| dc.description.abstract | Part I
Because of the cognitive symptoms in cerebellar patients, the cerebellum is currently thought to be involved in cognitive functioning. Cerebellar loops running to and from the cerebral cortex let researchers to the hypothesis that the cerebellum could, along with the basal ganglia, be part of a subcortical pathway involved in language.
There is evidence pointing towards cerebellar dysfunction in developmental dyslexics. They exhibit deficits in balance, speed in motor tasks and procedural learning. The procedural learning deficit hypothesis states that in dyslexia, procedural learning mechanisms (via the cerebellum) fail in the process of implicitly learning phonological rules in language acquisition, leading to reduced phonological awareness and poor word recognition. These mechanisms then give way to problems in reading and spelling. However, not much research has been done into this hypothesis with the cerebellum as the key to the influence of implicit learning on developmental dyslexia.
The Language, Learning and the Cerebellum study aims to prove that there is a network in the brain including the cerebellum that is involved in language acquisition through implicit learning mechanisms. The study therefore contains two participant groups, dyslexics and patients with known cerebellar disease. Both groups will be compared to matched control groups on a neuropsychological test battery, implicit learning tasks and (for dyslexics and their controls only) a structural MRI scan. The hypotheses are that a (relative) cerebellar deficit is related to impaired implicit learning (e.g. dyslexics and cerebellar patients perform worse on both the implicit learning tasks than their controls) and that the cerebellum of dyslexics varies significantly from the cerebellum of healthy controls in volume.
Part II
The Language, Learning and the Cerebellum study consists of a variety of different tasks. Each of the tasks has to be piloted first. The implicit learning tasks that will be used in the Language, Learning and the Cerebellum study are a serial reaction time task (measuring implicit learning of motor skills) and an artificial grammar learning task (AGL). In this AGL task, participants have to listen to center-embedded sentences consisting of pseudowords and created by a phrase-structure grammar to approach human grammar as closely as possible. After listening to these 'grammatical' sentences, they will listen to grammatical as well as 'ungrammatical' sentences (that cannot be made according to the grammar) and judge whether each sentence is grammatical or ungrammatical. Hypothesis is that they will judge right above chance, indicating they learned the grammar implicitly. For this AGL task, the sound files for the lexicon of pseudowords have to be made which will then be put into sentences according to a grammar that accounts for center-embedding. The pseudowords contain phonological and prosodic cues to address the center-embedding within the grammar and increase its learnability. In this case, phonological cues that are added are the first consonant and the vowel of each word. These cues are necessary as there are no other ways of finding out center-embedding with words that contain no meaning. Prosodic cues such as modulated pitch are added to the four-word sentences to approach prosody in center-embedded sentences in natural human language and increase learnability.
The aim of this pilot study is to create 32 sound files of nonsense words for a previously made artificial grammar and test the perceivability of their phonological and prosodic cues. Three requirements have to be met regarding the perceivability of the cues. (1) All 32 words have to be equally well intelligible. (2) All 4 variants (one for each place in the sentence) of each word have to be equally intelligible. (3) Words have to be intelligible at first as well as at non-first occurrence. 10 participants were made to listen to the 4 variants of the 32 randomly presented stimuli. They had to type the words they heard. Typed responses and accuracy (right/wrong) were obtained. Wrong responses that violated the cues of the artificial grammar language were marked as 'critical mistakes'. Analysis was performed by descriptive statistics and repeated measures ANOVA of these critical mistakes only. (1) Of the 32 words, three were perceived significantly worse than average, generating more critical mistakes (blong, blum, trul). (2) The number of critical mistakes of the four variants did not differ significantly. (3) Three words were perceived significantly worse than average at first occurrence (blong, plis, trul) and three words were perceived significantly worse than average at non-first occurrence (blim, blong, blum). Due to the number of critical mistakes generated by them, all four stimuli variants of blim, blong, blum, plis and trul will be resynthesized and tested again.
Most of the synthesized sound files of the nonsense words have proven to be highly applicable for the Artificial Grammar Learning task within the Language, Learning and the Cerebellum study. | |
