Searching the corpus of Sumerian composite texts

Introduction: the basics
Input and output: display conventions
Aliasing: coping with different transliteration systems
Grapheme separators
Boolean operators: using "and", "or", and "not"
Search scopes: defining the limits of a search

Introduction

The bibliographical search engine was written by Steve Tinney of the Pennsylvania Sumerian Dictionary Project at the University of Pennsylvania, and implemented with the aid of the University of Oxford's Academic Computing Development Team. It searches an XML version of the corpus of Sumerian composite texts for graphemes (i.e. transliterations of individual cuneiform signs), words, and phrases.

For instance, to find all occurrences of the word ur-gir₁₅ "dog", type ur-gir15 into the search box and click the Find button. (You can control the number of results displayed at once by entering a number in the box marked "Show" before you make your search.)

The results will appear in a new window, showing each line of each composition in which the word ur-gir₁₅ appears. The total number of examples found will be displayed [where?]. If that number exceeds the value you specified in the "Show" box, then click the "more results" at the bottom of the screen to continue viewing the items found. A double horizontal line marks the end of the list.

Clicking on the title of a composition in that second window will cause the whole composition to be displayed in a third window [? currently broken] at the bottom of the screen:

[new image needed]

To start a new search, simply go back to the search window. To view the translation or bibliography of a found composition, open another new window in your browser.

If your search finds no items, then the results window will appear blank except for a double horizontal ruling. If the search engine was unable to process your request, a red error message will appear in the results window.

Input and output

The search engine uses ASCII encoding for input, so to enter special Sumerian diacritics into the search box type:

c for "sh" or "shin"
j for "ng" or "g-tilde"
ordinary numerals 0-9 for diacritic numerals (as in the example above)

The search results are displayed just like the rest of the site (see Display conventions page) with some minor differences:

textual features such as Emesal and proper nouns are not displayed in colour
there are no hypertext links to the translations of the compositions found
substantive textual variants are shown in round brackets, separated by a semicolon (as in line 23 of the Cursing of Agade, Old Babylonian version, illustrated above: (tarah;kucu;ance)

Aliasing

It is not necessary to be familiar with the ETCSL preferred transliteration values, as the search engine recognises aliases or alternative transliterations for the same grapheme. For instance, in the first example above, entering the search term ur-gi7 instead of ur-gir15 produces the same set of results:

In fact, as you can see in this example, alternative transliterations have occasionally slipped into the corpus too, despite our best proofreading efforts! The full list of aliases is given on a separate page.

Grapheme separators

The search engine recognises three different ways of separating graphemes in the search input box. Use:

a hyphen (-) to find adjacent graphemes within a word, as in ur-gir15; use this to find determinatives and glosses too: d-utu, for instance, will find ^dutu
an underline (_) or a space ( ) to find graphemes separated by a space (i.e. in adjacent words); lu2_inim, for instance, will find the sequence lu₂ inim
a tilde (~) to find both adjacent graphemes within a word and graphemes separated by a space; this is recommended for finding lexicalised phrases whose hyphenation conventions may vary within the corpus (e.g. the Sumerian for "witness" is transliterated both lu₂-inim-ma and lu₂ inim-ma)

Boolean operators

You can use Boolean operators in the input search box to find collocations and non-collocations of graphemes or words. Use:

and to find occurrences of two graphemes (or words or phrases) within the same line; for instance, lu2 and ur-gir15 will find all collocations of the words lu₂ "man" followed by ur-gir₁₅ "dog" within the same line
or to find either or both graphemes (or words or phrases) within the same line; for instance, lu2 or ur-gir15 finds all instances of both words, whether alone or in the same line as the other
not to find instances of one grapheme (or word or phrase) that only occur without the second grapheme (word or phrase) specified; for instance, lu2 not ur-gir15 will only find occurrences of lu₂ which are not followed in the same line by ur-gir15

The and and not operators are both sequence-sensitive. That is, lu2 and ur-gir15 will find only instances in which lu₂ precedes ur-gir₁₅ in the same line, while ur-gir15 not lu2 finds only lines in which ur-gir₁₅ precedes lu₂.

Search scopes

You can narrow or widen the scope of a Boolean search by adding the following phrases to your search request, where n is any positive or negative whole number. Use:

within n signs or within n words to narrow your search down to a specified number of graphemes or words; for instance, lu2 and inim within 4 signs will exclude all collocations of the two graphemes more than four signs apart
within nlines or within n texts to broaden your search to a specified number of lines or compositions (as defined by the standard catalogue numbering of the corpus)

For instance, lu2 and ur-gir15 within -3 lines will find the following:

results of search for inim and ur-gir15 within 3 lines

For more complex ways of searching the corpus, see Advanced searching with regular expressions.

Back: Searching the consolidated bibliography

Top | Home | Search: composite texts | consolidated bibliography | help with searching

Page created on 29.iii.2002 by ER. Last revised on 12.iv.2002 by ER.