Invited Speakers

Aivars Glaznieks
(Eurac Research, Italy)
Think Global, Write Local – Patterns of Writing Dialect on SNS
Social Net­work Sites (SNS) claim that they are "on a mis­sion to con­nect the world". They facil­it­ate com­mu­nic­a­tion among people wherever they are loc­ated. Con­sequently, many users of SNS com­mu­nic­ate with a broad and het­ero­genic group of friends on dif­fer­ent occa­sions and thereby express vari­ous aspects of their iden­tit­ies (such as gender, age, eth­nic back­ground etc.). One aspect may also be a local identity.
Users of SNS can show their local iden­tity lin­guist­ic­ally by using a regional vari­ety. Some­times, the use of single region­ally marked words or sporadic regi­olectal spellings are suf­fi­cient to identify the regional back­ground of the writer; in other cases entire text mes­sages and con­ver­sa­tions appear in dia­lectal spellings mean­ing that the dia­lect appears as the main vari­ety of the con­ver­sa­tion. The extent of dia­lect use in com­puter­-­me­di­ated com­mu­nic­a­tion (CMC) may depend on vari­ous factors such as the indi­vidual dia­lect skills, the vivid­ness and prestige of the respect­ive dia­lect in the com­munity, emo­tional involve­ment in the given top­ic, age, gender, the inten­ded recip­i­ent, and other factors prob­ably inter­act­ing with each other.
The use of regional dia­lects in writ­ten CMC is one reason (amongst oth­ers) why lan­guage in CMC often dif­fers from the respect­ive stand­ard lan­guages. Since no ortho­graphic rules are usu­ally avail­able for writ­ing in dia­lect, it is up to the users to rep­res­ent their dia­lect in a proper but read­able and com­pre­hens­ible way. Users have to con­struct their regi­olectal lan­guage vari­ety on the basis of the ortho­graphy of the respect­ive stand­ard lan­guage, which usu­ally allows also for vari­ation. One reason for this may be vari­ous adequate pos­sib­il­it­ies to rep­res­ent a dia­lect word within a given writ­ing sys­tem (e.g. Ger­man). Another reason may be the (some­times very slight) phon­etic dif­fer­ences between region­ally close dia­lects that writers want (or do not want) to turn up in the dia­lect respelling. There­fore, dia­lect respellings are not always coher­ent (neither with respect to a group of dia­lect speak­ers nor with respect to indi­vidual writers) but usu­ally appear in vari­ous forms. However, uni­fic­a­tions of respellings in CMC are described for pidgin lan­guages  and also occur in dia­lectal CMC.
Over the last dec­ade, research­ers star­ted to com­pile cor­pora con­tain­ing dif­fer­ent genres of CMC. Such CMC cor­pora enable a sys­tem­atic ana­lysis of the way dia­lect fea­tures are reflec­ted in writ­ten com­mu­nic­a­tion. In my talk, I will focus on pat­terns of the regional dia­lect(s) in the DiDi Corpus, a col­lec­tion of Face­book mes­sages from around 100 South Tyr­olean writers. I will provide examples of regional fea­tures, ana­lyse the dis­tri­bu­tion of such fea­tures, and dis­cuss chal­lenges of identi­fy­ing local writ­ings on SNS.

A. Seza Doğruöz
(Independent Researcher)
Small vs. Big Data in Language Research: Challenges and Opportunities
Mobile com­mu­nic­a­tion tools and plat­forms provide vari­ous oppor­tun­it­ies for users to inter­act over social media. With the recent devel­op­ments in com­pu­ta­tional research and machine learn­ing, it has become pos­sible to ana­lyze large chunks of lan­guage related data auto­mat­ic­ally and fast. However, these tools are not read­ily avail­able to handle data in all lan­guages and there are also chal­lenges hand­ling social media data. Even when these issues are resolved, ask­ing the right research ques­tion to the right set and amount of data becomes cru­cially important.
Both qual­it­at­ive and quant­it­at­ive meth­ods have attrac­ted respect­able research­ers in lan­guage related areas of research. When tack­ling sim­ilar research prob­lems, there is need for both top-­down and bot­tom-up data-­based approaches to reach a solu­tion. Some­times, this solu­tion is hid­den under an in-depth ana­lysis of a small data set and some­times it is revealed only through ana­lyz­ing and exper­i­ment­ing with large amounts of data. However, in most cases, there is need for link­ing the find­ings of small data sets to under­stand the big­ger pic­ture revealed through pat­terns in large sets.
Hav­ing worked with both small and large lan­guage related data in vari­ous forms, I will com­pare pros and cons of work­ing with both types of data across media and con­texts and share my own exper­i­ences with high­lights and lowlights.


