Is guessing from context a load of XXXXXX?

Look at the following sentence, -what do you think the missing word is?

Juan’s teacher is always angry because Juan never does his XXXXXX

I’m sure many of us, myself included, have seen or taught a sentence like the one above to introduce the skill of “guessing from context”, or trying to infer the meaning of unknown words by looking at clues such as the grammar or context. The logic of this approach seems to be that students spend far too much time looking at dictionaries and not enough time XXXXXX listening. They shouldn’t try to understand every word but should just “guess” at the meanings. 

But is it even possible to guess the meaning of a word? and if it is, how good are students at guessing? Just for fun, here are two sentences to try. I’ve been reading “Going Clear” this week and luckily, I actually came across two words I didn’t know. I’ve included both sentences because you might know one of the words but probably not both, and so one can be a ‘cloze’ and one a real word. Let’s see how XXXXXX well you do at guessing:

Homes was an XXXXXX* with almond-shaped brown eyes.

The racist white cop who molests a tony upper-class black woman.

I decided to read the literature on “guessing” (so that you don’t have to!) and quickly discovered two things. Firstly there’s a XXXXXX lot of it, and secondly it’s XXXXXX messy. But before I get down to the details, it’s important to understand a key problem with this area. There is essentially a tension between the fact that humans must learn a huge amount of vocabulary by guessing/inferring, (after all, no one actively taught you most of the words you know) and the fact that in a large number of situations (like the two examples above) guessing seems almost impossible. The type of word trying to be guessed is clearly a factor. guessing the word “hammer” is probably a lot easier than guessing the meaning of the word “acknowledge”. But what do the experts think about ‘guessing’ and more importantly, what does the evidence say?

Paul Nation thinks it is a “a very powerful and useful strategy” and is “worth spending a few minutes on” every week (2009:55). Unfortunately, he offers no evidence to back this up. Walter and Swan aren’t so keen describing it as “the alleged ‘skill’ of guessing unknown words from context.” and adding that “research has shown, and it can easily be demonstrated practically that unknown vocabulary can rarely be successfully processed in this way” (2008:71) However a different Walters (note the S) has carried out a meta-analysis of inferring in which she concludes that “it seems clear from research that…drawing students’ attention to context when attempting to infer meaning of unknown words is worth the time effort in the language classroom.” (2004:250)

There were two other meta-analyses carried out on this subject, (though it should be said, mostly relating to L1 learners) which completely disagreed with each other. Kuhn and Stahl reviewing the literature conclude that “if these studies represent where the field is now, then we cannot recommend instruction in context clues.” (1998:135) but Fukkink and de Glopper disagree noting that “even a small improvement of the ability to infer the meaning of unknown words would result in a sizable number of words learned.” (1998:451)

One major issue is that the research seems a bit XXXXXX sketchy at times, with a lot of it coming from L1. Add to that small sample sizes, lack of control groups, difference in testing procedure etc and it’s not surprising that Kuhn and Stahl note “given the frequent recommendations that children be taught the use of context clues, the paucity of research evidence is quite disappointing” (1998:129) Almost 20 years earlier Nation described it as being “widely acknowledged as a useful skill” while pointing out that there was very little evidence to back that sentiment up. In 1994 Knight reiterated this noting “although in recent years, many researchers, teachers, and textbook authors have encouraged students to guess, to use inference as the strategy of first choice (30; 48; 49; 64), this advice appears to be based more on conjecture than on empirical finding” (Knight 1994:286)

But lack of evidece aside, are student any good at it? Well they certainly like it. Schmitt, refers to one study which “found that their university ESL students used inferencing in about 78% of all cases where they actively tried to identify the meanings of unknown words” (Paribakht and Wesche 1999 in Schimidt) and another study found students used it with 58% of unknown words (Fraser 1999 in Schmidt). However, despite their enthusiasm for guessing, students are not very XXXXXX good at it for example Grabe states:

Guessing meanings of words is not an efficient way of learning new words explicitly when it is used as a textbook or class exercises…In four studies Gough and Wren (1999) showed that when L1 students guess words from context they are accurate only 14 to 45 percent of the time.(Grabe 2009:73)

However he doesn’t write it off completely as a strategy, but notes a dictionary would yield better results. Incidentally, the results for dictionary use are clearer “subjects who used the dictionary not only learned more words but also achieved higher reading comprehension scores than those who guessed from context” (Knight 1994:295) There was no evidence in other papers that guessing improved reading scores or that students were even able to remember words they guessed correctly.

