Word Processors' Grammar and Spelling Assistance: Consequences for Second Language Learning and Teaching. by: Ingrid Fandrych

Abstract
Modern word processors do not only include spell checks and dictionaries but also grammar checks. Do these facilities help second language students to compose correct sentences and texts? Will they even make the grammar component that has traditionally been part of Academic Writing courses redundant?

This paper sets out to determine the usefulness of modern word processors in terms of their grammar and spelling assistance and discuss the consequences for second language learning and teaching. For this purpose, the most common word processors on the market at present (Word, WordPerfect) will be tested on a variety of texts ranging from students' essays to newspaper articles and literary and academic texts.

Introduction
With the use of computers and word processors becoming more and more common among both academics and university students, a closer look at the grammar and spell check facilities of widely used word processors is called for in order to determine whether, indeed, the 'blind faith' that many people put in modern software is justified. After a brief look at the place of grammar in Academic Writing courses at university level, the following study attempts to test the most frequently used word processors - Word and WordPerfect - on a variety of text types. Some of the texts analysed in this context undoubtedly represent what is commonly called 'good English style' (e.g., the literary text by Nadine Gordimer, the non-fiction text by Noam Chomsky), but we will also look at less sophisticated texts like newspaper articles, and finally we will look at some examples of authentic students' essays in order to determine the usefulness of the suggestions made by the word processors.

Grammar in Academic Writing Courses at Tertiary Level
The National University of Lesotho (NUL) admits students with the equivalent of 'O' Level Exams (COSC), and therefore, they have to take certain bridging courses to help them to cope with tertiary education. Central to this bridging programme is a "Communication and Study Skills Course". The grammar component of this Academic Writing course consists of the following main areas:

* Sentence Structures;
* Paragraph Structures;
* Subject-verb Agreement;
* Tense;
* Modal Auxiliaries;
* Conjunctions;
* Articles;
* Prepositions;
* Punctuation.

Furthermore, we teach 'Academic Style', i.e. no contractions, no colloquialisms, no idioms, and the avoidance of impersonal constructions, as far as possible.

According to previous research carried out at NUL (c.f. Fandrych 1999, in print), some of the problem areas for students at tertiary level are:

* quantity instead of quality: unnecessary repetition and a general tendency to produce long-winded explanations;
* errors in style (contractions, colloquialisms etc.);
* problems with sentence structures;
* problems with punctuation;
* no editing/proof-reading;
* students are not used to looking up spellings, collocations, usages; instead, they 'improvise', and they often confuse words, e.g. conflict - convict, conduct - consult;
* sentences starting with again, also etc. ;
* problems with tenses, modes;
* problems with articles;
* wrong use of whereby;
* spelling errors, partly due to lack of phonetic distinction: tend/turn, this/these, hash/harsh; mader (for 'murder'); penulty, pinishment, lithal, nagetive; crimminals, limmit;
* redundancy, e.g. all the remaining others, can be able to, convincingly argue and convince us, decreases or diminishes;
* omission of 'to be' in attributive phrases.

Settings of Word Used in This Study
The relevant settings can be selected in the Edit and Tools menus of Word 97. For the purposes of this study, 'English (South African)' was chosen. Furthermore, the options 'Check spelling as you type', 'Check grammar as you type', and 'Always suggest corrections' were switched on, and, finally, in terms of style, 'standard' was selected, including all the options offered, namely:

* Capitalization
* Commonly confused words
* Hyphenated and split words
* Negation
* Numbers
* Passive sentences
* Phrases
* Possessives and plurals
* Punctuation
* Relative Clauses
* Sentence structures
* Subject-verb agreement
* Style:

1. Cliches
2. Colloquialisms
3. Contractions
4. Gender-specific words
5. Jargon
6. Sentences beginning with And, But, Hopefully
7. Unclear phrasing
8. Wordiness

The 'don't check' option was selected for the following features:

* Comma before last item
* Punctuation with quotes
* Spaces between sentences

At first glance, the above list of features Word attempts to rectify looks quite impressive and a number of items in the list definitely try to address some of the problems experienced by students, e.g. confusing words, punctuation, sentence structures, agreement, colloquialisms, contractions, conjunctions in sentence-initial position, and wordiness.