Michael Beißwenger, Ciara R. Wigham, Carole Etienne, Darja Fišer, Holger Grumt Suárez, Laura Herzberg, Erhard Hinrichs, Tobias Horsmann, Natali Karlova-Bourbonus, Lothar Lemnitzer, Julien Longhi, Harald Lüngen, Lydia-Mai Ho-Dac, Christophe Parisse, Céline Poudat, Thomas Schmidt, Egon Stemle, Angelika Storrer and Torsten Zesch
Connecting Resources: Which Issues Have to be Solved to Integrate CMC Corpora from Heterogeneous Sources and for Different Languages?
The paper reports on the res­ults of a sci­entific col­loquium ded­ic­ated to the cre­ation of stand­ards and best prac­tices which are needed to facil­it­ate the integ­ra­tion of lan­guage resources for CMC stem­ming from dif­fer­ent ori­gins and the lin­guistic ana­lysis of CMC phe­nom­ena in dif­fer­ent lan­guages and gen­res. The key issue to be solved is that of inter­op­er­ab­il­ity – with respect to the struc­tural rep­res­ent­a­tion of CMC gen­res, lin­guistic annota­tions metadata, and anonym­iz­a­tion/pseud­onym­iz­a­tion schem­as. The object­ive of the paper is to con­vince more pro­jects to par­take in a dis­cus­sion about stand­ards for CMC cor­pora and for the cre­ation of a CMC cor­pus infra­struc­ture across lan­guages and gen­res. In view of the broad range of cor­pus pro­jects which are cur­rently under­way all over Europe, there is a great win­dow of oppor­tun­ity for the cre­ation of stand­ards in a bot­tom-up approach.

Steven Coats
European Language Ecology and Bilingualism with English on Twitter
Soci­etal and demo­graphic changes have con­trib­uted to increas­ing bi- and mul­ti­lin­gual­ism in European coun­tries in recent years, and com­mu­nic­a­tion on social media plat­forms such as Twit­ter reflects this lin­guistic diversity. While high rates of Eng­lish use online have been attested for many European coun­tries by sur­vey research, rel­at­ively little work has quan­ti­fied the extent to which Eng­lish is used on social media in European con­texts. In this study, Eng­lish use and bilin­gual­ism with Eng­lish in Europe are invest­ig­ated on Twit­ter.
A large cor­pus of Twit­ter mes­sages with geo­graph­ical metadata was cre­ated by access­ing the Twit­ter APIs. After lan­guage detec­tion and fil­ter­ing, lin­guistic pro­files for European coun­tries were determ­ined and the beha­vior of bi- and mut­lilin­gual users examined. The ana­lysis sup­ports some pre­vi­ous find­ings that sug­gest that a large-s­cale lan­guage shift towards Eng­lish may be ongo­ing in Europe. Geo­graph­ical dif­fer­ences shed light on the dynam­ics of this process.

Darja Fišer
CLARIN Survey of CMC Resources and Tools
Since it is the mis­sion of CLARIN to cre­ate and main­tain an infra­struc­ture to sup­port the shar­ing, use and sus­tain­ab­il­ity of lan­guage data and tools for research­ers in Digital Human­it­ies and Social Sci­ences, it is our goal to have a good over­view of the avail­able resources and tools, to offer sup­port to their developers to over­come the tech­nic­al, legal and eth­ical obstacles and deposit them to the CLARIN infra­struc­ture, as well as to the research­ers with diverse back­grounds, such as lin­guist­ics, media stud­ies, psy­cho­logy etc., but also to inter­ested parties from the edu­ca­tion­al, com­mer­cial, polit­ic­al, med­ical and legal sec­tors of the soci­ety who are inter­ested in using them.
The first step in this dir­ec­tion was an inter­dis­cip­lin­ary workshop on the cre­ation and use of social media which was organ­ized within the Hori­zon 2020 CLAR­IN-PLUS pro­ject on 18 and 19 May 2017 in Kaun­as, Lithuania. The aims of the work­shop were to demon­strate the pos­sib­il­it­ies of social media resources and nat­ural lan­guage pro­cessing tools for research­ers with a diverse research back­ground and an interest in empir­ical research of lan­guage and social prac­tices in com­puter­-­me­di­ated com­mu­nic­a­tion, to pro­mote inter­dis­cip­lin­ary cooper­a­tion pos­sib­il­it­ies, and to ini­ti­ate a dis­cus­sion on the vari­ous approaches to social media data col­lec­tion and pro­cessing.

