Differences between revisions 1 and 20 (spanning 19 versions)
Revision 1 as of 2016-01-11 19:27:24
Size: 465
Editor: 128
Comment:
Revision 20 as of 2016-07-07 18:33:52
Size: 5790
Editor: XinXie
Comment:
Deletions are marked like this. Additions are marked like this.
Line 10: Line 10:
 * The paper from Reinisch and Holt on generalization of s-f: Reinisch, E., & Holt, L. L. (2014). Lexically guided phonetic retuning of foreign-accented speech and its generalization. Journal of Experimental Psychology. Human Perception and Performance, 40(2), 539–55. doi:10.1037/a0034409. They claim that generalization happens when the talkers are "perceptually similar". [[attachment:reinisch-holt-2014.pdf]] -DK
 * If you're writing javascript for mturk experiments that depends on multiple libraries (or has code split over multiple files), I'd suggest using something like browserify or webpack. This, I now realize, has a few distinct advantages. -DK
    1. It makes dependencies explicit. Instead of putting a bunch of `<script src="library.js">` tags in an HTML file somewhere and then calling functions that are included therein, you `require('./library.js')` in the javascript files that actually ''depend'' on `library.js` (like `import`ing in Python or `library` in R).
    2. it lets you easily use any library that is available as a node package in your experiment code. e.g. if `purrr` inspired you want to write everything in a functional style, you can drop in [[http://ramdajs.com/0.21.0/index.html|`ramda`]]. Need a library to parse dates and times? Dropping in [[http://momentjs.com/|`moment.js`]] is just an `npm install --save moment` and `require('moment')` away.
    3. with a minimal amount of extra work (i.e., you have to run `browserify` to bundle everything up), you can get serious performance enhancement that's not possible by loading individual js files using `script` tags: the total javascript weight of my experiment went from 300KB to less than 300 '''bytes''' (''including'' JQuery and underscore). This is because if you have everything in one big bundle, there are good tools out there that automatically detect which bits your using and which your not, and then compress everything as much as possible. Not that performance is super critical, but still nice to get a 1000-fold reduction in size.
Line 11: Line 17:
 1. '''Edited''' and '''Submitted''' Masha's paper to ''Psych Science''. Shudder.
 1. '''Edited''' paper with Dan for submission to ''PNAS''.
 1. '''Edited''' abstracts for IWLP (3) and AMLaP (1?). Even wrote one myself ;-) .
 1. Read and evaluated more post-doc applications. Skype with more candidates.
 1. '''Feedback''' on paper with Juergen Bohnemeyer and Guillermo.
 1. '''Feedback''' to Darcy Rose on her plural pronunciation study.
 1. '''Feedback''' on study design for artificial language learning study by Scott Grimm and Chigusa (on plurality).
Line 13: Line 27:
 1. contacted ``LCN`` to gauge interest in introductory-style paper on ideal observer and ideal adapter approaches to speech perception
 1. prepared and gave cowplot tutorial in lab mtg
 1. prepared and gave lecture in Scott Grimm's data science for linguistics class
Line 15: Line 32:
 1. Tinkering around the edges of the ALL app to improve its look
 1. Started transitioning my SPR``Runner Flask app to use Blueprints on the way making it a general M``Turk experiment running framework, including adding an RESTful API with anonymization and where others can use the list balancing, etc. (Mostly just in planning at this point.)
 1. Cleaned up the Turk Payments Google spreadsheet and generated a report for Florian about how everyone is spending his money.
 1. Poked around trying to learn more about Red``Cap to see if I can reduce the Continuing Review load by improving our setup on it for easier subject count and demographic reports.
Line 20: Line 41:
 1. Went through another batch of annotations from RAs 😤.
 1. Did first pass analysis of follow-up to my JML paper with flo'tan. Looks like things are replicating 😌.
 1. Worked out some remaining issues with a pygaze study. Turns out something in pygaze/pygame/psychopy/??? does not like 32bit wav files but is ok with 16bit 😶.
 1. Met with Jill Warker, had a good chat about ideas for future work 😺.
 1. Submitted too many abstracts 😅.
 1. Finally did the first round of testing of autoVOT on mturk recordings. Not all that encouraging 😿.
Line 21: Line 49:

 1. Refactored my experiment code to use browserify for sane dependency management. (See above)
 1. Wrote a real simple node back end with list balancing using postgres for data storage.
 1. Worked out with Andrew how to deploy experiment on www in these modern times.
 1. Submitted an abstract on fMRI adaptation project to the Society for Neurobiology of Language meeting.
Line 29: Line 62:
 1. Spent a long time trying to understand "predict.merMod" in R and why it was bugging out on me.
 1. Wrote a more for subtitle project
 1. Coded up and demo'd the pupillometry project
Line 31: Line 68:
 1. More analysis for pen in mouth study.
 1. Ran additional experiment for pen in mouth study.
 1. Finished adding in mouse tracking to the Tower of Hanoi task. Solicited some helpful feedback from Susan Wagner-Cook.
 1. Put together tutorial on LateX for RA meeting.
 1. Noam Chomsky in the flesh with da booosh et al. :)
Line 35: Line 77:
 1. Started writing ditransitive definiteness paper. Writing is hard.
 1. Prepared & gave purrr tutorial in lab mtg
 1. Cleaned up simulation code some more for sharing with others. I feel like at this point I should just make it completely general and stick it on GitHub.
 1. Saw Chomsky in the flesh

