Assignment 1: due 2/1/2023

The background: Coronal Stop "Deletion", as discussed in MacKenzie and Taminga, "New and old puzzles in the morphological conditioning of coronal stop deletion"; and the nature of allophonic variation, as discussed in Liberman, "Towards Progress in Theories of Language Sound Structure", 2018

The goal(s): Make a quick practical (and superficial) survey of how lexical (= underlying/dictionary pronunciation) "coronal stops", /t/ and /d/, can be realized in English, and when they seem to be entirely absent, whether due to allomorphy or to a symbolic phonological substitution. We'll discuss the results in class, and design a way to do annotation and measurement consistent with your findings. Note that we will look only at two short samples from two American speakers performing in a formal (NPR) register -- a broader sample of varieties, speakers and registers would give us a different inventory of options as well as different distributions of contextual effects.

How to do it:

1. Create a suitable directory (=folder) on your laptop; download Assignment1.zip there; and unzip it to create

   101012_7.wav
   101012_7.TextGrid
   101416_15.wav
   101416_15.TextGrid
   ScriptFILE1.praat
   ScriptFILE2.praat
   praatnotesFILE1.txt
   praatnotesFILE2.txt
   

FILE1, 101012_7.wav, contains the audio for the 7th segment from an All Things Considered story, "Feeling The Fiscal Squeeze, EPA Seeks To Slim Down", 2/26/2014. The speaker is Karen Kellen.

FILE2, 101416_15.wav, contains the audio for the 15th segment from All Things Considered, "Afghan Security Agreement Is Still Unsigned — Who's At Fault?", 1/31/2014. The speaker is Tom Bowman.

These were selected at random from a corpus of 109,067 NPR podcasts, containing 3,199,859 such "segments" (basically, speaker turns).

2. (As shown in class...) In praat, click

   praat>>Open Praat script

   to open the ScriptFILE1.praat file (..and of course later the ScriptFILE2.praat file).
   
3. In the script window that opens, click

   run>>run

   so that praat will take you through the list of coronal stops in the selected file (15 in FILE1, 37 in FILE2). At each step, you'll see a "Pause" button like this
   
   
   
   which shows you where in the list you are, and offers a "Continue" button to take you to the next example.


4. In your favorite text editor, open the file praatnotesFILE1.txt (and later praatnotesFILE2).txt, which contains just a numbered list of locations, e.g.
   

   
   1 101012_7 2.8965
   2 101012_7 3.7255
   3 101012_7 4.5445
   4 101012_7 4.6985
   5 101012_7 4.9335
   6 101012_7 5.3405
   ...
   

   
   

As praat takes you through the file, edit the file to add your commentary on the example in questions, e.g.



1 101012_7 2.8965 "projects", 39ms silence between [k] and [s], onset sounds like [ks] = no [t]?
2 101012_7 3.7255 "and" 78ms [n] 40ms [xd] release into 100ms schwa + 66ms silence



Note that you can use praat's general operation to view or listen to longer or shorter audio segments -- you should be familiar with the relevant commands.



5. When you're done, upload the two .txt files to the course Canvas site.

UPDATE: For a peek at coronal-stop allophony in a different variety of English, download JamieOliver1.zip. This give you a 37-second speaking turn from this NPR podcast, by Jamie Oliver, who comes from southeastern England.

The cited .zip file should yield

You can load and run the Praat script as before, which should lead you through 43 (lexical) coronal stops in the .wav file. You can add relevant notes and measurements in the praatnotesJamieOliver1.txt file.

IMPORTANT: As usual in the early stages of such projects, there are no well-defined "right answers", although there are plenty of wrong answers, as well as greater or lesser degrees of insight and helpfulness. The goal here is not to test how well you've memorized stuff, or how well you can apply specific methods in cases where the answer is clear to a credentialed grader. There are skills involved that you should know by now, but this is science, not the GED or the MCAT.