There are an emormous variety of documents available to the historian, whether in print, manuscript or other media. Anything written in the past can constitiute a historical document or 'text' -- literary texts are as much historical sources as shopping lists, letters, charters, or acts of parliament. The form and content of each of these texts will differ, but there are certain questions that can be asked of any one of them:

The form and content of the text are the clues here. Different kinds of document are set out in different ways. Thus a will may begin with the formula 'In the name of God. Amen.', a letter will invariably begin with a greeting (e.g. 'Right worshipful husband, I recommend me to you'), and a tax record may contain a list of names each associated with a sum of money. Some kinds of source are more uniform and formulaic than others. Wills, for example, vary enormously in detail, but are remarkably consistent in terms of their basic structure. Tax records, on the other hand, vary considerably according to the nature of the tax. Two points follow:
1. The form and content of the text allow the nature of the source to be determined.
2. The nature of the source to no small extent dictates what is (and is not) recorded. Thus a will may record the testator's parish and will invariably name his or her executors, but need not record all members of the testator's immediate family and may well omit any words of endearment. To use wills as a source for the composition of the coresident family group or for evidence of affectivity would, therefore, be a hazardous exercise.
  A 'must' for the historian, the question of dating is not always easy to resolve with confidence. Some documents are, of course dated, though in the case of chronicles, for example, may be much more recent that the events they describe. Historically a variety of dating conventions may be found. A useful guide here is C.R.Cheney's Handbook of Dates. Where documents are undated, then there may be a variety of clues that allow an approximate date to be determined. These clues include names and events mentioned (and not mentioned), the form of the document, the style of the handwriting, and the language / phraseology used. Sometimes it is possible to say that a text must have been written after a certain date (terminus post quem) or before another date (terminus ante quem). Often it is possible only to say that the date is approxiamtely or around such and such a date (circa written as c.).

Since the form and nature of different kinds of documents tend to change over time, it is important to keep the dating of the text firmaly in mind when reading it. Conversely the form and nature of the document can be a clue as the date.

This question is important on a number of levels. Quite often you may be working with a translated text, so it is important to be aware of the original language and that you have access only to something of the meaning of the original, not to the actual words used. The language used may be dictated by the nature of the document. For example, in England legal texts tended to be written in French and clerical texts in Latin. Departures from the 'expected' language are thus worth noticing. Language also has implications for authorship and for readership. To state the obvious, documents written in Latin could only have been compiled by someone with a training in the language and could only have been read by the like. A Latin-literate writer is likely to have had a different outlook from someone lacking such an education. Thus the monastic chroniclers of the Peasants' Revolt (1381) described the rebels as 'rustici' and 'vilani' (rustics and villeins), whereas most scholars would now find this to crude a generalisation. Where texts purport to record the words of specific individuals, the language of record may provide a crucial clue as to how far this may have been the case. Clearly witness statements, wills or petitions not written in the vernacular must at least translate the actual words used. It follows that the 'translator' may have played a significant role in shaping or even determining what is reorded.
  This question follows directly from the last. Every author will have some sort of agenda to follow and this will shape the content and tone of the text, but authorship and writing do not necessarily coincide unproblematically. Sometimes it may be that the text is negotiated between more than one party. Thus a will is usually written by a professional, whether a cleric or a lawyer, who knows how a will should be set out and may have their own sense of what may or may not be done in a will. On the other hand the testator may have their own ideas and priorities. The eventual text represents a negotiation between these (sometimes conflicting) agendas. Similarly a letter may be written by a scribe on behalf of the sender, who may or may not be illiterate. It remains, therefore, an open question as to how far the sender represents the author or how far the scribe in fact intervenes in creating the final text. Here it can be helpful to have knowledge of the appearance of the original document, something that can be lost when working from a published edition. Thus an unpractised hand or an unconvenional layout can sometimes reveal that a will or a letter is actually the work of the person in whose name the document is written.
  This again follows from the questions already posed, but it is a crucially important question because it determines the content of the text. To use a metaphor, if the text represents a block of wood to be worked by the historian, the purpose represents the grain of the wood. The historian must try to work with the grain in order to achieve the most satisfactory result. The problem is that the historian is invariably asking questions that differ from the purpose of the documents interrogated. Letters, for example, may appear a good source for family relations, but it needs be remembered that letters were only written when the correspondents were apart and invariably with a particular objective in mind. For example, Margery Brews's famous valentine (from the fifteenth-century Paston letters) is not primarily, as it appears, a naive and touching expression of romantic love, but rather a carefully crafted instrument in her family's strategy to engineer a 'good' marriage. (The point is reinforced when is remembered that the letter was written by a scribe who no doubt played a substantial part in its composition.)