R. David Lankes
New librarianship, based on conversation theory, concerns itself with two levels of language being exchanged between conversants: L0 and L1. L0 is the language exchanged between two conversants where at least one of the parties has little knowledge of the domain being discussed. It tends to be very directional (do this, now do this). Most of the discourse is negotiating meanings and terms at a very simple level. L1, in contrast, is exchanged between two parties with a high understanding of the domain discussed. Here conversations tend to use special language, explore more “why” questions, and establish structural relationships between concepts.
Let’s use a simple example to illustrate the difference in language levels. Take the following sentence:
“Our catalog uses MARC to present our users with a great searching experience.”
If you are a librarian, this is a meaningful sentence. You might even ask “how can MARC impact a user’s experience?” However, if you are not a librarian, this sentence is a jumble. Are we talking about catalogs like from a store where I pick out sweaters—who is Marc and why is he so helpful—do we really want to make it easy for users—that is people who use drugs? As we saw in our previous word game, preexisting structures and contexts to words matter a lot. The more relationships and contexts of words that are shared by conversants (what we will call agreements), the higher the possible level of discourse. A high level of shared contexts equals L1.
These levels of language have real implications for the systems we present to users. Systems can either attempt to work at differing levels of language, to bring users from L0 to L1, or bring the system from L0 to L1 (see the example of a search engine interface in figure 166).
It does little to educate the user about how to interact using high-level language. It is built around an assumption that users will be communicating their needs in L0. The system will use complex algorithms and information-retrieval techniques to make up for the fact that the search engine will probably be getting a very anemic query. If you do take the time to learn the language of this system, you can actually use some rather advanced language to improve your results (in this case, using a query language with +, –, ~, and quotes).
Help desks and search are often used in systems where there is an anticipated difference in the languages of the users and the system builders. In libraries, for example, reference as a function came about because indexes and classification systems were too complex for many library users. The idea was to provide a human intermediary as a sort of bridge between a person with a question and the complex language used by library systems.
The second approach is what Pask would refer to as learning systems that systematically bring users from L0 to L1. One of the best examples of increasing language levels in systems can be seen in modern games. Where once games came with long and in-depth manuals, today complex games actually incorporate learning into the game itself. The first level is often a form of in-game tutorial, familiarizing players with the basic mechanics of the game while still advancing the game’s narrative.
Of course the approach of raising the user’s language level is founded on the rather dubious assumption that a system can change the user. It is also antithetical to the user-centered paradigm dominant in today’s system development world. A third approach is to assume the user is at L1 and it is the system that needs to catch up. The use of tagging and annotations in web systems demonstrates this approach. Here users incorporate their own language. Systems can then look for patterns in language use to provide information.
Nobody Is Born Speaking Dewey
It is worth stopping here for a moment to explore the real implications of these different languages. Think about the basis of a great deal of library work: classification or, more broadly, information organization. To make things easier to find, we have developed a whole host of common languages: the Dewey Decimal Classification and the Library of Congress Classification (LC), just to name two. The idea being that by mapping a body of knowledge to one system of terms, we can collocate items on shelves and make it easier (at least more efficient) to find something. We are using language to do this. In fact, Dewey and LC are the end result of an ongoing conversation centered on the question, “How can we organize the world’s knowledge?”
So why, then, doesn’t this approach always work? To be more precise, why does it work so well for librarians and so poorly for a mass of patrons? This is a great example of L0 versus L1. When patrons try to navigate Dewey (or LC), particularly in catalogs, they are working at or pretty close to a simple directed “click here” level. Once they get a call number, off they go to the shelves. Librarians, in contrast, understand the worldview of these classifications and so can manipulate the system much more effectively. So one conclusion that one could reach is that we need to bring patrons up from an L0 to an L1 conversation. How? Well, one wide-scale approach has been library instruction. Another approach, the one Pask spent most of his effort on, was building capabilities in the system to teach patrons domain knowledge and transition them to L1 use of systems. Never thought of your catalog as a learning system? It shows.
There are big implications in terms of digital systems and language. One approach to the L0/L1 gap is in interface design. By doing a good job of anticipating the user’s needs, we can have the system do sophisticated things with little input from the user. Google is a great example of this. The old way of doing search was to provide two interfaces: simple search and advanced search. L0 users go to simple, and L1 users head to advanced. Google even combined this into one interface. If you are at an L0 level, just type a term and Google uses clever algorithms like Page Rank1 to assume that simple use of language can be best answered by the most popular information. If you are an L1 user, in the same box you can add cool things like quotes, ~, +, –, and so on. The more sophisticated your knowledge, the more sophisticated interaction you can have with Google. However, note that the interface does not teach users or help them to transition from L0 to L1.
Another approach to the language divide is to build learning systems. This is an interface that attempts to teach as it works. Librarians attempt to model this behavior. Good reference librarians in person or online don’t just provide an answer; they help users understand how they found an answer and impart the skills to the users so they can selfserve. This approach also inspires things like “tool tips,” where little explanations pop up when you hover a mouse over a button in Microsoft Word. Word also shows how just trying to instruct is insufficient… think “Clippy.” Perhaps the most effective demonstration of this learning approach can be found in videogames as mentioned before.
A great number of Web 2.0 sites point to another approach all together. Minimize the functions and interfaces of the system to allow users to use their own language and domain understandings. This is the allure of tagging. By letting the users add their own terms or letting users link differing information in their own way (think mashups or FaceBook pages), users create their own L1 systems. Of course, these can create a sort of Tower of Babel, but it is effective in circumstances when there is already a set of common terms or agreements. At the least, by exposing the process of language development, users have a better chance to come up to an L1 level and become more effective and efficient users of systems.