An opportunity to immortalize yourself

« previous post | next post »

Below is a guest post by Andrew Caines:


There's been growing interest in recent years in crowdsourcing as a means of data collection: for example, asking workers on Amazon Mechanical Turk to rate sentences for grammaticality, implicatures, sentiment, etc. As part of a special session for this year's INTERSPEECH Conference on innovative uses of crowdsourcing, we're building a crowdsourced spoken corpus of English and German.

The recordings come in via Crowdee, a new service dedicated to mobile crowdsourcing, and are then transcribed by CrowdFlower workers, before we add some automatically-generated annotations such as grammatical relations and phone alignments. We'll present and publish a freely-available paper about our full method at the conference.

For now, we're looking to fill up our corpus with 10 more German/English bilinguals and 50 more native speakers of English. It's first come first served: once the places are taken, that's it! But if you still wish to contribute, please let us know, as we can then make a case for more funding and make the corpus even bigger.

To contribute, you'll need a mobile or tablet with Android OS, approximately 10 spare minutes, and a fairly quiet environment. Please download Crowdee for free from Google Play, then search for the English or German 'speaking task'. Note that Crowdee pays out €5 minimum, so you'll need to do both English tasks (or the Eng/Ger bilingual task alone), or one of the many other tasks offered by Crowdee.

Finally, I should add that the corpus will be made freely available for research purposes, so any contributions will not only help our particular project (on language learning), but also the linguistics and speech community more widely. If you'd like to be kept informed, please get in touch.

You can find out a little more about me, including contact information, here.


Above is a guest post by Andrew Caines.

I need to note that your immortality will be anonymous. But still.



7 Comments

  1. Jerry Friedma said,

    May 13, 2015 @ 9:06 am

    How is this going to work? Will the volunteers be wearing mikes?

  2. Jerry Friedman said,

    May 13, 2015 @ 9:07 am

    Was that LLOG's first typo in a commenter's name? (And why did I have to start re-entering my name and e-mail address every time? Did that happen to everyone?)

  3. Rodger CC said,

    May 13, 2015 @ 10:56 am

    It's certainly been happening to me.

  4. Brett said,

    May 13, 2015 @ 11:04 am

    I too have to enter my name every time I comment now, but it's easy enough, since the autocomplete remembers what I've typed in the boxes before, even if the browser is no longer storing the entries explicitly.

  5. Jerry Friedman said,

    May 13, 2015 @ 3:44 pm

    Thanks, Rodger CC and Brett. Sounds like it was a decision made for good reasons in the boardroom at Language Log Plaza. (By the way, there's no autocomplete on my computer at work, which would be relevant if, hypothetically, I commented while at work.)

  6. Rubrick said,

    May 13, 2015 @ 4:00 pm

    I believe the "need to reenter info every time" issue cropped up following a server outage/issue a month or two ago. I emailed Mark informing him of the problem at the time.

  7. Alan Palmer said,

    May 14, 2015 @ 3:41 am

    I'm glad to learn the ' re-enter info every time' issue is a server one. There I was blaming my own computer. Since autocomplete isn't available on this computer at work it's a (minor) nuisance.

    In the interests of getting a decent cross-section of users, how about Crowdee developing an app for iOS as well as Android so iPhone and iPad users can join in?

RSS feed for comments on this post