The low rate of success for guessing is a common finding:

Nassaji (2003) found that of 199 guesses, learners only made 51 (25.6%) that were successful, and another 37 (18.6%) that were partially successful. This low success rate is similar to the 24% rate that Bensoussan and Laufer’s (1984) learners achieved. (Schmitt 2008:350)

Frantzen’s results (N11!) show students were only successful in about 30% of cases. She also reports Kelly’s (1990) findings that even when just one word was unknown in a text “Contextual guessing alone seldom allows the reader to arrive at the correct meaning”(1990 in Frantzen 2003:169)

Another problem is that it isn’t always clear what exactly people are talking about when they talk about this. When teachers “teach” this skill what exactly is it that they are doing? Likewise, what is actually being tested? Knight (1994) suggests that the huge disparity in results could be due to a disparity in researcher testing methods. For example some use XXXXXX (cloze) while others use made up words and others researchers use real words. Fukkink and de Glopper suggest that giving advice on best practice with regard to this method is hard because “the empirical evidence is not unequivocal and the theoretical foundations of instruction are sparse or even absent.” (1998:462) All of this adds to the confusion.

But generally speaking it seems, slightly effective. Experimental groups almost always seem to outperform those not taught anything. However these are largely the results of L1 studies and “research in L2 contexts however, does not provide such strong support for lexical inferencing” (Nassaji 2006:397) Another caveat to this is many L1 studies also show that mere “practice may be equally effective as instruction” (Fukkink and de Glopper 1998:452) that is where a “practice only” group was included they did just as well as those who were taught. Kuhn and Stahl suggest that “merely practising deriving words from context would be enough to make students better at deriving words from context.” (1998:129) 

There is however a very strong correlation between language ability and the ability to guess word meanings (Frantzen 2003). Nassaji (2006:394) adds that, being able to understand the text “as a whole and most of the words in it” is a good indicator of success in inferencing and this fits with Nations finding that students should know 94% of a text to be able to understand it. It’s perhaps not XXXXXX rocket science to suggest that guessing from context is tough when you have no idea what the context is.

Kelly (1990) and Laufer (1997) question the value of guessing on the grounds that texts do not always provide adequate information to facilitate correct guessing of words. Stein (1993:204) suggests that “part of the problem is that the contextual clues themselves are largely insufficient to narrow in on a word’s meaning. The language itself allows for many unavoidable possibilities in interpretation, often many more than wanted.” This view is supported by Grabe and Carrell who note that “what may appear to be transparent ‘guessable’contexts to native speakers are often incomprehensible contexts to native speakers”(in Schmitt 2002:240) So you might think it’s easy for your students but it XXXXXX isn’t.
Don’t believe me? Try it for yourself:

A biomic approach by integrating three independent methods, DNA microarray, proteomics and bioinformatics, is used to study the differentiation of human myeloid leukemia cell line HL-60 into macrophages when induced by 12-O-tetradecanoyl-phorbol-13-acetate (TPA). (Juan H-F et al 2002)

Guess away!

What is slightly odd however, is students who presumably know how to do this in their L1 mysterious forget how to do it in L2. This should be something we don’t need to teach them. As Nagy (in Kuhn and Stahl 1998:133) notes “Learning from context is a natural process, as well as the way in which we have learned most of what we know.” Swan agrees:

Why should language students need training in making intelligent guesses?Are they less intelligent people, less good at guessing, than other groups in the population? Than language teachers, for instance? Is there any reason at all to suppose that they do not already possess this skill? And if they possess it, do we have any real evidence that they cannot in general apply it to learning a foreign language? And if we do not have such evidence, what are we doing setting out to teach people something they can do already?” (Swan 1985:8)

One suggestion is that lower-level students’ “processing power” is entirely taken up with trying to understand the language to the extent that when they improve they will be able to use this skill. “The ability to apply the skill is inversely proportional to the user’s linguistic competence. This, and all the other sub-skills and strategies, are things that the learners have mostly already got in their L1, but they can only progressively apply them to L2 as their linguistic abilities improve” (Stranks 2010)

So in short; students like it, but they XXXXXX suck at it. They can be taught it but the results aren’t much better than just practicing. and some words are just unguessable. The more words they know and the better their English is, the better chance they have of guessing correctly. All of which leads me to the XXXXXX astonishing conclusion that working on their English, rather than teaching them how to ‘guess’, might be a pretty XXXXXX good idea. 

*Ingenue: A naive, innocent girl or young woman
tony: marked by an aristocratic or high-toned manner or style

exceptio probat regulam in casibus non exceptis

If you follow my blog, and you’re a bit of a stickler for language use then you probably think something like “hey, this guy is way too lenient about language!” or “everything I think is wrong, he says is OK!” or something like that. 

With that in mind, today I’m going to talk about a piece of language I unapologetically loathe. I hate it so much that I don’t believe it has any place in the English language (as it is presently used) and should be banned.  The phrase in question is the hideously non-sensical “the exception that proves the rule“.

Whenever I see this phrase it confirms my suspicions that the world is largely insane made even more insane by the fact that people seem satisfied by this a response:

A: Have you noticed how all TEFL teachers are massive lefties?
B: Well, Neil’s a life long member of the Conservative party.
A: well the exception proves the rule.
B: Hmm I suppose you’re right.

We have pedants running around screaming about less/fewer or who/whom (all of which are perfectly reasonable and defensible) and yet when someone utters this illogical bumf -everyone around just sits there nodding.


It is to my mind quite spectacular that this phrase, having no meaning and being so obviously contradictory to all facets of common sense, continues to exist. How would science even operate if this were true? Image the scenario, a scientist discovers a new law of nature, another scientist points out that said law has an exception, the first scientist with tears in her eyes exclaims “well then, it must be true!”

Are we supposed to believe that rules without exceptions, are unproven? A few seconds of thought will confirm to anyone that exceptions do very little to strengthen rules. What has happen here is that the original Latin phrase (see title) has been misunderstood and is now being used in a totally nonsensical way. The original meaning is a (not very useful) phrase indicating that exceptions to rules, show that rules must exist. So if you said to your students “OK, you can speak (native language) in class today”, then what you’re also saying is that usually, the class is English only. The exception to this rule shows that there must be a rule.
Does this make any sense?

There are some folk (me at one time) who have latched onto a faulty etymology, in which it was claimed that “prove” means “test“. So exceptions test rules. This makes more sense but it  unfortunately has two problems. The first problem is that it probably isn’t true, -or at least I’ve seen no evidence it is correct. More importantly, even if it were true, that certainly isn’t how people are using it now. Nowadays it is almost certainly meant to mean “confirm”.
Semantic shift is natural and it’s not at all odd to see a word change meaning, for example uninterested and disinterested, swapped meanings, peruse now means the opposite of what it meant and the word decimate has almost entirely lost its original meaning. The difference here is however, all these words are no fulfilling a useful function in language or at least express something meaningful. “TETPTR” means absolutely nothing. It’s like saying “well, you can prove anything with facts“.
what on earth does this mean?

It is interesting to wonder why people so frequently use a meaningless phrase. One blogger suggests it may come from the deep human desire not to be wrong, -which is certainly possible. It kind of acts as a rebuff when inconstancy has been pointed out. You get to have the last word. describe it neatly as “ignore that thing that disproves my theory; it only proves my theory”.

It reminds me of the similarly silly conversation stopper “I’m entitled to my opinion.”  which seems to mean “I’m not interested in the facts, stop arguing with me dammit.”