What is the Text Encoding Initiative?|
1The Text Encoding Initiative (TEI) is one of the longest-lived and most influential projects in the field now known as the Digital Humanities. Its purpose is to provide guidelines for the creation and management in digital form of every type of data created and used by researchers in the Humanities, such as source texts, manuscripts, archival documents, ancient inscriptions, and many others. As its name suggests, its primary focus is on text rather than sound or video, but it can usefully be applied to any form of digital data. Crucially, it was created and is maintained by the scholarly community for the use of that community. In this little book, we will try to explain the nature and goals of the TEI and to provide some concrete examples of how it may be used in a variety of digital resource creation projects. But before that, we will try to answer a question which introductory texts such as this one tend to neglect: why bother with TEI in the first place? After all, almost everyone now knows how to use a word processor to create digital documents, or a database system to create and manage information, or a simple editor to create web pages. What advantage is there in learning XML or TEI to do the same job?
2There are many answers, but we will discuss three in particular. The first is that TEI XML focuses on the meaning of text, rather than its appearance. The second is that TEI XML is independent of any particular software environment. The third is that TEI XML was designed by and for the scholarly research community, which is also responsible for its ongoing development.
- Meaning before format
If you care about the words in your text and their meanings more than you do about the way they are laid out on the page you will soon encounter some frustrating limitations when using a typical word processor. Almost any such system will readily search through hundreds of pages of text and retrieve all the occurrences of the string “London” for you. But an intelligent search - one in which “London” as the name of a place in Canada is distinguished from that of a place in England, or the surname of an author - requires something more. As Coombs et al pointed out back in 1987, there is a significant difference between the way a scholar looks at textual resources and the way in which most automated document preparation systems text are designed. To use an often quoted phrase, What You See is What You Get means What You See is ALL you Get: that is, systems which are mostly concerned to narrow the gap between what you see on the screen and what you type at the keyboard tend to achieve this by reducing what you can express and store in digital form to information about the appearance of strings of characters rather than their linguistic properties or function. You can adjust the font, size, alignment, or colour of each occurrence of the word “London” in your document as much as you like; but you cannot reliably or consistently use word processing software to indicate which ones are personal names and which are place names.
- Software independence
If you want to share the textual resources you create with other people (or with yourself a few decades hence) you should also be concerned about the tendency of many computer systems to enforce their own idiosyncratic ways of storing information. How easily can a document or database prepared using your current favourite software be used by a different piece of software? Can it be transferred from your PC to her Mac to my Linux box without loss of information? Of course software exists to make transformations amongst the dozens of different file formats currently in use; that is the good news. The bad news is that such transformations are always necessary, and are often complex, lossy, or costly to perform. Even within the same style of software true interoperability remains problematic: a document prepared on one word processor (say Word) may or may not look the same when opened by another (say Open Office); a database created by Filemaker Pro may or may not be easy to convert to one using Microsoft Access, and vice versa.
The information captured in a TEI XML document looks exactly the same to every piece of software using it. There is no special magic, accessible only to one particular piece of software running in one environment. A TEI XML file, like other XML files, is just a sequence of readable characters, with no secret embedded codes. Indeed, a TEI XML document represents a text largely without any consideration of the software which might be used to visualise, analyse or store it. Instead, the content of a TEI XML document represents directly whatever its creator considers significant about its structure and content. If there is a string of characters which the encoder identifies as the name of a person or place, or as a foreign word, or as the title of a song, or as a passage of direct speech (for example), the TEI provides labels to identify them. It may be desirable to visualise song titles in italic when printing them, but to add links to a database entry when displaying them on a website. It might be desirable to include (or exclude) passages of direct speech or foreign words when doing a vocabulary analysis of a text. Texts in digital form can be used in many different ways, after all: that is why we create them. If all that is needed is a new copy of an old book, a digital photocopy will suffice. But if the goal is to make an old text accessible to a variety of new readers, then just reproducing its appearance on the page will hardly suffice. TEI-XML gives us a framework for representing whatever is considered of importance about the text, not just its appearance, so that software can act on the distinctions identified, generating new visualisations and new perspectives. Consequently, deciding what TEI-XML markup to use is always a scholarly activity, always involving conscious choice.
The TEI was first developed over thirty years ago, as a research project in the then-emerging field of “humanities computing”. The original idea was to propose a set of recommendations about how researchers should create “machine-readable” textual resources, which would be adequate to the needs of research, to the extent that consensus on the topic existed, but which would also be expandable as those needs changed and evolved. This combination of flexibility with a focus on the scientific needs of a particular community makes the TEI unusual among other standardization efforts, and is also one of the factors underlying its longevity. In a strictly Darwinian sense, the TEI evolves by responding to changes that are considered beneficial by its user community, while ignoring those which are not.
This close involvement with its user community also has beneficial effects for the novice: awareness of and expertise in the TEI can now readily be found in most disciplines in the humanities: it is not restricted to a recondite area of information science. Institutions like the TEI mailing list TEI-L provide vivid testimony of the heterogenous nature of the TEI community, as well as repeatedly demonstrating its openness, friendliness, and readiness to assist the beginner.
3This book aims to showcase a few of the many varieties of TEI markup without going into much detail for any of them. We hope it will be accessible to a wide readership, including both the practising digital humanist and those who have not yet fully succumbed to the lure of the computer. To understand it, you need only some very general notions about XML technologies and an interest in the traditional problems of textual scholarship. You should also be aware that using the TEI does not mean using the whole of the TEI : on the contrary, most TEI users use only a small subset of what it offers. But deciding which subset is best for your purposes is something that only you can do.
4The first chapter gives a little technical background introducing the XML concepts that underlie TEI markup. The second chapter describes how TEI documents are structured, followed by a number of typical examples in the third chapter “Varieties of textual structure”. The two following chapters then go into more detail about the “TEI Cornucopia”, discussing first some textual features common to most types of document and then a few of the more specialised possibilities offered by the TEI system. Chapter six describes the TEI Header, the mandatory digital title-page of a TEI document that provides its metadata. A seventh chapter “Customizing the TEI” discusses how to tailor the TEI system to the needs of a particular project. We conclude with some discussion of how the TEI itself is organized, its present, and its future.
© OpenEdition Press, 2014