Volume 56, Issue 2

Motivated by an increasing use of social media for the expression of personal stance towards a certain target, we analyse the language used to produce such opinionated content with expressions of sentiment, which represents the main data source for sentiment analysis. We use the first manually annotated corpus for sentiment analysis of the Serbian language developed for the service sector of higher education. Our study focuses on how various linguistic constructions, used in different context, influence the sentiment polarity of a text. Our findings indicate that sentiment expressions and negation have a most significant role in determining whether the text conveys positive, neutral, or negative sentiment, while intensifiers (words which either increase or decrease sentiment) have a considerable influence on sentiment intensity. We also present an analysis of the impact of conjunctions, conditional sentences, comparative and modal verbs, and pronouns on sentiment polarity and intensity. Based on the derived observations, we propose a set of rules that could be integrated with machine learning algorithms into an automated sentiment analysis system for the Serbian language. Our findings also make a much-needed contribution to the few currently available resources for natural language processing of Serbian.

This paper offers a formal analysis of three constructions in English: locative inversion , central deictic inversion and directional inversion . These constructions constitute thetic statements with a locative intentional base which sets a scene that (re)introduces an entity in the discourse; syntactically, they display a non-canonical word order and have a number of unusual grammatical properties which make them particularly interesting to show how syntax connects, and adapts, to discourse. I propose that they all obtain from a language particular mechanism which involves a functional category LocP that adjusts the computational requirement to have a preverbal subject to the intentional need to have the subject post-verbally. As for the differences among them, they are approached in terms of the features that head LocP and the lexical properties of the verbs that head each of the structures. Ultimately, the paper also serves to discuss the role of certain informational features (the so-called core intentional features ) in the syntactic derivation.

This paper represents a foray into the largely unexplored territory of the cognition of semantic approximations in first language acquisition. Current advances on cognitive modelling are applied to a corpus of 500 semantic approximations produced by 20 children between 1;06 and 5;00 years old. The results reveal that a large number of those semantic approximations are the output of a set of cognitive operations including those of comparison and correlation (i.e. metaphorical projections), domain expansion and reduction (i.e. metonymic mappings), and mitigation and strengthening (i.e. scalar operations). Far for being an impediment to communication, most semantic approximations in our data are found to help children capitalize on their incomplete lexical pool, maximizing its communicative potential. The set of cognitive strategies involved is available from an early age, underlying the use of language throughout our lifespan, and adapting its functions to diverse communicative needs in different stages of our lives.

This article is devoted to the problem of translating bad language in cases when the target audience recipient has direct access to the source text and the impact translational and editorial choices have on the overall meaning of the work. As an illustration of this point, it discusses a voice-over translation, in which case it is common practice to censor vulgarities, mainly by means of under-translating certain phrases, which are considered taboo, or omitting them completely. Such choices are justified for a number of reasons, including protecting vulnerable audiences or reducing semantic density because most taboo words exercise a phatic function. However, in certain cases censoring bad language has an impact on the semiotic make-up of the work and thus, changes the meaning intended by the original author. A good illustration is provided with the analysis of particular scenes from the Canadian drama American Heist .

Valency theory has been applied to investigate various languages, such as German, Chinese and English. However, most studies in this field were based on the linguistic materials produced by native speakers. The current research aimed to examine the valency structures in the interlanguage. Based on the English writing produced by L2 Chinese learners, we adopted the quantitative approach, trying to find out whether the distributional features of verb valency in the interlanguage also had regular probability distributions as those in the native languages, and whether there was a relationship between these valency distributional characteristics and L2 learners’ language competence. It was found that (1) verb valency in the interlanguage followed distributional regularities which had been identified in the native languages; (2) the valency features showed differences in the diversity of valency patterns, the use of valences and the complexity of forms of complements between the interlanguage and the target language; (3) the distribution functions and parameters related to verb valency could manifest the development of students’ language competence. The current research has extended valency theory to the study of interlanguage and the valency perspective has profound methodological and pedagogical implications for L2 learning. Its item-specific property and the integration of grammatical and lexical factors are conducive to analyzing the way various words combine with each other.