BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Data-Free\, Universal Prior Distribution for Syntactic Structure
 s - Dr Fermin Moscoso del Prado Martin - Department of Computer Science an
 d Technology\, University of Cambridge
DTSTART:20240228T150500Z
DTEND:20240228T155500Z
UID:TALK208855@talks.cam.ac.uk
CONTACT:Ben Karniely
DESCRIPTION:Across linguistic theories\, human language structures are rep
 resented by graphs. Much research has focused on the linearisation of such
  graphs into actual sequences expressing utterances\, but less attention h
 as been paid to the shapes that the graphs themselves take: their topology
 . A current hypothesis from psycholinguistics argues that the structures i
 n human language are primarily shaped by the nature of human language prod
 uction processes. Utterances are planned in an incremental manner: success
 ively incorporating chunks –either single words\, phrases\, or even full
  clauses– into partial syntactic structures. Incremental construction sh
 ould constrain the plausible probability distributions of syntactic struct
 ures in predictable ways. I show that the topologies of actual syntactic g
 raphs exhibit the precise deviation from randomness that incremental const
 ruction predicts. This is a previously unknown universal regularity of hum
 an languages: Syntactic structures are constrained to a predictable topolo
 gical distribution –that generated by sublinear preferential attachment
 – constant for all 124 languages studied\, across language families and 
 modalities (spoken\, written\, and signed). It supports the hypothesis tha
 t syntactic structures are mainly shaped by language production. Furthermo
 re\, it demonstrates how the observed efficiency of languages might just b
 e epiphenomenal. Crucially\, this finding implicitly defines a data-free u
 niversal prior distribution for parse structures\, with possible applicati
 ons in language technologies.\n\nLink to join virtually: https://cam-ac-uk
 .zoom.us/j/81322468305\n\nA recording of this talk is available at the fol
 lowing link: https://www.cl.cam.ac.uk/seminars/wednesday/video/
LOCATION:Lecture Theatre 1\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR