Step by Step Guide to Pause Analysis

1

Step by step guide to pause analysis

July (2005)

Background – using SFS

Appendix 1, adapted from the “UCLASSCA using SFS and CHAT manual” (To, 2005) includes guidelines on how to create an SFS file from a sound file and how to create and edit annotation lines in SFS. Throughout this guide the relevant sections of Appendix 1 will be pointed out.

Step by step guide to pause analysis

The aim of pause analysis is to measure the number of pauses that occur at phonological (PW) boundaries and the number that occur at syntactic boundaries. Thus in addition to producing a phonetic or orthographic transcription of your recordings you also need to include annotation lines which show where PW and syntactic boundaries occur.Pause analysis has already been carried out for the persistent/recovered oldest files and you can view these completed files on the Z drive in the “syntax” folder. To enable you to prepare your files in the same way you need to complete 4 steps:-

Step 1 converting a sound file into SFS format

Step 2 phonetically or orthographically transcribe your file

Step 3 marking PW boundaries on the transcriptions

Step 4 Marking syntactic boundaries on the transcriptions

Each of these are covered in depth below:

Step 1 converting a sound file into SFS format

• Converta sound file into an SFS file (see APPENDIX 1 section 1 – “using SFS” for how to do this). The Spanish adult and child recordings/ the persistent and recovered recordings (all three ages) and the adult English fluent and disfluent recordingshave already been converted into SFS format (CDs available from Steve Davies)

Step 2 phonetically or orthographically transcribe your file(discuss with Pete whether you should use phonetic or orthographic transcriptions). You need an orthographic transcription to carry out the syntactic parsing so unless you need a time aligned phonetic transcription for any other reason I would suggest that you complete a time aligned orthographic transcription.

Many of our files have already been transcribed. The Spanish adult and child recordings/ the persistent and recovered recordings (all three ages) and dysfluentadult English recordingshave already been phonetically transcribed and annotated. The fluentEnglish recordings have been orthographically transcribed and annotated (CDs available from Steve Davies).

If you need to transcribe your file

• APPENDIX 1 section 2 – “using the annotation line” tells you how to create new annotation lines on SFS.

• If you need to phonetically transcribe new files you need to consult the how to transcribe section on our website: You should transcribe at the syllable level.

• For tips on how to add an orthographic annotation line (see APPENDIX 1 section 3 – “orthographic transcriptions). As with phonetic annotations it is essential that you transcribe pauses in the recordings.Consult for how to annotate pauses. Contractions e.g. “don’t”/ “can’t” should be transcribed in full as “do not” and cannot.

• For both phonetic and orthographic transcriptionsfluent items should be transcribed using lower case. Dysfluent items should be transcribed using upper case. contains detailed instructions on how to transcribe both fluent and dysfluent items.

• Both orthographic and phonetic transcriptions should be punctuated.

• Each word should be marked as being either a function or a content word has a detailed account of how to categorise words as function or content words and how to indicate the category of word on your annotations(it is important to note that content words are preceded in transcriptions by the symbol : function words are preceded by the symbol / in transcriptions)

Step 3 marking PW boundaries on the transcriptions.

• Copy your phonetic or orthographic annotation line (see APPENDIX 1 section 2 – “using the annotation line” tells you how to copy annotation lines on SFS) and name it PW. If you have a phonetic transcription your PW annotation will be created by going through the copied phonetic annotation (which you have named PW) and deleting everything except the first syllable of each PW and the x marking the end of the last syllable of any PW (see APPENDIX 1 section 2 – “using the annotation line” tells you how to delete items on SFS).If you have an orthographic transcription your PW annotation will be created by going through the copied orthographic annotation (which you have named PW) and deleting everything except the first word of each PW (see APPENDIX 1 section 2 – “using the annotation line” tells you how to delete items on SFS).

• How to identify PW boundaries. PWs consist of a content word (nucleus)and any function words (satellites) that are semantically related to the nucleus. Each nucleus may have any number of satellites and satellites may precede or follow the nucleus. Some content words may not have any satellites in which case the PW is made up of just the one content word.

References:

Au-Yeung, J., Howell, P., & Pilgrim, L. (1998). Phonological Words and Stuttering on Function Words. Journal of Speech, Language and Hearing Research, 41, 1019-1030

Howell, P. (2004). Comparison of two ways of defining phonological words for assessing stuttering pattern changes with age in Spanish speakers who stutter. Journal of Multilingual Communication Disorders, 2, 161-186 give more details.

• Splitting utterances into PWsIdentify each nucleus (content word) and then decide which function words are semantically related to that content word (this comes down to making a judgment about which content word each function word is most closely linked to semantically).

