Aren't I? - only for question tags
that's what Michael Swan says in Practical English Usage, OUP, 1995.
The question should be asked: Why am I not a movie star?
However, it is quite common hearing the sentence as you mentioned above.
Note added at 2003-01-31 09:56:32 (GMT)
Why can we use \"aren\'t I\" in a tag, but not in any other utterance in the first person?
You mean why can\'t we say \"I are,\" right?
The logical tag would be \"amn\'t I,\" since we say \"I am,\" not \"I are.\" However, in Southern British English both
\"amn\'t I\" and \"aren\'t I\" would be pronounced the same--as \"an\'t I.\" The /m/ and /r/ would both delete for syllable
structure simplification. \"Amn\'t I,\" spoken as \"an\'t,\" was reinterpreted as \"aren\'t I\" by most American and British
speakers early in the twentieth century. The natural development of \"amn\'t I\" to \"an\'t I\" to \"ain\'t I\" has been
rejected in the last stage. So, we are left with a logical absurdity, \"aren\'t I,\" to avoid the \"ear splitting\" \"ain\'t I.\"
By the way, many hardline prescriptivists would argue that you are wrong to say, and certainly wrong to write,
\"aren\'t I,\" unless you are representing in writing naturalistic dialogue. These hardline critics would prefer that you
say and write \"am I not,\" avoiding the contraction and the natural progression of phonological change altogether.
English teachers also need to distinguish between grammatical errors and usage errors. Usage is a \"social\" issue: certain grammatical constructions or word uses are appropriate in one context but not another. It is fine and grammatical for someone to say \"She be working,\" or \"Me and him went fishing\" or \"Ain\'t I right?\" or \"He did it hisself,\" when sitting around the dinner table at home, assuming that these constructions are part of the person\'s dialect. But those expressions can be quite inappropriate in certain formal contexts, even if they represent a more consistent grammar than what is appropriate (be working is a consistently applied, durative form; ain\'t I may be more \"correct\" than aren\'t I, since the general rule for tag questions requires agreement between the verb and pronoun, and are doesn\'t agree with I; hisself is consistent with the possessive forms myself, yourself, ourselves, but formal usage calls for the \"ungrammatical\" object forms himself and themselves in the third person).
Language change inevitably leads to variation, and variation within a speech community often leads to social valuation of particular features as \'good\' or \'bad\'. \'Good\' variants are typically believed to be characterized by logical superiority or venerability, or both; \'bad\' variants must then be illogical and/or recent inventions by the vulgar.
But neither logic nor great age plays a significant role in the labeling of variants. Consider \'ain\'t\', which may be the English word most despised by schoolteachers and pundits. Far from being illogical or recent, \'ain\'t\' is a legitimate phonological descendant of \'amn\'t\', which was the original contraction of \'am not\'. It isn\'t clear how \'ain\'t\' fell into disrepute, but once there, it left an awkward gap in the system of negative contractions: We have \"You\'re going, aren\'t you?\", \"She\'s going, isn\'t she?\", and so on, but surely no real person actually says \"I\'m going, am I not?\". Instead, people say \"I\'m going, aren\'t I?\", in part because they have been taught to avoid \'ain\'t\' like the plague; and here logic shudders, because while \"You are going, She is going,\" etc., are fine, \"I are going\" is impossible for native speakers of English. The point of this example is not to urge rehabilitation of ain\'t\'--legislating language change is generally a losing proposition--but to illustrate the linguistically arbrary nature of social valuation of the results of language change.