Purpose Statement for PAWS

As the name of the starter kit indicates, the Parser and Writer for Syntax (PAWS) has two basic purposes and two kinds of outputs:

• The Writer produces a rough draft of a grammar write-up of the syntax of your language, based on the answers to the questions in each section of PAWS. A single file is output for each interface language, which can produce a practical style descriptive grammar of the language and/or a descriptive grammar comparative to the interface language, via a simple parameter change. This output file is in both a web page format and in XLingPaper XML format. See also Simons and Black (2009), Black (2009), and Black (2017) for why this is a good format to use.
• The Parser produces all the files needed for using the PC-PATR syntactic parser (McConnel 2002, included in the Resources section of the Help menu item) for parsing and disambiguation within FieldWorks Language Explorer (aka FLEx) or CARLAStudio.

Those who wish to use the Writer only, be sure to uncheck the "Output PC-PATR grammar file and test example files" box on the Output Options section. (This will also hide all the instructions in red in each section that relate to annotating your lexicon with the features needed for using the PC-PATR parser.) Once all the sections are completed, the writer output file may be edited and polished. (If your interest is solely in the Writer, you may return to the Contents page now. Return to Contents) Those who wish to do parsing will also get at least the English grammar write-up as a bonus.

The purpose of the Parser is to aid the user in setting up the initial phrase structure rules, lexical entries, feature templates, and constraints needed for using PC-PATR . Those characteristics of the language which are general, usually about 3/4 of the rules, can be modeled via this starter kit. The language-specific rules will need to be added later using the interface tools to be provided for user support and/or with consultant help.

Setting up a complete syntactic parser for a language is a monumental task. Experimentation was therefore done to determine how much was necessary to achieve adequate coverage of the language for the parser to be able to disambiguate texts in FLEx or AMPLE output within CARLAStudio. On the sliding scale between the simplest and most the complex, this parser is in about the middle. Specifically, it is simpler in some ways and more complex in others than the analysis put forth in Bickford's (1998) text Tools for Analyzing the World's Languages: Morphology and Syntax (which is used in many SIL Grammar I courses). The major differences are:

• Only very limited, partial subcategorization is used, since this is the point which requires the most time and complexity in a complete parser.
• An articulated phrase structure following X-bar theory is used (see Black 1999, included in the Resources section of the Help menu item, for an introduction), since flat structures with lots of optionality are much harder to properly constrain and require more memory and processing time.
• A parser necessarily works on the surface structure, so the phrase structure rules must model the output of transformations, not the underlying structure. This parser does not include traces or other null elements which would be necessary in a complete parser (see Black 1997 for an example; it is included in the Resources section of the Help menu item).

This starter kit is designed in sections, working from the smaller phrases which may be included in other phrases up to the more complex phrases and clauses. This allows the user to work on one section at a time, test hypotheses for each type of phrase and at the same time become familiar with using PC-PATR for disambiguation within FLEx or CARLAStudio. The entire starter kit must be completed before attempting to parse regular texts, however.