Robert Eklund’s Linguistics Research Page

My main research activities over the years are listed and described below. Open, still ongoing, activities are listed first where as no longer active projects are listed below. Within these two main categories all projects are listed according to start year for the activities, with the most recent ones listed first (i.e. “First In, Last Out”, LIFO). Results and published work can be found Publications page.

01 – Overtone singing

Med start år 2019 studerar jag tillsammans med Anna-Maria Hefele, världsberömd ”polyfonisk” övertonssångerska, och Anita McAllister (se Kulning) övertonss˚ng, dels akustiskt, dels hur man kan arbeta med övertonssång som ett pedagogiskt verktyg inom fonetik- och akustisk analys-undervisning. Se artikel [98]. Se även:
https://anna-maria-hefele.com/
https://en.wikipedia.org/wiki/Anna-Maria_Hefele/
YouTube: Polyphonic Overtone;singing – explained visually
https://www.sygyt.com/en/

02 – Melody in human-cat communication (Meowsic)

In December 2015, Stiftelsen Marcus and Amalia Wallenbergs Minnesfond awarded 4 MSEK to our project proposal Melody in human–cat communication (Meowsic); see:
http://vr.humlab.lu.se/projects/meowsic

Dr Susanne Schötz of Lund University was the main force behind this application and is the project lead, with me and Dr Joost Van de Weijer, Lund University, as co-applicants. The project started up in September 2016, and has already received enormous media attention world wide; see paragraph 16 Media appearances and the Research page for more information, sound files of e.g. radio interviews etc. For more information on media coverage, project presentations and so on, see:
http://vr.humlab.lu.se/projects/meowsic/thanks

Media attention

Meowsic has received extreme media attention, world-wide. By June 2016 we (me and Susanne Schötz had done over thirty interviews which had resulted in over 100 articles in magazines, newspapers and websites. Just to mention a few of these, we have been covered by National Geographic, The Washington Post, The Guardian, BBC 5 Live, The Telegraph, Daily Mail, as well as by Swedish radio, TV and press, Norwegian radio, and also online all over the world, including Chile, China and Vietnam. However, we take special pride in being covered in a special issue “Cats (and Dogs)” of Annals of Improbable Research (2015, volume 21, number 3) – i.e. the people behind the Ig Nobel Prizes – where three of our previous papers are mentioned, and my own purring website is also referred to.

Below, a small (!) sample of selected media appearances are listed. More information about the project and the media coverage it has received is found on the project website:
http://meowsic.info
... and for an overview of media coverage, project presentations (etc):
http://meowsic.info/thanks

I appeared on the Swedish public broadcasting program Vetenskapsradion Forum (“Science Radio Forum”) on 23 August 2016, where the program host Urban Björstadius interviewed me about Meowsic and other felid vocalizations in general (this is also covered below, section 04). According to Urban Björstadius this show was one of the most popular and listened-to in the history of Vetenskapsradion Forum and it was re-broadcast several times. The show can be streamed here but can also be downloaded as an mp3 file below

Vetenskapsradion Forum, 23 August 2016. (23.3 MB, 24:26 minutes.)

[mp3]

Ig Nobel Prize

Susanne Schötz, Joost van de Weijer and myself received the 2021 Ig Nobel Prize in Biology for our work on cat vocalisations; see:
ttps://www.improbable.com/2021/09/09/announcing-the-2021-ig-nobel-prize-winners/
https://en.wikipedia.org/wiki/List_of_Ig_Nobel_Prize_winners#2021

03 – Kulning, Swedish cattle calls

Since 2012 I have started to study the acoustics of kulning (Swedish cattle calls). The work is a collaboration with the speech–language pathologist (SLP) Anita McAllister and the kulning singers Fanny Pehrson (also an SLP) and singer Kajsa Dahlström. During 2015 we augmented our studies to include Anne-Maria Laukkanen, University of Tampere, and Ahmed Geneid, Helsinki University. Together we have combined acoustic, EGG and high-speed photography studies of kulning, and in 2017 MRI studies were performed at Oulo University, Finland.

Radio interviews

Swedish Broadcasting’s Urban Björstadius interviewed me, Anita McAllister and Kajsa Dahlström about kulning for the science program Vetenskapsradion Forum, which was broadcast on 26 October 2016. The programme (in Swedish) can be streamed here, but an mp3 file can also be downloaded below.

Vetenskapsradion Forum, 26 October 2016. (23.3 MB, 24:30 minutes.)

[mp3]

04 – Cheetah and African Wild Dog vocalizations

In December 2013 I recorded wild cheetahs and African wild dogs at N/a’an ku sê Foundation in Namibia, see: http://naankuse.com

The project is officially sanctioned by the Namibian Government and is carried out collaboration with the (then) local head of research Dr Florian Weise (presently in Botswana). We are presently in the phase of organizing and classifying our collected data.

05 – Cheetah purring and other felid vocalizations

Since felid purring occurs on an alternating egressive and ingressive airstream, it is related to my research on ingressive speech and phonation, which led to several new projects and activities.