Grammar and Spell Check Options in WordPerfect
WordPerfect is in many respects more flexible than Word. There are a number of 'checking styles' from which the user can choose, e.g. 'Very Strict', 'Formal Memo or Letter', Technical or Scientific', Advertising'. In addition to these options, computer users can create their own styles, change the formality of the styles, and predefine the number of elements to be allowed in noun phrases, consecutive prepositional phrases, sentence length and the maximum number of elements allowed in a split infinitive.

'British English' was chosen (South African English is not offered as an option), and the standard settings in terms of style were applied, as the author wanted to be able to compare the different reactions of WordPerfect towards the range of texts under discussion. Similarly, and especially in the light of the mechanistic treatment of 'long sentences' and 'wordiness' in Word 97 (c.f. below), no restrictions were stipulated in terms of sentence length etc.

Analysis of the Texts with Word 97
Apart from some unfamiliar names and the term back-tracking, Word 97 only criticises the two newspaper articles for the use of contractions.

The extract from Nadine Gordimer's novel (literary text) is criticised for lack of commas (for example after morning in the second paragraph), 'wordiness' (absolutely no), the use of the passive voice (are said to be - no suggestions), long sentences (In the marble foyer …; I would hand the key …), the 'plural' (sic) of so-and-so's, and the use of and in sentence-initial position in the last sentence of the second paragraph. In the last two paragraphs, the contractions are criticised.

Chomsky's text (non-fiction) also shows signs of 'wordiness' (drawing to a close - suggestion: ending), long sentences (first sentence, second paragraph and the sentence beginning with The factors … in the third paragraph). Furthermore, Word finds irregular sentence structures (first sentence, third paragraph and first sentence, last paragraph), lack of punctuation (suggestion: comma after declined in the last sentence of the third paragraph).

Machobane (academic/historical text) is in good company with his long sentences (third sentences in the first and the second paragraphs, last sentence in the third paragraph, first and last sentences in the fourth paragraph), his comma use (Yet in the beginning of the second paragraph), and his 'wordiness' (at the moment, third paragraph).

Finally, Reagan (academic/linguistics) also produces some long sentences and, obviously, his American spelling is commented on.

Let us now turn to our examples of students' writings. The only problem Word has with Student A's text is the subject-verb agreement in the last sentence. Word suggests either to replace are with is, or to pluralise speech.

Student B: Word suggests replacing inorder with one of the following: ignored, ironed, ironware, intruder, inured - all of which would, obviously, be syntactically inappropriate. Instead of recognising the problematic sentence structures and lacking full stops (e.g. in the first line, third paragraph and second line, fourth paragraph), Word only comments on the "Extra Space between Words". Neither the really problematic sentence structures in the first and last paragraphs, nor the abbreviation and the use of need, the comma after also in the third paragraph, or the repeated use of some in the fourth paragraph are criticised.

Student C: Several wrongly spelled words are indicated. Several long sentences (first paragraph; second sentence, second paragraph; third paragraph; first sentence, fourth paragraph; last sentence, fifth paragraph; last paragraph) are discovered, but no suggestions are made. Furthermore, Word comments on the "Extra Space between Words" (line 2, fourth paragraph), suggests commas after problem and Thus (fourth paragraph), and after Law (fifth paragraph). In the latter case, Word suggests replacing which with that as an alternative. Finally, 'wordiness' is noted in the first sentence of the fifth paragraph, and it's should be replaced with it is (sic).

Student C's essay is rather problematic, but Word neither comments on the use of would (line 2), nor this offices (second paragraph), and ignores the syntactic structure of the third paragraph (apart from its length), the pronoun redundancy in the same paragraph, the position of also (second sentence, fourth paragraph), the use of where (last sentence, fourth paragraph), the spelling of can not (fifth paragraph), and the semicolon and the possessive in the first sentence of the fifth paragraph.

Finally, Student D is criticised for the contraction in line 1, and Word suggests replacing serve with serves in the second sentence. However, neither under (for undergo) nor the sentence structure of the sentence beginning with Also … (first paragraph) are commented upon.

