Cerberus, you’ve really hit the mark with intuition. More than anything the selection of tense/aspect combination relies on the intuition of native or near-native speakers. What most fail to recognise though, is what characteristics of the language are governing that intuition. English utilized three types of temporal attribute in every expression (you could also say English has three grammatical categories based on time). These are tense (temporal reference), aspect (temporal nature of the utterance), and aktionsart (temporal nature of the verb).
Tense is simply a reference between the b[/b] time of the utterance (which is the time that the construction is spoken or written, or in reported speech, the time in which is is said to occur), and either the b [/b]time of assertion (time that the verb occurs – used for progressive constructions, and for non-progressive constructions except for habituals and generalizations), b[/b] time of completion (point at which the verb is completed – used for perfect and perfect progressive constructions), or b[/b] the time of evaluation (the time at which the assertion is evaluated as true – used for habituals, generalizations, and most modal future forms). All that tense is, is a contrast between these references. If the Tutt occurs concurrently with the other references, the sentence is present tense. If Tutt occurs after these references, the sentence is past tense. And, if Tutt occurs before the other reference, it’s in the future.
Aktionsart: Verbs in English (and especially other Germanic languages) have an inherent temporal nature. In German, this is very strong, so that the verb itself carries inherent in its meaning a nature of being an action, a state, etc. In fact, the word aktionsart is German for verb-type. Aktionsart is a strong force in German and governs the temporal nature of the sentence. This is why German only has one aspect while English has two – they don’t need a separate aspect, because the verb itself determines whether the sentence expresses action or not. Some English verbs have a strong inherent aktionsart that affect how the verb behaves but most verbs in English can be used in any combination of tense or aspect.
Aspect: Aktionsart in English, unlike in German, is not a strong governing force. The temporal nature of the verb itself rarely determines the temporal nature of the utterance as a whole. English instead uses aspect which utilizes the structure of the utterance to establish its temporal nature. This allows English to use verbs which may have a non-active aktionsart in an active way. Consider the current McDonald’s slogan “I’m loving it.” Love is a non-active verb and the normal usage would be “I love it” because love is generally thought not to have duration in the same way that run or drive does. Yet by using an active structure (be + -ing), non-active love can be used as as active “I am loving,” as in, “right now, this food is so great that I am loving it as we speak,” because the progressive aspect connotes duration as part of its structure regardless of the content verb used. This is the key difference between English and most other languages – English does with structure, what most verbs do with meaning.
The combinations of aspect and tense are themselves, very clear cut. The confusion of when to use which form comes from the relationship between aspect and aktionsart, especially with those verbs in which their aktionsart has a specific nature expressing duration, or inherent termination. These two attributes are normally handled by the progressive aspect and the perfected forms. But when a content verb’s aktionsart (inherent temporal nature) is already progressive or perfect, perfecting a perfect or making a progressive progressive becomes awkward, unnecessary, or even wrong. This is what’s happening in the examples Swann uses.
Swann fails to recognize this (as most grammar book authors do), and instead attempts to create rules based on observations of forms without reference to what is happening linguistically within the sentences. These ‘rules’ end up only working under certain conditions or having as many exceptions as applications – a total ‘fail’ in any scientific system.
Native speakers recognize when aktionsart overrides the need for aspect to establish the temporal nature of a sentence. Non-natives must learn this, and since most dictionaries don’t discuss aktionsart, learning about this is very difficult.
(I hope that made sense and wasn’t too technical)