Vocabulary File Format

For a vocabulary file each first word in each line of the file is taken to be a vocabulary word. Comments are lines that start woth a single comment char -- usually a ";"
This character does not have to be the first character in the line, but it has to be seperated from anything else in the line by at least one white space. In other words: it is not possible to have a word that has the spelling , but words that start with that character are possible.
Possible files are wordlists or dictionary files. As we're using tcl to read the vocabulary files, avoid lines beginning with quotes. If in doubt, use brackes to ensure correct interpretation.

Example 1

;
; This is only the beginning of the dictionary that was used
; for the 1995 VERBMOBIL EVALUATION. If you write comments,
; remember the whitespace between the comment sign and the rest of
; the comment.
;
{"Ogai}                      {? OEH G AI}
{"Ogai-Jahrestagung}         {? OEH G AI J AH R E2 S T AH G U NG}
{"Ogai-Tagung}               {? OEH G AI T AH G U NG}
{"Ol}                        {? OEH L}
{"Uberlegen}                 {? UEH B ER2 L EH G E2 N}
{"Uberlegen(2)}              {? UEH B ER2 L EH G N}
{"Ubernachtungskosten}       {? UEH B ER2 N A X T U NG S K O S T E2 N}
{"ahnlich}                   {? AEH N L I CH}
{"ahnliche}                  {? AEH N L I CH E2}
{"argerlich}                 {? ER G ER2 L I CH}

Example 2

;
; This is also a valid vocabulary file.
;
{"Ogai}
{"Ogai-Jahrestagung}
{"Ogai-Tagung}
{"Ol}
{"Uberlegen}
{"Uberlegen(2)}
{"Ubernachtungskosten}
{"ahnlich}
{"ahnliche}
{"argerlich}

Note on Variants

Variants can be added to the vocabulary by adding a string in parentheses to the main form. Only if the variant is listed in vocabulary and dictioanry it will be used in the search. Very important: always list the main form (without parentheses) first. The search will ask any language model penalty from the lm module using the only trunc (everything up to the first open bracket) of the variant, thus ensuring that they get the same probability as the base form.

Example 3

;
; this is a vocabulary file with pronounciation variants
;
einen           whatever I write here does not matter
einen(ohne_en)  as currently only the first entry of each line
einen(ohne_ein) is used for the vocabulary