The first markup is a simple annotation of how many translations are identical. Only absolute typographical identity counts -- thus differences of punctuation or capitalization or spelling count as non-identical. This markup is shown in parentheses. The first figure is the number of identical readings, the second figure is the total number of readings.

Thus, when there are only two translations compared, the possible codings are either (1/2) -- different or (2/2) -- identical. We also have comparison among three texts, and four texts. With four texts, the annotation (2/2) is slightly ambiguous. It could refer to a case where there are two identical texts and two additional identical texts or two identical texts and then two others that differ among themselves as well.

The second annotation is only for those segments that are not completely identical -- that is, not marked (2/2), (3/3), or (4/4).

There are three possible codings for the set of segments: 1 is a mistaken translation; 2 are equivalent translations -- no meaning difference; and 3 are non-equivalent translations -- there is some difference in the denotation, connotation, or implications of the translations.

Codings of 1 can only be marked by comparison with the original. Since we have not included the original in the set of comparisons, nothing is currently marked as 1. Following the one is an indication of which text(s) are in error. Texts are labelled consecutively A, B, C, D, etc.

Codings of 2 indicate equivalent translations. Thus a coding of 2 must be followed by a mention of at least two texts that are equivalent. Identical texts are indicated by immediately consecutive letters, equivalent texts are separated by commas, sets of equivalent texts are separated by a semi-colon.

2[AB,C] would indicate that texts A and B are identical, and equivalent to text C.

2[A,B;C,D] would indicate that texts A and B are equivalent and C and D are also equivalent.

Codings of 3 indicate meaning differences. Equivalent texts are indicated by consecutive letters, texts that differ in meaning are separated by commas. If there is no coding of 2, then consecutive texts are understood to be identical.

Thus, there is some ambiguity and some redundancy in the coding.

3[ABC,D] indicates that texts ABC are equivalent and differ in meaning from text D. If there is a 2 coding, that would indicate how the texts are equivalent, e.g., 2[AB,C] indicates A and B are identical and equivalent to C, while 2[A,B,C] indicates they all differ, but are equivalent. If there is no 2 coding, then A,B, and C are identical. It should usually be possible to deduce the 3 coding from the 2 coding.

At this point, clearly most of the 2 codings should be handled by our IL2. Possibly some of the 3 codings, but probably most of the 3 codings should not be handled by merging at IL.