If we call something descriptively grammatical, we mean that it obeys the usual practice of native speakers. Conversely, something that is descriptively ungrammatical violates the usual practice of native speakers. When linguists use the term ungrammatical by itself, they almost always mean descriptively ungrammatical. By convention, we mark something that is descriptively ungrammatical with an asterisk (*). Sentences (2) and (3) above illustrate sentences that are descriptively ungrammatical.
A common way that linguists determine whether something is descriptively grammatical is simply by asking a native speaker if the utterance sounds right or not. If we are studying our native language, we can be our own informant—in other words we can use our intuition and knowledge of how we and others speak to decide what is grammatical. This introspection makes it seem as if judgments about grammaticality are simple. But there are several factors that complicate the process. First, it isn't the case that anything which comes out of someone's mouth counts as descriptively grammatical. People do make slips of the tongue, or change their minds about what they were going to say halfway through the utterance, and the result can be word salad. The point is that such slips are not regular. They do not form an internally consistent set of rules. Remember that grammar, whether descriptive or prescriptive, implies following a rule of some kind. Second, native speakers will occasionally disagree in their intuitions. For example, the presence of regional or ethnic dialects means that something can be grammatical in one variety of English and ungrammatical in another. Despite such differences, native speakers do show a large measure of consistency for most features of language.
If we call something prescriptively grammatical, we mean that it obeys the rules of usage which are listed in handbooks and taught in school. Conversely, a prescriptively ungrammatical utterance violates those rules. In almost all cases, if something is descriptively ungrammatical, it is also prescriptively ungrammatical. That is, the authors of traditional grammar books would object to it too. In these cases, we can simply call the statement ungrammatical and mark it with an asterisk. But often prescriptive rules do not represent the way people ordinarily speak. In some cases, prescriptively grammatical utterances will sound formal, uncommon.
(4) !Whom shall I say is calling?
People may not commonly speak this way, but we still recognize sentence (4) as part of one register of English. We will indicate such sentences, which are only found in formal contexts, with an exclamation point (!).
Detecting a prescriptive violation, when it is not also a descriptive violation can be hard because you cannot trust your instincts. Unless someone has taught you the rule, you may not notice the violation at all. Moreover, different handbooks differ in the rules they present. What may be prescriptively ungrammatical according to one book is just fine according to another. If we need to indicate something that is prescriptively ungrammatical but descriptively observed among native speakers, we mark it with the pound sign (#).
(5) #Give Al Gore and I a chance.
The final logical relationship between prescriptive and descriptive grammar—a statement that is prescriptively grammatical but descriptively ungrammatical—is possible, although rare. Such sentences follow prescriptive rules literally, but the result is something that no native speaker would ever utter. For example, many traditional grammar books claim that you must use a singular pronoun to refer to the indefinite pronouns. The following rule is found in Warriner's English Grammar, one of the most widely used high-school grammar books of the second half of the twentieth century:
"The words each, either, neither, one, everyone, everybody, no one, nobody, anyone, anybody, someone, somebody are referred to by a singular pronoun—he, him, his, she, her, hers, it, its."
Following this rule, we are supposed to write sentences such as (6a) rather than (6b), the latter of which reflects the way that most people actually use English:
(6a) !If someone calls, tell him I'll be out of town until Tuesday.
(6b) #If someone calls, tell them I'll be out of town until Tuesday.
But obeying this rule can, in some instances, lead to nonsense.
(7) ?Everyone in the room was speaking Spanish, so I spoke Spanish with him.
There is no way that him in (7) can be taken to refer to everyone. If we're approaching this rule from the point of view of a linguist, it appears flawed because it makes a false prediction that (7) should be grammatical. There is no special symbol for this situation, although the question mark can be used for any sentence whose grammatical status is ambiguous.
 One complicating factor in deciding whether or not something is prescriptively grammatical is that grammar books and English teachers often disagree about what is correct. For simplicity's sake, we will assume that any rule which is widely found in many usage books counts.
 This sentence, uttered by Bill Clinton, violates the prescriptive rule that the case of the pronoun inside a coordination ("Al Gore and I") should be the same that you would use if the pronoun appeared alone (i.e., "give me a chance", hence "Give Al Gore and me a chance.").
 John E. Warriner and Francis Griffith, English Grammar and Composition, rev. ed., Harcourt, Brace, and World (1965), p. 100.
 Warriner's suggests that for such sentences, it's acceptable to ignore the rule, but begs the question of why we should believe that the rule is correct in the first place if it creates such problems.