1 Introduction

Grammatical accounts of na in Akan identify two different forms. These are nà with a low tone (henceforth lt-na) and ná with a high tone (henceforth ht-na). It has been argued in the literature that lt-na functions in two ways, namely as a focus marker or a conjunction, the latter of which can take a prefix and be realized as ɛna ‘and then’. Consider the slightly modified examples below from Schwarz and Fiedler (2007: 274) (upper case letters in the English gloss mark a focused element).1

    1. (1)
    1. a.
    1. àdùá
    2. beans
    1. nà
    2. lt-na
    1. ɔ̀-dí-ì-yɛ́
    2. 3sg.subj-eat-pst-Yɛ́2
    1. ‘He ate BEANS.’ (Schwarz and Fiedler 2007: ex. 13)
    1.  
    1. b.
    1. Maame
    2. Maame
    1. nòá-à
    2. cook-pst
    1. àdùá,
    2. beans
    1. nà
    2. lt-na
    1. n’-àdámfò
    2. 3sg:poss-friend
    1. dí-ì-yɛ́
    2. eat-pst-Yɛ́
    1. ‘Maame cooked beans and her friend ate them.’ (Schwarz and Fiedler 2007: ex. 14a)

The NP àdùá ‘beans’ in (1a) is fronted and followed by lt-na. The translation suggested by Schwarz and Fiedler shows that àdùá ‘beans’ is marked as the focus of the sentence. The two clauses in (1b) seem to be conjoined by lt-na. While scholars like Boadi (1974) treat the lt-na in examples like (1a) as a different particle from the lt-na in (1b), others like Bearth (1999a, 2002) and Schwarz and Fiedler (2007) point to a commonality between the two. Our contention is in accord with the latter studies that consider the two to be related (see discussion in §3 below).

ht-na has been analyzed by Boadi (2008) as three homophones, marking past tense (as in 2a), future tense (as in 2b) or logical connector (as in 2c):

    1. (2)
    1. a.
    1. mè
    2. 1sg
    1. bá-è
    2. come-pst
    1. nó
    2. tp3
    1. ná
    2. pst
    1. ɔ̀-rè-dìdí.
    2. 3sg.subj4-prog-eat
    1. ‘When I came, he was eating.’ (Boadi 2008: ex. 62)
    1.  
    1. b.
    1. (wó
    2. 2sg
    1. bɛ́-bá
    2. fut-come
    1. nó)
    2. tp
    1. ná
    2. fut
    1. Kofi
    2. Kofi
    1. hó
    2. skin
    1. á-tɔ̀
    2. perf-settle
    1. nò.
    2. 3sg
    1. ‘(By the time you come back), Kofi will have enjoyed a state of mental calm.’ (Boadi 2008: ex. 74)
    1.  
    1. c.
    1. sɛ̀
    2. If
    1. wó
    2. 2sg
    1. hù-ù
    2. see-pst
    1. nò
    2. 3sg
    1. áà5
    2. cond
    1. (ɛ́ńdèɛ̀)
    2. (in.that.case)
    1. ná
    2. it.follows.from.the.aforesaid.that
    1. ɔ̀
    2. 3sg.subj
    1. bà-è.
    2. visit-pst
    1. ‘If you saw him, then it logically follows that he visited.’ (Boadi 2008: ex. 79)

We have modified the glosses of these examples to conform to Leipzig glossing rules but kept Boadi’s glosses of ht-na to show that he treats them as separate morphemes. We will suggest in Section 4 different glossing according to how we interpret these sentences.

In this paper, we argue that all uses of ht-na (as illustrated in 2a–2c), along with the uses of lt-na (illustrated in 1a and 1b), are subcategories of one super-category. We propose that the super-category is a non-tonal na, call it Root-na, with a common basic function which explains all five seemingly unrelated interpretations. Our contention is that Root-na links the na-clause with something in the common ground, i.e., to something that appeared in the previous context or is presupposed. It is spelled out as a lt-na or ht-na, depending on the kind of linking.6

The paper is organized as follows. In §2, we show that the behavior of tone in Akan supports our proposal that there is one na with different realizations. In §3, we show that lt-na marks focus (as in 1a above) or discourse coherence relations such as narrative-sequence (as in 1b). In §4, we discuss ht-na, showing it to be an intensional marker which links times (as in 2a and 2b above) or possible worlds (as in 2c). In §5, we summarize and conclude the findings of the previous sections.