During the winter of 2009 I had the opportunity to make one of the best recordings ever of cheetah purring at the Dell Cheetah Centre in South Africa, and initiated collaboration with Dr Gustav Peters at Forschungsmuseum Alexander Koenig in Bonn, Germany, with a paper (in 2009) that also included cheetah breeder Elizabeth D. Duthie as co-author.

Cheetah recording session

Short excerpt from the recording session with the (constantly) purring cheetah Caine. Recorded by camera operator Elizabeth D. Duthie and microphone operator, linguist Robert Eklund. Recording date: 5 December, 2009, at the Dell Cheetah Centre, Parys, South Africa. Nota bene!, in case the video does not show up in your browser, here is the link:
https://www.youtube.com/watch?v=ZFvULxbN3NM/

06 – Ingressive speech and phonation

Since the end of 2000 I have been interested in ingressive speech, i.e. speech produced on a pulmonic ingressive (inhalation) airstream. So far, this research has resulted in a number of conference papers and a major journal review paper [49]. To summarize my findings, I show that – very much contrary to popular belief – ingressive speech is not “typically Swedish”, but is instead found all over the world, in taxonomically different languages, with basically the same (paralinguistic) linguistic function, and that it consequently might even constitute a neglected linguistic universal.

I also maintain a website devoted to ingressive speech, which is constantly updated with new information:
http://ingressivespeech.info/

I have also initiated research on pulmonic ingressive phonation in mammals, in collaboration with Dr Gustav Peters and primatologist Marina Davila-Ross at University of Portsmouth, United Kingdom. (This project, however, is on hiatus.)

Radio interview

I appeared on the Norwegian public broadcasting program Språkteigen on Sunday, 6 September 2009, where the program hostess Gøril Grov Sørdal interviewed me about ingressive speech, evidently believed by Norwegians to be a typical Norwegian phenomenon (see Swedish below). The interview can be downloaded below. (Nota bene! Gøril speaks Norwegian, and I speak Swedish; the languages are sort of mutually understandable.)

Språkteigen, 6 September 2009. (3.5 MB, 7:31 minutes.)

[mp3]

07 – Disfluency in spontaneous speech

My PhD work focused on the study of disfluencies – i.e. phenomena like hesitation sounds like “eh”, prolonged sounds, word fragments, pauses, recapitulations etc.) in spontaneous speech.

The studies were encouraged and funded by Telia Research and was partly carried out in order to obtain more robust speech and language models for improved speech recognition and understanding (see 06.15 Automatic speech recognition below).

Beginning in 1999 (at University of California at Berkeley) a series of international workshops devoted to non-pathological disfluency have been organized, the so-called DiSS workshops.

In September 2003, I organized DiSS 2003 in Göteborg (Sweden), with the help of Åsa Wengelin and Professor Jens Allwood.

In August 2013, I organized DiSS 2013 in Stockholm, in collaboration with Jens Edlund, Joakim Gustafson and Sofia Strömbergsson, all at the Royal Institute of Technology (KTH), Stockholm, Sweden.

In August 2017, I organized DiSS 2017 in Stockholm, in collaboration with Robin Lickley, Jens Edlund, Joakim Gustafson and Ralph Rose. Note that the DiSS 2017 workshop website contains the proceedings of all previous workshops (DiSS 2015 proceedings are still pending).

Mirror website found here:
http://roberteklund.info/conferences/diss2017/

Radio interviews

I appeared on the Norwegian public broadcasting program Språkteigen on Sunday, 22 November 2009, where the program hostess Gøril Grov Sørdal interviewed me about disfluency. The interview can be downloaded below. Note that Gøril speaks Norwegian, and I speak Swedish, two sort of mutually understandable languages.

Språkteigen, 22 November 2009. (5.3 MB, 11:25 minutes.)

[mp3]

I appeared on the Swedish public broadcasting program Språket on Tuesday, 1 June 2010, where the program hostess Anna-Lena Ringarp interviewed me about disfluency in general and filled pauses in particular. The interview can be downloaded below. The beep indicates where a cut-edit has been made. The interview was also on felid purring, and that part of the interview can be downloaded from my sister site at http://purring.org.

Språket, 1 June 2010. (3.3 MB, 3:31 minutes.)

[mp3]

08 – Musicology

Click Musicology above

I have intermittently returned to musicology since I commenced my studies within the field in 1984 (with a first publication in 1988). For the moment I am working on an invited journal paper based on my MA thesis. This work is carried out as a collaboration with internationally renowned lutenist Jakob Lindberg, who was also my lute teacher at the Royal College of Music in London (1986–1987).

09 – Felid larynx postmortems (2013–2016)

During the period 2013 through 2016 I collaborated with Dr Andrew Kitchener and Georg Hantke, National Museums Scotland, at Edinburgh. Together we worked on performing post mortems of the larynxes (and related other anatomy) of all species of felid available – which could eventually be all 37 (or so) species. The goal of this project is that by matching the physiology of the felids and their vocalizations, we hope to be able to explain the long-known difference between cats that either roar (tigers, lions, jaguars, leopards) or purr (all the others, possibly except the snow leopard). This project is now on hiatus.

