Obligatoriness is often presented as an essential property of the grammar of a language (Boas 1959: 492). Some languages are claimed to possess a grammatical category because it is encoded by ‘obligatory’ markers, such as number in English, gender in Spanish, evidentiality in Quechua, etc. (Athanasopoulos & Kasai 2008, Lew-Williams & Fernald 2007, Aikhenvald 2007). Obligatorification also is often considered one of the definitory processes of grammaticalization, even though Lehmann (1995) stated that its importance should not be overemphasized (Lehmann 1995[1982]: 123-128; see also McGregor 2013). The relative obligatoriness of the subject in German, be+V-ing in English, or a gender suffix in a Romance language cannot be analysed as one phenomenon.
This epistemological study will review what is usually meant by obligatoriness, and to what extent forms may grammaticalize without acquiring this property (Heine & Kuteva 2007: 34). Basing myself on a critical analysis of theoretical works and first-hand cross-linguistic data, I will argue that obligatoriness is a cover term which may only be investigated by distinguishing three levels, i.e. morpho-syntax, semantics, and pragmatics. For the first two levels, the constraint may be intra-paradigmatic or extra-paradigmatic (Lehmann 1995[1982]: 123-128). Furthermore, the degree of systematization of a notion may determine how obligatory this grammatical category is in a given language.
This review will lead me to delve into the causes of obligatorification, which might appear contradictory to the principle of economy as one of the main motivators of grammaticalization (Gelderen 2004). I will argue that chunking, zero grammaticalization, and the Gricean maxim of quantity are necessary concepts to explain how a form may obligatorify (Ellis 2003, Bybee 1994, Grice 1975).