The concepts of mood and modality are still largely ill-defined in applied linguistics, and so are their applications in the various languages. This is why "working definitions" are temporarily used while attempts are being made towards a deeper understanding and a satisfactory systematisation of the area. The study of moods and modality allows two fundamental levels of analysis: morpho-syntactic and semantic. The latter is mainly adopted here.

There are two distinct and complementary approaches in semantic studies:

• the onomasiological approach goes from concepts to forms; for example, we might explore the area of duty and obligation to see the various forms that are used: the modals must, shall, should, ought, then other verbs like to have to, to be to, to compel (usually in the passive: to be compelled), and so on — the analysis might also be extended to nouns like duty or obligation, to adjectives like compulsory, etc. depending on the objectives of the research;
• the semasiological approach goes from forms to concepts; we start from a word or expression and explore all its possible uses and meanings.

The latter approach will be adopted here in order to clarify and exemplify a few aspects of modality in English and it will be based on the modal should. The preference for the semasiological approach arises from the complexity of some onomasiology-based analyses. Mathesius, for instance, mentions a number of modalities (indicative, optative, voluntative, expectative, potentialis, concessive, dubitative, irrealis) but they are not clearly defined and their description is partially based on the use of English modals:

Besides, some passages that are presumably clear in the original language (Czech) become circular and tautological in the English translation:

Notice that the definiendum ought to is also used as a key word in the definition.

Before the analysis of should, however, we need some preliminary remarks on the main modalities.

By default (that is, unless there are signals to the contrary), readers assume that each assertion is an expression of certainty. In other words, a simple sentence like

is understood to mean

Now compare:

with:

As we proceed, we find various degrees of uncertainty: the writer no longer guarantees that the helpfulness of computers is a known fact and that he is sure of it.

The truth value of a statement is not assessed on the basis of a binary choice between true and false but along a scale going from It is absolutely certain that.. to ...is totally to be excluded / out of the question. This scale is also known as epistemic gradient.

Epistemic modality, then, is the expression of the mode of knowing, from verified to undecided to falsified. Like all modalities in English, it is often expressed by modal verbs like must, may and might, but also in many other ways, as the examples above show. The other main modalities are:

deontic modality, expressing degrees of obligation, from obligatory to permitted to indifferent to forbidden;

alethic modality, pertaining to the modes of truth, from necessary to possible to contingent to impossible. (5)

The sentence:

is ambiguous because must can be interpreted as having an epistemic, a deontic or an alethic value. In the first case the writer is saying that he or she is convinced that the project is relevant (epistemic). In the second case, the writer is giving an order to someone who is to ensure that the project be relevant (deontic). Finally, the sentence may mean that being relevant is the logical consequence of the project's coherence with some premises: As all the conditions have been met, then it necessarily follows that the project is relevant (alethic).

The modal may is also often ambiguous, allowing for either an epistemic or a deontic modality:

can be interpreted as It is possible that Hilda ... (epistemic) or as Hilda is permitted... (deontic).

The avoidance of ambiguity in modality is an essential feature in most texts (possible exceptions are literary passages where the writer chooses to be ambiguous to achieve some special effects).

There are fields that raise deontological issues (6) and where, accordingly, deontic modality is highly relevant. But even in everyday discourse we frequently give orders, express suggestions, advise others, give or deny permissions, promise to do something and so on; in all these cases there is an underlying idea that somebody wants to do something or wants somebody else to do something: all the concepts of duty, obligation, permission, suggestion, advice, etc. imply an idea of volition in connection with an idea of what is good and correct, either from a moral or from a practical point of view.

A semasiological analysis of should will provide an example of the wealth of problems connected with English modals.

2.3 Uses of the modal should

The choice of should is prompted by the wealth of its different values, in particular within deontic modality. In Mathesius's book — and in almost all the dictionaries, textbooks and manuals of applied linguistics — the description of should is introduced starting with the meaning that seems to be the most ‘normal' and frequent, that is as a modal used for suggestions and advice. The following analysis will show that in several texts the cases where should corresponds to a form of condizionale presente of dovere (‘tu dovresti', ‘egli dovrebbe'...) are a minority.

In another passage of his book, Mathesius remarks:

This is called ‘putative should' by present-day grammarians(8); the Danish linguist Jespersen had already highlighted this as a remarkable case:

Here, again, painting January and August together is given as a fact — a fact that was surprising about seventy years ago (now it seems almost conservative, given the further developments of modern painting) and should is an attitude marker showing the speaker's emotional reaction. The few examples we have seen so far are already enough to clarify that it would be misleading to treat should as the past form of shall, or anyway as an auxiliary of the past tense. Most of the times should refers to the present or to the future. When it is used to give advice, as in:

the action of the main verb (here, ‘seeing a doctor') is always necessarily projected into the future, which sometimes may be an immediate future, as in:

The case of should have + V is different. We saw it used to refer to past events that are perceived as strange or unusual, as in the example about the civil war in Spain, but should have is mostly used to say that something necessary or desirable did not happen in the past:

Only the "simple form" of should will be analysed below, while the uses of should have followed by V-en will be omitted.

A comparative analysis shows frequent cases where should is translated into Italian with an indicativo:

or with expressions that somehow convey the speaker's attitude — in the following example, an idea of surprise:

The modality here is clearly epistemic. Most of the times, and particularly after the conjunctions if and that, the most appropriate Italian translation contains the congiuntivo presente or imperfetto of the main verb in :

Several examples are found in argumentative texts where the meaning of should approaches the one of dovrebbe:

Here should retains one of its main traits, i.e. advisability; however, the cases where the communicative purpose of the utterance leads to other interpretations are more frequent. In the Helsinki Declaration on clinical research involving human subjects (1964), should occurs in a number of fundamental points:

This is not a norm or a rule that researchers dovrebbero follow if they like: this is one of the rules that devono be followed in order not to violate the principles guiding the Declaration — which, anyway, begins with an Introduction to the Principles that explicitly states and clarifies that the document has a binding value. Here should is as categorical as must, as can be seen in the following paragraph where both modals occur: