by Tyler F. Williams

Search Capabilities

The General Search Window

Accordance's General Search window is more than adequate for even the most complex searches. Search arguments ranging from a single term to a wide variety of commands and options can be entered into the argument entry box either directly, or by means of menus and dialog boxes. Words can be entered directly in the Search argument box or can be entered from lists of inflected words or lexical forms (limited to tagged texts). This comes in handy when unsure of a word's spelling, or when making sure all forms of a word are entered with English searches (e.g., Walk, Walked, Walking, Walks). Another nice touch are the English glosses provided for the Greek and Hebrew lexical word lists. It is also possible to display biblical references from the Search window by toggling the search entry box to Verse reference mode, and then typing in a verse reference. Only as much information as is necessary to identify a biblical book needs to be entered. So, for instance, to display Obadiah 12 all you have to enter is "O 12" as no other biblical book starts with the letter "O" (Common abbreviations for most biblical books are also recognized).

There are a number of easy to use features that give users the ability to determine an almost endless variety of options and parameters for searches. The text to be searched, the search range and field are all set by pop-up menus, while most other search options are entered via menu commands or dialog boxes. Search ranges are defined by the user. The entire Bible, the OT or NT, a biblical book or collection of books, or any noncontiguous portions of a text can be defined and stored as the search range. This is an extremely versatile feature, making it possible to limit a search to any number of verses or books such as the Yahwist's narrative in the Pentateuch, the Deuteronomistic History, the psalms of lament, the gospel passion narratives, Paul's letters, etc. Similarly, search field or boundary options are easily set via the search field pop-up menu. This allows users to define the context or field for searches that include more than one element. Thus, the results of a search for "Lord <and> God" can be limited to places where they occur together in the same clause, verse, chapter, or book, etc. Not all text modules have the same boundary options, however, as the fields are determined by the text's punctuation and tagging (Details on boundary-related punctuation of specific texts are clearly delineated in an appendix). For English versions the search boundaries can be set to verse, clause, sentence, chapter, paragraph, or book; while for the Greek NT they can be set to verse, chapter, clause, sentence, or book. Unfortunately, the search boundaries for the Hebrew MT and the LXX are limited to verse, chapter, and book. Despite these limitations, this is an extremely powerful feature as it permits searches across verse and chapter divisions. Even though boundary options are more limited with the Hebrew Bible and the LXX, there are ways to get around these limitations. For instance, to ensure that a search for a combination of terms or a grammatical construction is not limited by verse divisions, the field can be set to "chapter" or "book" and used in conjunction with a proximity command.

Entering search criteria and arguments is straightforward with Accordance. Entering single or multiple terms is facilitated by word list dialog boxes (as noted above), while the syntax for search arguments is clear and easy to master. Commands are indicated by surrounding brackets: angle brackets (< >) for relational operators, and square brackets ([ ]) for stand-alone commands. With Accordance a full complement of relational operators are available. This includes the standard Boolean operators (AND, OR, XOR, NOT), as well as a number of additional commands:

• <FOLLOWED BY> will find every occurrence within the selected field where the first search term (or expression) is followed by the second.

• <PRECEDED BY> will find locate place where the search terms are in the same field, and the first term is preceded by the second.

• <WITHIN ? Words> limits the field of search between the two search expressions to a user-specified number of words. For example, "Lord <WITHIN 1 Words> God" will find all occurrences where "Lord" and "God" are adjacent, whether "Lord God" or "God Lord."

In addition to these relational operators, Accordance also has three very useful stand-alone commands:

• [FIELD Begin/End ] can be used in combination with the <WITHIN ? Words> command to specify the number of words from the beginning or end of the search field that the search term will be found. For instance, "David <WITHIN 2 Words>[FIELD Begin]" will find each case where "David" occurs no less than two words from the beginning of the search boundary (i.e., verse, phrase, chapter, etc.).

• [LINK - ] will include the arguments from another window, either a Search or a Construct window, and can be used in combination with other arguments and operators.

• [CONTENTS - ] will substitute verses from another Search or Reference List window (this is an extremely useful feature if you want to compare, for example, the results of two searches).

All of the above operators or commands can be typed in manually, selected from the "Enter Command" submenu of the "Options" menu, or entered by intuitive keyboard equivalents (e.g., shift-command-A for <AND>, shift-command-N for <NOT>, etc.). In addition, like verse references, it is only necessary to type enough to identify the command. Thus, entering <A> is the same as entering <AND>, <W 5 W> is equivalent to <WITHIN 5 Words>, etc. The operators can also be combined on the same search terms or expressions, with some obvious limitations. For instance, <OR> and <XOR> cannot be used in combination with other relational operators.

