Poster Session Program


Technical note: Posters will be displayed all throughout the three days of the conference. Short (3-5 minutes) presentations will be given by the author(s) in the first hour  of the Poster Session (Saturday, 11:30-13:00), which will be followed by 30 minutes of discussion. (But of course, discussions are expected to continue in the coffee breaks, lunch breaks etc.)

Fitch and Hauser (2004) have tested monkeys and humans with artificial grammars in both FSG and PSG categories. They have shown that humans can easily learn both structures but monkeys fail to learn sentences which have recursive structure. They use (AB) n sentences as FSG and AnBn as PSG grammar where A’s are CV syllables pronounced by a man speaker and B’s are CV syllables pronounced by a woman speaker. This is very easy for humans to learn because they can just listen to the transitions between man voice and woman voice. We have tested the same structures on human adults but using  more difficult materials. In our experiments A group is consisted of Cv syllables and B group is consisted of CVC syllables. If humans own a system to detect dependencies in recursive structures they should also learn sentences of our experiments.
Serially ordered consonants (Cs) of an artificial language stream constituted exclusively by CV syllables are used to segment speech on the basis of their transitional probabilities (TPs), see Newport et Aslin (2004) and Bonatti et al (in press). Here we explore more extensively this claim.
First, we show that the TPs among consonants are also employed to parse an artificial speech stream when more complex syllables than CV (i.e., CCV, CVC) are used. Then we explore the relative importance of syllables and consonants to extract the constituent items in an artificial speech stream. We demonstrate that the TPs between Cs are useful only if the Cs of test items maintain the same syllabic position they occupy in the habituation stream.
We show that syllables act as organizing structures, determining which consonants are used to compute statistical dependencies during segmentation.
The results of two ERP experiments with semantic and syntactic violations in Croatian sentences will be presented. In the first experiment subjects were presented with semantically correct and semantically incorrect sentences such as "The dog ate a question." In the second experiment there was a syntactic violation in the target stimuli - the target word violated the agreement in gender, e.g. "The nurse (f) is nice (m)."
The results were similar to the results in similar experiments in other languages. The N400 component was observed in sentences with semantic violation while both left anterior negativity (LAN) and P600 was elicited in sentences with agreement error.
In comparison to other languages the N400 component showed delayed latency preceded by another negative wave that could reflect morphosyntactic processing of the target word.
Finally, Croatian specific formal features such as aspect will be discussed in light of finding electrophysiological evidence in favor of particular models of language comprehension.
Universal to human language is the distinction between function words (e.g. the, he, up etc.) and content words (e.g. table, dog, run, nice etc.). These two major categories play different roles in language. The former encode syntactic relations, while the latter carry semantic meaning. Moreover, languages seem to show a tendency to prefer either of the two possible relative orders of the two categories. In English, for instance, auxiliaries precede main verbs (will go), prespositions precede noun phrases (under the table) etc., whereas in Hungarian, the opposite is true (menni fog 'go will'; az asztal alatt 'the table under'). It is, therefore, essential for language learning infants to distinguish these two broad categories. They are aided in this task by a number of "low level" cues, available from the speech signal, that the two categories differ in. First, function word tokens seem to be universally more frequent than content word tokens. Secondly, the former appear to be shorter, typically monosyllabic, and phonologically more reduced than the latter.  Werker and colleagues have shown in a series of experiments that young infants are able to tell the two major categories apart on the basis of these phonological cues.
In a series of artificial grammar experiments, we asked, as a first step, whether adult subjects are able to do the required classification on the basis on the first kind of cue, i.e. the difference in frequency. The evidence suggests that the answer to this question is positive. As a second step, we investigated whether there are native language-related differences  in what subjects take to be the basic order of the two categories. We tested speakers of five typologically different languages: Italian, French, Hungarian, Japanese and Basque.