2 Tone in Akan

Akan is a (New) Kwa language of the Niger-Congo family spoken in Ghana (cf. Boadi 2005). Its dialects include Akuapem, Asante, Fante, Wasa, Agona, Akyem, Bron/Abron, Kwahu and Gomua (cf. Dolphyne & Dakubu 1988). Dolphyne & Dakubu (1988) note that the major non-Fante dialects (i.e., Asante, Akuapem, Akyem and Kwawu) are called Twi. Unless otherwise specified, the examples for this paper are taken from the Asante Twi.

Akan is a tone language with two basic tones, i.e., High and Low. Tone plays lexical as well as grammatical functions. The examples below in which tone distinguishes between lexical items are from Dolphyne (1988: 53):

    1. (3)
    1. a.
    1. pápá ‘good’
    1.  
    1. b.
    1. pàpá ‘father’
    1.  
    1. c.
    1. pàpà ‘fan’
    1. (4)
    1. a.
    1. dá ‘day’
    1.  
    1. b.
    1. dà ‘never’
    1. (5)
    1. a.
    1. ɔ̀bɔ̀fó ‘hunter’
    1.  
    1. b.
    1. ɔ̀bɔ́fó ‘creator’

High and Low tones also occur on the same lexical item and make a grammatical distinction (Dolphyne 1988). Examples are given below:

    1. (6)
    1. a.
    1. ɔ̀-hyɛ́
    2. 3sg.subj-wear:ht
    1. àtàdéɛ́.
    2. garment
    1. ‘She wears a dress.’ (Dolphyne 1988: ex.70)
    1.  
    1. b.
    1. ɔ̀-hyɛ̀
    2. 3sg.subj-wear:lt
    1. àtàdéɛ́.
    2. garment
    1. ‘She is wearing a dress.’ (Dolphyne 1988: ex.71)
    1. (7)
    1. a.
    1. Kofi
    2. Kofi
    1. dá
    2. lie:ht
    1. m̀pá
    2. bed
    1. só.
    2. top
    1. ‘Kofi lies in bed.’
    1.  
    1. b.
    1. Kofi
    2. Kofi
    1. dà
    2. lie:lt
    1. m̀pá
    2. bed
    1. só.
    2. top
    1. ‘Kofi is lying in bed.’
    1. (8)
    1. a.
    1. tòá
    2. bottle
    1. nó
    2. def
    1. sí
    2. be_on_base:ht
    1. pónó
    2. table
    1. nó
    2. def
    1. só.
    2. top
    1. ‘The bottle is usually on the table.’
    1.  
    1. b.
    1. tòá
    2. bottle
    1. nó
    2. def
    1. sì
    2. be_on_base:lt
    1. pónó
    2. table
    1. nó
    2. def
    1. só.
    2. top
    1. ‘The bottle is on the table.’

The High tone on the verbs in the (a) examples of (6)–(8) is analyzed in Akan linguistics as the Habitual while the Low tone is analyzed as the Stative (Dolphyne 1988; Boadi 2008) or the Continuative (Osam 1994). For our purposes, these examples show that the same lemma can take a different tone and yield a different interpretation. We argue that lt-na and ht-na also involve the same lemma, and they receive their specific interpretation based on their respective tone.

An anonymous Glossa referee has pointed out that our argument that lt-na derives from the underlying Root-na (just like ht-na) may be challenged by Ofori (2011). According to Ofori, na is derived from ne /nɪ/ which he treats as the basic focus marker in Akan. Ofori (2011) derives na from a construction that involves a relative clause headed by a generic noun (e.g., ‘The girl is the person who took my pen’, where ‘person’ is the generic noun). Ofori argues that when the relative clause is headless (i.e., it does not have a generic noun head), the contiguity of ne and the relativizer gives rise to na. He provides the following morphophonological rule: “/ɪ/➔/n_Ø /n_ +a/ ([a] being the relativizer), and the fusion of [n] and [a] to derive na.” (2011: 251). According to Ofori, the high tone of ne is dropped when the vowel is deleted leaving the tone of the relativizer only. While this is an intriguing proposal, it is problematic because the tone of the relativizer is falling, not low. Saah (2010: 93) writes: “Akan uses the particle áà (said with a falling tone) to mark the beginning of the relative tone” (referring to the relativizer). He adds in a footnote: “The relative complementizer is represented by the letter a in Akan orthography but I will follow Schachter (1973: 23) in representing it as áà to reflect its actual phonetic realization” (Saah 2010: 9, fn. 2). We can conclude that contrary to what Ofori suggests, lt-na does not derive from [ne] and argue that it derives from Root-na.7

The parallelism between the focus marker and the conjunction is not peculiar to Akan alone. Fiedler and Schwartz (2004) discuss similar parallels in two other Kwa languages (i.e., Ewe and Lelemi) in addition to Akan and two Gur languages (i.e., Buli and Dagbani). In section 3.3, we suggest that a “narrative hypothesis” which they propose to account for the parallelism supports our proposal.

