Roberto Franzosi is Professor of Sociology and Linguistics at Emory
University. He earned a BA in Literature from the University of Genoa (Italy) and a PhD in Sociology from Johns Hopkins University. After a postdoctoral year at the University of Michigan, he taught at the University of Wisconsin-Madison, Oxford University (with a fellowship at Trinity College), and University of Reading. His main research interests have been in social protest (e.g., The Puzzle of Strikes: Class and State Strategies in Postwar Italy, 1994). He has had a long-standing interest in issues of language and measurement of meaning, with several articles published and two books, From Words to Number: Narrative, Data, and Social Science (2005), and Content Analysis (2008). [SAGE INTRODUCTION]
Q: What is quantitative narrative analysis?
It is a technic that I developed, I started working on it in the mid-1980, during my Ph.D. work at Johns Hopkins. I did mostly statistical work and mathematics, that’s all I did, very quantitative. But at the end of my dissertation, which was on Italian strikes, I realized that there were more to strike than economic coefficients, like there were actors, social actors, which had disappeared, what did the state do, what did the workers do, what did the unions do.
And I started coming up with the idea of using newspapers as sources of social historical data. I was postdoc at the University of Michigan and with Charles Tilly, and lots of people at Michigan were using newspapers as sources of data to go after the actor, and that’s when I started working with the idea of using text to extract information of social actors.
So, quantitative narrative analysis started from this, the idea of extracting who dose what when where and why. And quantitative because by counting the frequencies of these words in relation to other words then become a quantitative problem. And at the beginning of 1980s, and even in the mid-1990s when I published the book , I thought of the problem as, you know, going from words to numbers. And only when I published the SAGE book that I use the slogan ‘QNA’, because it is easy to remember than ‘from words to number’.
Q: So, you developed the word?
Well, it is interesting. Yeah, I mean all traditions of narrative analysis, right? I call it quantitative narrative analysis because mostly narrative analysis is qualitative. But because I turn these words into numbers, then I call it quantitative narrative analysis. In reality I didn’t develop anything. I mean this 5Ws, the who what where when why, has been there since this 5th century before. But at the time when I developed QNA in code, I didn’t know any it, it’s only afterwards I started.
Q: What is the difference between quantitative and qualitative analysis?
Qualitative analysis they would never try to get numbers out of the text. They analysis the text very closely, they use quotes, they use passwords, but they don’t like quantitative analysis.
Q: Then what do you think is narrative analysis?
So narrative analysis deals with a very specific type of text, a text that is narrative. So we deal with the genre, which is called stories, narrative. While content analysis deals with any types of text that is a text that could be analytical, could be philosophy, could be history, or could be just any newspaper. So, narrative analysis by definition deals mostly with narrative, with story. So the basic type of narrative analysis, the 5W of journalism, the who the what the where the when and the why, and not necessarily would you have those in a more general type of text.
Q: Why do we need to take QNA specifically out?
That was my own development, because I was interested in quantifying text. But I was interested in social actors and their actions. Hence I came up with this structure, all the 5Ws, and I was very lucky because if I look at any other type of text genre, text types, I would not have found 5Ws. And Alexandra already talked about the 5Ws. So it was my development because I was trying to extract social actors but I want to quantify because I develop all that idea at my Ph.D. based on computers and math. So when I wanted to go after the social actors, I wanted to quantify them.
Q: What does narrative mean, and what does narrative contain?
I suppose that the qualitative narrative analyst will also focus on the 5Ws, for the simple reason that this is found in the narrative, right? But there is more in narrative that justify 5Ws.So if you think of William Labov’s classification of the type of functions that you found in narrative, that is a part of narrative. So there is an abstract, like the summery at the beginning, then there is the narrative proper, description, and description means like: ‘It was a beautiful day in London, it was spring, it was cool, there were people in the park'. So that is description. And then there is narrative proper, and narrative proper is about 5Ws.So Rebecca is talking, Rong is also talking, or listening or recording, so those will be the narrative proper, it is about people. Then there are people playing football, some people are sleeping, people are holding hands. Then there is the evaluation. And evaluation is for instance if in this kind of narrative it would be ‘it was a pleasant surprise to be with Rong.’and just to see the difference between an English culture and the Chinese culture. You know, then you evaluate the conversation and the event of you and I being here. And then about culture coda, that is the end. Coda is like tail, literally coda means tail.
So, the 5Ws cannot capture evaluations, because it is about abstract concept. So the 5Ws would not explain to that. The 5Ws would not be interested in the coda than the abstract because whatever is in the coda and in the abstract this probably going be in the body of the narrative. So you focus on the body of the narrative. And the 5Ws would not capture description. ‘It was a beautiful day, there is anybody doing something.’ It is more likely that people doing qualitative narrative analysis may focus on description or on evaluation. Although they will focus on actors or actions but they would not quantify it. To them it is not the numbers that matter.
I am trying to see if it is possible to get to see with QNA to description and evaluation. In other words, can we figure out the way which there are certain verbs that are descriptive and certain words that are evaluative. And you can then use this to just abstract the evaluations and the descriptions. Because next year they are taking an independence course with me and I am thinking what could we do, and this is one of the things that we could develop the computing routines. Because I work on narrative, all my life I work on narrative and so the routines that we deal with narrative like the shape of the stories really excite me. Because they have beautiful shapes, it really does turn the game from a stupid table of numbers to beauty. And I have an obsession sort of beauty.
Q: What is new in this area?
Dynamic network, dynamic GIS (Geographic Information System), and I think that now with the conversion of all these narratives into text files, we can process natural language in tool, that is why I became so obsess with QNA. To see what they can tell me, it is like trying to get more data in other words. Because doing quantitative narrative analysis all you care about is 5Ws.
And now you have many more information it is like how is the black man characterize. It is really funny I am not started to analysis most obsess with developing ever more new tools. Develop ever more ways of looking at things you know. Sometime I wonder whether I am just methodologist you know, but the project on this is coming to an end. I mean it is basic finished now. I am a methodologist there is no doubt, but my interests are always substantive.
Q: Should I call you a story researcher?
You know this is really funny because through my work on stories, which are the object of my own study. Now they have become the tools of my own work. So from object of my observation, they have become tools for me to write beautifully. I was just using them as object of study, but now I figure out how to use stories.
Q: What is your opinion on mixed methods?
Now quantify the text through QNA which is called the distance reading, you have the computers do the reading. But in reality, I have down so much more close readings. Because the distance reading brings up stuff, they bring up some terms, some relationships, and then you go in and read those documents very closely. I combine them so I draw a pattern and I zoom in. I also combine qualitative and quantitative outcomes in visual.
Q: What is the difference between PC-AC and NVivo.
When I started working on quantitative narrative analysis back in 1984-1985, there were no computer system. So I started developing my own software. And then in any case, NVivo, and others cannot do quantitative analysis. It is impossible to do it, we wrote a paper showing that you could not do it. So I just have to continue developing the stuff. And now I am at the point where it is almost good enough, you know I always use it myself, and a couple of other people.
PC-ACE is capable of lots of functions, but as you saw, it is very difficult to install for one thing. So it is much more capable of doing and it can do a lot of computation linguistic stuff. There are more and more software to do this stuff very easily.