Word search

A word search matches a pattern against one token at a time. For sequences of two or more words, see Multi-word search; the two are not separate modes — typing a space is what moves you from one to the other.

Development status

The search engine is under active development. As things stand:

queries are case-insensitive by default; the Case sensitive switch changes this (see the search form);
results are grouped by normalised word form — that is, by the sequence stripped of editorial markers such as square brackets;
determinatives (DIŠ, DINGIR, MUNUS…) may still produce unexpected sorting.

Query input and pattern matching

The search box accepts any sequence of characters, whether a complete word form (i-ša-am) or a fragment of one (ša-).

Matching is by substring

By default a pattern matches anywhere inside a token. Searching for the preposition a-na therefore also returns the toponym a-ia-la-na-za, the personal name ḫa-na-ak-ka₄, and every other form that happens to contain that sequence. To match a whole word, anchor the pattern — see below.

Searching by dictionary headword

The same box also searches lemmata, once include lemmata is enabled in the form. This retrieves every attested form of a lexeme without your having to enumerate them: searching šâmu (or shamu) "to buy" returns a-ša-am, i-ša-am, i-ša-am-mu, i-ša-am-šu, ta-aš-ta-ma and the rest of the paradigm, whatever their spelling.

It is an extension of the same search, not a different one: the pattern rules below apply unchanged.

Regular expression syntax

With Regex search enabled, the pattern is read as a regular expression, which is what makes the operators below available. For the syntax in general, standard references such as the Wikipedia article on regular expressions apply — what follows is the part that earns its keep in a cuneiform corpus.

1. Anchors

Anchors tie the pattern to the beginning or the end of the token:

^i-ša matches sequences beginning with i-ša (e.g., i-ša-am), excluding pa-ni-ša;
a-am$ matches sequences ending with a-am;
^a-na$ matches exactly the preposition a-na, excluding longer forms.

2. Wildcards and quantifiers

These match a variable amount of text:

. (dot) matches any single character (letter, digit, whitespace);
? (question mark) indicates zero or one occurrence of the preceding element;
* (asterisk) indicates zero or more occurrences of the preceding element;
+ (plus sign) indicates one or more occurrences of the preceding element.

Combining them

Wildcards earn their keep in combination. Take the pattern -ru.?$, which matches:

The reading -ru for the sign RU;
Followed by any single character (.);
Where that character may occur zero or one time (?)—i.e., optionally;
At word boundary ($), with no subsequent characters.

One expression thus retrieves maḫ-ru, maḫ-ru₃ and maḫ-rum together, covering both homophonic sign variants¹ and forms with mimation².

3. Alternation

Alternation matches any one of several patterns:

a|b matches either a or b;
(šu|su)-nu matches both šu-nu and su-nu;
[aeiou] matches any single vowel character.

Examples:

^(a-na|i-na)$ retrieves both prepositions a-na and i-na simultaneously;
ma-[ḫh]i-ru matches both ma-ḫi-ru and ma-hi-ru, accommodating orthographic variation;
DUMU(\.MEŠ)? matches both DUMU (singular) and DUMU.MEŠ (plural).

Reading the results

Aggregation and ordering

Results are aggregated and sorted alphabetically. They are clustered by normalised word form and by class (PN = personal name, GN = geographic name, DN = divine name, and so on). The Additional groupers control adds further criteria on top; word form and class remain as the baseline.

Occurrences

Each attested form is followed by the list of its occurrences, where the notation of the line number is itself informative:

plain (2, 15) — the word stands in an undamaged passage;
in square brackets ([2]) or half-brackets (⸢2⸣) — the passage is damaged, wholly or in part; the reading depends on a restoration;
followed by an asterisk (3*) — the form is attested only in an editorial note, as an alternative reading or a restoration, and not in the text as edited.

Clicking a tablet siglum opens the text, with the term you searched for highlighted in yellow. Sigla of texts you are not allowed to open are shown without a link — see the note on corpus-wide searching.

Search results

Seminara 1998, L'accadico di Emar, p. 90-97. ↩
"Etymologically justifiable": the reading mahrum is inadmissible, even in post-Old Babylonian contexts, as no Akkadian dialect attests third-person plural masculine verbal suffixes with mimation (Seminara 1998, L'accadico di Emar, p. 91). ↩