Go to WCCFL 2025 Conference homepage‘Irrealis’ particle ji in Gitksan
A particle ji in Gitksan (Tsimshianic; VSO)1 and a neighbouring language, Nisga’a, has been glossed as an ‘irrealis’ marker (Rigsby, 1986; Tarpent, 1987), but no formal account of its exact semantic contribution is available. Ji occurs in a variety of semantic contexts, only some of which are traditionally associated with irrealis morphemes (Von Prince et al., 2022), with others seemingly involving consideration of alternatives. This paper offers a unified account of semantic contribution of ji by combining ingredients from the Inquis- itive view on conditional antecedents and imperatives (Starr, 2014, 2020) and the literature on mood and modality (Schlenker, 2003). I will argue that ji generates a ranked set of alternatives, and presupposes that each alternative is a live possibility in a doxastic state of some salient agent. Puzzle Part of the challenge posed by ji is its occurrence in a wide variety of semantic contexts. A subset of those contexts suggest that ji involves consideration of alternatives. In particular, ji is obligatory in embed- ded polar interrogative clauses (1) and conditional antecedents (2). A corpus of the language also includes five instances of ji in disjunctions. In fact, Matthewson (2024) and Brown (2024) suggest that ji (and its counterpart) creates polar alternatives in Gitksan and another Tsimshianic language, Sm’algyax. (1) A: “Does Michael like tea?” B: “I don’t know but ask Lisa...” Wilaay[-t]=s Lisa ji anooḵ=s Michael=hl dii know[-3.II]=PN Lisa IRR like=PN Michael=CN tea ‘Lisa knows whether Michael likes tea.’ (2) Ji bax-t nee=dii gina hetxw-t IRR run-3.II NEG=FOC late-3.II ‘If he ran, he was not late.’ However, ji also has occurrences that do not appear to involve alternatives at a first glance. Such cases include imperatives and hortatives (3),2 which may have motivated the ‘irrealis’ label (Rigsby, 1986; Tarpent, 1987). Moreover, ji is licensed in attitude complements when the attitude holder or the speaker considers negation of the embedded proposition to be possible. In (4), the attitude holder is uncertain about the truth of the embedded proposition.3 In (5), the speaker knows the embedded proposition to be false. (3) Am ja ha’w-i’m good IRR go.home-1PL.II ‘Let’s go home!’ (Rigsby 1986:315) (4) Context: The speaker ate the berries. Lisa suspects that that is the case, but isn’t certain. Ha’niig̱oot[-t]=s Lisa (ji) nii’y an=t gup=hl ma’ay. think[-3.II]=PN Lisa IRR 1SG.III AX=3.I eat=CN berry ‘Lisa thinks that I ate the berries.’ (5) The speaker knows that Prof. Henry Jackson is a man. A new student comes in and says “I’m looking for Miss. Jackson.” Ha’niig̱oot-t (ji) hanaḵ[-t]=s Henry think[-3.II] IRR woman[-3.II]=PN Henry ‘He thinks that Henry is a woman.’ Analysis The obligatoriness of ji in embedded polar interrogatives (1) suggests that it is an alternative gen- erator responsible for taking a proposition p and generating a set {p, ¬p}, akin to whether and if (Kart- tunen, 1977). Moreover, a polar interrogative analysis of conditional antecedents has been developed by 1Unless otherwise noted, data are obtained by elicitation with two fluent speakers using the standard methodologies in semantic fieldwork (Matthewson, 2004). 2The morpheme is pronounced ji or ja depending on the speakers and dialects. 3Although the verb ha’niig̱oot is glossed as ‘think’, the literal translation of x ha’niig̱oot-s p is ‘it is on x’s heart that p’. I assume that the modal force of the verb is weaker than strong necessity. 1 Starr (2014), who argues that such an analysis explains the conditional-interrogative link observed across a number of unrelated languages. How can the polar interrogative analysis account for the modal meanings of ji in (3)-(5)? In fact, Starr (2020) proposes an account of imperatives using alternatives, in which an imperative p! ranks p as being preferred to its alternative, ¬p. Taking a hint from Starr (2020), I propose that the common denominator of various occurrences of ji is indeed alternative generation, with a ranking of p over ¬p and a presupposition that both p and ¬p are considered possible by some salient agent. A proposed denotation of ji is in (6), where the index 1 refers to a set of salient belief agents via the assignment g. Ji takes a proposition p with a presupposition that, for both p and ¬p, there is a salient belief agent in the set g(1), whose doxastic state does not rule out a possibility of the proposition being true. If defined, ji(p) returns a set of ordered propositions {p, ¬p}, where the underline on p indicates that it is ranked first. The idea that a single particle can evaluate the status of a proposition with respect to belief states of different agents depending on the context has a precedence in Schlenker’s (2003) analysis of the indicative in French, though in a different implementation. (6) Jji1Kw,g=λpst: ∀q[q∈{p, ¬p}→∃x[x∈g(1) & ∃w′[w′∈DOXx,w & q(w′)=1]]]. {p, ¬p}