Listening Assignment

Preparation stage

Before you start, make sure you have downloaded and familiarized yourselves with the Praat software (see Working in Praat).

You will receive your recording (.wav file) by email within two weeks from the recording session, as well as the TextGrid (.TextGrid file). You can open the sound and TG in Praat.

Please do not make any changes to the sound file – you only open it, never save it. You will only be changing, saving and sending to us the TG. Do not rename any of the files (specifically, do not add things like „final“, „evaluated“, „really-final“ or anything else to the file name, nor to the extension).

The automatic aligner which marks the boundaries of words and segments works with the original text. It may thus occasionally make mistakes which must be corrected manually. Therefore, before doing any other work, go through your files and see whether there are any “problems”.

  • long pause: just insert a new boundary, thus creating an empty interval, and shift the relevant boundaries accordingly. You can do the same thing with your hesitations, repetitions or any other interruptions of the flow of speech.
  • unknown word: this happens rarely, but if the algorithm encounters an unknown word (e.g. an uncommon name), it skips it and writes all subsequent text in capital letters. Please add the missing word(s), replace any text in capitals with small type, and shift word boundaries if needed.

In addition, the alignment and subsequent post-processing may lead to faulty phonemes transcribed in some words; if these include some of the target sounds for analysis, please correct them.

Analysis (winter semester)

Once you’ve prepared the TG, you can start working on the analysis. First, create a new point tier which will be called “evaluation”. To do this, click on TierAdd point tier… in the Editor window. In the dialogue window which pops up, enter “2” for Position, and “evaluation” for Name. That’s it. Don’t forget to save the modified TG.

You will be evaluating three kinds of speech sounds:

  • The vowels which correspond to /e æ ɒ ə/ in Standard British English. Please be aware that the automatic aligner “speaks” American English. You will therefore have to change some vowels to their British values, especially /ɒ/ which is usually /ɑː/ in AmE (e.g., rock).
    You will evaluate your vowels in the second half of the text (from And then there was the food onwards).
  • The dental fricatives /θ ð/ and the velar nasal /ŋ/.
    You will evaluate these consonants in the whole text.
  • Fortis plosives /p t k/, regarding their aspiration.
    You will evaluate aspiration in the first half of the text (finish with … obviously we didn’t use it).

For each evaluated sound, insert a point into the evaluation tier by clicking into the little circle at the top of the tier (or as described on the Working in Praat page). Then specify your evaluation in the newly created point.

The vowels and the three consonants will be evaluated on the following scale:

  • “2” = native-like realization of the target sound. The standard against which you compare your pronunciation is Standard British English. If your pronunciation of the given sound is native-like American, enter “2 Am” instead of “2”.
  • “0” = target sound pronounced with a strong Czech accent (or your native accent).
  • “1” = target sound pronounced in a manner which is intermediate between native and strong non-native accent (e.g., pronouncing /ð/ as a dental plosive [d̪] or /æ/ as [ɛ]).

You will handle aspiration differently. Instead of evaluating the correctness of your pronunciation, you simply mark the presence or absence of aspiration in your pronunciation (“Y” equals presence, “N” equals absence of aspiration in the given sound).

An example sound and TG can be downloaded here (all phenomena are marked in the initial portion of this file).

The course of work is up to you, but we strongly recommend that you follow the rough guidelines below, and/or our announcements in classes.

NOTE: If you pronounce a completely different sound, don’t evaluate it! Especially in altered or otherwise modified words.

Listening assignment guideline (winter semester):
Week 4: do the preparation stage (see above)
Week 5: start evaluating your /e/ and /æ/ vowels
Week 6: start evaluating the other two vowels; all vowels are checked in the first Progress Check
Week 8: start marking aspiration
Week 9: evaluate your dental fricatives
Week 10: evaluate your velar nasal sound

Analysis (summer semester)

You will work with the same recording and TextGrid as in the winter semester.

First of all, delete the evaluation tier from the winter semester. Then you will need to create two new point tiers, one for linking analysis (as the first tier) and one for the analysis of weak forms (between the phone and word tiers). To do this, click on TierAdd point tier… in the Editor window.

You will be evaluating two connected-speech phenomena:

  • linking – in the Linking tier, add a point between the word-final sound and word-initial vowel, wherever linking is possible, and enter ? for a glottal stop (no linking),res for pseudo-resyllabification, r for linking r, j for transient j and w for transient w.
    You will analyze the way you linked in the whole recording.
  • weak forms – the point tier will be called weak. You will focus only on words which we selected as those where weak forms are most likely to occur in native English and where advanced Czech speakers of English may pronounce a full vowel. You will find the text, with the target vowels highlighted in yellow, here. Insert a point into each target vowel, and assess whether you pronounced it as really reduced (mark R) or as a full vowel (mark F).

See an example sound and TG here to make sure you’re doing the task correctly.