Unfortunately, Word 97 cannot keep all its promises. This is not entirely surprising, as some grammatical features are rather too complex and context-sensitive for a computer programme to discover. In terms of the spell check, Word certainly has its - limited - usefulness, depending on the dictionary used. In terms of its grammar check facilities, Word is rather mechanical: whether something counts as a 'long sentence', even if it is correct (c.f. the analysis of literary, non-fiction and academic texts), seems to depend on word counts performed by the software; 'wordiness' apparently depends on certain phrases in the dictionary. Finally, the fact that some wrong suggestions are made is more regrettable than the mere oversight of certain problems Word claims to be able to solve.

We can conclude that the grammar component of EAP courses cannot be reduced or even done away with as far as Word 97 is concerned. What the software can do, is draw the attention of its users to certain problem areas, e.g. long sentences (which, however, do not necessarily constitute a problem per se, especially in academic texts), as syntactical errors might be lurking there.

Users who intend to make extensive use of the spell check facilities are well advised to establish their own, customised dictionaries. Academics who intend to rely on Word's grammar and spell check facilities must be warned that Word 97 can be extremely stubborn as far as the arrangement of lists or sectioned passages (e.g. legal articles) is concerned, and Word's constant criticism of 'fragments' in the references section can be irritating, too.

Analysis of the Texts with WordPerfect 8
Like Word 97, WordPerfect 8 does not have any serious problems with the two Mail & Guardian articles, apart from the use of the number 12 000, the spelling of ARs in the second article, and the unfamiliar word Accessline (all in the second article).

Nadine Gordimer's text is criticised for the use of words like wormholes, sun-set, bell-hop, show-cases (all of these in the second paragraph). Furthermore, WordPerfect informs us that but (second paragraph, middle) "is usually awkward" after a semicolon.

Chomsky's text is criticised for the use of the hyphen in post-industrial (first paragraph), a comma is suggested after Meanwhile (last sentence, second paragraph), and his American spellings are noted and commented upon.

In Machobane's historical writings there are, of course a number of unfamiliar names, some leading to rather amusing suggestions, such as Muchachas for Moshoeshoe, and dulcification for dutchification. Apart from these, the spelling of south-westerly is corrected to southwesterly or south westerly.

In Reagan's language, the use of the semicolon is explained as usually joining "two independent clauses", and, as an alternative, the use of a comma is suggested.

As far as the students' texts are concerned, WordPerfect also has less to say than Word 97. In Student A's first line, WordPerfect suggests a comma after performance, to make the sentence clearer. Otherwise, no comments. In Student B's text, the spelling of Inorder is criticised, and the lack of full stops between people and And (third paragraph) and between project and It (fourth paragraph) is corrected. However, the insertion of a comma after Language (fourth paragraph) does not clarify matters, as suggested by the programme. Student C is criticised for spelling errors (e.g. dicided, decesion, imformation, diffent, thier, interpretors, can not, stylisticans, diskets), and a full stop is suggested between meaning and This (fourth paragraph). In addition, the mistake in it's (fifth paragraph) is recognised and corrected to its. No other comments are made to improve the texts.

Comparing the analyses of the grammar checks of a variety of texts by Word 97 and WordPerfect 8, we can conclude that WordPerfect seems to be rather more flexible than Word 97, in particular as it gives the user several options to choose from in terms of the more mechanistic counting processes (which seem, at this stage at least, the best word processors have to offer as far as syntax is concerned). Another useful option for (advanced?) computer users is the choice between a range of styles, from 'very formal' to rather informal.


Word 2000

Word 2000 offers improved spelling and grammar checkers. For example, the spelling checker now recognizes a much broader range of names for people, organizations and companies, cities and countries, Internet and file addresses, and more. The grammar checker now flags mistakes and offers more effective user-friendly grammar, and rewrite suggestions. ("What's new with proofing tools", Word 2000 Help Function).
Word 2000 further announces that it can now "automatically correct spelling and grammar", and that the thesaurus offers synonyms; it can also check texts written in other languages, and detect hyphenation language-specifically.