Laura Herzberg and Angelika Storrer
Investigating Interaction Signs across Genres, Modes and Languages: The Example of OKAY
The paper presents res­ults of a case study that com­pared the usage of OKAY across genre types (Wiki­pe­dia art­icles vs. talk pages), across media (spoken vs. writ­ten inter­ac­tion), and across lan­guages (Ger­man vs. French CMC data from Wiki­pe­dia talk pages). The cross-­genre study builds on the res­ults of Herzberg 2016, who com­pared the usage of OKAY in Ger­man Wiki­pe­dia art­icles with its usage in Wiki­pe­dia talk pages. These res­ults also form the basis for com­par­ing the CMC genre of Wiki­pe­dia talk pages with occur­rences of OKAY in the Ger­man spoken lan­guage cor­pus FOLK. Finally, we com­pared the res­ults on the usage of OKAY in Ger­man Wiki­pe­dia talk pages with the usage of OKAY in French Wiki­pe­dia talk pages. With our case study, we want to demon­strate that it is worth­while to invest­ig­ate inter­ac­tion signs across genres and lan­guages, and to com­pare the usage in writ­ten CMC with the usage in spoken inter­ac­tion.

Lisa Hilte, Reinhild Vandekerckhove and Walter Daelemans
Modeling Non-Standard Language Use in Adolescents' CMC: The Impact and Interaction of Age, Gender and Education
The present paper deals with Flem­ish adoles­cents' informal com­puter­-­me­di­ated com­mu­nic­a­tion (CMC) in a large cor­pus (2.9 mil­lion tokens) of chat con­ver­sa­tions. We ana­lyze devi­ations from writ­ten stand­ard Dutch and pos­sible cor­rel­a­tions with the teen­agers' gender, age and edu­ca­tional track. The concept of non-stand­ardness is oper­a­tion­al­ized by means of a wide range of fea­tures that serve dif­fer­ent pur­poses, related to the chat­speak max­ims of oral­ity, brev­ity and express­ive­ness. It will be demon­strated how the dif­fer­ent social vari­ables impact on non-stand­ard writ­ing, and, more import­antly, how they inter­act with each oth­er. While the find­ings for age and edu­ca­tion cor­res­pond to our expect­a­tions (more non-stand­ard mark­ers are used by younger adoles­cents and stu­dents in prac­tice-ori­ented edu­ca­tional track­s), the res­ults for gender (no sig­ni­fic­ant dif­fer­ence between girls and boys) do not: they call for a more fine-­grained ana­lysis of non-stand­ard writ­ing, in which fea­tures relat­ing to dif­fer­ent chat prin­ciples are examined separately.

Tobias Horsmann, Michael Beißwenger and Torsten Zesch
Reliable Part-of-Speech Tagging of Low-frequency Phenomena in the Social Media Domain
The object­ive is to imple­ment a tag­ger that tags this phe­nomenon with a high degree of cor­rect­ness in order to be able to use it as a cor­pus query tool on plain text cor­pora, so that new instances of this phe­nomenon can be eas­ily found in plain text.
We focused on avoid­ing manual annota­tion as much as pos­sible and exper­i­mented with alter­ing the fre­quency weight of the PoS tag of interest in the small train­ing data set we have.

Taja Kuzman and Darja Fišer
Corpus-Based Analysis of Demonyms in Slovene Twitter
This paper reports on a cor­pus-­based ana­lysis of demonym men­tions in the cor­pus of Slov­ene tweets. First, we ana­lyze the fre­quency of men­tions for the demonyms for the inhab­it­ants of the European and G8 countries. Then, we focus on the rep­res­ent­a­tion of demonyms for res­id­ents of Slov­e­ni­a’s neigh­bor­ing coun­tries: Aus­tria, Ita­ly, Hun­gary and Croa­tia. The main topic of the tweets men­tion­ing Croa­tians, Aus­tri­ans and Itali­ans is sport, whereas Hun­gari­ans occur most often in rela­tion to the Euro­vi­sion. Some eco­nomic and polit­ical issues are also rep­res­en­ted, such as the selling of Slov­ene com­pan­ies to for­eign firms, the refugee crisis and the arbit­ra­tion pro­ced­ure between Slov­e­nia and Croa­tia. A col­loc­a­tion ana­lysis revealed a highly ste­reo­typ­ical treat­ment of the neigh­bor­ing nations and hos­til­ity of some Slov­ene Twit­ter users to inhab­it­ants of Slov­e­ni­a’s neigh­bor­ing countries.

