Word Search Help 0.61

Main Pattern

Enter a pattern in the "Basic Pattern" field and press Submit. You can use "standard" wildcards, * or ?. For example, F* to find all words starting with F or ??F* to find all words with the third letter F. You can specify groups of letters with brackets; for example, [AHIMOTUVWX] for letters with mirror symmetry or [a-m] for letters in the first half of the alphabet. The special character @ can be used for any consonant, # for any vowel, or & for any standard Roman numeral.

If you check the "any order" box, it will find words with the letters in any order. For example, BE* will find any words containing both B and E. Only one set of brackets can be used when this box is checked. You can also use the "Repeat letters checkbox" to allow the specified letters to be repeated one or more times. For example, EHORS with both boxes checked would give results including HEROES, HORSE, HORSES, HORSESHOE, and SHORE.

Additional Constraints

Regular Expression

You can search for the intersection of those results with any regular expression. For example, if you enter B* for the basic pattern, then press the #Add Constraint# button and enter f[aeiou]+f and select Regular Expression, you'll get all words starting with B and containing two F's separated by one or more vowels. If you check the Not box at the beginning of the line, it will allow only words which do not meet the criterion. For example, if you check the box and use a regex of e, it will find only words with no letter E. (You can also do this with a regex of ^[a-df-z]*$.) Normally, these searches are done against just the letters in the result. For example, José M. Rodríguez would be considered as JOSEMRODRIGUEZ. If you check the Raw box, the original text would be searched.


You can also intersect with another pattern, using the same syntax as in the basic pattern field. The Raw checkbox works the same way as described above.


You can filter by a subword or secondary word. Enter a series of specifications of the following forms: Separate numbers by commas. For example, if the main word is BOLSTERED, 4,2L-1:-3 would be SOLDER and would match.


This option will look at the weight of each letter, which can either be its value in Scrabble® (e.g., H=4) or its position in the alphabet (e.g., H=8), possibly with a multiplier, added up over the whole word. If the search field is left blank, a multiplier of 1 will be used for each letter. If a series of digits is entered (e.g., 3112), the multipliers will be used for the corresponding letters--x3 for the first letter and x2 for the fourth letter. These digits can be followed by a plus sign to use 1 for the remaining letters in the word or a minus sign to use 0 (skip). Finally, a second set of digits can specify weights for letters at the end of the word (e.g., +31 to use a weight of 3 for the next-to-last letter). Some full examples on the word EXAMPLE: 3111 with Scrabble chosen will be 15: 3x1 + 8 + 1 + 3. 12+21 with alphabet chosen will be 112: 5 + 2x24 + 1 + 13 + 16 + 2x12 + 5. After this specification, put either <, =, or > followed by a number. +>50 will give all words with total weight greater than 50.

You can press the Wizard button to get a popup form to help you with this.

Letter Matches

The option allows you to specify that certain characters within the word must match or have another relationship. The simplest case is something like 3=8, which means that the third and eighth letters are the same. A more complicated example is 3>-3+^5 which says that the third letter has to be more than five places later in the alphabet (+^5) than the third letter from the end (-3).;

You can press the Wizard button to get a popup form to help you with this.


The option allows you to specify a pattern of matching and nonmatching letters: a word which might be a solution in a cryptogram for the other word. For example, ELLISVILLE would match REENGINEER and ABCABC would match words such as ATLATL, BONBON, and TSETSE. Use * to specify that the word has no matching letters.;


This option allows you to specify a pattern of word lengths and punctuation. A simple example would be 5 4, to represent a five-letter word followed by a four-letter word. You can also use # to represent any digit (e.g., #### 3 3 4 would match 1776 and All That), ' and - (quote and hyphen) to represent themselves (6 2 '## would match Spirit of '76), * to represent a word (sequence of letters) of any length (6 * 9 would match Ludvig van Beethoven or George Albert Boulenger, or ? to represent any punctuation mark other than quote and hyphen. Note that all Unicode quote-like and hyphen-like characters are interpreted as ordinary quotes and hyphens here.

Other Filters

You can filter the final list of results with the minimum and maximum length boxes.


You can choose among Wikipedia, Wiktionary, and any user-defined word lists to which you have access. For both, by default, redirects are not searched; most of these have little logological interest. You can search the content of the pages of matching entries (but not currently of redirects), either for an ordinary pattern or a regular expression, just as for the original string. For example, you can search for articles that match "*germany*" but the content does not match either "east" or "west". You can also search by the size of the page. Note that this feature may be slow, as the page content must be retrieved over the internet. In this case, queries may time out and display only partial results with a "more" link.

You can also search for categories in one of three ways: an exact match, the category and descendants (to five levels), or contains, which is normally the most useful. Exact matches and contains are faster. There is a lookup button which allows you to search for categories. It is currently a bit slow because it tries to give you more useful matches.

Finally, you can filter based on the number of incoming links. This feature a useful proxy for finding well-known answers. Links which are not in the main namespace count as only ⅓.


The download is approaching the end, with 0, A-Iso, and M-Th complete as of September 26.


I have done only a very little bit of the download, enough to test the code. By default, only English terms are searched; there is a checkbox to allow foreign words as well. If you want a specific foreign language, you can search for categories beginning with the language name.

There are also checkboxes to filter to capitalized or uncapitalized entries. (Wikipedia does not make this distinction.)


The main page should be stable and functional. There are links to go to the previous version and the beta version.