Additional verse features

poem stanza speaker line number brace line group part line epigraph argument
l lg head epigraph argument part next prev I M F

Besides the verse lines and their groupings, poetry often contains other features which require attention from the encoder.

1. Headings and stanza numbers

The title of the poem as a whole should be encoded with head. The location of the head element will depend on how much other information is present; if the head appears at the top of a division that also includes a headnote followed by the poem, then the head will be nested within the div or text element that contains all of this material. If the head simply introduces the poem, then it should be grouped inside the element (div or lg) that encloses the whole poem.

Stanza numbers should be encoded as head, within the lg that encloses the stanza. TEI provides an n= attribute on lg which could also be used to encode the stanza number. However, if (as is frequently the case) the poems in your collection sometimes have stanzas with non-numerical headings, you’ll have a more consistent encoding if you capture all stanza headings with head instead of putting numerical headings on n and non-numerical headings in head. Duplicating the content of head on the n attribute is unnecessary and rarely carries any benefit. The only exceptions are:

  • where the stanza numbering is in roman numerals or some other non-arabic numbering system, in which case encoding a regularized (arabic) numbering on the n= attribute would enable you to display or use a regularized stanza number where that is useful (e.g. in searching).
  • where there are errors in the stanza numbering, in poems where the stanza numbering is a significant navigational aid (for instance, long poems accompanied by commentary keyed to stanza numbers). In such a case (admittedly rare) it might be useful to encode a regularized version of the stanza numbering on the n= attribute.

2. Speaker labels

Poems in dialogue often include speaker labels, even when the poem has no other dramatic features such as stage directions. If using an unmodified TEI DTD, you will need to encode the verses of the poem as drama, using the sp element to enclose each speech and using speaker for the speaker labels.

The extended TEI DTD that accompanies this Guide allows speaker as the first child of lg, without requiring sp at all. Our assumption is that poetic dialogues are not drama in any useful sense, and that treating them as a set of speeches in the dramatic sense is not useful; however, the speaker labels correspond exactly to the function of the speaker element in dramatic markup, and it makes sense to enable this element to be used in other contexts.

The distinction between simple dialogues of this sort, and true dramatic texts, is one of degree. We can imagine a continuum of dramatic texts with full-blown drama at one end, and simple dialogues at the other. Full-blown drama probably includes a cast list, speaker labels, stage directions, and clearly demarcated acts and/or scenes: its presentation marks it as a text intended for performance. As the dramatic text moves away from performance, it may lose some of these features (for instance, stage directions, cast list). Speaker labels are almost always the last to go, since they are the minimal signal that a spoken exchange is taking place without an intervening narrative.

3. Epigraphs and arguments

An epigraph is a short quotation (often with a citation) that prefaces a text or a section of a text. An argument is a brief description of the contents of the section of text to follow, or of the occasion that prompted its writing.

Epigraphs and arguments that are associated with individual poems should be grouped closely with the poem, but like headings their specific position will depend on how the poem as a whole is being encoded. If each poem is treated as a div, the epigraph or argument element should be enclosed within the div but outside the lg that contains the poem proper. If each poem is being encoded within an lg, then the epigraph or argument may go within the outermost lg.

4. Line numbers printed in the original text

Most projects will find they prefer to ignore the original line numbering of the text, since line numbering can so easily be generated automatically and since the original line numbering rarely carries any additional information of interest. However, if for any reason your project wishes to capture this aspect of the original document, we recommend using the fw element, with a type="lineNum". The fw (for forme work) element is intended to capture aspects of the document which are part of the infrastructure of the printed page, rather than part of the flow of the text itself, and line numbering seems to fall well within this category.

In the extended DTD that accompanies this Guide, the fw element has been renamed mw (for meta work), to reflect its slightly broader usage to capture any feature of the document which is not part of the flow of the text: including running heads, line numbers, table headings repeated after page breaks, and similar features.

5. Fragmented verse lines

Particularly in dramatic verse, a single verse line may be divided into two or more parts which together form a single metrically complete unit. The TEI provides the part attribute on l and lg as a way of indicating that two or more part-lines form a single whole. For example:


<sp who="ALB">
   <speaker>Albert</speaker>
   <l>The brilliant <name>Hesperus</name> that leads the day</l>
   <l>Is not so cheering to the pilgrim’s sight</l>
   <l part="I">As she to mine!</l>
</sp>
<sp who="MON">
   <speaker>Montalva</speaker>
   <l part="F">Now, Albert, hear me speak:</l>
   <l>When last I saw her, on the tender theme...</l>
</sp>

The part= attribute has three possible values: I, M, and F (for initial, medial, and final). The first part of the line must be tagged as part="I", and the last part must be tagged as part="F"; there may be as many medial parts as necessary, or none at all (as in the example above). This fragmentation strategy is also useful when dealing with overlapping structures in verse.

6. Braced line groups

In long poems made up of couplets, occasional tercets are often marked with a marginal brace ({). In texts where all tercets are marked with braces, that fact can be marked (if it seems interesting) in a note within the notesStmtin the TEI header, and if you wish to reproduce the braces in your printed or online display, this can be accomplished through a stylesheet. If the braces are intermittent (so that simply rendering all tercets as braced is not possible), then you should record the braces using the rend attribute of the lg element enclosing the braced group of lines. For instance:


<lg type="tercet" rend="braced(yes)">
  <l>It is enough, as much as we should mind,</l>
  <l>As if we were for nothing else design’d,</l>
  <l>But made, like Puppets, to divert Mankind.</l>
</lg>