Paola Leone
Developing a protocol for collecting data in Higher Education: assessing natural language metadata for a Databank of Oral Teletandem Interactions
The cur­rent study addresses the defin­i­tion of a pro­tocol for col­lect­ing, stor­ing data and describ­ing (in a simple and gen­eric way) a databank. Par­tic­u­larly, the trans­par­ency of a form aimed at gath­er­ing inform­a­tion about the ped­ago­gical con­text of oral tele­col­lab­or­a­tion for lan­guage learn­ing named Teletan­dem (TT; Tell­es, 2006) will be tested before it is spread more widely. To uncover prob­lems in sub­mit­ting inform­a­tion, data-in­put-trig­gers qual­ity and reli­ab­il­ity have been tested inter­view­ing pro­fess­ors and lan­guage instruct­ors who will be involved in a pre­lim­in­ary phase of Teletan­dem cor­pus imple­ment­a­tion. Gen­eral goals of the study are to enlarge the research group, to increase data and to improve effi­ciency in data collection.

Julien Longhi, Claudia Marinica, Nader Hassine, Abdulhafiz Alkhouli and Boris Borzic
The #Idéo2017 Platform
The #Idéo2017 plat­form allows cit­izens to ana­lyze the tweets of the 11 can­did­ates at the French 2017 Pres­id­en­tial Elec­tion. #Idéo2017 pro­cesses the mes­sages of the can­did­ates by cre­at­ing a cor­pus in almost real time. By using tech­niques from lin­guist­ics sup­plied with tools, #Idéo2017 is able to provide the main char­ac­ter­ist­ics of the cor­pus and of the employ­ment of the polit­ical lex­icon, and allows com­par­is­ons between the dif­fer­ent can­did­ates.

Harald Lüngen, Michael Beißwenger, Laura Herzberg and Cathrin Pichler
Anonymisation of the Dortmund Chat Corpus 2.1
As a con­sequence of a recent cur­a­tion pro­ject, the Dortmund Chat Cor­pus is avail­able in CLAR­IN-D research infra­struc­tures for down­load and query­ing. In a legal expert­ise it had been recom­men­ded that stand­ard meas­ures of anonymisa­tion be applied to the cor­pus before it could be repub­lished. This paper reports about the anonymisa­tion cam­paign that was con­duc­ted for the cor­pus. Anonymisa­tion has been real­ised as cat­egor­isa­tion, and the tax­onomy of anonymisa­tion cat­egor­ies applied is intro­duced and the method of apply­ing it to the TEI files is demon­strated. The res­ults of the anonymisa­tion cam­paign as well as issues of qual­ity man­age­ment are dis­cussed. Finally, pseud­onymisa­tion as an altern­at­ive to cat­egor­isa­tion is dis­cussed in gen­eral as a method of the anonymisa­tion of CMC data, as well as pos­sib­il­it­ies of a (par­tial) auto­mat­isa­tion of the process.

Márton Petykó
“You’re trolling because…” – A Corpus-based Study of Perceived Trolling and Motive Attribution in the Comment Threads of Three British Political Blogs
This paper invest­ig­ates the lin­guist­ic­ally marked motives that par­ti­cipants attrib­ute to those they call trolls in 991 com­ment threads of three Brit­ish polit­ical blogs. The study is con­cerned with how these motives affect the dis­curs­ive con­struc­tion of trolling and trolls. Another goal of the paper is to exam­ine whether the mainly emo­tional motives ascribed to trolls in the aca­demic lit­er­at­ure cor­res­pond with those that the par­ti­cipants attrib­ute to the alleged trolls in the ana­lysed threads. The paper iden­ti­fies five broad motives ascribed to trolls: emo­tional/mental health-re­lated/so­cial reas­ons, fin­an­cial gain, polit­ical beliefs, being employed by a polit­ical body, and unspe­cified polit­ical affil­i­ation. It also points out that depend­ing on these motives, trolling and trolls are con­struc­ted in vari­ous ways. Finally, the study argues that par­ti­cipants attrib­ute motives to trolls not only to explain their beha­viour but also to insult them.

