Computational methods allowed us to discover patterns in research questions of this particular corpus of interviews For example, 88 distinct titles were mentioned in the interviews. However, there is an interesting disconnect between genres and specific titles. Several interviewees mention European art cinema and specific directors, such as Fellini, Antonioni, Bertolucci, yet they do not mention any specific film titles. Rather, these names appear as repeated lists. One conclusion we draw from this finding is that these names serve as an affirmation of cultural taste and a shared “cultured” identity, not a memorable individual experience of watching a film. The same interviewees list specific titles of popular culture films.
We chose to tag references to “Foreign Cinema” (зарубежное или иностранное кино) as a separate category. These references occur mostly in the speech of older interviewees. Likewise, in addition to references to specific European national cinemas (French, Italian, German, etc.) some interviewees mention “European cinema” as a category distinct from Russo-Soviet or American. Our hypothesis, based on the frequency of references to particular directors and actors, is that “European cinema” most likely refers to the New Wave and art cinema of the 1950s-60s, especially French and Italian. The interviews contain very few references to Eastern European cinemas and usually in the context of not having watched too many films from the socialist block countries.
Data mining allowed us to see larger patterns, especially in network analysis where co-occurences and paths are impossible for humans to calculate. Likewise, the data on the uses of pronouns and hesitation support our hypothesis but also modifies it. (1) Collective and plural pronouns are used more often by Soviet generation interviewees. It is important to note that currently we have interviews only by female interviewees who grew up during the Soviet era. The addition of interviews by male interviewees who grew up during the Soviet era might change our conclusions. (2) Hesitation marks are also more prominent in the interviews by female interviewees who grew up during the Soviet era, but are also contingent on age and remoteness of memories. (3) Their use also depends on how extensive the answer is; male responses are usually shorter in our sample of interviews.
Finally, there are some conclusions concerning the nature of oral history data analysis, prompting us to go back and look at the context. The results of the pronoun and hesitation analyses in particular need to take into account such additional factors as the topic of conversation (childhood memories vs. recent experiences) and the type of discourse produced. The latter might include a variety of modes: on-the spot attempts to remember; shared narratives of the past; stories that the interviewees told many times and that sound “rehearsed”; answers that are produced for a particular audience / interviewers (young Americans), etc. What kind of stories do the interviewees tell influences the language they use? For example, one of the male interviewees, Vadim Leventhal, tells a story of taking his friends to a film festival and their shock at the Taiwanese film which turned out to be “an art porno.” He does not hesitate a single time. The conclusion is that we all have ready-made stories we told many times vs. on-the-spot memories that use a very different discourse and more hesitation, connective words or pauses. See for example, Irina Leventhal’s interview.
The future of the project is contingent on adding more texts to the corpus of interviews. We have 17 interviews to tag, validate and add to our collection for computational analysis. We might also want to use linguistic statistical and topic modeling tools to analyze the interviews.