3 lt-nà: inferential relations

In this section, we show that all the uses of lt-na observed in the literature (see, e.g., Amfo 2007; Schwartz & Fidler 2007) have one property in common, namely inferential linking. In subsections §3.1 and §3.2, we discuss its function to mark discourse coherence relations and focus (linking), respectively. In the concluding sub-section §3.3, we include a short account for the “narrative hypothesis” of Schwarz and Fiedler (2007) as support for our proposal.

3.1 Marking discourse coherence relations

In §3.1.1, we discuss the notion of discourse coherence relations in language, and in §3.1.2, we show that lt-na in Akan is used for marking such relations.

3.1.1 Discourse coherence relations in language

Asher (1993), Asher (1998), Lascarides and Asher (1993) and Asher and Lascarides (2003) have developed Segmented Discourse Representation Theory (SDRT) to account formally for discourse coherence relations.8

SDRT suggests an analysis of the discourse relations holding between propositions that a (coherent) text consists of. This analysis takes into consideration not only linguistic knowledge, with its logical entailments, but also world knowledge, whose inferences may be defeasible. The following examples from Asher & Lascarides (2003: 6) illustrate:

    1. (9)
    1. a.
    1. Max fell. John helped him up.                                                                                                       (Narration)
    2. (Asher & Lascarides 2003: ex. 5)
    1.  
    1. b.
    1. Max fell. John pushed him.                                                                                                       (Explanation)
    2. (Asher & Lascarides 2003: ex. 6; Lascarides & Asher 1993: ex. 2)

As Asher & Lascarides note, discourses (9a) and (9b) have the same tense forms and aspectual classes (aka ‘Aktionsarten’), yet they seem to imply different coherence relations between their sentences. In (9a), the sentences are interpreted as relating a story in which a certain sequence of events is described; hence they are understood to be temporally ordered. The coherence relation between them is therefore determined to be Narration. In (9b), the event reported by the first clause (Max’s falling) is interpreted as having happened after (and as a consequence of) the event reported in the second (John’s pushing Max). Accordingly, the coherence relation between the first and second clause is understood to be that of Explanation. Other relations include Contrast, Parallel, Elaboration, etc.

Since discourse relations are based in part on world knowledge, they are defeasible, and “one and the same proposition can have different discourse roles in different contexts” (Asher & Lascarides 2003: 136). The relationship between Max fell and John pushed him, for example, is Explanation in (9b) above but Narration in (10) below:

    1. (10)
    1. John and Max were at the edge of a cliff. Max felt a sharp blow to the back of his neck.
    2. Max fell. John pushed him. Max rolled over the edge of the cliff.
    3. (Asher & Lascarides 2003: ex. 16; Lascarides & Asher 1993: ex. 20)

The different relationships in (9b) and (10) affect the interpretation of the events’ temporal order. While in (9b) Max’s falling is understood to have followed John’s pushing, in (10) it is the other way around.

Crucially for the points made in this section, the analysis can be applied to conjoined sentences. Consider the sentences in (11a) and (11b) below:

    1. (11)
    1. a.
    1. Max fell and John helped him up.                                                                                                 (Narration)
    2. Max loves Mary and she hates him.                                                                                              (Contrast)

The conjoined clauses in (11a) are understood to hold the Narration relation between them, as the event depicted in the second clause is understood to follow in time the one depicted in the first. In contrast, the conjoined sentences in (11b) are understood to depict two simultaneous states that contrast each other.