Damjan Popič and Darja Fišer
Fear and Loathing on Twitter: Attitudes towards Language
The paper deals with the soci­o­lin­guistic concept of prestige imbued in the notion of stand­ard lan­guage, and the social status con­nec­ted to the inher­ent lan­guage skill (or lack there­of). To this end, we ana­lyse Slov­e­nian tweets per­tain­ing to lan­guage use and the (in-)­cor­rect­ness of other users’ use of lan­guage, pro­pose a typo­logy, espe­cially in cases where lan­guage use is used as an argu­ment against someone’s qual­i­fic­a­tions or beliefs.

Muhammad Shakir and Dagmar Deuber
A Comparative Study of Computer-mediated and Spoken Conversations from Pakistani and U.S. English using Multidimensional Analysis
Present study com­pares four com­puter­-­me­di­ated con­ver­sa­tional registers – com­ments, FB groups, FB status and tweets – and spoken con­ver­sa­tions from Pakistani and US Eng­lish using Biber's Mul­ti­di­men­sional Ana­lysis frame­work on three dimen­sions of vari­ation, i.e. (i) Inter­act­ive versus Descript­ive Explan­at­ory Dis­course, (ii) Expres­sion of Stance, and (iii) Inform­a­tional Focus versus 1st Per­son Nar­rat­ive. Spoken con­ver­sa­tions have a high score on dimen­sion 2, while CM con­ver­sa­tions show register and regional vari­ation on dimen­sion 1 and 3. FB groups are sig­ni­fic­antly dif­fer­ent in both regional vari­et­ies, fol­lowed by FB status, com­ments and tweets. Pakistani FB groups dis­cuss self-help related top­ics, and appear to be slightly inter­act­ive and highly inform­a­tion­al, while the US ones are inter­act­ive and nar­rat­ive dis­cussing com­munity and polit­ical issues. Pakistani FB status and tweets use Eng­lish mainly for inform­a­tional pur­poses, while the US coun­ter­parts have inter­act­ive and per­sonal ori­ent­a­tion indic­at­ing a wider func­tional role of Eng­lish.

Stefania Spina
Emoticons as multifunctional and pragmatic Resources: a corpus-based Study on Twitter
The aim of this study is to invest­ig­ate the use of emoticons in Twit­ter by Italian users, and to veri­fy, by rely­ing on cor­pus data and on stat­ist­ical meth­od­o­lo­gies, some of the pre­vail­ing opin­ions on the use of emoticons: that they are tech­nic­ally-driven resources, that they are mostly used by young people, and more often by females, and that they are super­fi­cial and easy ways of express­ing emo­tions using images instead of words.
A mixed-ef­fects model ana­lysis has shown that the use of emoticons on Twit­ter is affected by a com­plex inter­ac­tion of cul­tur­al, tech­no­lo­gic­al, situ­ational and soci­o­lin­guistic variables.

Lieke Verheijen and Wilbert Spooren
The Impact of WhatsApp on Dutch Youths’ School Writing
Today’s youths are con­tinu­ously engaged with social media. The informal lan­guage they use in com­puter­-­me­di­ated com­mu­nic­a­tion (CMC) often devi­ates from spelling and gram­mar rules of the stand­ard lan­guage. There­fore, par­ents and teach­ers fear that social media have a neg­at­ive impact on youths’ lit­er­acy skills. This paper exam­ines whether such wor­ries are jus­ti­fi­able. An exper­i­mental study was con­duc­ted with 500 Dutch youths of dif­fer­ent edu­ca­tional levels and age groups, to find out if social media affect their pro­duct­ive or per­cept­ive writ­ing skills. We meas­ured whether chat­ting via What­s­App dir­ectly impacts the writ­ing qual­ity of Dutch youths’ nar­rat­ives or their abil­ity to detect ‘spelling errors’ (de­vi­ations from Stand­ard Dutch) in gram­mat­ic­al­ity judge­ment tasks. The use of What­s­App turned out to have no short-term effects on par­ti­cipants’ per­form­ances on either of the writ­ing tasks. Thus, the present study gives no cause for great con­cern about any impact of What­s­App on youths’ school writ­ing.