• Example The words in the utterance “I have to travel up to the language centre by the tube” would have been categorized into content and function words as follows (remember content words are preceded in transcriptions by the symbol : function words are preceded by the symbol /) “/I :have /to :travel /up /to /the :language :centre /by /the :tube”. I would split them into the following PWs:

/I :have (nucleus is “have” and the semantically related “I” is a preceding satellite.)

/to :travel /up (nucleus is “travel” and the semantically related “to” is a preceding satellite and the semantically related “up” is a following satellite.)

/to /the :language (nucleus is “language” and the semantically related “to” and “the” are both preceding satellites)

:centre (nucleus “centre” has no satellites)

/by /the :tube (nucleus is “tube” and the semantically related “by” and “the” are both preceding satellites)

Step 4 Marking syntactic boundaries on the transcriptions.

• If you have phonetic transcriptions orthographicallytranscribe your phonetic transcription -one sentence per line in plain text files(there is no need to time align this in SFS for the purpose of pause analysis)

• If you have orthographic transcriptions you can output your text file (see APPENDIX 1 section 2.6 –tells you how to output annotations as text on SFS). It will look like this:

0.77001 I

1.09931 have

1.35804 to

1.38156 travel

1.60894 up

1.62462 to

1.79711 the

1.86767 language

2.47923 centre

3.64746 by

4.26686 the

4.29822 tube.

You need to delete the numbers and convert the text into one sentence per line plain text files. You can do this manually. It’s quicker (but fiddly) to open your text file in excel. In the “original data type” box check delimited and return. In the next screen check space in the delimited box and return. Your file will open and you can delete the time column.This will leave you with the words only which you can then convert into one line per sentence. Save the excel file as a text file.

Examples of one sentence per line plain text files

I have to travel up to the language centre by the tube.

Before that, I used to live and work in Dublin, in Ireland.

I should not really have left, but, I, did not have a job.

And, so that was my first job.

Before that I come from England.

You can probably tell from the accent.

Semantic Parsing You should consult Peter Howell for guidance on how to parse your files.

Once your files are parsed, you need to copy your phonetic/orthographic annotation (APPENDIX 1 section2 “using the annotation line” tells you how to copy annotation lines on SFS). This new (copied) annotation line should be called Syntax. Go through this new (copied) annotation line and delete each word replacing it with it’s syntactic marker.

Appendix 1.

1. Using SFS

To begin using SFS you first need to have a sound file. This can then be imported and turned into an SFS file.

1.1 Importing a Sound File

1)Open SFS

2)Left click on the item icon, located on the tool bar at the top of the window.

3)Move the cursor over the first option, import.

4)Left click the option speech (copy), from the drop down list, to select it.

5)A second window will appear SFS Program:conv2sfs

6)Left click the browse button.

7)Select a file to be imported into SFS and left click open.

8)Left click ok in the SFS Program con2sfs window.

9) A speech file will then appear in the SFS window.

1.2How To Display a Speech Wave

The speech file can then be opened to display a speech wave.

1)Check the box next to the speech file and left click the seventh picture icon on the tool bar.

2)An eswin window containing your speech wave and audio file will open.

3)To select a smaller section to zoom in, left click the beginning of the section and right click the end of the section. You will see a blue and a green line.

4)To zoom in on the section that has been selected, left click on the reddown arrow on the icon bar, which looks like this:

5)To zoom out of a section left click the red up arrow on the icon bar.

6)To move the section of speech forwards or backwards use the light blue left and right button at the bottom of the screen.

1.3 How to Play Back an Audio File

The selected section can then be played back by left clicking the green play button on the icon bar.

To stop play back left click the blue stop button on the icon bar.

1.4How to Save a File in SFS

1)Left click file on the icon bar.

2)Left click save as from the drop down list.

3)In the save as window select where to save the file and left click save.

4)After saving for the initial time, the save icon can be used instead.

1.5 Shortcuts and Keyboard Functions

In SFS there is a list of shortcuts and keyboard functions found in the help option on the icon bar.

1)In the eswin window left click help on the icon bar

2)Left click keyboard help. The following window should appear contain shortcuts and functions.

I found the space bar shortcut particularly helpful, it plays the section that is selected without having to use the mouse to click on the play button.

The [Shift]--> Move annotation right and

[Shift]<-- Move annotation left

function was also useful for when I needed to adjust the position of my annotation backwards or forwards.

1.6 The Time Line

In the eswin window you will see a time line along the top and the bottom of your speech wave.

These tell you the time in the recording where that particular part of speech is.

For example, the word “we” in the above file is approximately twelve and six tenths of a second into the recording.