3.1.2 lt-na and discourse coherence relations

Amfo (2007) shows that lt-na in Akan may imply the same coherence relations as the English ‘and’ in sentences like (11a) and (11b) above. In particular, she shows that various discourse relations may exist due to the appearance of lt-na: Narration (Temporality, in her terminology), Causal, Parallel, Contrastive as well as Explanation. Consider some of her examples:

    1. (12)
    1. Narration (= Temporality)
    1.  
    1. a.
    1. Agyeman
    2. Agyeman
    1. dì-ì
    2. eat-pst
    1. àdùàné
    2. food
    1. nó
    2. def
    1. nyìnáá
    2. all
    1. nà
    2. lt-na
    1. ɔ̀-dá-é.                    (Asante)
    2. 3sg.subj-sleep-pst
    1. ‘Agyeman ate all the food and (then) he slept.’ (Amfo 2007: ex. 1)
    1.  
    1. b.
    1. ὀ-yì-ì
    2. 3sg.subj-take.off-compl
    1. nè
    2. 3sg:poss
    1. m̀pàbòwá
    2. sandal
    1. nyìm̀fá
    2. right
    1. dzè
    2. own
    1. nó
    2. 3sg
    1. nà.
    2. lt-na
    1. ɔ̀-dzé
    2. 3sg.subj-take
    1. nè
    2. 3sg:poss
    1. náń
    2. foot
    1. tsíá-à
    2. step-compl
    1. dó                    (Fante)
    2. top
    1. ‘He took off his right sandal and then stepped on it.’ (Amfo 2007: ex. 7)
    1. (13)
    1. Contrast
    1.  
    1. a.
    1. mè-kɔ̀-ɔ̀
    2. 1sg-go-compl
    1. nè
    2. 3sg:poss
    1. fíé
    2. house
    1. nà
    2. lt-na
    1. m’-à-ǹ-kɔ̀
    2. 1sg-compl-neg-go
    1. tó
    2. meet
    1. nò.
    2. 3sg
    1. ‘I went to his house and (but) I didn’t meet him.’ (Amfo 2007: ex. 2)                                             (Asante)
    1.  
    1. b.
    1. áhyɛ̀sé
    2. beginning
    1. nó
    2. def
    1. nná
    2. then
    1. ǹkyɛ̀
    2. mm
    1. mè-m̀m̀-pɛ́
    2. 1sg-neg-like
    1. nà
    2. lt-na
    1. sèèsé!í
    2. now
    1. dze
    2. fm
    1. mè-pɛ̀.                                  (Fante)
    2. 1sg-like
    1. ‘In the beginning, (then) I didn’t like (the idea), but now I do.’ (Amfo 2007: ex. 12)
    1. (14)
    1. Explanation
    1. kɔ̀
    2. go.imp
    1. prá
    2. sweep
    1. dáń
    2. room
    1. nó
    2. def
    1. mú
    2. inside
    1. hɔ́
    2. there
    1. nà
    2. lt-na
    1. mama
    2. Mama
    1. bɛ́-bá
    2. fut-come
    1. sèèsé!í
    2. now
    1. árá.
    2. just
    1. ‘Go and sweep the room because Mama will come right now.’ (Amfo 2007: ex. 15)

Amfo glosses all the lt-na occurrences in examples (12)–(14) as ‘and’, as she understands their respective clause to be syntactically coordinated to the previous clause. Even when a lt-na clause appears detached from a previous clause, she takes it to be still conjoined. Consider her example given in (15) below:

    1. (15)
    1. nà
    2. lt-na
    1. mó
    2. 1sg:poss
    1. wùrà
    2. lord,
    1. wóáná
    2. who
    1. nà
    2. lt-na
    1. ɔ̀-bɔ́-bòá
    2. 3sg.subj-fut-help
    1. hèǹ
    2. 1pl
    1. wɔ̀
    2. LOC
    1. fíé?      (Fante)
    2. home.
    1. ‘And my lord, who will help us at home?’ (Amfo 2007: ex. 26)

In this case, too, Amfo believes that the lt-na in sentence initial position functions as a connective that correlates with the English ‘and’. Her contention is that its clause is coordinated with a clause uttered previously by the interlocutor who uttered (15). Based on these and other similar examples, Amfo (2007: 682) concludes:

[The] semantics of nà is univocal. It is simply a coordinating connective. Specific communicated inferential relations between conjuncts is the result of collaboration between the univocal lexical semantics of nà and the syntactic structure of conjuncts, general world knowledge and other contextually determined information.

While Amfo is adequate in indicating that lt-na in examples like (14) and (15), “tells the interlocutor to look out for certain kinds of inferential relations between the conjuncts” (p. 666), she does not connect this function to the focus marker. In sub-section §3.2 below, we discuss the notion of focus, showing it to have linking properties, and in sub-section §3.3, we show that because of that lt-na may function as a focus marker.