Michael Beißwenger, Marcel Fladrich, Wolfgang Imo and Evelyn Ziegler
MoCoDa 2.0: Creating a Database and Web Frontend for the Repeated Collec-tion of Mobile Communication (WhatsApp, SMS & Co)
The paper reports about an ongo­ing pro­ject in which we cre­ate a data­base and web fron­tend for the repeated col­lec­tion of CMC inter­ac­tions from smart­phone apps. The col­lec­tion strategy and the design of the col­lec­tion pro­cess is focused on involving the users of mobile phone apps such as What­s­App, Threema and oth­ers in the cre­ation of a data­base with detailed metadata which may serve as a resource not only for quant­it­at­ive but also for qual­it­at­ive approaches. For rep­res­ent­a­tion and annota­tion of the data the pro­ject builds on best prac­tices from pre­vi­ous pro­jects in the field. The cor­pus will be made avail­able for online query­ing and will be suc­cess­ively integ­rated into the Ger­man Ref­er­ence Cor­pus DEREKO at the Insti­tute for the Ger­man Lan­guage in Mannheim.

Leo Muelle
A corpus-driven approach to pseudonymity and the expression of emotions in Computer Mediated Communication in French and English
The pur­pose of the study is to observe the influ­ence of pseud­onym­ity on the use of these fea­tures. Indeed, pseud­onym­ity is a new notion that appeared with CMC, linked to digital iden­tity, which plays an import­ant role in the way emo­tions are expressed from a socio-­lin­guistic point of view. Like François Perea (2010) did, Fanny Georges (2009) divided the digital iden­tity in three: declar­at­ive iden­tity, effect­ive iden­tity and cal­cu­lated iden­tity. The declar­at­ive iden­tity, being what inform­a­tion the user chooses to provide on CMC, is rep­res­en­ted by the pseud­onym (Bechar‐Is­raeli, 1995). On the con­trary, the effect­ive iden­tity will be con­struc­ted by the actions the user will do on CMC. For this study, the effect­ive iden­tity con­cerns everything a user writes and shares with other users, which will add pieces of inform­a­tion on his per­son­al­ity and iden­tity. However, in con­texts where users meet for the first time, this effect­ive iden­tity is not developed yet and users can only rely on their declar­at­ive iden­tity: the pseud­onym.
Two main ques­tions emerged from these obser­va­tions: is pseud­onym­ity a bar­rier for express­ing emo­tions because of sus­pi­cion and the lack of inform­a­tion on the other users? Or, on the con­trary: is pseud­onym­ity a boost for express­ing emo­tions in Com­puter Medi­ated Com­mu­nic­a­tions by help­ing to cope with shyness?

Daniel Pfurtscheller
Public Service News on Facebook: Exploring Journalistic Usage Patterns and Reaction Data
The social media data are explored using stat­ist­ical research meth­ods to identify and com­pare dif­fer­ent usage pat­terns and to visu­al­ise the reac­tions of Face­book users. This provides an over­view over the dif­fer­ent forms of con­tent (i.e. types of posts) and the basic com­mu­nic­at­ive prac­tices that can be observed in the con­text of the Face­book Pages (i.e. num­ber of com­ments, shares, likes and Reac­tion types). To allow deeper insights an explor­at­ory case­-study approach is used. Draw­ing upon media lin­guistic research the focus is on the micro level of the media texts and their mul­timodal design. The in-depth ana­lysis aims to char­ac­ter­ise dif­fer­ent forms of news report­ing via Face­book and looks at the dif­fer­ent usage of mul­timodal res­sources in the con­text of the Face­book posts and comments.
This com­bin­a­tion of qual­it­at­ive and quant­it­at­ive meth­ods should allow a bet­ter under­stand­ing of how Face­book is used as a means of news dis­tri­bu­tion by pub­lic ser­vice media pro­viders on a large scale and how tech­nical afford­ances shape the design of news con­tent and fol­low-up inter­ac­tions. This know­ledge is crit­ical for the dis­cus­sion of the emer­ging role of social media in the con­text of pub­lic opin­ion and polit­ical decision-­mak­ing. The poster presents the pro­ject as work in pro­gress and shows pre­lim­in­ary findings.

Simone Ueberwasser
The graphic realization of /l/-vocalization in Swiss German WhatsApp messages
/l/-­vocal­iz­a­tion is a fea­ture nor­mally found in a rather clear cut area in the west­ern part of Switzer­land. Its geo­graph­ical bound­ar­ies are well doc­u­mented as well as social influ­ences on the real­isa­tion of the vari­ants. This study, based on a cor­pus of authen­tic What­s­App mes­sages, takes another approach by doc­u­ment­ing isol­ated forms of /l/-­vocal­iz­a­tion out­side the area tra­di­tion­ally attrib­uted to the fea­ture.