Search expressions and commands can be grouped with parentheses, and in English texts no special delimiters need to be used to mark off a phrase (in tagged texts if a phrase does not have quotation marks it will search for the same combination of lexical forms, not inflected forms). Search terms can also employ wild-card symbols both within single words or phrases. An asterisk (*) is used as a global wild-card (i.e., it stands for any number of letters in a term), while a question mark (?) indicates any single character in a word. Thus, "b?d" will find any three letter word that starts with a "b" and ends with a "d" (e.g., bad, bed, bud), while "b*d" will find any words that begin with the letter "b" and end with "d" (e.g., , etc.). It is even possible to exclude certain words in wild card expressions by the use of a hyphen. For instance, the search expression "b*d-bed" would find all words that begin with a "b" and end with "d" except "bed." Similarly, the characters for the (?) wild card can be specified by placing them in parentheses after the question mark, or limited by preceding the characters with a hyphen. For example, "b?(eu)d" will find all three letter words that begin with "b" and end with "d" where the middle letter is an "e" or "u" (e.g., bed, bud), while "b?(-u)d" will find the same words that do not have a "u" as the middle letter (e.g., bad, bed).

Results of searches are displayed in a "text window pane." All the verses that meet the search criteria are displayed with the matching terms highlighted (assuming that the text in the window pane is the same as the search text). The number of matches (or "hits") is displayed in the top right hand corner of the window pane, along with a fraction that indicates the current verse over the total number of verses in the text pane. While the data these figures represent is helpful, they can be a bit confusing. For example, the number of "hits" does not necessarily mean the number of matches for a search argument. This is due to the fact that the "hits" number represents the total number of matches for each expression in the search argument. So if the search argument included two expressions, the number of "hits" would equal the number of times both expressions are matched (e.g., the number of hits for a search for "Lord <AND> God" would include the sum total of the occurrences of "Lord" and "God"). Navigating through search results is effortless. The scroll bar can be used to move through the verses (with the current verse conveniently displayed in the reference box at the bottom right corner of the window), or text access buttons at the bottom left of the window can be used to move the results up and down one verse, chapter, or book. A specific reference can also be typed manually into the reference box to go directly to the verse or passage.

Displaying search results (or select passages) in parallel columns with different translations and versions is only limited by the number of versions installed, memory, monitor size, and the display font size. For instance, with a standard 14-inch monitor and 10 point Geneva as the display font, it is possible to have four parallel columns displayed (leaving room for the Amplify palette). When more than one text is displayed in parallel Text window panes (e.g., the Hebrew MT and the NRSV), the texts scroll simultaneously aligning parallel references at the top of the window. On the whole the parallel scrolling works without any hitches, even when scrolling through books or passages that are significantly different between versions, such as, for example, the Hebrew and Septuagint versions of Jeremiah. Occasional problems do arise at some places, however. For instance, psalms superscriptions that make up only part of the first verse in the Hebrew text (e.g., Psalms 50; 73; 74; 78; 79 etc.), are not be displayed in most English versions. This is due to a number of factors. Accordance's simultaneous window scrolling is based on a one-to-one match between verses. This, along with the fact that most English versions separate the superscriptions from the Psalm text proper (e.g., NIV, NRSV, NASB, etc.), makes it necessary to choose between linking the superscription or the first verse in the English text with the first verse in the Hebrew text. Understandably, the developers of Accordance decided to line up the text of the psalm proper, rather than the superscriptions. This annoying problem can be worked around by selecting an English version as the search text, or by showing the context of the English text via the "Context" button on the Amplify palette, and then tieing it to the Hebrew text.

Hebrew and Greek Grammatical Searches

Accordance's powerful search engine and its intuitive, easy to use interface makes Hebrew and Greek grammatical searches a pleasure to perform. There are three original language research modules available for Accordance: the GRAMCORD NA27 Greek NT; the GRAMCORD Westminster Hebrew MT; and the GRAMCORD Greek Septuagint. All offer complete grammatical research capabilities with parsing, statistical and graphical analysis, and come with the appropriate Greek or Hebrew font.

