Course Info
Syllabus
Handouts
Questions and Answers
Textbook Errata
Bibliography
Internet Resources
Labs
Home

CAS LX 522 Syntax I
Labs 6 and 7: Linger and statistics

This page has three parts:

Introductory notes

  • This lab makes use of the psycholinguistic experimentation framework Linger, written by Doug Rohde of Tedlab. Linger also comes with some analysis tools, but we will not be using them (mainly because they're somewhat difficult to install on the public machines in the lab), but instead trying to make do with Excel. However, if you do later end up using Linger, check out Lingalyzer and Lingrapher (Lingrapher is rather painful to install under Mac OS X, I hope someday soon to post together some helpful advice on the subject).
  • In the past, PsyScope was the standard for creating online psycholinguistic experiments (judging reaction time, etc.), but development has ceased. Although PsyScope was a Mac-only product, it has been ported to Windows and is now being sold as a Windows-only product called E-Prime, which is prohibitively expensive. There is currently a project underway to port PsyScope to Mac OS X, although it is still in the alpha stages.
  • Another option, somewhat similar to Psyscope is PsyScript, which again works only for Macs, but is currently under active development. PsyScript uses AppleScript as its basic scripting language. The language reference is here on Apple's web site.
  • Alternatives include Presentation (for Windows only, still I think free, but very soon to become commercial ) and the PsychToolbox project, which is an add-on for MATLAB (and thus requires MATLAB, which we do not have access to on the Mac). PsychToolbox does exist in a Windows version, but it is less "perfect" than the Mac version.

Take-home exercises (first set, due Nov. 22)

Your task: Suppose you want to find out whether Principle A filters out candidates before they have a chance to affect slowdown.

1. Design some stimuli to test this. Each sentence should come in four variations

  • A. John thought that Beth owed him another opportunity to solve the problem.
  • B. John thought that Bill owed him another opportunity to solve the problem.
  • C. John thought that Bill owed himself another opportunity to solve the problem.
  • D. Jane thought that Bill owed himself another opportunity to solve the problem.

2. Create four such items. Keep that frame: Name verb that name verb pronoun/anaphor trailing words.... For each condition/item come up with a yes-no question to ask.

3. Create an items file for Linger using the format discussed in class.

4. Next time we'll pool them together and run them to get some data to analyze.

In-class exercises

The task: Walk through the example experiment. We're going to pinch the files from Colin Phillips' psycholinguistics page again.

Step one: Get the experiment files and set up the experiment

  • Find Linger, find the Experiments folder, and duplicate the Example folder within.
  • Change the copy's name to Binding3.
  • Open the folder and throw out the items file, we'll replace it.
  • Based on the items I got from several of you over the weekend, I have compiled together an Linger items file (in some places edited from what you'd provided). Control-click (or right-click) on the link (Linger items file) and save this file as items inside the Binding3 folder.
  • This should be enough for you to be able to run the experiment, which we'll do first before continuing the discussion further.

Step two: Run the experiment

  • Start up Linger as before, select the Binding3 experiment.
    • On the Mac, you will want to "hide the dock", which you can do by pointing to the faint vertical line to the left of the trash can (lower right corner of the screen, on the far right of the dock) and Control-clicking. Choose "Turn hiding on" if it is an option (if it say "Turn hiding off", it's already set up correctly). Under this setting, the dock "hides" beneath the screen until you put the mouse down at the very bottom of the screen, at which point the dock comes up to meet it.
  • We'll come up with a system for deciding what your subject number will be. Do not leave it at 1!
    • It is important that we move sequentially through the item numbers because this is how the "version" of the items is chosen. Because we have 4 conditions, subjects 1, 5, 9, 13, etc. see one version of the test, subjects 2, 6, 10, 14, etc. see a second version of the test, and so forth (there are four versions).
  • Clear your mind, take a deep breath, temporarily forget everything you know about linguistics, and start the test. It should be pretty quick, under 10 minutes I think. In fact, if it goes quickly enough, we might actually run it twice to increase the amount of data we have available to work with. If we do this, we will again come up with a system for deciding what your subject number will be on the second run.
    • Are you using a Mac and seeing a "type 1 error" message? If so, this will very likely fix the problem: Find the items file in your Binding3 folder. Choose Get Info... from the File menu, and look partway down the information window for Name & Extension. If you see that the filename is items.txt and the checkbox to Hide extension is checked, change the name to items (remove the .txt) and uncheck the Hide extension checkbox.

Step three: Collect the data together

  • When your experimental runs are over, your data will be in the Results folder of the Binding3 folder. We need to get that together in one place to analyze it. What I would suggest is that you email it to me using BU webmail. Then, during the class midpoint break, I'll try to collect it together and put it in a file you can download after the break.

Step four: Set up Lingalyzer

Once we have the data collected, we'll talk a bit about data analysis. There is a companion program to Linger called Lingalyzer that is very helpful in making sense of the data. We'll talk about what it does conceptually, but you wouldn't really want to do this in Excel.

Lingalyzer should already be installed on the machines in the lab, since it is part of the Linger download. However, Lingalyzer relies on a couple of statistics programs to run. Specifically, anova and regress from the |STAT package (written by Gary Perlman, tweaked by Doug Rohde). Here's what you do on a Mac (there is a way to do it on the PC, but I haven't tried it out).

Go to the |STAT page and click on MacOS X binaries to download them. If it doesn't automatically decompress, double-click on the archive to decompress it to a folder. The resulting folder will have a folder called bin within it. For our purposes, we should be able to simply drag the two files anova and regress from within that bin folder into the Lingalyzer folder inside the Linger folder.

Now we'll talk through how to create the condition file for Lingalyzer so we can analyze our data, including details about the items file and what we're looking for, the kinds of things we're going to use Lingalyzer to do, etc. This is where the handout will come in.

 

Take-home exercises (second set, due Nov. 29)

Your task: We'll talk about this in class. Perhaps there won't be anything.