The latest version of the program can be found here.
The slide shows for all keynotes are contained in this archive.
Martti Vainio - University of Helsinki/Finnland
Tone and quantity: the psychoacoustic trade-offs in prosodic signaling
Many languages exploit suprasegmental devices in signaling word meaning. Tone languages exploit fundamental frequency whereas quantity languages rely on segmental durations to distinguish otherwise similar words. Traditionally, duration and tone have been taken as mutually exclusive. However, some evidence suggests that, in addition to durational cues, phonological quantity is associated with and co-signaled by changes in fundamental frequency in quantity languages such as Finnish, Estonian, and Serbo-Croatian. It is also known that the fundamental frequency of a complex sound modulates the perceived duration of a sound. Higher pitch sounds are perceived longer compared to lower pitch sounds as shown by several independent studies since 1973. In my talk I will present a series of experiments that explore the interactions between tonal features of speech and both temporal and energy based features, and the ways these trade-offs are used to signal both phonological and functional aspects of language. Our research ranges from production studies to measuring complex auditory brainstem responses using participants from several different language groups.
Professor Martti Vainio is the head of the Phonetics and Speech Synthesis
Research Group and the Department of Digital Humanities at the University of Helsinki.
Before his professorship prof. Vainio workded as a university lecturer in the discipline of
language technology at the Department of Linguistics at the University of Helsinki.
After that he held a Finnish Academy Fellowship and worked on emotional speech synthesis.
He has also worked at the Aalto University and is currently collaborating with Academy
Professor Paavo Alku on statistical speech synthesis.
In addition to basic scientific research on speech he has been working on speech synthesis
actively for more than twenty years. His doctoral dissertation examined the use of artificial
neural networks in controlling prosody in text-to-speech synthesis. He has published more than
a hundred papers on phonetics and speech synthesis research and is a former board-member of the
board of the ISCA special interest group on speech synthesis (SynSIG).
Prof. Vainio has been a principal investigator in several Finnish Academy and privately funded projects
and has taken part in international projects with the Simple4All synthesis project coordinated by
the University of Edinburgh being the latest. His most important research links are with Aalto
University and the Center of Speech Technology Research at the University of Edinburgh.
Benjamin K Dichter - University of California at San Francisco/USA
Using ECoG to understand the representation of vocal pitch in humans
The flexible control of vocal pitch during speech production and the perception of pitch during listening are
fundamental aspects of human oral communication. Intonation patterns created by changing vocal pitch are a rich
source of information for conveying and understanding meaning. Still, it is currently unknown how the brain generates
the complex laryngeal motor commands that allow for prosody or how listeners are able to perceive it. Here, we used
direct high-density cortical recordings from the human brain to determine the encoding mechanisms of vocal pitch
control during natural speech production. We found neural activity at electrodes over the dorsal laryngeal motor
cortex (dLMC) that was highly selective to vocal pitch encoding, but not for other features in speech articulation.
Using a computational model of vocal pitch contours, we found that neural activity at a subset of dLMC electrodes was
selective for producing pitch accents, but distinct from those that encoded voicing. The same neural populations
showed similar pitch encoding in a non-speech singing task, suggesting a general control mechanism. Finally, we
confirmed the causal, feed-forward involvement of dLMC in pitch production by using direct cortical stimulation
to evoke laryngeal electromyographic responses and vocalizations. Together, these results reveal a neural basis
for the voluntary control of vocal pitch in human speech.
When listening to speech, humans understand prosodic patterns across a variety of sentences, speakers, and vocal ranges.
In order to investigate how prosodic patterns are perceived during listening, we synthesized sentences that varied across
speaker gender, emphasis pattern, and syllabic content. In the superior temporal gyrus, interspersed among
electrodes representing syllabic/phonemic information, we found electrodes that encoded absolute pitch
as well as electrodes that encoded relative pitch. Most electrodes were remarkably specific in which of the
three features it encoded. The separation of representation provides a key insight into how the human brain
processes speech prosody.
Dr. Benjamin K Dichter studied under Dr. Edward Chang at UCSF, using electrocorticography in humans to uncover the neural mechanisms of speech production and perception. He is an NSF GRFP fellow, and received his PhD from the UC Berkeley - UCSF Joint Program in Bioengineering in 2017. Dr. Dichter recently joined the lab of Dr. Ivan Sotlesz at Stanford as a data scientist.
Frank Kügler - University of Cologne
Intonation in tone languages
In this talk I will focus on the issue of how much and which kind of intonation is found in a tone language. Phonological tone has F0 as its phonetic correlate to manifest lexical and/or grammatical distinctions in a tone language. F0, among duration and intensity, is also the most important correlate of intonation. So the general question is how much F0 is 'left over' in a tone language to signal pragmatic meaning usually expressed by intonation in a non-tone language. I will discuss the prosodic expression of focus and the prosodic expression of sentence mode as two core areas of pragmatic meaning conveyed by intonation.
Frank Kügler holds currently a Heisenberg research fellowship from the German Research Foundation (DFG).
He is affiliated with the Phonetics Institute at Cologne University, Germany. He received his Ph.D. degree
from Potsdam University, Germany, in 2005. He then continued at Potsdam University as a postdoc and assistant
professor. In 2017, he replaced the chair of phonology at Goethe University of Frankfurt, Germany.
His research is primarily on prosody, studying tone and intonation from a cross-linguistic and typological
perspective. Among others he investigates different linguistic factors such as syntax and information structure
that influence the tonal realization and the phonological representation of tones in tone languages and non-tone
(intonation) languages. He worked on typologically diverse languages such as German, Czech, Swedish, Hindi,
Yucatec Maya, Mandarin Chinese, Akan, and Tswana.
Nancy C. Kuhla - University of Essex/UK
Tonal patterns and phonological processing in Bantu
The talk will investigate High tone spreading patterns in two dialects of Bemba (central Bantu) contrasting differences in bounded and unbounded spreading. Two contrasting patterns of binary versus ternary spread will be illustrated, focusing on how these interact with the OCP and downstep. The talk further presents results of a perception experiment that investigates the perceptual robustness of the tonal patterns discussed to test whether speakers can distinguish binary, ternary and unbounded spreading, shedding some light on the status of derived tones in contrast to lexical tones..
Nancy C. Kula is professor of Linguistics in the department of Language and Linguistics at the University of Essex. She did her PhD at the University of Leiden, The Netherlands on the phonology of Bemba. Her research focuses on Bantu Languages where she primarily works on phonology and the phonology-syntax interface, but has also worked on various aspects of Bantu morphosyntax. She has published on a range of topics in various international journals and is co-editor of the Bloomsbury Companion to Phonology (2013).