3.2 Focus marker

It has been observed in linguistic literature that one of the characteristics of focus is implying some relationship with some element(s) in its respective discourse. In §3.2.1 below, we will discuss the notion of focus in language, concentrating on its discourse linking characteristics. Being in general a discourse linking morpheme, we argue that lt-na in Akan appears with a focused element to alert the addressee to look for some discourse linking suggested by that element (§3.2.2).

3.2.1 Focus in language

Different kinds of focus have been observed in the literature, usually classified according to their function. (A classification according to a different parameter is suggested by Rooth 2008 – see discussion on example 19 below.) The sentences in (16a–16e) below illustrate (where the focused phrases are capitalized):

    1. (16)
    1. a.
    1. (Who painted the shed yesterday?)
    2. JOHN painted the shed yesterday. (Rooth 2008: ex. 1)
    1.  
    1. b.
    1. (When did John paint the shed?)
    2. John painted the shed YESTERDAY. (Rooth 2008: ex. 2)
    1.  
    1. c.
    1. (Mary bought some lentils for the soup.)
    2. BEANS Mary bought for the soup (not lentils).
    1.  
    1. d.
    1. (I know Mary likes lentils, so I made a pot full of lentils.)
    2. Mary likes BEANS (not lentils).
    1.  
    1. e.
    1. It was A HAT that Mary picked up for herself. (Kiss 1998: ex. 8)

The focused phrase in each of the question-answer pairs in (16a) and (16b) is referred to in the literature as ‘completive focus’ or ‘information focus’ (see, e.g., Rooth 2008), as it aims to fill missing information. The sentences in (16c) and (16d) illustrate what is usually called ‘contrastive focus’, as it stands in contradiction with what is stated or presupposed in its respective discourse. And (16e) is an example of ‘exhaustive focus’, which “represents a subset of the set of contextually or situationally given elements for which the predicate phrase can potentially hold” (Kiss 1998: 249). In these and other examples, a focused phrase may be distinguished from the other phrases in its respective clause phonologically, syntactically, and semantically/pragmatically.

Phonologically, a focused phrase is distinguished from the other phrases in its respective clause by having stress prominence (Jackendoff 1972), and, in some African languages, by pitch accent (Ameka 2010). Syntactically, a focused phrase may remain in situ, as in (16d), or be moved to the front of its respective clause, as in the English example (16c) and (all) the Akan examples in the preceding sections.9 We will not discuss word order here, as it does not affect the semantic/pragmatic points we are trying to make.

It is well acknowledged that a focused phrase adds some new information to its respective discourse. Rooth (1985; 1992, etc.) shows that such a phrase, furthermore, evokes a set of alternatives, which he refers to as “focus semantic value” (Rooth 2008: 283) and Harbour (2008) calls it “(alternatives) focus set”. By way of illustration, compare example (17a) below with (17b = 16c) and (17c):

    1. (17)
    1. a.
    1. Mary bought beans.
    1.  
    1. b.
    1. Speaker A: Mary bought lentils for the soup.
    2. Speaker B: BEANS Mary bought for the soup (not lentils).
    1.  
    1. c.
    1. Speaker A: I know Mary doesn’t like beans, but that’s all I have.
    2. Speaker B: Mary LIKES beans (but she’s allergic to them).

By fronting the NP object ‘beans’ and giving it prominent stress, Speaker B in (17b) induces a set of alternative propositions of the form ‘Mary bought x’: {Mary bought beans, Mary bought lentils, Mary bought peas …}, from which one element is picked out (‘Mary bought beans’). The prominent stress on the verb ‘like’ in (17c) functions to focus it, inducing an alternatives-set of the form ‘Mary x beans’: {Mary likes beans, Mary hates beans, Mary is allergic to beans…}. In contrast to (17b) and (17c), the sentence in (17a), with the original word order and with no prominent stress on any of its constituents, does not evoke a set of alternatives and merely asserts a proposition, adding some new information. To put it in Harbour’s (2008: 863) words, “focus both asserts the sentence and evokes a set of alternatives, the focus set.” The (open) proposition with the variable, e.g., ‘Mary likes x’, is referred to as the “focus skeleton” (Rooth 2008).10

3.2.1.1 Focus and linking

As reflected by the examples given so far, a speaker may focus a phrase only against some background within its respective discourse, where the speaker answers a question (as in 16a and 16b), expresses disagreement with the interlocutor (as in 17b and 17c), etc. In other words, focusing a phrase is only felicitous if its clause is linked to some element(s) in its respective discourse. It is not surprising, then, that Bearth (1999a: 148) asserts that “The primary contribution of focus to discourse is to ensure its coherence.” Building on Halliday’s (1967) analysis, Rooth (2008) suggests a syntactic-semantic interface account for the focus’ property of being linked to some element in its respective discourse.