Accordance's original language research modules employ the best electronic texts currently available. In addition, the GRAMCORD Institute's involvement in the official editing and revision of all three research modules illustrates their continuing commitment to providing the most reliable original language research modules. Even now the Greek NT module -- updated to NA27 -- is considered the best machine-readable Greek NT text produced, both in regards to its accuracy and the philosophy behind its grammatical tags, which tend to be more formal than functional (for a discussion of the machine-readable Greek NT texts available see Harry Hahne's paper "Interpretive Implications of Using Bible-Search Software for New Testament Grammatical Analysis"). The Hebrew text module is based on the latest version (2.0) of the Westminster Hebrew Morphological Database. This is a reliable text, though there are some problems associated with its tagging. While the majority of the problems have more to do with the consistency of the tagging (for example, some compound names are lemmatized with a maqqef while some are not), there are some true errors. For example, plural suffixes are tagged as singular in Josh 3:4 and 24:7, while a search for y'd, "to appoint" in the Hebrew MT only yields 28 hits instead of the expected 29, as the occurrence in Amos 3:3 is incorrectly tagged as yd', "to know." The grammatically tagged text for the LXX is the least reliable of the research modules. It is a modified version of the text produced by University of Pennsylvania's Center for Computer Assisted Tools (CCAT) (The tagging was modified to bring it in line with the tags for GRAMCORD's Greek NT). The base text for the LXX module is A. Rahlfs' Septuaginta (Stuttgart: Deutsche Bibelgesellschaft, 1935, 1979). Ideally, it would be nice to have the critical Göttingen editions as base texts for the LXX module. Nevertheless, considering that the critical edition is incomplete and uneven, it is understandable that Rahlfs' edition was used. The LXX module includes two texts, one that provides the main text found in Rahlfs' edition ("LXX"), and another that provides the alternative version of some books ("LXX2"). So, for instance, both the Vaticanus and Alexandrinus version of Judges are available, as are LXX-Daniel and the so-called kaige-Theodotion revision of Daniel, as well as the short and the long Sinaiticus versions of Tobit.

Some knowledge of the philosophy behind, and the details of, the grammatical tagging is necessary for effective use of the research modules. For instance, it is important to know that Greek accents and breathing marks and Hebrew vowel points are ignored, or that the Hebrew tagged text does not distinguish between homographs (i.e., two or more lexemes that have the same spelling). Thus, a search for "mashal" will not distinguish between mashal I, "to repeat a proverb," and mashal II, "to rule." In this respect the documentation on the tagging details is excellent. Two appendixes provide comprehensive listings of all the options that are available when using the grammatically tagged Greek and Hebrew texts.

Both electronic texts underlying the Hebrew MT and LXX research modules are currently undergoing editing and revision under the direction of the GRAMCORD Institute. The revisions will increase the accuracy of the texts as well as add a number of other features (e.g., the tagging of homographs in the Hebrew MT -- a questionable move considering that many so-called homographs are debated). Nevertheless, until these revisions are complete, all results should be double-checked by conventional means before employing results in any serious scholarly research.

Like English text searches, Hebrew and Greek searches can be executed through the general Search window or the search construct window. Searches can be performed for any combination of inflected or lexical forms, or grammatical data (part of speech, verb tense, aspect, mood, person, case, number, etc.), using all of the search commands and operators noted above. Including grammatical details in search arguments is almost fool proof. Grammatical tags can be entered directly into the Search Argument window, or can be entered via a pop-up dialog box. The latter prevents searches for impossible constructions like plural infinitive absolutes in Hebrew or feminine active imperfect verbs in Greek by dimming invalid options. Even if impossible grammatical tags are entered manually into the Search argument box, a dialog box highlights the error, and even suggests using the tag dialog box to generate correct tags.

The Search Construct Window

A truly unique feature of Accordance is its Search Construct window. This window allows the user to graphically build complex search constructs in English, Greek, or Hebrew. The Search Construct window consists of a description box, connection area, element columns, and a construct palette. A construct is created by dragging any number of elements (i.e., a word or a phrase, or a grammatical element for tagged texts) into different element columns and then defining relationships between the elements. While there is definitely a learning curve involved in gaining competence with the Search construct window, it is actually quite easy to use. For example, to include a word into an English search construct it is only necessary to click on and drag the "WORD" option from the handy "construct palette" to the desired element column. Then a dialog box will open so that the appropriate word may be selected. Other words, lexemes, and connecting items are all entered in a similar fashion. To execute a search using the Construct window it is necessary to link it to a Search window that will display the search results. This is done automatically if a Search window is active when opening a new construct, or can be done manually by choosing "Link" from the "Enter Command" submenu of the "Options" menu. While this may be seen as unnecessarily complex, it actually allows for powerful searches where multiple Construct windows are combined in the same search.

The Construct window makes it possible to execute sophisticated searches with English or grammatically tagged texts. It is possible to exclude or require intervening terms or expressions between items, and define precise word order, among other things. While the Construct window allows for unprecedented English language searches, it's features shine when working with grammatically tagged texts. For instance, with tagged texts, agreement between search expressions can be required for any grammatical element (lexical form, inflected form, part of speech, gender, number, case, person, etc.). Thus, searching for grammatical or syntactical constructions like Greek genitive absolutes (a construction where a genitive participle occurs with a genitive noun or pronoun with agreement in gender and number), or Hebrew infinitive absolutes followed by verbs of the same root (i.e., an intensifying infinitive) is quite simple (see the example construct above). All these powerful search features make Accordance second-to-none in sheer searching power, flexibility, and ease of use.

CONTINUED ON NEXT PAGE