Thursday, May 8, 2014

Blog Post # 8 PLS Project


For this project, I am attempting to use text analysis software to identify any possible differences in the “My Goals” statement written by “successful” students vs. “unsuccessful” students in their ePortfolios. It might be useful if such goal statements, written at the outset of students’ careers at the college, were to prove predictive of future success or lack thereof, perhaps enabling the college to provide greater support to students less likely to succeed.

For the purposes of this project, we gathered 2841 goal statement “pages” that were created by students between 9/16/2010 and 6/15/201 from our Digication ePortfolio system. We wanted statements written closer to the beginning of students’ time here, so we limited selection to students with 0-20 credits (which yielded 1065). Of these 1065, only 511 had actual text in them.

This data was sent to the office of Institutional Research to identify two groups: students who have been successful (defined as graduated or transferred early) vs. unsuccessful (defined as no longer in attendance, but neither graduated nor transferred early).  The complete statistics generated for each students included the number of semesters enrolled starting in Fall 2010, final GPA, whether graduated (1), whether transferred (1), whether graduated or transferred (1), and credits attempted, credits earned and GPA for Fall 2010 and Spring 2011.
 
Professor Dragan then provided massive assistance by creating an “interface” called ePortfolio Explorer that allowed the data base to be searched and easily loaded into text analysis software.  

The goal statements (combined into 1 document for successful students vs. 1 document for unsuccessful students) were then run through Voyant.

 

 OUTCOMES

 Successful students (=graduated or transferred):

 
There is 1 document in this corpus with a total of 73,081 words and 5,993 unique words.

Most frequent words in the corpus: career (574), college (495), work (358), goals (323), degree (319).

 

 
Unsuccessful (=neither graduated nor transferred):

There is 1 document in this corpus with a total of 73,020 words and 5,997 unique words.

Most frequent words in the corpus: career (569), college (558), goals (470), want (365), like (300).



 
DISCUSSION
At this point, I haven't found any particular insights, and am not quite sure how to pursue the analysis this further (any thoughts would be welcome!). For both successful and unsuccessful students, 3 of the top five words were career, college and goals. While degree was in the top five for successful students (323 occurrences), it was mentioned fairly often by unsuccessful students as well (9th with 248 occurrences). The other 2 top five words for unsuccessful students, want and like, also appear in the top 10 words for successful students. So maybe what I can say at this point is that both successful and unsuccessful students are entering college with goals that are articulated in similar ways (though the individual goals themselves differ widely).
 
 

Thursday, April 10, 2014

Blog Post #6

I decided to annotate Google Earth for one of my favorite places on the globe - Kyoto, Japan. I am a garden fanatic, and as many of you will know, Kyoto is famous for its many traditional Japanese gardens. Here is my garden tour of Kyoto (I just put placemarks in for 3 gardens):

Sunday, April 6, 2014

Blog Post #5  I spent time looking through the Anterotesis site and was impressed by the range of projects. One that particularly piqued my interest was the linguistic atlas project, which maps English dialects across the United States. I would certainly consider using this the next time I teach Intro to Linguistics. The site contains audio files of speakers from around the US, illustrating all the regional dialects in the country. However, I don't really get how this site qualifies as much of a digital mapping project. While there is a map of the US, it does not seem interactive. The audio files are a valuable resource, but are just on the site as a listing, so I don't find it to have a particularly interesting user interface. Here is the link: www.lap.uga.org.

Thursday, February 20, 2014

Blog Post #3 Exploring Text Visualization Tools

I took 3 sample student goal statements from their ePortfolios as my "corpus" for this exercise.  Had a number of technical difficulties, including how to get the screenshots to work so that I could post the images here. Still haven't figured that out, so no images posted yet; the amount of time involved to "play around" with the tools, particularly "Voyant" has been frustrating. (Just as an aside, I find that as a learner, I have little patience to explore technology tools on my own, and much prefer to have a "mentor" at my side).  BTW, is it possible to attach a pdf or Word document here? Another thing I haven't figured out...

Although I did not find the Wordle word cloud and Voyant visualizations to be particularly illuminating, I am viewing this exercise as completely exploratory. I am wondering what exactly we can learn from these data. What can they tell us about the students who wrote them? Is it possible to discern goal commitment? Why they are coming to college? What they hope to achieve? I am not really sure yet. Also, a certain amount of the data may be irrelevant: for example, does it mean anything that the word "goal" appears most frequently in an essay entitled "my goals?" Should that word be "thrown out" so to speak, like a "stop word" (e.g., to, in, a. the) in Voyant? 

I will update this post after I figure out how to add the images.... More to come....  Paul
  

Sunday, February 2, 2014

Blog Post #1 Beyond Two Cultures

I can't say that I bridge the two cultures - humanism and science - in my everyday life, not exactly. But I do balance my intellectual interests (linguistics, pedagogy and learning) with more physical, hands-on pursuits. Foremost is gardening, which I have avidly done since I was a young boy. I love designing, planting, and caring for the garden, which I find a total change of pace from my "mental" work at the college. My other physical pursuit - which most people are surprised to hear - is carriage driving! Yes, it's actually a competitive sport (I have some ribbons to prove it!) and it's how I get the opportunity to be around horses. I took up the sport in my 50s, and believe me, it was difficult to be a total and complete novice at something - and it's not that easy to do! Just as some people view science and the humanities as two cultures, it's interesting to me that the mental and physical realms are also often viewed as two distinct worlds.
Blog Post #2 - Challenges and Opportunities in the Digital Humanities

Franco Moretti's article discusses a quantitative history of literature, specifically looking at a genre (the novel)  in terms of a series of cycles - showing not one rise of the novel, but multiple rises. For me, the following statement was most illuminating: "Quantitative research provides a type of data which is ideally independent of interpretations... and that is also its limit: it provides data, not interpretation." The question arises: is data alone sufficient? Don't we expect research to also involve interpretation, hypotheses, explanations of data? Moretti's answer seems to be that while "quantification poses the problem...and leaves you often with a perfectly clear problem - and no idea of a solution," this lack of explanation (solution) is exactly what is needed, at least in his field, because we "are used to asking only those questions for which we already have an answer." Nevertheless, Moretti also seems to acknowledge in other parts of the article that he, as a researcher, wants to account for the data (and thus has published explanations and hypotheses). I am not sure yet how to reconcile this in my mind, and look forward to my own project - to see if I will be content with quantitative data, or feel compelled to provide explanations (conjectures??).


Friday, January 10, 2014

Brainstorming a DH Project

I'm not teaching right now, but much of my attention and focus as an administrator is on improving student success, retention, and graduation. Also, as you also know, to these ends the college is in the middle of implementing a new advising model. I'm thinking that these endeavors could possibly be supported by an analysis of our student ePortfolios. For example, we have a huge corpus of student writing in these portfolios related to their academic and career goals. Would this information be useful to an advisor? to a program director?

We also have been attempting, over the past few years, to come up with metrics to "predict" the likelihood of a student completing a degree vs. dropping out - but without much success. Is it possible that looking at /analyzing students' reflections/writings in their first semester's ePortfolio could give us an indication which students might be at more risk?"

I would have to use a textual analysis tool, but am not sure yet if that would be sufficient. Suppose, for example, Wordle or Voyeur indicated that the word "career" was prominent in the reflections/writings of first year students - so I would know that the topic of careers is on their minds. But how would I know what the appearance of that word meant? Does it mean they didn't know what career they wanted, or that having a career (getting a good job) was their primary goal, or what? So it seems the tool would point me in a direction, but another level of analysis would be needed.