According to www.duden.de (online German dictionary) we can define 'genre' as an "aggregate of types of things, individuals and forms, whose fundamental characteristics are in line with one another". Relevant here, though, is the following usage: "beautiful literature divides itself up into the three genres of lyric, epic and dramatic." I can accept such a genre system without too much fuss, since the differences between these three forms are normally clear, uncontroversial and unambiguous. This classic three-way division goes back to Aristotle's Poetics, but it was also extremely popular with Johann Wolfgang von Goethe and other poets and thinkers, and despite centuries of debate it is still in use. Gradually it has become somewhat more complicated (should we perhaps introduce a fourth category, which we could call "non-fiction"?) but nevertheless this three-way division is useful in the purely literary sphere.
As I say, I have few problems with such a system. Let's take for an example Love-Song, my favourite poem of Rainer Maria Rilke's. Let us read lines like „Auf welches Instrument sind wir gespannt?/Und welcher Geiger hat uns in der Hand?/O süßes Lied“, and here we can clearly see that we are reading a poem; there is no dialogue, no monologue, no prose, no narration. There is only lyric, only poetic images, only poetry in verse form. Or we could take as an example of the epic the Song of the Nibelungs, where we read: „Ez wuohs in Burgonden/ein vil edel magedîn,/daz in allen landen/niht schoeners möhte sîn“. It is with these words that the narrator begins his story; he lays great emphasis on the musicality of the words, on the flow of speech; and most importantly of all the text itself is not divided up into different roles. Such texts are not written directly for the stage, but can take various forms: novels, epics, autobiographies, sagas, stories, novellas, short stories, fairytales, legends, fables, and so on. Thus "epic" is a particularly broad genre, since it encompasses both the Song of the Nibelungs as well as Stefan Zweig's short story Incident on Lake Geneva and Charlotte Brontë's Jane Eyre. Our last example, in this case a drama, is Danton's Death, written by Georg Büchner in 1832, and as expected we find scenes, acts, stage directions, etc. The text is divided up into different roles, because the dramatist intended that this play would premiere on the stage.
These three types of genre are obviously indispensable. How could literary scholars distinguish between a four-line poem and a thousand-page novel without this classical three-way division of lyric, epic, and dramatic? They are just as different from one another as the three primary colours, and we couldn't really begin a conversation about them without using the corresponding useful terms. Such a system doesn't split literature off into brackets, rather it broadens our understanding of what literature is.
The further divisions (comedy, tragedy, romance, historical novels, fantasy, sci-fi, picaresques, bourgeois tragedies, etc). are somewhat more controversial. In some cases I agree with the original premise - that is to say that literature does not allow itself to be easily split off into such brackets. These categories can sometimes lead to a better understanding of literature, but they are also frequently a hindrance. Literature scholars can't even reach a consensus about the genres of Shakespeare's plays, let alone the whole literary canon.
Shakespeare's plays are a classic example for the aforementioned controversy. In some cases the genre is quite easy to identify: you basically never hear that Much Ado About Nothing or A Midsummer Night's Dream aren't comedies, or that we ought not to refer to Hamlet, Othello, King Lear or Macbeth as tragedies, or that King Henry IV Part II should be classed as anything other than a historical drama. But on the other hand Shakespeare wrote various pieces which defy easy categorisation; the first four acts of Cymbeline are tragic, graphic and bloody... but the play ends with a happy ending (weddings, reunions, reconciliations, and so on). Is Cymbeline then a tragedy or a comedy, or, as some claim, a romance? Is The Tempest, his last play, a romance or pure fantasy? And what about the group of so-called "problem plays"? When you're sitting in the theatre such scholarly questions are of minimal importance, but these dramas which are so hard to categorise continue to cause headaches for many experts.
The problem is not just limited to Shakespeare's plays. How should we categorise the novels of E.T.A. Hoffmann or Franz Kafka? The Sandman and The Metamorphosis are perhaps the two best examples. The Sandman is in the tradition of an old fairytale, almost a folk ballad, and yet is told like a Gothic novel, and yet also plays with our perceptions of reality. The Metamorphosis is a stylistic mix of realism, fairytale and horror, with the result that the novel is sometimes described as an "anti-fairytale". The most striking thing about these novellas is exactly this contradiction, this unusual mixture, this surprising juxtaposition of different genres. That's their secret of success. If Hoffmann or Kafka had written in a more conservative and conventional style, these two novels would not have been anywhere near as effective, as entertaining, as powerful.
This is not to say that the entire concept of genres is completely irrelevant. I freely admit that terms like romance or young adult fiction are often pretty helpful; when you're standing there in a bookshop and you have to choose what Christmas present to buy for your granny, but you know that she is only interested in gardening books or crime thrillers, the shelving system that's split up by genres saves you lots of time and effort. But hopefully I have shown that it is a limitation to put too much emphasis on genre. Such systems exist as an aid, yet sometimes present a hindrance. The variety and richness of literature has more to do with its flexibility than its neat categories.