on 94 participants. The pictures depicted a transitive action taking place between an inanimate agent and an animate patient (animate condition) or between an inanimate agent and an inanimate patient. The subjects were given handouts with the pictures and were asked to write down what is happening in each picture. There were 60 Turkish learners of L2 English and 14 English participants in the study. Turkish learners of English belonged to level-1 (16 students), level-2 (25 students) and level-3 (19 students). In addition, 20 Turkish speakers were consulted for their knowledge of Turkish. It was hypothesized that in L2 English, animate entities would be accessed first and this will directly affect sentence structure choice through grammatical subject assignment or through word order. Thus, it was expected that when the learners are shown a picture depicting a transitive action taking place between an animate patinet and an inanimate agent, they would tend to use the passive in English, which assigns both a sentence-initial position and a subjecthood role to the animate entity. L2 proficiency level and native language were expected to play a role in determining the role of animacy on sentence structure choice. Chi-square analysis and odds ratio calculations were made. The results showed that animacy of the patinet affected sentence structure choice in L2 English by triggering the passivce usage in only level-3 (the most advanced group). Animacy of the patient affected native speakers of English in the same way, i.e. native English speakers tended to use the passive voice in the animate condition. No such effect was found in lower level learners of L2 English (i.e. level-1 and level-2) and Turkish native speakers. It was found that in the animate condition, Turkish native speakers tended to use the OSV word order more frequently than they did in the inanimate condition. This result suggested that in Turkish, animacy of the patient triggers the use of the OSV (Object, Subject, Verb) order rather than the passive voice. In short, the research results suggested that L2 proficency level and native language could play a role in determining how animacy affects sentence structure choice in L2.

Word order universals constitute a well-known problem in language typology, first outlined in Greenberg (1963). It has been firmly established in databases of over 1500 languages that languages with verb-object ordering are very likely to have prepositions and noun-genitive ordering, while languages with object-verb ordering are very likely to have postpositions and genitive-noun ordering (Dryer and Haspelmath 2011). This thesis attempts to give a historical explanation for these facts in terms of the origin of syntactic categories: adpositions have historically developed from nouns and verbs (Givon 1984, Aristar 1991); and verbs often develop from nominalizations used with a genitive object. These types of grammaticalization can explain why adpositions retain the ordering of their source nouns or verbs, and why verb/object ordering often parallels noun/genitive ordering. This historical explanation is elaborated on, with data from different language families. Examples of verbs grammaticalizing from nominalizations used with genitive objects are given, drawing on historical work such as Salanova (2007) on Brazilian Jê languages and Starosta, Pawley and Reid (1982) on Austronesian. Different languages show varying degrees of 'nominalism', the morphosyntactic resemblance between verb forms and noun phrases/nominalizations. Other languages show a less developed distinction between adpositions and verbs/nouns. These examples of gradience in syntactic categories are argued to be behind resemblances in word orderings. Language contact is responsible for preserving word order types, when languages undergo change in more than one word order (e.g. Greenberg 1969); and the difference in rates of word order change across constructions is argued to be behind hierarchies such as Hawkins (1983)'s Prepositional Noun Modifier Hierarchy. This explanation of word order universals contrasts with more mainstream accounts such as Hawkins (1994) in terms of processing efficiency, and Kirby and Christiansen (2003) in terms of learnability. While these explanations are perhaps compatible with the historical explanation, they are argued to be redundant; grammaticalization arguably is not driven or constrained by learnability and processing efficiency, with memetics, 'typological poise' (Enfield 2003) and language contact given as alternatives. Instead of reflecting functional biases, word order patterns are argued to reflect language history, both the history of language contact, and the history of syntactic categories developing through grammaticalization. / published_or_final_version / Linguistics / Master / Master of Philosophy