Reading is an activity that most literate people realize daily without a lot of effort. For this reason we are not aware of the huge complexity that brings about this neurocognitive proces. When we read a word our brain has to perform a series of complicated operations to eventually recognise ink marks and connect them with the meanings stored in our memory. These computations can include the definition of the limits of the words, letter recognition, determination of the letter positions, detection of word forms by groups of letters, activation of phonological codes, and the competetion of potential meanings. For example to recognise a word it is necessary to not only recognise each of the letters seperately but also their relative position (Muñoz, García-Orza, Perea y Barber, in press). Additionally on the orthographic level it was proposed that letters can be grouped into intermediate units like syllables (Barber, Vergara y Carreiras, 2004) or morphemes (Barber, Domínguez y de Vega, 2002). Therefore, words can be segmented into intermediate units that contribute to activate possible meanings of which the correct one is selected. In the last few years research has favoured these processes with different techniques including electrophysiology, and the obtained results are used to elaborate detailed theoretical models and computations about word recognition (see review Barber y Kutas, 2007).
Scanning printed text involves continuous fast eye movements called saccades, but words are only perceived during fixations — the period between saccades when a reader’s gaze lands on a specific word. It is well known that not every word in text is fixated during reading; many words (especially short words and function words) are perceived only parafoveally. Moreover, parafoveal perception could influence normal reading in at least two different ways: For one, when a word n is fixated, the next word in the sentence n+1 can be perceived in the parafovea. Then when the same word n+1 is fixated after a saccade, its processing can be sped up because of the parafoveal preprocessing. A second and more controversial role for parafoveal perception in reading is the putative parafoveal influence (of upcoming inputs) on the processing of the currently fixated word (n). Determining the specific contributions of saccade programming, lexical processing, and semantic integration on mean fixation times is difficult. Much remains unknown about the exact type and amount of linguistic information garnered from the parafovea, under what circumstances, and how this information is integrated in real time with the foveal information. In our lab we use both eye-tracking and ERP techniques to determine the factors that constrain the amount of information extracted from the parafovea during reading (Barber, Doñamayor, Kutas, & Münte, 2010; Barber, Shir, Bentin, & Kutas, 2011).
An important aspect of cognitive functioning of the human brain is the way the different mental contents are produced, stored and retrieved. This issue is also central to the study of language processing in relation to how to represent and process the meanings of the words. In several previous ERP studies we have looked at the role of different semantic dimensions of words when we are recognizing and processing them. For instance, we have shown that some processing differences between nouns and verbs can be reduced to the semantic features associated to those types of words, as the amount of motor or sensory features (Barber, Kousta, Otten, & Vigliocco, 2010; for a general review including neuropsychological and neuroimaging data see also Vigliocco, Vinson, Druks, Barber, & Cappa, 2011). In a related ERP study we compared the processing of concrete and abstract words keeping constant other semantic variables like their imageability or context availability. After these controls we could replicate the previous ERP concreteness effect but obtaining a behavioural advantage for abstract words, in contrast to previous reports (Barber, Otten, Kousta & Vigliocco; submitted). Such abstractness effect could be partially explained by the emotional valence of the words, because many abstract words have associated emotional information. All together these studies highlight that semantic variables are tightly interconnected in meaning representation and word processing.
Across the world, languages make extensive use of agreementto signal the structural relation between words in an utterance. Morphological marks are important and necessary information for computing dependenciesbetween a noun and other words in a sentence, such as determiners,adjectives, past participles, pronouns, etc., especially in richly inflectedlanguages. For example, in Spanish, nouns are markedas either masculine or feminine, with a natural criterion for assigning gender to human beings (based on the biological sex of the referent) and an arbitrary criterion for objects, roles, and abstract entities. They also can carry explicit plural markers signalling one aspect of quantity of the semantic referent. The gender and number of determiners, adjectives, pronouns, and past participles must always agree with the entity to which they refer. Therefore, agreement poses a challenge to sentence processing systems both in production and in comprehension. During comprehension, the processor needs to figure out which items agree and which do not, because long distance agreement relations often occur.In our lab we have carried out several studies in which we analyzed the cognitive operations associated to the processing of different types of agreement relationships during language comprehension. We have defined the main ERP correlates of the processing of gender and number agreement violations (Barber & Carreiras, 2003; Barber, Salillas & Carreiras, 2004; Barber & Carreiras, 2005; see review in Molinaro, Barber & Carreiras, 2011), and we have tried to determine the main brain structures associated to these processes using bothfMRI(Carreiras, Carr, Barber & Hernandez, 2010) and the TMS techniques (Carreiras, Pattamadilok, Meseguer, Barber & Devlin, in press).
Language production is a language skill that most of us learn without any formal training. The average speaker knows about 50,000 words, and therefore, in order to produce fast and error-free speech, there need to be efficient memory access procedures that allow the retrieval of the correct word. Research has shown that these access procedures rely primarily on the meaning of a word, but also use information related to the form of the word (Janssen & Caramazza, 2009), and seem to rely less on formal aspects of the word (Janssen, Melinger, Mahon, & Caramazza, 2011). There are also differences in the way linguistic units known as open-class and closed-class elements are retrieved (Janssen, Schiller, & Alario, in press). Fast and efficient language production means that not only single words are stored in memory, but that also morphologically complex words (e.g., cars, doghouse; Janssen, Bi, & Caramazza, 2008), and multi-word phrases (e.g., the red car; Janssen & Barber, 2012) are stored. Recently, neuro-imaging techniques that exploit the high-speed nature of language production such as electroencephalography have provided further insight into the cognitive mechanisms that underlie language production (Janssen, Barber, & Carreiras, 2011).
The late second language learners is an interesting group to study, because it is getting more and more common to learn a language at school even later as an adult, and also because the study of second language learners can provide information on how the process of language acquisition in general develops. For example, whether characteristics of the first language (L1) transfer to the second language (L2), or how and on what level the two languages function independently on the long term. To put it briefly: whether or not the process of L2 acquisition follows a similar pattern as L1 learning. In a series of experiments we have looked at the electrophysiological correlates of grammatical gender and number processing in highly proficient late L2 learners of Spanish from different L1 backgrounds (English and Chinese), to explore the roles of age of acquisition, L2 proficiency and language transfer effects in L2 morphosyntax processing. The results from these studies indicate transfer effects, or in other words, that L1 grammar is used to support L2 learning, at least in late L2 learners (Gillon-Dowens, Vergara, Barber & Carreiras, 2010; Gillon-Dowens, Guo, Guo, Carreiras & Barber, 2011). Bilingualism also results in people that eventually have to manage switches between languages. In other related experiments, we analyzed this language switching phenomena when the languages are not balanced in proficiency; one is more dominant than the other. The results of these experiment show that language switching involves highly skilled cognitive control and its costs are clearly asymmetrical depending of the direction of the switching (Van der Meij, Cuetos, Carreiras & Barber, 2011; Van der Meij, Hernández, Carreiras & Barber, submited).