Rooth (2008: 282) hypothesizes that “the semantics and pragmatics of focus involves a relation to context, which is a kind of anaphora.” Following Jackendoff (1972), Rooth considers focus to be a syntactic feature F, such that F marks the focused phrase as in examples (18a) and (18b) below, corresponding to (16a) and (16b), respectively:

    1. (18)
    1. a.
    1. [S[NP John]F [VP[VP painted [NP the shed]] [NP yesterday]]]
    1.  
    1. b.
    1. [S[NP John] [VP[VP painted [NP the shed]] [NP yesterday]F]]

Crucially for the current discussion, “The point of the F feature is to link the phonology of focus with the semantics and pragmatics of focus” (2008: 278). This linking is done by matching the variable in the focus skeleton, e.g., ‘x’ in ‘Mary likes x’, to the focused element in the discourse. Rooth suggests adding an index to match the focus with its “antecedent”. Consider one of his examples.

The following example (Rooth’s ex. 20) contains an instance of contrastive focus, which Rooth calls “substitution focus”, to reflect the fact that the speaker substitutes a phrase in the “antecedent” with another:

    1. (19)
    1. Speaker A:
    2. Speaker B:
    1. [John wrote the report]4
    2. [No [MaryF4 wrote it]]
    1. (Rooth 2008: ex. 20)

According to Rooth‘s analysis, the focused NP (‘Mary‘) of Speaker B is coindexed with the whole clause of Speaker A. As B‘s utterance contradicts the part of A‘s assertion that attributes to John the writing of the report, Rooth claims that “John in the antecedent substitutes for the F-marked Mary” (p. 282).

3.2.1.2 Wh-questions

As noted by Titov (2019), a number of scholars have noted that wh-categories exhibit a linguistic behavior identical to that of foci, proposing that wh-elements are inherently focused (Rochemont 1978; Culicover & Rochemont 1983; Beck 2006; Sandler & Lillo-Martin 2006; to mention just a few).11 As summarized by Eckardt (2007: 211), in a number of typologically unrelated languages, the same means mark wh-phrases in questions and focused constituents, including prosody, position and particles. What is most relevant for the present paper is the semantic/pragmatic parallel observed for focus and questions.

According to the standard semantic theory of questions (Hamblin 1973; Karttunen 1977), the denotation of a question is the set of answers to that question. According to the standard theory, wh-phrases, like focus, introduce a set of alternatives (hence the term ‘alternatives semantics’ given to the framework within which Hamblin’s analysis of questions and Rooth’s analysis of focus were developed). For instance, the question ‘where did Mary go’ introduces the set {Boston, New York, Miami …}. The semantic parallel has inspired semanticists to derive the interpretations of questions and focus the same way (Hamblin 1973; Ramchand 1997; Rullmann & Beck 1998; Kratzer & Shimoyama 2002; Beck 2006; and others). Accordingly, some authors offer a semantic motivation. To give one example, Sandler & Lillo-Martin (2006: 434) argue that “[s]emantically, the interpretation of WH-questions involves a presupposition (the known information) and a gap (the question). Thus, it is not surprising that WH-phrases should occur in structures used for focus” (emphasis added). In other words, like focus, a wh-question is linked to something in its respective discourse. As shown by Titov (2019), this explains the appearance of the lt-na in wh-questions in Akan (see discussion in §3.2.2.2 below).

3.2.2 lt-na as a focus marker

Studies such as Boadi (1974), Bearth (1999a, 1999b), Schwarz and Fiedler (2007), Ameka (2010), among others, acknowledge the use of lt-na in Akan as a focus marker (henceforth fm). Most of these accounts, notably Bearth (1999b), show that a phrase followed by this fm is linked to some other phrase in the previous discourse or to some implied set of propositions. As this was already shown in §3.2.1.1 to be a general characteristic of focus in language, in this section we can just illustrate the phenomenon in Akan.

3.2.2.1 lt-na in focus constructions

The discourse in (20) below is interpreted as containing some disagreement between the interlocutors, suggesting that utterance (20b) is an instance of a contrastive focus. In this case, the focused phrase in (20b) substitutes a phrase within the other speaker’s utterance in (20a):

    1. (20)
    1. Contrastive focus (= Substitution focus)
    1.  
    1. a.
    1. Akwasi
    2. Akwasi
    1. dì-ì
    2. eat-pst
    1. ɛ̀mó
    2. rice
    1. ‘Akwasi ate rice.’
    1.  
    1. b.
    1. dààbí
    2. no
    1. àdùá
    2. beans
    1. nà
    2. lt-nà
    1. ɔ̀-dí-ì-yɛ́
    2. 3sg.subj-eat-pst-YԐ
    1. ‘No, he ate BEANS.’

