During this week off from translating I decided to put on a new hat, an auditor's hat. I scheduled time to finish collecting my job stats for the quarter. I filled out a spreadsheet with my time spent on projects, what the service provided was, whether it was a translation, editing or other quality management activity, and how it was billed (my hourly or word rate).
Much of the information I have put on paper is not news on its own, and still needs to undergo some processing to give the whole activity some purpose beyond information collection as an end in itself.
I want to know:
What does my "best" project look like? That is, where was I most efficient in terms of time management and amount billed. Is it a project where there was little research to do? Where was the time on research spent? in locally saved TMs, glossaries, web, databases on the web? This leads into another question which I found relevant to my professional development.
What project was the most intellectually stimulating? the one that solidified my understanding of a field I was interested in before. In my case, it is chemotherapy drugs. During this quarter I worked on several of them, but one in particular stood out. Why? It provided empirical data about the trial rather than just the information on how the trial was structured. I have not yet recovered my web search queries that I made during the job, but this should be my next step to reflect on my thought and research process and to understand what worked and why and see how efficient I am at using the web. We could all benefit from this because whether we are under the gun or not to finish, we just want to do what works.
My hypothesis is that we can use programming and text processing/NLP techniques tools as a shortcut to a good search query on items that will prove problematic for understanding or present translation doubts. The tools are out there (e.g. NLTK, Scrapy using Python) but these need to be welded together to create a working text pipeline. that gets you from source text chunk > explicit doubt about translation > effective search query > applicable search result that resolves the doubt.
The benefit of strategizing our research is time saved. Yes, I expect I will still save time by first reading the document and analyzing the structure and determining the audience and then later launching into web search with our learned parameters (audience, terminology, structure, context in mind) to find what we I am looking for:
I expect my web searches fall into a few categories
What? predict to be create combinations that are better formed than the last time around, so that you get the most relevant source or target webpage/document from your final web search, thus increasing your chances you are finding and using the most natural target equivalent.
Of course, it's best to restrain yourself from running a search query until you've gained a more thorough understanding of the document as a whole and even after you have worked and translated through the document's other sections, but my goal is to reduce time spent in the search browser, and craft a strategy or tool that does the heavy lifting of web search and extraction.
Write, post, publish, prophet!