Previous evidence has shown that word frequencies calculated from corpora based on film and television subtitles can readily account for reading performance, since the language used in subtitles greatly approximates everyday language. The present study examines this issue in a society with increased exposure to subtitle reading. We compiled SUBTLEX-GR, a subtitled-based corpus consisting of more than 27 million Modern Greek words, and tested to what extent subtitle-based frequency estimates and those taken from a written corpus of Modern Greek account for the lexical decision performance of young Greek adults who are exposed to subtitle reading on a daily basis. Results showed that SUBTLEX-GR frequency estimates effectively accounted for participants’ reading performance in two different visual word recognition experiments. More importantly, different analyses showed that frequencies estimated from a subtitle corpus explained the obtained results significantly better than traditional frequencies derived from written corpora.

For more than a decade, a large body of empirical evidence has shown that, for the recognition of a written word, subword units are accessed at early stages of visual word processing, and the properties of these subword units have an effect on reading behavior. Researchers have repeatedly reported data showing how letters and phonemes (e.g., Pelli, Farell, & Moore, 2003;Perea, Duñabeitia, & Carreiras, 2008b;Rastle & Brysbaert, 2006), syllables (e.g., Conrad, Stenneken, & Jacobs, 2006Perea & Carreiras, 1998), and morphemes (e.g., Duñabeitia, Perea, & Carreiras, 2007a, 2008Rastle, Davis, & New, 2004) constitute the building blocks of word processing. However, how polysyllabic words are segmented into their syllables during reading is still an open question, and further research is needed in order to shed some light on this issue. The present study reports the process of creating a database of Spanish and Basque syllables: SYLLABARIUM, a Web tool for psycholinguistic experiments that includes features for material selection and syllable analyses. Syllables As Processing UnitsMuch research has been done to explore the influence of the syllable in word processing, focusing on transparent languages like Spanish, in which orthographic representations (i.e., graphemes) map to phonological representations (i.e., phonemes) almost in a one-to-one manner (see Ál-varez, Carreiras, & Perea, 2004;Carreiras, Álvarez, & de Vega, 1993). Two findings have been repeatedly reported in the literature: the syllable-congruency priming effect (Carreiras & Perea, 2002) and the inhibitory effect of the first syllable's positional frequency (Carreiras et al., 1993).The term syllable-congruency priming effect refers to the fact that, when a word is preceded by a string containing the same orthographic or phonological syllable, this word is recognized faster and more accurately than when it is preceded by a string in which the initial syllable is not the same. Carreiras and Perea (2002) were the first authors to report a syllable-congruency priming effect, showing that a Spanish word like PASTOR (shepherd), syllabified as PAS.TOR, was recognized faster in a Spanish lexical decision task when it was preceded by a string like PAS *** than when it was preceded by a string like PA **** . In the same experiment, Carreiras and Perea also showed that a word like PASIVO ( passive), syllabified as PA.SI.VO, was recognized faster in the lexical decision task when it was preceded by the syllabic-congruent string PA ***** than The present article introduces SYLLABARIUM, a new Web tool addressing the needs of linguists, psycholinguists, and cognitive scientists who work with Spanish and/or Basque and are interested in retrieving information about several syllable-related parameters. This new online syllabic database allows the user to generate complete lists of Spanish and Basque syllables with information about the syllable frequency. Among other measures, for a given orthographic syllable, SYLLABARIUM provides its number of occurrences (i.e., the typ...

Medical images from different clinics are acquired with different instruments and settings. To perform segmentation on these images as a cloud-based service we need to train with multiple datasets to increase the segmentation independency from the source. We also require an efficient and fast segmentation network. In this work these two problems, which are essential for many practical medical imaging applications, are studied. As a segmentation network, U-Net has been selected. U-Net is a class of deep neural networks which have been shown to be effective for medical image segmentation. Many different U-Net implementations have been proposed. With the recent development of tensor processing units (TPU), the execution times of these algorithms can be drastically reduced. This makes them attractive for cloud services. In this paper, we study, using Google's publicly available colab environment, a generalized fully configurable Keras U-Net implementation which uses Google TPU processors for training and prediction. As our application problem, we use the segmentation of Optic Disc and Cup, which can be applied to glaucoma detection. To obtain networks with a good performance, independently of the image acquisition source, we combine multiple publicly available datasets (RIM-One V3, DRISHTI and DRIONS). As a result of this study, we have developed a set of functions that allow the implementation of generalized U-Nets adapted to TPU execution and are suitable for cloud-based service implementation.

No abstract