Sara Harland's Blog
This is the portion of the site where I will post weekly updates on my progress with the project."
This was our first week working on the Intonation and Evidence project. Our group tasks consisted of setting up our website. This week we also began exploring PRATT and the TOBI tutorial. As a group we discussed our goals and jobs for the project as well as looking at various sound files in PRATT.
Throughout this week the main task was to work on the first portion of the TOBI tutorial. We were introduced to the structure and process of labeling sound files. This included the diffrences between tones, boundary tones and break indices. The first three chapters began an inventory for different types of tones and boundary tones including H*, L* tones and H-H% and L-L% as boundary tones.
This week I completed the ToBi tutorial to 2.4, where the boundary tones of L-H% and H-L% were introduced. Throughout this section of the tutorial examples were given of different patterns that could be heard. I found the activities that were given in this chapter to be helpful, in then sense that I was able to label them myself and then go through and check if my answers were the same as the tutorials.
This week I completed 2.5 & 2.6 of the Tobi tutorial. This included the introduction of bi-tonal pitch accents as well as a downstep pitch accent. After first being exposed to these elements I had a little bit of difficult truly differentiating from other pitch accents, but after our group meeting today I feel that I have a better understanding of the concept. In our team member meeting this week I found it helpful to put sound files on the projector and label together as we could talk through discrepancies that were presented.
This week my time mainly consisted of labelling the FM News Radio corpus. I found that labeling this was relatively straightforward in the sense that it was easy to distinguish the tone and pitch accents that the speaker presented. I found it very common that for H* pitch accents to be present. I also continued to work on reviewing the Tobi Tutorial.
This week I finished working on labelling the FM news radio corpus. I am looking forward to discussing these labels with the rest of the group to make comparisons. My individual task this week consisted of reviewing the updated material on labelling. This extends the information in the Tobi tutorial. When I have completed learning this material, I will work on teaching this material to Emily and Karina. This week I enjoyed how we each had an individual task and will collaborate this week.
This week was for the most a continuation of the previous week. I am still working on learning the labels, to allow for me to begin labelling previous and future data sets. I am finding that these updated labels are fairly similar to the previous ones, however I need to continue practicing until I feel fully confident in them .
This week I made a chart that outlined the new labels. One of my concerns was that during labelling sometimes the pitch track shows something different than what I am hearing. This was something that I brought up in the group meeting and I feel a little more confident in how to approach this. In the group meeting we discussed what should be completed by the end of this semester in order for us to be on track to run the second data set.
This week I mainly focused on working on the experimental design and set-up as well as the IRB proposal. This needs to get approval in order to run the experiment at Simmons. We assigned different tasks and parts of the IRB to work on. I also formatted and set up the powerpoints with the new comics that will be used to run the experiment.
Moving forward into this week it was mainly an extension of last week in terms of the IRB being the main priority. I also familiarized myself with the experimental design and set-up and how we previously collected and organized data when the experiment was run the last time.
This week I continued to help in finalizing the IRB document. Specifically, this week I added appendices into the IRB that showed how the labelling conventions that we use work, as well as an example of the comic strip that will be used in the experiment.
This was the week of out Thanksgiving break so we did not meet as a group. We are continuing to wait for our IRB approval.
Throughout the last two weeks Emily constructed randomization for the comic strips that will be used for the experiment. After she sent me the output for the randomization I then used that information to make the different versions of the comic strips with the randomized order.
This was our last meeting before winter break. We have finished the IRB and and necessary changes have been made. In regards to next semsester we are on track to start running subjects as soon as we return from break. Our subjects will be Simmons Undergraduate students and will receive course credit for their participation in the study. We are hoping to have a large subject pool, and would like to run these subjects early in the semester so that we can then proceed to analyze the speech and go from there.>
This was our first week back meeting as a group. We had prepared before the break to be able to collect data. Our IRB was approved before break and we were able to start scheduling people for the experiment. In order to organize the sign up sheet we used the website WeJoin, where we were able to post the available time slots. The first week we had seven participants and the overall the process ran smoothly. We recorded the speech in three separate audio files in PRAT and made each participant a folder with their audio recordings. After listening through the participants recordings, it seemed that they understood how to properly read the comics as if they were the person in the comic. Moving forward it is necessary to recruit more participants to get a larger subject pool and spread the word around campus on how to sign up.
This week was a continuation in collecting data from participants. This week I felt more comfortable in running subjects and feel that the system we used worked very well. I am looking forward to using this data to analyze the speech.
This week was the final week for data collection for this round of participants. Our next steps will be to edit the files and then focus on analyzing the data.As a group we were in contact with the psychology department and confirmed that we would be able use subjects from that pool for future experiments.
This week a large portion of my time was used to edit and chop the PRAT files. At first I was unsure on how to approach this process but figured that it would be easiest to have the names for each file in an excel spreadsheet and then copy and paste that name to the corresponding WAV file. Once I started to work through these files I found the process to be easier and feel that I have a system that works well.
This week throughout looking through our subjects data we found a small issue in our data. The comics did not include punctuation which effected how the participants said the phrases. We are wondering if this is any way to related to the participants age and how texting language could influence the prosody. Our new hypothesis includes that when there is no punctuation present in the comic the participant will use a declarative boundary tone rather than a higher type boundary tone if applicable.
This week Karina set up an aligner which made the annotating the sound files much easier. This week I reviewed the Tobi tutorial to have a refresher on the labels and looking towards the end of this week and after break will include the labelling portion of this experiment.
This week Nanette was at at conference so instead of doing our group meeting we had a skype meeting with ZL who is also in the process of labelling files at Princeton. I thought this meeting was very helpful and it was nice that we were able to begin labelling together and collaborate with another labeler.
This week marked the official start of labelling. Karina's aligner was very helpful and allowed for a more organized way of labelling. I began with labelling the A participants and will complete this labelling in sections.
In order to prepare for the upcoming conference we needed to label our data so that it could be analyzed for our poster. Karina and I worked together to label the files which was very helpful. Once we finish the labels Emily is going to analyze the data in R Studio.
This week we worked on finalizing the details for the next round of subjects. We will be using Audacity to record and the comics and directions will be altered slightly and we will compare these participants to the previous ones. I am looking forward to continue to run subjects and gather more data.
This week was largely spent running subjects. Each subject took about 15-20 minutes and we made sure to tell the participant that although there was no punctuation in the comic to insert whatever they felt was appropriate. Using Audacity was easier this time because it allowed for us to create just one sound file rather than three which made organizing the data easier. The next two weeks will be used to recruit and run subjects.
After finishing a majority of the subjects Karina and I met to begin the labelling process. We put the sound files on a projector and worked through them putting the labels in an excel spreadsheet. We will need to finish the labelling so that we can analyze the data and work on the poster for the Undergraduate Symposium.
Back to the top