Introducing Dexter

A set of tools for coding linguistic data

Gregory Garretson

Boston University

What is Dexter?

Dexter is a suite of tools for performing custom analyses of language data.

It's designed principally for the analysis of spoken data, but it can also be used for written data.

At present, there are two tools in the suite:

Who makes Dexter?

Dexter is designed and written in Java by Gregory Garretson at Boston University.

The project is run by Cathy O'Connor and Gregory Garretson and is funded by the Spencer Foundation.

To get Dexter, or for more information:

How is Dexter different?

Dexter is the only coding tool with this set of features:

What can you do with Dexter?

Dexter makes it easy to code a transcript or written text for whatever features you're interested in, using any coding taxonomy you like. It allows you to:

Dexter's search engine

The Coder's powerful search engine allows you to perform complex searches including any of the following information:

It also allows you to:

What can't you do with Dexter?

Here are a few things Dexter isn't:

So how do you use Dexter?

Here's the typical sequence of events you go through when using Dexter:

  1. You start out with a document, such as a transcript you've made.
  2. You feed the document into the Dexter Converter, which walks you through the process of converting it to XML.
  3. You take the resulting XML document and open it in the Dexter Coder.
  4. You enter into the Coder a set of codes you would like to apply to the document. They can represent anything at all.
  5. Examples: the discourse moves made by the participants; the various ways topic changes; regionalisms in the participants' speech; attempts at humor made; and so on
  6. You apply codes by selecting portions of the text and clicking. The codes show up as colored regions. They work like an overlay in the form of a transparency, and can be shown or hidden.

So how do you use Dexter?

  1. You save your codes in a code file. These are separate from the document itself, so you can create an infinite number of them (in theory, of course). This is what we mean by "stand-off markup".
  2. To perform an analysis, you use the search function to answer questions about the codes, the text, info on the speakers, or some combination of these.
  3. Examples: Show me all instances of a female speaker interrupting a male speaker who had himself interrupted someone; Show me all instances of a speaker using both the ING and the IN' variants in a turn when they're speaking about childhood experiences.
  4. You can output your search results in various formats or browse through the hits in the document itself. We find that seeing things in context can be a big advantage.
  5. You can go back and forth between adding codes and searching. We have found that this is the best way to develop a robust coding taxonomy; until you've really worked with the data, you don't know what you'll find there.

How can I get Dexter?

Installing and running Dexter is a snap. You have two options:

Java Web Start

The first option, which we recommend, makes use of a technology called Java Web Start. This makes it very easy to distribute software over the Internet. It allows you to install Dexter on your computer with approximately three clicks. And it helps keep your copy up to date.

Manual installation

The second option is to download the program files in a zip file, which you then open on your hard drive. You can run the program by clicking on the program files (which are Java .jar files), or you can create shortcuts to them in a convenient place. However, this option makes it harder to keep your software up to date.

Future features

Here are some things we plan to include in Dexter in a future version:

Why XML?

One thing that users will probably wonder at first is why it's necessary to convert their documents to XML. We'll explain very briefly.

XML is the new standard for digital data exchange and storage. It's a way of describing or "marking up" data so that it can be read by computers. It's a lot like HTML, but it's intended for data description, not formatting. We call the Dexter version DexML. You'll never actually see any DexML (unless you open the data files in a word processor) because Dexter presents the data in a much nicer-looking format, but behind the scenes, DexML is what allows Dexter to do everything it does.

The one downside is that XML, because its purpose is to allow computers to understand data, has a very strict syntax. This means that the formatting of your document must be totally clean and unambiguous by the time it's converted to DexML. And we all know that inconsistencies and typos happen. So how do we deal with this?

Why XML? (continued)

That's where the Dexter Converter comes in. This is a very smart interactive conversion program. It asks you questions about how you've formatted your document and tries to adapt to your conventions as much as possible. It then walks you through the conversion process step by step, doing most of the work itself but asking you to correct any inconsistencies or critical typos in your document. We think this reduces the hassle of document conversion to a reasonable level, and it results in a sparkling clean transcript, ready for intensive analysis. So even though the conversion to XML is a big extra step, it's well worth it.

Dexter is non-profit

We should make clear that Dexter is a not-for-profit project. That means that future development and support of the software depends on:

Obviously, the more users there are, the more likely we will be to secure funding. And the more people are willing to work on Dexter (remember that it's open-source), the less dependent we will be on funding. So we're looking for people to get involved, as users or as developers.

What's the current status?

As of May 2006, the Dexter tools are in beta version.

This means that we are actively seeking people who are willing to try Dexter and give us feedback. We hope there won't be too many bugs, so it should be quite usable. If you're interested, please let us know.

This would also be an opportunity for you to suggest new features to add to the tools, so that they'll be even more useful to you.

That's it!

For more information, please see the rest of the Dexter website:

We hope you'll give Dexter a try.