Continuing the dive into modality, I’ve been looking at the scope of irrealis marking in different languages. Irrealis is a contentious label, with positions ranging from:

• Irrealis is a coherent category that recurs in multiple languages
• Irrealis is a semantically/diachronically linked set of meanings that form some kind of semantic map, and languages vary in which parts of the map they group together
• Irrealis is not a meaningful category

The second position appeals most to me for most linguistic phenomena (coherent cross-linguistic categories being the product of universal functional pressures over time), so let’s roll with that.

The Semantic Range of Irrealis

Syntax, Volume I by Givón gives an illustration of the range an irrealis marker can have using the prefix mag-from Bikol. The following examples are given:

Future

Subjunctive-imperative (polite)

Hortative

Non-implicative modality-verb complement

Non-implicative manipulation-verb complement

Non-factive cognition-verb complement

Irrealis Adv-clause

Although examples are not given, Givón states that:

• In Fijian, a single irrealis marker covers a similar range but also including habitual
• In the Northern Iroquois languages, a single irrealis marker covers a similar range but including negative clauses

Many of these categories are more commonly thought of as tense (future), aspect (habitual), or non-TAM (negation), so the use of a single marker to cover all of them is quite interesting.

It’s worth point out at this stage that grouping together these categories as irrealis doesn’t preclude having separate markers for them as well. Some languages do have an irrealis marker that can be used alone in a vague/ambiguous way, but many combine a general irrealis marker with more specific markers.

Irrealis semantics by frequency

Plugian’s Irrealis and Modality in Russian and Typological Perspective summarises the frequency of different uses for an irrealis mood marker as follows:

Always marked as irreal: Counterfactual Optative and conjunctive Intentional and volitional Probabilitive and dubitative	Less often marked as irreal: Indirect evidential Negative Interrogative Imperative and prohibitive Habitual and past imperfective Remote past
Most often marked as irreal: Future (especially remote or uncertain) Prospective Conditional and concessive Purposive	Never marked as irreal: Past perfective Immediate or resultative past Present progressive

The top-left box basically corresponds to the semantic core of subjunctive moods in European languages. In some European languages, subjunctive-like moods may also have conditional, purpose, or imperative functions, but typically not future or any of the other categories in the top-right (negative, habitual, and even remote past!). The typical European subjunctive therefore looks like an irrealis, but not a maximal one.

What do these categories share? Plugian suggests that they share a lack of at least one of the following properties of typical realis situations:

1. ‘Having taken place’, i.e. to occur or have occurred sufficiently close to the moment of speech
2. ‘Being certain’, i.e. the speaker must have a “strong belief” with regard to what is asserted
3. ‘being specific’, i.e. to represent a clearly identifiable situation with definite or specific arguments and precise location in time (preferably, single and completed)

For example, futures fail (1), dubitatives fail (2), and habituals fail (3). A similar approach can be seen in this interesting dissertation on Nafsan by Ana Krajinović Rodrigues. She reproduces a table from Van Gijn & Gipper, which I in turn have reproduced below:

Language	Counterfactual	Possible		Factual
		-SC	+SC	-TR	+TR
Bininj GW	IRR
C. Pomo	IRR
Alamblak	IRR
Caddo	IRR
Amele	IRR
Bargam	IRR
Yurakare	IRR

Here, SC = speaker commitment, TR = temporal. She quotes them as saying:

“Irrealis minimally includes counterfactual events; if it contains possible events, it will also contain counterfactual events; within the category of possible events, languages may also draw a boundary between [–speaker commitment] and [+speaker commitment] events. Inclusion of the later into the irrealis category implies inclusion of the former; if a language includes habituals [-temporal], it will also include possible and
counterfactual events.”

If I understand the terminology correctly, typical future tense clauses are possible with speaker commitment, whereas habituals are factual but without specific time reference. This typology therefore aligns with Plugian’s frequency table, which places future before habitual in terms of frequency of coverage by irrealis markers.

A more detailed semantic map of modality would be useful here, but I’m not completely happy with the ones I’ve found so far. If I find more good stuff, that can be the topic for next time.