The speaker of (20a) states that Akwasi ate rice. Disagreeing only with the ‘rice’ part, the speaker of (20b) “corrects” it, claiming that what Akwasi ate was rather beans. Hence the fronting of the NP àdùá ‘beans’, and the focus (linking) marker lt-na following it.

The example in (21) from the play Guasohantan- (Koranteng 199l: 35) discussed by Bearth (1999b; ex. 2) illustrates an inferentially derived set focus (see Rooth 1992 for a discussion on this kind of focus)

    1. (21)
    1. Inferentially derived set (Glosses are Bearth’s, ibid)
    1.  
    1. a.
    1. mé
    2. lsg
    1. nà
    2. lt-na
    1. mé-ré-hyíá
    2. lsg-PRG-meet
    1. wò
    2. 2sg
    1. àyèfóró
    2. wedding
    1.  
    1. b.
    1. nà
    2. lt-na
    1. ɛ̀-n-yɛ́
    2. 3sg:INAN-neg-be
    1. wó
    2. 2sg
    1. nà
    2. lt-na
    1. wó-ré-hyíá
    2. 2sg-prog-meet
    1. mè.
    2. 1sg
    1.  
    1. c.
    1. éntí
    2. therefore
    1. wó
    2. 2sg
    1. dè
    2. as.for
    1. hwɛ̀
    2. look-to
    1. mé
    2. 1sg
    1. árá.
    2. simply
    1. ‘It is me who is marrying you, and not you who are marrying me. As for you, just depend on me.’ (Koranteng (1991: 35) as given in Bearth (1999b: ex 2))

As Bearth (ibid) notes, a grammatical focus in (21a) is indicated by fronting (a copy) of the focus element – mé 1sg – and attaching the focus marker lt-na to its right. As he further indicates, the F-marking invites the focus skeleton (though he does not use this term) ‘x organizes the wedding’, where the variable (x) takes its value from the possible candidates in this discourse, namely the male protagonist Osei Bonsu and his fiancée Akyere. In other words, we may conclude that the “antecedent” of the focused phrase mé in (21b) is Osei Bonsu. The second lt-na in (21b) is also indicated by Bearth to function as an F-marker. In this case, the “antecedent” of the focused phrase ‘you’ is Akyere. (The first lt-na in 21b is adequately glossed by Bearth as ‘and’, reflecting its linking to 21a). As Bearth adequately argues, the focus sets of both the first and second focused phrases are inferred from the context.

    1. (22)
    1. Completive focus
    1.  
    1. a.
    1. [hwáń
    2. Who
    1. nà
    2. lt-nà
    1. ɔ̀-kɔ́-ɔ̀
    2. 3sg.subj-go-pst
    1. páátì
    2. party
    1. nó
    2. def
    1. ásé?]6
    2. under
    1. ‘Who went to the party?’
    1.  
    1. b.
    1. [[Akwasi]6
    2. Akwasi
    1. nà
    2. lt-nà
    1. ɔ̀-kɔ́-ɔ̀
    2. 3sg.subj-go-pst
    1. páátì
    2. party
    1. nó
    2. def
    1. ásé]
    2. under
    1. ‘AKWASI went to the party.’12

The focus skeleton suggested by the question in (22a) is ‘x went to the party’. The value of x in (22b) is the proper name Akwasi, which means that it is the focused phrase, to be coindexed with the question in (22a). As expected, it is followed by the fm lt-na.

Note that the wh-word in question (22a) is also followed by a lt-na. In light of the discussion in section 3.2.1.2 above, this is to be expected. The next section is devoted to the lt-na in wh-questions.

