I am currently a research fellow at the Basque Centre for Cognition, Brain and Language where I lead three independent research projects: 1) “Speech rhythm in language acquisition and speech evolution”, funded by Ikerbasque (2015-2020); 2) “Bilingual Metacognition: The effect of linguistic experience on the neural mechanisms of metacognitive processing in the linguistic domain” funded by the Spanish Ministry for Research and Development MINECO (2017-2020); and 3) “Cognitive and neural mechanisms for splitting continuous sensory input into discrete constituents in different domains and modalities”. Below, I outline the main objectives and premises for these three research lines which I propose to continue investigating.
For more information on each research line, click the image following the outline of each research line
Speech rhythm in language acquisition, speech evolution and language change
Rhythm is fundamental to every motor activity. The mechanisms for controlling motor rhythm have been inherited by mammals from their distant evolutionary ancestors and are shared by all vertebrates. As speech production is a kind of motor activity, investigating speech rhythm can provide insight into how general motor patterns have been adapted for more specific use in articulation and speech production. Studies on speech rhythm may further provide insight into the development of speech capacity in humans. As the capacity for vocal communication is putatively a pre-requisite for developing a language faculty, studies on speech rhythm may also cast some light on the mystery of language evolution in the human genus.
Although every human individual possesses the same neural, anatomical, and physiological mechanisms for controlling speech rhythm, rhythmic patterns differ across languages. These language-specific patterns of motor activity are based on general physiological mechanisms, yet shaped by the peculiarities of the ambient language and by cultural transmission between populations sharing a given language. As speech rhythm patterns are simultaneously naturally-occurring and culturally transmitted, studying speech rhythm opens a window onto nature–nurture interactions in language use and acquisition.
The best methodological approach for investigating such naturally-occurring yet culturally-transmitted behaviors is Tinbergen’s framework. I have modified Tinbergen´s ethological approach to address research on human behavioural patterns related to speech activities. This includes four different directions for investigation:
1. Development of behavioral patterns in ontogenesis (considering how rhythmic patterns are acquired by individuals);
2. Development of behavioral patterns in phylogenesis (considering how language-specific rhythmic patterns emerge in populations sharing the same language(s), and which components of rhythmic cognition humans share with non-human animals);
3. Functional load (considering how rhythmic patterns facilitate communication and how better capacity for rhythmic cognition might enhance the fitness of individual organisms);
Emergence and mechanisms (considering the underlying mechanisms that lead to the emergence of speech rhythm and allow for culture-specific rhythmic modulations within biological constraints as well as cognitive constraints on learnability and processability).

Metacognition and multi/bilingualism
Metacognition – the ability to monitor and evaluate one’s own cognitive performance – has been the object of intensive investigations in memory and perception. By comparison, metacognitive processes have received considerably little attention in bilingualism research. Moreover, there is neither positive nor negative evidence regarding the possible effects of bilingualism on cognitivehigher-level decision making in non-linguistic domains. It is possible that metacognitive processing in a specific domain may be influenced by individual experience in that domain. Consequently, bilinguals might display improved metacognitive awareness in tasks that are linguistic in nature, because they have more experience with speech and language processing. Besides, if experience indeed matters, then early bilinguals should outperform late bilinguals in metacognitive processing in linguistic tasks. Potentially, enhanced metacognitive processing in the linguistic domain can be transferred to non-linguistic tasks because empirical evidence suggests that neural circuits involved in metacognitive processing are task-independent and operate across modalities (visual and auditory) and domains (e.g., linguistic, logical, creative, and others). I will investigate whether individual experience in language processing leads to enhanced metacognitive processing in the linguistic domain and can be transferred to non-linguistic tasks.
Such studies might have important societal implications. Differences in metacognitive processing also lead to differences in confidence when making decisions and expressing opinions. When group decisions are made, disagreements may frequently occur between experts; in such cases, more importance will be assigned to the opinions of those who express themselves with more confidence. Individual confidence levels, at the same time, can vary due to task-irrelevant – including social, cultural and psychological – factors. Multilingualism is an integral characteristic of contemporary society. Bilingualism, age of acquisition of a second language, exposure to bilingual environments, i.e., living in bilingual societies and exposure to a wider range of linguistic cues, and even proficiency in second/foreign languages can affect confidence judgements in linguistic tasks, which engage a broad set of speech and language processing mechanisms, including those involved in making metalinguistic judgements. Assuming that metacognitive confidence is affected by linguistic experience and can then be transferred across domains and modalities, we need to know to what degree and in which domains individual linguistic experience may affect decision making. Note that this may be important not only at the individual but also at the group level: Understanding metacognitive confidence during decision making and the influence of factors such as bilingualism and bilingual environments can place constraints on the extent of human reasoning across different social groups.
To address these scientific and societal challenges, we need to answer five research questions:
- How does bilingualism influence metacognitive processing in linguistic tasks in the visual modality?
- How does bilingualism influence metacognitive processing in linguistic tasks in the auditory modality?
- How does bilingualism influence metacognitive processing in non-linguistic tasks in the visual modality?
- How does bilingualism influence metacognitive processing in non-linguistic tasks in the auditory modality?
- What is the interaction between bilingualism per se and living in bilingual environments? Bilingualism entails having acquired two languages from birth or at an early age (under 2 years) and using both languages on a daily basis in a full range of communicative situations. Exposure to bilingual environments entails exposure to linguistic stimuli with greater variability and may lead individuals to adjust the trade-off between exploitation/exploration strategies in information processing.

Cognitive and neural mechanisms for splitting continuous sensory input into discrete constituents
Although sensory input is continuous, cognitive systems operate on discrete constituents. Splitting continuous sensory input into discrete units is called segmentation. Segmentation is based on statistical learning, the process of finding statistical regularities, such as transitional probabilities (TPs), in continuous sensory input in order to structure this input into processable units. Statistical learning mechanisms operate across all modalities and domains (splitting speech into words and phrases, musical compositions into rhythmic groupings, continuous actions into isolated sequences of events). However, it is still unclear whether statistical learning results in representations of extracted constituents, or rather in a set of statistical regularities which can then be used to test the statistical coherency of newly encountered exemplars. The cognitive mechanisms for extracting, learning and exploiting conditional statistics are highly debated. Two possible classes of mechanisms, suggesting different roles of TPs in segmenting a continuous input, have been proposed: 1) clustering mechanisms and 2) boundary-finding mechanisms. Boundary-finding mechanisms are based on detecting high-entropy transitions, when one element predicts the following element with low probability. This implies identifying rarely co-occurring elements: the high-entropy transitions between these elements signal boundaries between discrete constituents. By contrast, clustering mechanisms are based on detecting high transitional probabilities between consecutive elements, i.e., detecting clusters of elements within which one element predicts the transition to next element with high probability. This mechanism is based on identifying frequently co-occurring elements and merging them together to form a new constituent. It is also important to identify neural substrates and networks that support learning statistical patterns and/or discrete units, and the recognition of extracted constituents vs. recognition of mere statistical patterns. Finally, I aim to ascertain whether the neural and cognitive mechanisms engaged in statistical learning are domain- and modality-general, and if not, which aspects are adapted specifically for more efficient speech processing in the auditory and visual modalities. This will help us explore the evolutionary roots of such mechanisms, and their role in the emergence of language faculty in humans.
