I couldn't imagine how reading comprehension wouldn't have a high correlation with verbal comprehension; clearly, the greater one's reading comprehension the more likely he or she will preform better on vocabulary and information; after all, the more information one comprehends while reading, the more likely he or she will store that information in his or her long-term memory. Also, the greater one's reading comprehension the more terms he or she is likely to comprehend and thus the larger his or her vocabulary is likely to be. Not only that, but according to Dr. Arthur Jensen, for adults a measure of reading comprehension relative to the general population is essentially as accurate a measure of g as a certified IQ test. Clearly, there is a very strong correlation between verbal comprehension and reading comprehension.

As far as not being convinced is considered; I would also pose the contrary example: consider a person who reads extremely rapidly but as a result comprehends little of what he or she is reading. He or she may preform very highly on processing speed, but is likely to preform comparatively poorly on verbal comprehension in particular.

One is able to estimate the g-factor using a method developed by Spearman known as factor analysis; as a general rule, factor analysis shows that the more diverse and different tasks measuring mental abilities one administers, and the higher the g-loading on those tasks, the greater the accuracy to which one can be sure one is measuring g.

If you are unconvinced of what I'm saying, you would have to believe that certain personality traits can alter one's IQ significantly, in which case you would have to believe that IQ tests are not a reliable measure of g. I would argue quite vehemently that one would not score well on all or even a large number of the sub-tests on the Wechsler battery without having a very high level of g and a very high level of g only. What are the chances that one, by sheer serendipity, would have all of the personality traits that are conducive to proficiency at tasks measuring abilities as different as vocabulary, block design, digit span, coding, and the like? Not only are these tasks only positively correlated due to the g-factor, they tend to be negatively correlated when g is factored out (for example, verbal comprehension and processing speed have a negative correlation when g is negligible for the aforementioned reasons).