|
If birds could talk... would we notice?| old_uid | 15160 |
|---|
| title | If birds could talk... would we notice? |
|---|
| start_date | 2015/02/27 |
|---|
| schedule | 11h30-13h |
|---|
| online | no |
|---|
| location_info | Pavillon Pasteur, RdC, salle de réunion |
|---|
| details | Séance organisée conjointement avec le séminaire LINGUAE |
|---|
| summary | There is only one species we know of that uses language, so how could we expect to recognise language from another species if it exists? This problem is very real for those involved in the Search for Extra Terrestrial Intelligence (SETI), as well as for those investigating the evolution of language on Earth. Are there mathematical properties of language that are universal? If so, can these properties be inferred without any semantic knowledge of the language itself? I will examine some popular metrics such as entropy, and Zipf's law, that do not in themselves appear to provide a useful discrimination between language and non-language. I will also use artificially generate sequences, both from arbitrary context-free grammars, and from neural models of birdsong generation, to examine other ways to assess the possible linguistic nature of sequences from an unknown origin. |
|---|
| responsibles | Schlenker |
|---|
| |
|