Lab Meeting, Spring 2016, Week 15

Awe- and Aw-inspiring readings (or just stuff you think deserves a mention)

  • The paper from Reinisch and Holt on generalization of s-f: Reinisch, E., & Holt, L. L. (2014). Lexically guided phonetic retuning of foreign-accented speech and its generalization. Journal of Experimental Psychology. Human Perception and Performance, 40(2), 539–55. doi:10.1037/a0034409. They claim that generalization happens when the talkers are "perceptually similar". reinisch-holt-2014.pdf -DK

  • If you're writing javascript for mturk experiments that depends on multiple libraries (or has code split over multiple files), I'd suggest using something like browserify or webpack. This, I now realize, has a few distinct advantages. -DK
    1. It makes dependencies explicit. Instead of putting a bunch of <script src="library.js"> tags in an HTML file somewhere and then calling functions that are included therein, you require('./library.js') in the javascript files that actually depend on library.js (like importing in Python or library in R).

    2. it lets you easily use any library that is available as a node package in your experiment code. e.g. if purrr inspired you want to write everything in a functional style, you can drop in `ramda`. Need a library to parse dates and times? Dropping in `moment.js` is just an npm install --save moment and require('moment') away.

    3. with a minimal amount of extra work (i.e., you have to run browserify to bundle everything up), you can get serious performance enhancement that's not possible by loading individual js files using script tags: the total javascript weight of my experiment went from 300KB to less than 300 bytes (including JQuery and underscore). This is because if you have everything in one big bundle, there are good tools out there that automatically detect which bits your using and which your not, and then compress everything as much as possible. Not that performance is super critical, but still nice to get a 1000-fold reduction in size.

Florian

  1. Edited and Submitted Masha's paper to Psych Science. Shudder.

  2. Edited paper with Dan for submission to PNAS.

  3. Edited abstracts for IWLP (3) and AMLaP (1?). Even wrote one myself ;-) .

  4. Read and evaluated more post-doc applications. Skype with more candidates.
  5. Feedback on paper with Juergen Bohnemeyer and Guillermo.

  6. Feedback to Darcy Rose on her plural pronunciation study.

  7. Feedback on study design for artificial language learning study by Scott Grimm and Chigusa (on plurality).

Kodi

  1. contacted LCN to gauge interest in introductory-style paper on ideal observer and ideal adapter approaches to speech perception

  2. prepared and gave cowplot tutorial in lab mtg
  3. prepared and gave lecture in Scott Grimm's data science for linguistics class

Andrew

  1. Tinkering around the edges of the ALL app to improve its look
  2. Started transitioning my SPRRunner Flask app to use Blueprints on the way making it a general MTurk experiment running framework, including adding an RESTful API with anonymization and where others can use the list balancing, etc. (Mostly just in planning at this point.)

  3. Cleaned up the Turk Payments Google spreadsheet and generated a report for Florian about how everyone is spending his money.
  4. Poked around trying to learn more about RedCap to see if I can reduce the Continuing Review load by improving our setup on it for easier subject count and demographic reports.

Olga

Esteban

  1. Went through another batch of annotations from RAs 😤.
  2. Did first pass analysis of follow-up to my JML paper with flo'tan. Looks like things are replicating 😌.
  3. Worked out some remaining issues with a pygaze study. Turns out something in pygaze/pygame/psychopy/??? does not like 32bit wav files but is ok with 16bit 😶.
  4. Met with Jill Warker, had a good chat about ideas for future work 😺.
  5. Submitted too many abstracts 😅.
  6. Finally did the first round of testing of autoVOT on mturk recordings. Not all that encouraging 😿.

Dave

  1. Refactored my experiment code to use browserify for sane dependency management. (See above)
  2. Wrote a real simple node back end with list balancing using postgres for data storage.
  3. Worked out with Andrew how to deploy experiment on www in these modern times.
  4. Submitted an abstract on fMRI adaptation project to the Society for Neurobiology of Language meeting.

Sarah

Dan

Amanda

Zach

  1. Spent a long time trying to understand "predict.merMod" in R and why it was bugging out on me.
  2. Wrote a more for subtitle project
  3. Coded up and demo'd the pupillometry project

Linda

  1. More analysis for pen in mouth study.
  2. Ran additional experiment for pen in mouth study.
  3. Finished adding in mouse tracking to the Tower of Hanoi task. Solicited some helpful feedback from Susan Wagner-Cook.
  4. Put together tutorial on LateX for RA meeting.
  5. Noam Chomsky in the flesh with da booosh et al. :)

Maryam

Wednesday

  1. Started writing ditransitive definiteness paper. Writing is hard.
  2. Prepared & gave purrr tutorial in lab mtg

  3. Cleaned up simulation code some more for sharing with others. I feel like at this point I should just make it completely general and stick it on GitHub.

  4. Saw Chomsky in the flesh

LabmeetingSP16w15 (last edited 2016-07-07 18:33:52 by XinXie)

MoinMoin Appliance - Powered by TurnKey Linux