For the 2021-2022 academic year, I had the opportunity to work with classmates on various projects, some student-run and some individual, all related to localization in some way, shape, or form. This past year has taught me a lot about my professional interests and what areas I excel in, as well as need improvement in. With that, I’ll introduce the projects I worked on and everything I managed to accomplish through them.

WIPO Terminology Collaboration

First and foremost, my main practicum project was the Middlebury Institute of International Studies (MIIS)-World Intellectual Property Organization (WIPO) terminology collaboration research project. The purpose of this collaboration was to work individually to research a given field, identify, and later propose potential terms for submission into the WIPO Pearl multilingual terminology database. Several of my classmates worked on this project, but in general we all chose to work in different fields, making this a largely individual project.

Before I continue, I should first explain some considerations/requirements when recommending terminology to WIPO. Ideally, the source of the proposed term would come from a patent already in WIPO’s database. This patent should be written by a native speaker of the source text language, in this case English, and should originate from a country in which the language is the official working language within the country. Finally, an appropriate context sentence should be included with the term. This context sentence should ideally provide a defining context (a sentence such as “x is…”), or some other suitable context (“x is used to achieve…”).

We started things off in September 2021. We spent time taking a look at WIPO Pearl and exploring any domains that interested us. In my case, I ended up working with nanomedicine, focusing on nanotechnology and its applications to cancer treatment. I spent a great deal of time searching through patents on  WIPO’s database but, unfortunately, the vast majority of my search results were patents written by non-native English speakers. This left me with little choice but to give up on identifying terms through existing patents.

Since I couldn’t work with patents, I chose the next best thing: academic research articles published in peer reviewed journals. The benefit of doing this was that I could select articles written by non-native English speakers, and even articles published in non-English speaking countries, so long as the journals they were submitted to had extensive peer reviewing processes. To better guarantee the linguistic quality of the articles, I did my best to find only those articles that had an extensive number of citations. The large quantity of the articles I worked with had over 500 citations, which, in my opinion, is a fairly large number. As I further narrowed my search, getting more niche with my search terms, the number of citations decreased to about 200. Another limitation I worked with when searching for potential articles was the timeframe. I limited myself to working with articles published within the last two decades, which I hoped would give me the most recent advances in the field. Many of the texts I used came from the last 10 years.

Once I had all the research articles I wished to work with, I built a language corpus using Sketch Engine. The more words you include in your corpus, the richer the term variety will be when performing analysis, so I included a total of one million words in my corpus from a total of 20 research articles. Sketch Engine made it very easy to identify terms used in my corpus, which I then cross-referenced with WIPO Pearl. Unfortunately, most of the terms that Sketch Engine did identify already existed in the term base. My only choice after that was to manually skim through the articles I included in the corpus and identify potential terms myself. This was incredibly time-consuming, so by the end of the semester I only had 15 English terms to propose.

From the second semester onwards, our focus was on finding equivalents for the terms in our non-English language, which in my case was Spanish. At first, I tried applying the same criteria that I used in English (highly cited papers, published within the last 20 years) but I soon ran into an issue: there were not many papers written on the topic with a high number of citations. This indicated (to me) that either the topic is still fairly new to Spanish-speaking researchers, or that the majority of the research is published in English.

The process I used to find equivalent terms in Spanish was simple: I would look up related concepts in Spanish and skim through articles to identify any terms that seemed to be equivalents. For most of my terms, there seemed to be some sort of equivalent, but for others, it was incredibly difficult to find something within the existing body of research. As a workaround to that, I chose to look into non-research articles on the web. Terms from these types of sources can’t be included in WIPO Pearl, but they could still serve as a reference for when Spanish research in the field of nanomedicine grows.

Idem Translations

I worked as a QC specialist for Idem Translations throughout the academic year. The content I worked with was in the medical domain, and I was responsible for checking consistency in formatting between source and target documents, matching id/serial numbers, and following any client style guides, if provided. The texts I worked with were not always in my language pair, which surprisingly helped me focus more on the functional aspect of my job. I found that if I was working with texts written in any of my working languages, I would spend too much time actually reading the texts. This of course slowed me down, but I was able to improve on this issue with more practice. The texts I worked with were typically PDFs or Word doc files. PDFs were not complicated, but some Word files had layout issues that fell to me to fix. Issues such as spacing or incorrect font usage were simple to fix, but issues such as table formatting, while also simple, were incredibly time-consuming and mentally taxing.

LocReady

My role in LocReady was more limited in scope. I mostly did the QA for the transcripts we include alongside our video interviews, but I did participate as an interviewer for one video.

Miscellaneous

In addition to the projects listed above, I also helped a classmate with their task for the Fruit Vendor project. I helped my colleague, Grizelda Ambriz, with the formatting and translation of the Fruit Vendor cards into Spanish. The formatting of the Photoshop assets was a little difficult to work with, since there were limitations placed on the file itself, which prevented us from being able to drag around assets as needed. Sometimes the tags on the cards would expand with the translation, and we would have to manually expand the background boxes. Individually this wasn’t an issue, but when multiple cards have the same tags, and the tags are located in different spots each time? There was a lot of manual work involved in that, which I think could be streamlined in future iterations: since the single Photoshop file hosted the assets for all the cards in Fruit Vendor, it was difficult to scroll back and forth between the assets to get the tags I needed. As a solution, I would recommend splitting the cards among different Photoshop files by category. The assets are already split by category (Telecom vendors, Legal, etc.) so spreading the assets into different files would be an easy task. It would also be less resource-intensive on computers, and scrolling between the assets would be easier as well. For the tags, I would suggest creating a separate folder in the Photoshop file to house those assets. This would make it easier to copy any tags I need for the cards, while helping to maintain consistency in size.