Summer Institute Of Linguistics

Working With SIL: What It Actually Looks Like

The Summer Institute of Linguistics, now officially called SIL International, is one of those organizations that exists in a weird overlap zone between academic linguistics, biblical translation, and anthropological fieldwork. If you have ever tried to document an undocumented language, you have probably run into their materials. Their orthography development workflow, specifically, is the thing most people encounter first. The rest is secondary. I spent a few years dealing with SIL resources while working on a language that had never been written down before. The orthography design phase alone took about three months because the community had strong opinions about how certain sounds should look on paper, and nobody there cared about theory. They cared about what would actually get taught in a village school where the only printing resources were a hand-cranked mimeograph machine and a stack of reams of cheap paper.

What the Summer Institute Of Linguistics Actually Does

SIL International trains field linguists, develops writing systems for languages that lack one, and produces Bible translations in roughly two thousand languages. They publish a lot of reference grammars through the same channel, and those books tend to be some of the most useful practical grammars you will find because they are written by people who actually lived in the language community long enough to notice the weird edges. Their field methodology manuals, especially the ones on phonology and morphology, are not academic exercises. They are step-by-step guides written by people who have sat in a tin-roofed hut in Paraguay and tried to figure out what was happening with a tone system. Their most useful tool for serious fieldwork is FieldWorks Language Explorer, commonly called FLEx. It is free software designed for building dictionaries, glossed corpora, and morphological analyses in parallel. FLEx runs on Windows and also has a macOS/Linux path through Wine, though the Wine version has always been a bit fragile. You do not need a license. You download it from the SIL website and start entering data. If your work involves sound inventory analysis and orthography recommendation, SWISH is the other tool they built for that purpose. It helps you organize IPA data and generate suggested alphabets based on typological norms. It is not flashy, but it saves you from making embarrassing mistakes like recommending a diacritic that no keyboard in the world can produce.

How the Orthography Development Process Actually Works

The SIL approach to creating an orthography follows a fairly standardized pipeline, but it is not automatic. You cannot just feed raw phonetic data into a machine and get a writing system out the other side. Here is the process as it typically goes. First you do a phonological analysis. You collect a solid set of minimal or near-minimal pairs and determine which sounds are contrastive. This step usually takes a few weeks for a language with a simple inventory and considerably longer for something like a language with tone, vowel length, and a consonant inventory that includes ejectives. Once you have the phonemic inventory, you decide on grapheme choices. SIL has specific guidelines for this part, mostly revolving around clarity and printability. One grapheme per phoneme, avoid diacritics if a simple digraph works, and make sure the system is compatible with standard typewriters and later, basic computing environments. Then you run the proposed orthography past native speakers through reading tests and writing tasks. This is where things either go smoothly or fall apart. I had a case where the proposed orthography looked perfectly reasonable on paper, but when the community tried to read texts written in it, they consistently misread a particular vowel distinction. The phonological analysis had been correct, but the cognitive mapping between the graphemes and the existing oral tradition was wrong. We switched to a different vowel notation and it took about two weeks for reading speed to normalize.

After that you produce teaching materials, a primer, and eventually a complete text. SIL provides templates and style guides for this phase, and using their templates is worth it because everything lines up correctly when you send files to their typography team or to a local printer.

Practical Advice From Someone Who Has Done This

Don't skip the community acceptance phase. You can have the most elegant orthography theoretically, but if the people who are going to use it every day don't trust it, nothing else matters. I saw a project derail completely because the linguist assumed acceptance based on a single focus group, when in reality the broader community had major objections that only surfaced when actual literacy classes started. FLEx is powerful but it has a learning curve that most people underestimate. The database structure is relational, which means you need to understand how gloss fields, lemmas, and example sentences connect before you start entering data. I wasted about a week entering raw data before I realized I had built the database wrong and would have to restructure it. Spend the first two days reading the manual and watching the tutorial videos, then start working. Another thing nobody tells you: tone marking in FLEx is possible but tedious. If your language has phonemic tone, you will end up spending a significant amount of time manually tagging each morpheme. There are automation features, but they are not reliable enough to trust completely. I ended up writing a simple script to batch-process tone marks from my phonological notes into the FLEx import format, which cut the tagging time from roughly forty hours down to about six.

Common Problems And When SIL Tools Fail You

FLEx does not handle complex morphophonology gracefully if you are trying to represent underlying forms versus surface forms at the same time. The glossing layer is good for interlinear text, but it was not designed for detailed phonological rule representation. If your work requires that level of detail, you need a separate tool or a custom workflow, which is why some linguists pair FLEx with a basic annotation system like ELAN for phonological data while keeping FLEx for the dictionary and corpus components. The organization's translation approach also has limitations if you are working outside their standard framework. Their Bible translation workflow assumes a Christian theological context and specific source text traditions. If your project is secular or non-scriptural, you can still use their methods and tools, but you need to adapt their workflow yourself. There is no off-the-shelf support for that.

Where to Get Their Tools

FLEx, SWISH, and the other SIL field linguistics tools are available for free from the SIL website. You register, confirm your email, and download the installer. There is no subscription. The manuals are also freely available as PDFs, and they are actually readable, which is rare for technical documentation in this field. One thing to watch for: SIL occasionally changes the naming and grouping of their software products. The Paratype line was folded into FLEx years ago, so if you are searching for Paratype you will not find it as a separate download. Look for FLEx instead, and check the SIL field linguistics tools page for the current catalog. If you are doing language documentation work, start with their "How to Describe a Language" series. It is not comprehensive, but it covers the basics in a way that is immediately usable in the field without requiring a linguistics degree to parse.