Normally, ‘therefore’, ‘so’, etc. are classed as adverbs, ‘because’ and ‘since’ as conjunctions. This just shows that inference can be indicated in natural language by more than one grammatical construction, one in which the reasons are prefixed, the other the conclusion. Because the usual convention for formalising natural-language (NL) argument is to list the conclusion after or below the premise(s), the ‘so’ form tends to be the standard one – although it is worth noting that Hitchcock (2007: 107) apparently takes the view that ‘since’ is the more perspicuous. I, on the other hand, take the view that the standard form of NL argument is conjunctive – P and so-C.
conversely, all it takes to indicate argument in a random set of sentences is to replace a punctuation mark with a conventional inference indicator. A lot of bad arguments can be composed in that way. They have the surface form of natural arguments, but are no more natural for that than the ‘made up example’ to which Fisher refers as the antithesis of his concept of ‘real’ argument (of which more shortly).
Besides, not all uses of ‘therefore’, ‘so’, or ‘since’ have an illative sense. In many cases these words are found in the context of explanation which, although it is a form of reasoning, differs in kind from the reasoning in an argument. Take: ‘She was bored, so she left’, or ‘So did I’, or the line from the rock-song: ‘Since you’ve been gone, I can breathe for the first time.’ In the first of these the subject’s alleged boredom would explain why she left, but is no sort of warrant for concluding that she left. Nor is ‘so’ an illative in the second case, where it has the meaning of ‘also’ or ‘likewise’. The role of ‘since’ in the song is again neither explanatory nor illative, but relates to a lapse of time. It is not the connectives here that give meaning to the sentences, but the meaning of the sentences that determines the specific sense of the connectives. It would not be a relevant objection that the temporal ‘since’ and the causal or illative ‘since’ are different words – as are seitdem and da in German – and their sameness in English is merely an orthographic coincidence. The point here is that the words fail as indicators because they can indicate different things. ‘So’, ‘since’ etc. are recognisable as argument indicators only in an argument.
Many expressed arguments have no explicit inference indicator. Recall example (1) from the previous chapter:
(1) It’s dark, and so it’s dangerous to be out. The intended force of this would be retained if uttered as
But without knowing its history or context it now becomes a matter of speculative interpretation whether the author was indeed arguing from the first sentence to the second. It would be bizarre to interpret it as the reverse, though not because of the order of the sentences. In general the order of sentences in a natural-language argument does not reliably indicate the logical order. (1’) could be rewritten
(1′′) It’s dangerous to be out; it’s dark
without being a different argument (if it is argument). But as far as linguistic form is concerned there are no cues by which to settle whether (1′) means the same as (1), or something else – say, a pair of unrelated claims. Alternatively the two sentences might themselves be understood as premises to an implicit warning not to go out.
(2) It’s dark. It’s dangerous to be out. [So stay in.] 9. Interpretation
Theoretically speaking it can be asked of any list of sentences whether one of them is, or is not, a consequence of the others. Partitioning the list into premises and conclusion designates the list as an argument, and renders it apt for appraisal. We might say that it sets the complex up as a suitable object for appraisal as an argument. But that conception of an argument is both counter-intuitive and unrewarding for anyone who is looking for some difference between strings of sentences which can, and strings which cannot, be recognised as arguments ahead of being subjected to evaluation. Under the standard definition there is no such distinction. Any string can count as an argument, so long as it consists of
declarative sentences, or expressions which can be recast as declarative.39
Any sentence can be claimed to follow from another or others. Logic could not get started if it were incumbent upon logicians to first sort sets of sentences into those
39
Hitchcock (2007: 107) goes further that this and characterises a conclusion as any expression ‘concerning’ a proposition. This is too vague and inclusive in my view. Clearly there are arguments that have rhetorical questions as conclusion. Some have commands: ‘She’s armed and dangerous so what are you waiting for: shoot her!’ If you ask what the speaker is inferring (from the premises) it is that the addressee should shoot without delay. In this respect the non-indicative sentences
‘concern’ propositions by giving expression to them. But Hitchcock means it to be much more accommodating, stretching as far as attitudes, feelings, etc.
whose conclusions exhibited some actual purport or propensity to follow from their premises, whatever the signs of that might be. Besides, the question could not even be put without first identifying a conclusion; yet, by the standard definition, a conclusion is just a sentence that is claimed to follow from others. The circularity in that would nullify the distinction at the boundary between bad arguments and non- arguments – indeed, between arguments and non-arguments full stop. The only hard line that can be drawn is between valid arguments and all other sets of declarative sentences, randomly selected or arbitrarily assembled. Yet, that said, language speakers have apparent confidence in their ability to identify reasoning and inference intuitively in natural discourse. Some combinations of sentences are seemingly recognisable as arguments; that is, some sentences in a given text are recognisable as reasons, others as conclusions.
But what does it mean for a sentence (or its meaning) to be a reason, or to be a conclusion? Neither a reason nor a conclusion is a kind of sentence, except insofar as it is a declarative. In practice the identifying of some claims as reasons and others as conclusions seems to come down to a question of how ‘good’ an
argument formed from those claims would be, if propounded. But there are clear problems with that qualitative yardstick. Not all collections of sentences, with one designated as the conclusion, are valid. That is obvious. But nor are some more valid than others – validity does not admit of degrees. Example (1), which was considered just now is certainly not valid. It does not have a valid form.40
It is no more valid than, say:
(3) It’s dark; it’s Friday.