Overall Changes

The 11th edition has 156 Sentence Correction problems (including 18 Diagnostic problems). In the changeover to the 12th edition, 55 problems were removed, leaving 101 repeats (including all Diagnostics), and 57 problems were added, yielding 158 Sentence Correction problems in the 12th edition – 2 more than in the 11th edition.

The proportion of new Sentence Correction problems (36%) is very nearly the same as the average for all problems (33%).

Topic Distribution

Classifying Sentence Correction problems under one category is a simplification, since most problems test more than one topic. That said, we have identified the primary topic tested by each problem and computed the total in each category, as shown below.

The proportion of problems in various categories has changed, but only slightly.

The number of Pronoun and Modifier problems has risen somewhat, while the number of Comparisons and Idioms problems has fallen by a similar amount. Parallelism has increased by 2 problems. Otherwise, the net number of problems in various categories has remained the same. Parallelism is still the clear leader as a primary topic, but Modifiers replaces Comparisons in the #2 position.

Difficulty Distribution

This graph displays the difficulty level of problems that were removed, repeated, and added. Excluding problems in the Diagnostic exam, higher-numbered problems are more difficult, according to the GMAT. On the left, red problems were removed from the 11th edition. On the right, dark-green problems were added to the 12th edition. Light colors represent repeated questions.

Problems were removed from the 11th edition at all levels of difficulty. In some cases, groups of 3 or 4 problems in a row were deleted.

Problems were also added to the 12th edition at all levels of difficulty, but in an extremely regular pattern, with no more than 1 at a time. This pattern is the same as that for new Critical Reasoning problems.

The difficulty of various topics, as measured by position, shifted somewhat. The most substantial changes were among Comparisons, which increased moderately in difficulty, and Meaning/Concision, which fell in difficulty. Odds & Ends has a very small sample size, so the large swing of its position is not that meaningful.

Qualitative Observations

1. Like the other problem types, Sentence Correction has not changed greatly from the 11th to the 12th editions. Do not forget this point as you study the nuances of new problems.
2. The GMAT continues to look for ways to penalize unwarranted shortcuts. Certain awkward expressions are often but not always wrong, and the GMAT has included problems that force you to choose these expressions, such as being (in #101) or the fact that (in #83). Remember that the right answer is not necessarily graceful, but it must be grammatically correct.
3. Explanations for repeated problems have generally remained the same, but occasional changes have been made that may indicate a slight refinement in the GMAT's thinking on particular issues. For instance, the explanation for problem #7 in the 11th edition condemns the use of which in restrictive clauses and instead insists on the use of that. This problem has been preserved verbatim as #10 in the 12th edition, but now the explanation says that the rule against which in restrictive clauses is "controversial." As it turns out, none of the wrong answers in the problem fail solely on this basis. Nevertheless, with its new explanation, the GMAT has signaled that it will almost certainly never write wrong answer choices that can only be eliminated with the which rule.
4. A few idioms have been added or returned to the working lexicon of GMAT Sentence Correction. Some of these, such as not just X but also Y, are variations of well-known idioms. Of course, you must be extremely careful not to draw too many analogies with idioms, which by definition resist classification. For instance, one new problem (#140) uses the word dated, which must go with at, not to be, when you express an age: This fossil has been dated AT 10,000 years old, not TO BE 10,000 years old. In contrast, a seemingly similar word, estimated, must be used in precisely the opposite way with ages (This fossil has been estimated TO BE 10,000 years old, not AT 10,000 years old), as is tested by problem #27 (#21 in the 11th edition).