The studies investigate how bilingual and monolingual children understand situations that involve mental states of others. We explore whether bilingualism promotes success in theory of mind tasks. This may be possible because bilingual children switch between their languages as a function of the addressee’s language. According to our proposal this language selection requires an insight into the others’ mind. If this is the case, bilinguals develop a theory of mind earlier than matched monolinguals. In the first study 3 years old bilingual and monolingual children performed a standard theory of mind task, a modified theory of mind task and a control task for general information processing. Results show that success in theory of mind develops faster in bilinguals than in monolinguals while they do not differ in the control task. The second study investigates the mechanisms, which might underlie the observed phenomena. The questions we address are the following: 1.Are bilinguals really better in ToM or they are just better in inhibiting a prepotent response (the reality bias)? In order to study this, in addition to the standard ToM tasks we used tasks that lack the strong bias by decreasing the saliency of last event or increasing the saliency of previous event. 2. Are the differences in ToM ability due to better grammatical abilities of the bilinguals? We measured performance in understanding tensed complements that involved mental or communication verbs, with or without a prepotent response. 3. How long does the bilingual advantage last? (comparing 3 and 4 years old) The results show that the bilingual advantage is not due to better grammatical abilities and the differences between monolinguals and bilinguals disappear at the age of four. Furthermore, it seems that the young bilingual advantage is only restricted to the standard ToM tasks, which involve a prepotent response.
Restoration of hearing after cochlear implantation is based principally on two factors: i) degree of bypassing damaged structures in the inner ear by the surgical insertion of a neuroprosthetic device and ii) intrinsic properties of the cerebral cortex to process novel auditory sensory input (plasticity). Animal studies show that after frequency-selective deafening at the sensory periphery by a partial cochlear ablation, auditory cortex reorganizes its tonotopic maps (Schwaber et al, 1993). Namely, after some adaptation period, deprived parts of the cortex become responsive to intact cochlear regions. However, little is known about the mechanisms and timeline of the adaptation to the new stimuli in both adults and children cochlear implant (CI) users.
This study explores the outcome over 5 months of the changes in the regional brain activity associated with a language discrimination task, in a profoundly deaf adult following cochlear implantation, using Optical Topography (OT). OT is a novel non-invasive neuroimaging technique that records the hemodynamic (vascular) response of the superficial areas of the brain in response to sensory stimulation and cognitive tasks.
A study was performed involving phonological priming and tip of the tongue states (TOTs) where participants took either 200mg of caffeine, or a placebo. Results show a clear positive priming effect produced for the caffeine group when primed with phonologically related words. When primed with unrelated words the caffeine subgroup produced a significant increase in the number of TOTs.  This contrasting effect provides evidence that the positive priming of caffeine was not because of caffeine's well-known alertness effects. For placebo, a significant negative effect occurred with the related word priming condition. The results support the novel hypothesis that the blocking of A1 adenosine receptors by caffeine induces an increased short-term plasticity (STP) effect within the phonological retrieval system.  To test further the time dependent STP-caffeine hypothesis, in this study the prime-target interval is lengthened.  We show that when the prime-target interval is increased, the priming effect is weaker.
Research into segmenting fluent speech has revealed that infants and adults can use the transition probabilities (TPs) between syllables in order to achieve segmentation; syllabic sequences with high internal TPs are treated as possible word-like units, while 'dips' in TP are taken as evidence for segmentation points.
In theories of prosodic phonology, words are invariably aligned to larger prosodic constituents in speech. Thus, phonological phrase and intonational phrase boundaries are invariably also word boundaries.
In previous research we asked if Italian subjects would be sensitive to such segmentation information inherent in Italian IP characteristics. More precisely, we asked if artificial, trisyllabic 'words' that span IP contours are equally well accepted as similar 'words' that lie internal to IP contours. We found that contour-straddling 'words' were significantly dispreferred. In addition, we found evidence supporting the view that even in such artificial speech streams, subjects perceive 'chunks': 'words' that occurred at the leading edges of prosodically defined chunks were better recognized than 'words' in the middles of such chunks.
Here we ask if the result we obtained was due to Italian IPs being used with Italian subjects. In order to establish the potential universality of such a segmentation strategy, we examined the effect on the segmentation of statistically well-formed 'words' as in the previous experiments, but using Japanese IP characteristics instead of Italian.
The results suggest that even with Japanese IPs, subjects continue to reject 'words' with high internal TPs if they straddle IP boundaries, even though the IPs are from a foreign language.
We tested a group of patients with unilateral left- or right- hemisphere damage on an action imitation test. Actions were either meaningful (MF) or meaningless (ML) and were presented either in separated blocks or intermingled. Overall patients’ performance was worse than that of controls. No difference between MF and ML actions emerged when presented intermingled. However, the blocked presentation allowed dissociations between MF and ML actions to emerge in some patients. Results are interpreted based on a two-route model. In the mixed condition patients used the non-semantic route to imitate both kind of stimuli whereas in the blocked condition they applied the semantic route for MF and the direct route for ML actions.




[ UBC Department of Psychology ] [ Send Feedback/Report A Problem ]