Indeed, a fresh analysis of our sample texts shows a number of changes from Word 97 to Word 2000:

No more comments on 'long sentences' or 'wordiness'. So, apparently the mechanistic approach was abandoned as not very helpful.
Frequent reminders and instructions about hyphenation (as expected after the announcement summarised above), and concerning capitalisation.
The suggestion that not only should always be followed by but also.
In the first text about Chomsky, the sentence fragment in paragraph 10 is criticised.
In Student C's text the wrong number agreement this offices (second paragraph) is noted and explained, and the wrong spelling of can not is noted.
However, many of the other errors discussed above still remain uncorrected.

WordPerfect 9 (2000)
The most recent version of WordPerfect is more critical than its predecessor, but this also entails a number of problematic suggestions and wrong analyses, especially in the area of agreement and word classes.

In "Measuring the Collective Mind", the software identifies incidences of "Comma Splice of Fused Sentence" and suggests the addition of conjunctions or changes in punctuation in the second paragraph and in the second sentence of the 20th paragraph. Wrongly analysing the noun phrase one way mirrors, it suggests the singular form for mirror. Similarly, tracks (8th paragraph) is identified as a noun, which leads to the criticism of the adverb consequently, and the reverse for possible (same paragraph), sees (9th paragraph) is pluralised, miss (19th paragraph) is changed into misses, and so on. On the other hand, WordPerfect 9 reasonably warns of the homonymy of it's and its and finds the typo is for its (both in the 12th paragraph) - but wrongly suggests are instead.

In addition to the suggestions made by its predecessor concerning Nadine Gordimer's text, WordPerfect 9 suggests a comma after morning, the indefinite article before coloured beadwork (!), and, even more worryingly, was in there were two crossed assegais (all in the second paragraph). In the third paragraph, brassy and brand should be replaced by their adverbial forms, according the WordPerfect, and the position of the two prepositional phrases, with their lips and in the place, is criticised for "separating the verb from its direct object". Finally, the penultimate sentence of the passage "may not be complete".

Checking Chomsky's text, we are given more suggestions to replace "adjectives" with adverbs, for example in the first paragraph, power centered. An abstruse comment concerns the third paragraph: we are asked to consider replacing the "American spelling" (sic) of plead with British lead. Furthermore, the indefinite article is inserted before public subsidy (third paragraph), and in the final paragraph, we are told that "much of is usually followed by an uncountable noun. Check for agreement error."

Machobane's text is also subject to several incorrect subject-verb agreement suggestions, WordPerfect 9 cannot parse the complex noun phrase in line 2 of the opening paragraph, and in Reagan's passage most in most important (third paragraph) is wrongly analysed, which leads to the inappropriate suggestion to pluralise event.

Student A's text gets a comma after performance (first sentence). In "B's text, the indefinite article is criticised, as it should not precede the uncountable noun work (second sentence in the third paragraph), and a comma is suggested wrongly after Language in the fourth paragraph. The most worrying comment, however, concerns the third paragraph of B's text: "Normally an adjective like people doesn't modify a verb. Try the adverb form (usually - ly), or check for missing words."

In Student C's text, this offices is corrected to these offices (second paragraph), and can not and it's (both in the fifth paragraph) are corrected - but then which is deleted from the same sentence. Even more alarming are the changes from is to are after imformation (sic) in the second paragraph, and the suggested possessive for people in the last sentence of the fourth paragraph. Finally, in D's writings, one another (in the second sentence of the first paragraph) gets the comment: "One is not usually used with a pronoun", and WordPerfect concludes its proofreading process with another subject-verb misanalysis of crime rate (second paragraph).

Conclusion
It becomes clear from the above analyses that word processors cannot yet solve all the problems many users face when composing texts, especially if English is not their first language. It would certainly be premature to reduce the grammar component of Academic Writing courses and to rely instead on the grammar and spell checks of modern software. Furthermore, it comes as no surprise that many first language users even switch off the grammar and spell check options of their word-processors because they do not want to be bothered by the numerous less helpful - and frequently even wrong - suggestions and/or corrections made by the programmes. Personally, I consider any comments made by my software, but I do not always follow its advice.

Finally, we can conclude that the various products on the market all have their particular strengths and weaknesses - it would be impossible to recommend the use of one particular brand or even a certain version of that name. However, even if we cannot state any preferences, one valuable outcome of the above study is the realisation that there is no perfect grammar checker at present, and it is to be hoped that users of word processors are forewarned and will be aware of the individual shortcomings of the particular software they work with.