10 – Lion roaring

I have also conducted research on lion roars, in collaboration with Dr Gustav Peters, Evans Mabiza then at Antelope Park in Gweru, Zimbabwe and Ananthakrishan at the Royal Institute of Technology (KTH), Stockholm, Sweden, with the kind assistance of Jennie Westander then at Parken Zoo in Eskilstuna, Sweden.

11 – Neurocognitive correlates of disfluency

During the period March 2006 through March 2012 I pursued research at Karolinska Institute/Stockholm Brain Institute (during the period March 2006 through April 2009 as a postdoc, and from May 2009 through March 2012 as an affiliated researcher), where I studied neurocognitive correlates of disfluency in the perception of natural speech. The work was supervised by Professor Martin Ingvar.

I received an fMRI ”scanning licence” (at Karolinska Institute) during the summer of 2006, and have also taken both the Basic and Advanced Course in Cognitive Neuroimaging at the F.C. Donders Centre for Cognitive Neuroimaging in Nijmegen, The Netherlands (during Spring and Summer of 2007).

Given legal problems with regard to data ownership (I left Telia in 2008) this study was very much delayed. but our results were finally officially published in 2016, in publication [81].

Although the work focused on fMRI scanning, but EEG studies were run in parallel with Francisco Lacerda and Iris-Corinna Schwarz at Stockholm University.

A few pictures of my own brain are shown below.

12 – Meeting speech analysis

During my postdoctoral studies at the International Computer Science Institute (ICSI) in Berkeley, California, I collaborated with Professor Rebecca Bates, Elizabeth Willingham and Chad Kuyper at  Mankato University, Minnesota, where we attempted an alternative approach to analysis of meetings which aimed at developing and test a taxonomy for “importance”, i.e. stretches of speech in meetings that labelers consider important without being given any detailed instructions or definitions. We explored three types of importance in meetings: content (topic related), social interaction, and meeting flow and asked whether labelers consistently annotated regions as belonging to both the same type of importance and the same level of importance, and compared the results of this importance labeling with group interaction labeling as carried out at Mankato University. The work was supported by a grant from the European Union AMI (Augmented Multiparty Interaction).

13 – Automatic call routing

After having finished my thesis, I worked on a pilot project on automatization of the TeliaSonera call center customer service (90 200), i.e. to allow incalling customers to use free speech to describe their business, rather than being greeted with a typical menu-based tree-structure (e.g., “For X, say Y”). After a continuation of the pilot project, a free-speech call center customer service was deployed in 2006. During the pilot, we developed a novel extension to the traditional Wizard-of-Oz methodology (to collect data), which included what we call a “Prompt Piano”, described in publication [47].

14 – Language typology (field work)

Starting as a hobby project, I have also studied the language spoken in the village of Lakurumau, New Ireland, Papua New Guinea. So far, I have visited the village twice, and I would really like to return in the future. As part of those activities, I also learnt and studied Tok Pisin, and collected data used in my disfluency studies. Thanks to a recent contact with Lidia Mazzitelli, who is presently studying this language, this activity is now revived.

15 – Xenophones

During the years 1996–2000, me and my colleague Anders Lindström did research on so called xenophones i.e. “foreign” speech sounds in Swedish, such as the “lisp” sound in English names/words like “Thatcher” (“dental fricatives” in phonetic jargon). These sounds are normally not included in descriptions of Swedish, but are nevertheless used by most Swedes in normal Swedish conversation and consequently need consideration in the development of both speech synthesizers and speech recognizers.

The research has been presented in a number of articles that describe to what degree Swedish speakers have incorporated such sounds into their Swedish phone inventory, and the observations have served as the basis for an expanded polyphone material for a Swedish speech synthesizer.

The term “xenophone” was introduced in 1998 (in publication [24]) and has gained ground and has been frequently used in papers dealing with the same phenomenon in other languages, and the term is now established, being included in a major, international encyclopedia, published by SAGE; see Publications [095].

For more information on our research, go to our Xenophones Page.

16 – Machine translation

Within the The Spoken Language Translator, I mainly worked with the development of a Swedish grammar using the formalism The Core Language Engine. I also worked with transfer rule writing between Swedish, English and French, as well as with evaluation of system-generated translations. New approaches to evaluation were developed within the project. The machine translation work was carried out in collaboration with SRI International, Cambridge, England.

17 – Automatic speech recognition (ASR)

Within the Spoken Language Translator project, I was responsible for the generation of phonetically and phonologically balanced training material for Swedish, as well as for deciding which Swedish dialects we needed to record/cover. The speech recognition work was carried out in collaboration with SRI International, Menlo Park, California, and served as the basis for the Swedish speech recognizer now offered by Nuance (www.nuance.com).

18 – Speech synthesis

During the period 1994–1995 I was hired by Telia Research to develop the first concatenative synthesizer for Swedish. The work included generation of all possible Swedish demisyllables (polyphones) and recording of these, as well as evaluation of the results. The research part focused primarily on what phonetic and phonological processes are crucial within the framework of concatenative synthesis.