The study addresses cultural influence in the recognition of moderately expressed emotions in a second language (L2) and foreign speech. The web-based listening test consisted of context-free sentences drawn from the Estonian Emotional Speech Corpus. The task was to recognize the emotion (joy, anger, sadness) or neutrality of each sentence without seeing the speaker. Three adult groups participated: (1) 36 Estonians, with Estonian as mother tongue; (2) 16 highly educated Russians living in Estonia, with Russian as their mother tongue and Estonian as a second language; (3) 16 highly educated Russians living in Russia, with Russian mother tongue and no knowledge of Estonian. The results showed a significant difference between Estonians and Russians living in Estonia in their recognition of joy and neutrality; Russians living in Russia differed significantly from Estonians and Russians living in Estonia on all emotion scores. This confirms that cultural norms are mastered through interaction: to recognize vocal emotions expressed in another language it is necessary to live in the culture and communicate in its language.

Abstract. We investigated the influence of culture and language on the understanding of speech emotions. Listeners from different cultures and language families had to recognize moderately expressed vocal emotions (joy, anger, sadness) and neutrality of each sentence in foreign speech without seeing the speaker.

While Sentiment Analysis aims to identify the writer's attitude toward individuals, events or topics, our aim is to predict the possible effect of a written text on the reader. For this purpose, we created an automatic identifier of the polarity of Estonian texts, which is independent of domain and of text type. Depending on the approach chosen -lexicon-based or machine learning -the identifier uses either a lexicon of words with a positive or negative connotation, or a text corpus where orthographic paragraphs have been annotated as positive, negative, neutral or mixed. Both approaches worked well, resulting in a nearly 75% accuracy on average. It was found that in some cases the results depend on the text type, notably, with sports texts the lexicon-based approach yielded a maximum accuracy of 80.3%, while over 88% was gained for opinion stories approached by machine learning.

Abstract. Paragraphs of four genres are analysed to detect their emotional colouring, while a lexicon-based approach of linguistic analysis is weighed against reader opinion. The aim is to find out the prospects of automatic detection of emotion in any text by using a very small lexicon of about 600 frequent emotion words.*

A pleasant voice is an asset not only in various professions but also in speech technologies. This article addresses the correlation between voice likability and phonogenres. Men and women of different age groups were asked to evaluate voice likability in 5-second speech passages arranged into two web-based listening tests, presenting 50 female voices and 60 male voices, respectively. The passages represented three phonogenres: radio commentaries, lectures and radio talk shows. The results demonstrated the impact of a phonogenre on voice likability scores. The phonogenres were analysed acoustically using the extended Geneva Minimalistic Acoustic Parameter Set (eGeMAPS). Out of the 88 parameters, 27 turned out to be relevant for describing phonogenres (six parameters coincided with male and female voices). The analysis revealed that one of the three phonogenres-lectures-displayed a consistent difference from the other two. Lecturing voices were considered the least likable, irrespective of the age and gender of the speaker as well as the listener.