Saturday, February 16, 2013

As falls spelling, so falls grammar

I've been doing intensive research on both calculators and spell-check, and the changes brought on by their thorough integration into American society and its learners.

There is no question, I think, that pure spelling ability has declined since young people started relying on spell-check to do it for them. That is, if you took 1000 people over 50, and asked them to spell a variety of words from separate to definitely, more of those people would get them right than the same random group of 1000 20-year-olds. The 20-year-olds would protest that why, for example, should they be forced to know something the computer can look up for them; in other words, they may not value the skill of "knowing" the correct spelling, or may just feel that as a practical matter there's little point learning something that can be taken care of so easily 99% of the time they write anything.

Now, a fair number of people have taken that general reasoning, and are now incapable of knowing the difference between their and there, definitely and defiantly, etc. We see this a lot more because of the explosion of informal writing; I am now reading what my young nieces write (on Facebook) almost every day, whereas I would not have had this opportunity in the past. Nevertheless, my suspicion is that of these 1000 randomly chosen 20-year-olds, the scores are much lower; I believe this will be played out in any major statistical project measuring spelling abilities of Americans of all ages. By the way, I encourage a large-scale comprehensive test, because I have not established concrete data support for this yet.

Let's look at the ways learning is actually hampered.

1. When spell-check gives you the right word, you are encouraged not to learn it or memorize the correct spelling of the word you wanted (because it will always be provided for you).
2. You may be taught how to spell as a child, but you are rarely taught how to use spell-check, how to choose the right word, etc. Therefore you have worked out the processes of choosing the right word, entirely by yourself, without guidance.
3. You trust the computer to catch every error, because it has acted capable of doing that, and in fact has caught errors that weren't even errors. This trust may be your fatal error, since in fact you may have made some wrong choices.
4. Microsoft's spell-check is improving gradually but slowly. Some are better (Google, for example, catches definately and changes it to definitely, unlike spell-check, which still turns it into defiantly). Why is this? I assume that it's because Microsoft has less economic motivation to perfect its spell-engine; its revenue comes not from having a perfect spell-check, but from having Word, in the first place. Your average student, and international student, is not going to avail him/herself of Google or of the commercial correctors, I infer, because it's too much trouble. This may change, or I may be wrong. But I am still seeing a lot of defiantly on the computer.Most people are not getting advanced guidance on a regular basis. Bottom line: we're stuck with what's there, at least until they change it.*

Now let's apply these rules to the development of grammar-check software, particularly to Word grammar-check and its influence on non-native writers. We can assume that they've been communicating with Word grammar-check since they first got on Microsoft Word, but that could have been in another language, or with software that was similar but not the same. They develop habits of checking the green line, perhaps reading what the computer says about a structure, but they may not understand it, or they may make wrong choices in trying to correct something. One big difference lies in the choices: with spell-check, you can check each choice, looking it up to see if the meaning matches up, so that when you look up defiantly you know for sure that that wasn't what you intended; people who respond to grammatical advice do not have the luxury of this option, since you can't look up grammatical structures, or at least have to find some ingenious way of checking your choice. Suffice it to say that people don't. When the machine advises them to avoid passive, they do. When it advises them to use commas before "which" and no commas before "that," they follow that advice. Whatever the level of grammatical support exists on that spot is a crucial factor in what is produced.

One of Word grammar-check's most persistent critics is Dr. Krishnamurthy, who has made a hobby of collecting problems caused by Word grammar-check. He says that grammar-check is really pretty good if you already know grammar and can understand the conversation they provide over what you have made and whether it says what you want. The problem really is for people whose grammar is mediocre, or people who are just learning English grammar, or people who have absolutely no self-confidence. These people have little to go on in using the advice to come up with an adequate resolution; we really would have to follow their line of reasoning as they go along, reading and responding to the advice of a machine that not only doesn't know what they intended, but has only what they've written already, as a clue to what they really wanted to say or where they wanted to end up. Many of these conversations (between Word grammar-check and the developmental writer) are exercises in futility. The end product is basic, simple, very bland and possibly miles from what the writer originally intended.

Keep in mind that we are talking about grammatical constructions here, not just the errant definitely/defiantly quandary. One question is whether grammar-check can spot errors like should of or there/their/they're; with increasing grammatical sophistication, they more often can. If they can explain it to the developmental writer well enough, the problem will presumably go away; it won't matter if the developmental writer doesn't know, won't learn, or can't hear, the difference between should of and should have. We will still have to deal with people whose system of choosing, or discerning the right alternative, is fundamentally flawed. As with spell-check, you will have people who go in fundamentally wrong directions with their choices, and it's hard to explain why exactly they would choose one thing, or override the computer in certain cases, etc. I have found people like this, on the outside of what I'd imagined would be standard behavior.

More often, I think, with grammar, you get a baseline acceptable flood of the simplest sentences, with basic words, fit together in standard ways, with only an occasional clue that the writer really had no idea what he/she was doing. The computer has literally fought this person down to a level where the sentences are so simple that they can't be called wrong, and things that require more complex expression either don't get expressed, or are hidden in simple sentences where we, the reader, can no longer understand the writer's exact meaning. This kind of writing is too bland and unexpressive, but it will be the way things are for many, many writers and this could last for years. Keep in mind that I think grammar-check programs will get steadily better, so that it may only be in this sandwich generation that they get into habits essentially formed by the grammar-check software of the days that they wrote their most papers. It is the habits that they form that influence their writing for years to come. For example, they find ways of avoiding passive constructions, or ways to express what we know as present perfect, recent past, without using or creating verb forms that cause arguments with grammar-check. These habits become engrained and entire generations avoid certain grammatical constructions.

Wait a second, you might say. Isn't the primary way people learn spelling, by reading? Similarly, don't they learn good grammar, and good writing, by seeing how it's done correctly, and applying it to their own situation? The answer here is that the whole reason we're talking about it, is that somehow this system has broken down. Those students who do read, and read a lot, obviously, know how to spell definitely. The vast majority don't. Or at least a large number don't.

*Increased grammatical awareness is being built into spell-check and grammatical programs. I use the typical Chinese misspelling of from (form) as a bellwether. For example, a simple spell-check could not determine that form in the sentence I come form Xian is a misspelling, unless it does grammatical analysis. But if it analyzes the sentence, it should know that what is written as form should in fact be from. And newer, better programs can do this. In fact, they can even build in statistical likelihoods such as the following: if he/she is Chinese, he/she may also make a mistake on modern/morden; be careful to notice and flag that...

I think we should expect steady improvement in these machines, and give credit where credit is due. One thing I'll point out about the development of grammatical awareness in spell-check programs. It will not be entirely gradual. Instead, they will work on programs to determine the grammatical category of words in a sentence, and this will take years, and when they finally are ready to implement it, there will be a sudden jump in the general ability of programs to handle situations. Watch, and see if I'm right. See also, if Microsoft itself is the last to change.


Post a Comment

<< Home