At the very bottom right hand side of the screen you will see some letters and numbers.

The numbers next to “L=” represent the time in the recording marked by the blueline (left click).

The numbers next to “R=” represent the time in the recording marked by the green line (right click).

The numbers next to “T=” represents the time between the blue and green lines.

The numbers at the far right represent the time in the recording where your cursor is currently placed.

This function of SFS will be very useful for speech rate analysis.

2. Using the Annotation Line

2.1 How to Createan Annotation Line

1)Open an SFS speech file.

2)In the eswin window, left click the annotation option on the icon bar at the top.

3)Left click create/edit annotations from the drop down list.

4)In the create/edit annotations window name the annotation line (for the purpose of the pause analysis these should be labeled as syllables(for phonetic annotations)/orthographic/syntax/PW) and left click ok.

5)A blank annotation line should then appear below the speech wave.

2.2 How to Make an Annotation

1)To make an annotation, first select the point on the speech wave that you want to annotate by left clicking on it. A blue line will appear at the point selected.

2)Once the point of annotation has been selected press a on the keyboard and the cursor will flash in the annotation bar, located near the top of the window on the icon bar.

3)Type your annotation in the annotation bar and press enter. Your annotation should appear on the red annotation line below the speech wave.

4)Alternatively you can make an annotation by right clicking where you want to annotate and a green line will appear.

5)At step 2, press binstead of a, then follow the above steps.

2.3 How to Edit an Annotation

To edit an annotation when your annotation line is already open, do the following:

1)Left click on the annotation you want to edit, be sure you click on the red annotation line.

2)The existing annotation should then appear in the annotation bar, highlight in blue.

3)Press enter after editing the annotation and the new annotation should appear on the annotation line.

To edit an annotation after you have closed the file, do the following steps:

1)Open the speech file you want to edit.

2)In the eswin window left click the annotation option on the icon bar.

3)Left click create/edit annotations.

4)The create/edit annotations window should appear. Press down arrow to view the drop down list, and left click orthographic (or the name of the annotation line you want to edit).

5)The selected annotation line should then appear below the speech wave in red.

6)Then follow steps 1-3 for editing an open annotation line.

2.4 How to Delete an Annotation

Please note you can only delete an annotation if the red annotation line is open.

1)Left click on the annotation that you want to delete (on the red annotation line) and it should appear in the annotation bar, highlighted in blue.

2)Press delete followed by enter.

3)The deleted annotation should be removed from the red annotation line.

2.5 How to Remove an Annotation Line from an SFS File

For each speech file you may have more than one annotation line, these can be deleted individually or all at once.

In your SFS file window you will have your speech file and a list of you annotations.

1)To delete an unwanted annotation, left click in the left hand side box beside the line you want deleted. The box should now be checked.

2)After you have checked one or more of the annotation lines to be deleted, left click item on the icon bar and select delete from the drop down list.

A prompt window will appear asking if you want to delete, left click yes.

2.6 How to View Your Annotations as Text Output

1)in the SFS file window left click the tools option on the icon bar.

2)Highlight annotations.

3)Highlight export.

4)Then left click export annotations.

5)The SFS Program: anlist window will then appear. Left click ok.

6)An SFS output window will appear with your annotations listed with the times matching the speech file.

2.7 Copying an Annotation Line

You may need to copy a particular annotation line in order to add annotations for a different purpose.

1)In your SFS File windows, which contains all your annotation lines, check the box next to the annotation line that you would like to copy

2)Left click item

3)Left click copy

4)To paste it onto the same SFS file, uncheck the annotation line you had selected previously.

5)Left click item.

6)Left click paste copy.

You should now have a copy of the annotation line in your existing SFS File.

7)To copy the annotation line to a new SFS file, uncheck the annotation line you have selected.

8)Left click the blank file icon at top left hand side of the icon bar.

9)A blank SFS file will open.

10)Left click item

11)Left click paste copy. You copied annotation line will then appear in the new SFS file.

12)Save the new file.

For both the above you will need to rename the annotation line. To do this check the box next to your new annotation line and the follow steps 2-6 in section 5.1.

3. Orthographic Transcriptions

Before starting your orthographic transcription, you need to create an annotation line for it (see 2.1).

3.1 Finding Words

The first thing you need to do is find the beginning of the word you are transcribing on the speech wave.

1)Select four or five waves and zoom in.

2)Play back these waves so you can hear what is being said.Finding the exact beginning and end of the word may take a few tries to begin with, but with practice you will start recognising the wave patterns, which will make it easier.

3)Once you have found the beginning of the word you are transcribing, you can then add it on your annotation line. Repeat this process for the rest of the sound file.