[LOGOS] Virtual Seminar, Wed May 28, 10am -- Saying it or writing it on social media platforms: A comparison of written and audio messages on the Yellow Vests Telegram channel in Haute-Garonne

[Emiliano De Cristofaro]

Starting now

--
Sent from my mobile device


On Mon, May 26, 2025 at 12:37 PM Emiliano De Cristofaro <
emilianodc at cs.ucr.edu> wrote:

> Hi Everyone,
>
> The next LOGOS meetup will take place on Wednesday (May 28) at *10am PDT.*
>
> We will host a virtual seminar by Brigitte Sebbah and Pierre Ratinaud
> (from Toulouse Paul Sabatier and Jean Jaurès Universities) on Zoom. Please
> see details below.
>
> Cheers,
> Emiliano
>
> *ZOOM*
> https://ucr.zoom.us/j/92699528206?pwd=oo5ujmYE79Wqywca0CaaQBd1WAhUFz.1
> (Meeting ID: 926 9952 8206, Passcode: 076472)
>
> *TITLE*
> Saying it or writing it on social media platforms: A comparison of written
> and audio messages on the Yellow Vests Telegram channel in Haute-Garonne
>
> *ABSTRACT*
> Those involved in the Yellow Vests movement have used numerous
> communication channels, most of which come under the heading of
> socio-numerical networks (Facebook, Twitter, WhatsApp...). We gained access
> to the Telegram feed of the Haute-Garonne yellow vests and set about
> analyzing the discussions held there. Initially, we looked at the 50,000
> written messages, seeking to determine the different themes addressed and
> their temporalities of expression based on an analysis with the Reinert
> method (Sebbah & Ratinaud, 2023) in the IRaMuTeQ software (Ratinaud, 2020).
> The availability of the Whisper model (Radfort & al., 2022), which enables
> the automatic transcription of oral discourse, also enabled us to analyze
> the 2201 audio messages present on the discussion threads. After
> highlighting some of the limitations of the automatic transcription
> produced by Whisper, we will focus on the differences and similarities
> between text and audio messages. Our first remarks will concern the
> statistics associated with these corpora (average message size, word
> frequencies, etc.), then we'll compare the preferred themes and their
> chronology.  This experience of analyzing audio messages enables us to
> describe an approach which, despite its limitations, should help to broaden
> the practice of analyzing socio-numerical networks and the media in
> general, by simplifying and amplifying the use of textual statistics to
> analyze the discourse of radio, television and the countless videos put
> online every day.
>
> *BIOs*
> Please see attached
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ucr.edu/pipermail/logos/attachments/20250528/67a0d237/attachment.htm>