Groyser Verterbukh fun der Yidisher Shprakh online

The Groyser Verterbukh fun der Yidisher Shprakh is a Yiddish-Yiddish dictionary that was only partially completed. Between 1961-1980 four volumes appeared which covered the letter alef and were edited by Judah Joffe and Yudel Mark. See here for a short history from A history and guide to Judaic dictionaries and concordances. The existing four volumes can be downloaded in the Digital Yiddish Library here. Raphael Finkel (see Refoyl's Yidish Veb-Bletl) has now created a searchable version of the 4 volumes, see the link below. He has added the following notes: "1. The search term must be in Yiddish letters. There are some clickable letters near the top that you can use if you don't have a Yiddish keyboard. 2. If you put diacritics on your Yiddish letters, they will be required in any match, but if you leave them off, the letter will match with or without diacritics. For example, you can use a bare fey to match either a rofe-fey or a pintl-fey = pey, but if you use a rofe-fey, it will not match a bare fey or a pintl-fey. You do need to distinguish final letter forms from medial letter forms. You can use two yuds, two vovs, or a vov followed by a yud instead of the corresponding ligatures tsvey-yudn, tsvey-vovn and vov-yud. Or you can use the ligatures. 3. My software currently limits its output to 10 matches. It shows exact matches (full words) with a yellow background and partial matches with a pink background. It expands some (but by no means all!) abbreviations to fuller text with a green background. It tries to show exactly one dictionary entry per match, but the technique it uses to determine the start and end of a dictionary entry is sometimes wrong. 4. The groyser verterbukh usually spells words with a shtumer alef whereYIVO would use diacritics such as a melupm-vov or a khirik-yud. However, for citations before about 1800, it uses the original spelling, whatever that may have been. My software does not attempt to adjust spelling. 5. The OCR (optical character recognition) program is my own invention (I sound like the white knight in Alice through the Looking Glass!). It is by no means perfect. I hope to keep improving it, and as I do so, I will every so often re-generate the data files behind the search page. It takes about 4 hours of computer time to generate those files."