Agenda

Speaker: Rodolfo Delmonte

The talk is organized into three parts: in the first part I will present the book about Francis Webb, the poet and the man with his terrible predicament in psychiatric hospitals. I will read two of his poems and comment extensively their poetic and rhetoric devices. The second part will start with a short presentation of quantitative data on Webb's Collected Poems, comparing its overall vocabulary with the most renown British and American poets. This will be followed by a more extended section dedicated to computational analysis. I will show visual representations of a number of poets generated by the system APSA (American Poetry Style Analyzer) which I will also discuss in some detail. Then I will dedicate more time to the system SPARSAR (System for Poetry Automatic Rhythm and Style Analyzer), created to improve on APSA deficiencies, that I presented for the first time at a conference in 2013. I will focus on the newly (2023) introduced concept of "Sound-Sense" Harmony as an inspiring principle for poetry, discussed at length in two papers that just appeared to be found here.(https://www.mdpi.com/2078-2489/14/10/576, and https://www.mdpi.com/2076-3417/14/24/11949)
The final part of the talk will be dedicated to AI and platforms like ChatGPT where text reading is also provided. I will show how the total lack of semantic knowledge and the non-existence of publicly available phonetically transcribed texts determines the inability of LLMs and derived platforms to read poetry in a semantically acceptable manner. This will be proven by playing speech files produced by best speech synthesis companies suggested by ChatGPT after trying unsuccessfully to read a short poem. This will be followed by a reading of the same poem made by SPARSAR available on its website, sparsar.wordpress.com.

Link: https://unive.zoom.us/j/81876672725