Reader is a program which is designed to help you eliminate errors from your text documents. As anyone concerned with proof-reading knows, it is very hard to spot those last few mistakes. One must simultaneously be on the look-out for misspellings, incorrect punctuation, poor choice of words, over-long sentences and other stylistic matters and, last but not least, grammar errors of various kinds – duplicated words, missing main verbs, and so on. Another pair of eyes is a help in all this, so we often give our work to a colleague to check. Reader is offered as another such helper.
Reader is designed for checking the grammar, punctuation, spelling and word usage of any kind of English text, including literary, business and technical genres. The program is driven from a graphical user interface, with menus to facilitate selection of the various options and context-sensitive help available at the touch of a key. Input in the form of a plain ASCII text file is analyzed a sentence at a time, and the text, annotated with comments where necessary, is written to a report file. At the end of the report is a summary giving letter, syllable, word, sentence and paragraph statistics, a number of industry-standard readability measures, and totals of words by part-of-speech and by commonness in the language as a whole. (For an example see later.)
Like all programs of its type – and human readers too – Reader is not infallible. However, unlike other grammar checkers, this new version of Reader makes a full syntactic analysis of each sentence, and this enables it to find certain kinds of errors which no “string-matching” program can ever find.
Why don’t other programs do this? The answer is that it is a very hard thing for a computer to do. Among the problem areas in English are the multifarious properties of words like “all”, “as”, “how”, “more”, “over”, “so”, “than”, “what”, “whether”, etc., and the way in which conjunctions like “and” can appear almost anywhere. To solve these problems requires a comprehensive lexicon, a broad-coverage computer grammar, and an efficient parser. A major implementation issue is that, as the coverage improves, the memory and time requirements for processing can escalate out of all bounds: full analysis of a sentence of a dozen or so words can soak up many megabytes of RAM and minutes of CPU time on even a high-end workstation.
At the core of this new version of Reader is a state-of-the-art parsing engine, which is integrated with a very large computer grammar of English (several thousand phrase-structure rules). It has been highly optimized for both speed and space, allowing the program to run on any 80×86-based micro in a normal DOS partition. As a result, only a second or two on an 80386 machine is needed to analyze a typical (18 word, say) sentence.
The first version of Reader – lacking only this global sentence analysis – was released in July 1991. Among its most appreciated features were the straightforward user interface, and the unrivalled coverage of both American and British English vocabulary provided by its lexicon (some 400,000 entries, giving a word recognition accuracy of around 99.5%). Here are some reviewers’ reactions:
“Reader’s comments aren’t as good as a human’s, but they’re pretty good.” – BYTE, September 1991 “Reader is a good, simple package. It is easy to use and fast; it took five seconds to check a 500-word file. The comments on style are helpful and clear… It found all the mistakes in our test files.” – PC PLUS, January 1992 “The program is extremely simple to use and the manual easy to follow.” – The Times Higher Education Supplement, February 1992