3.2.2.2 lt-nà and wh-questions

Titov (2019) discusses the appearance of lt-na in wh-questions. She shows that only when the wh-word is moved from its original position does the lt-na appear. She compares the question-answer pairs in (23) and (24) below (hers: 25 and 26, respectively), from Saah (1994: 136–137):13

    1. (23)
    1. Q:
    1.   wó-hù-ù
    2.   2sg-see-pst
    1. héná?
    2. who
    1.   ‘Who did you see?’
    1.  
    1. A1:
    1.   mè-hù-ù
    2.   1sg-see-pst
    1. Adwoa
    2. Adwoa
    1.  
    1. A2:
    1. #Adwoa
    2.   Adwoa
    1. ná
    2. lt-na
    1. mè-hú-ù
    2. 1sg-see-pst
    1. nó
    2. 3sg.obj
    1.   (Titov 2019: ex. 25)
    1. (24)
    1. Q:
    1.   héná
    2.   who
    1. nà
    2. lt-na
    1. wó-hú-ù
    2. you-see-pst
    1. nó?
    2. 3sg.obj
    1.   ‘Who was it that you saw?’
    1.  
    1. A1:
    1.   Adwoa
    2.   Adwoa
    1. nà
    2. lt-na
    1. mè-hú-ù
    2. I-see-pst
    1. nó.
    2. 3sg.obj
    1.   ‘Adwoa was who I saw.’
    1.  
    1. A2:
    1.   mè-hù-ù
    2.   1sg-see-pst
    1. Adwoa.
    2. Adwoa
    1.   ‘I saw Adwoa.’
    2.   (Titov 2019: ex. 26)

The position Titov takes is that the wh-questions in (23Q) and (24Q) differ with respect to the nature of the alternatives set that the wh-phrases quantify over.

Adopting Pesetsky’s (1987) [±D-linked] feature, Titov claims that the in-situ and the left wh-objects in (23Q) and (24Q), respectively, have distinct values of the [±D-linked] feature. The ex-situ wh-object in (24Q), she argues, is [+D-linked] in that it consistently quantifies over a discourse-salient set of alternatives. This means that (24Q) “does not merely ask to provide a value for x in ‘I saw x’ … [but] to select a candidate that is the most likely member of its set to fulfil the relevant proposition” (p. 13). Echoing Saah (1994), Titov continues: “There is a feeling that the speaker is already aware of potential candidates that can fulfil this proposition and insists on the hearer telling which exactly candidate of this set can do so” (ibid). (Cf. Kiss 1998.) This discourse linking requires the appearance of lt-na. In contrast, according to Titov, the in-situ wh-object in (23Q) may open a new unlinked set, and, therefore, the lt-na is not licensed.

Titov’s analysis finds support in examples like (25) below, analyzed by Bearth (1999b: 259).

    1. (25)
    1. a.
    1. hwán
    2. who
    1. nà
    2. lt-na
    1. wó-bɛ́-wàré
    2. 2sg-fut-marry
    1. ɔ̀nó?
    2. 3sg:ind
    1. ‘(Of the two), whom do you want to marry?’
    1.  
    1. b.
    1. ↄhéné
    2. chief
    1. nó
    2. def
    1. nà
    2. lt-na
    1. mè-pɛ́
    2. 1sg-want
    1. sɛ́
    2. that
    1. mè-wáré
    2. 1sg-marry
    1. nó-ɔ̀
    2. 3sg-tp
    1. ‘It is the chief whom I want to marry.’
    2. (Bearth 1999b: ex. 9)

Bearth glosses the lt-na both in the question part (25a) and the answer part (25b) as F, suggesting that it is an fm in both cases. The lt-na in the answer (25b) is licensed because it follows a phrase that seems to be an exhaustive focus. As for the lt-na in (25a), Bearth claims that it is licensed because the girl in question faces two suitors, one of which is a chief. In other words, the wh-question in (25a) is linked to what is known from the context.

3.3 Summary

In §3.1.2, we showed that lt-na may be used to indicate some discourse coherence relation with previous or inferred clauses, where in some cases it corresponds to the English coordinator ‘and’. In §3.2.2, we showed that lt-na functions as a focus marker. The two functions are seemingly two different phenomena. However, as shown by Asher (1998) and Asher & Lascarides (2003), SDRT-based accounts of the two phenomena show that they display similar discourse constraints. In the case of sentences with no focused elements, an SDRT-account shows them to interact via discourse coherence relations like Narration or Explanation, whether they include the conjunction ‘and’ or not. Adopting Rooth’s (1992, etc.) claim that a focused element induces an alternatives-set, they suggest an SDRT-account of focus, considering the set of alternatives to be anaphoric in nature. Accordingly, (Asher 1998: 263) claims that the set of alternatives “must be either bound anaphorically or accommodated”. The idea that coordination has something in common with focus constructions finds some syntactic support in Fiedler and Schwarz’s (2005) and Schwarz and Fiedler’s (2007) “narrative hypothesis”.

Examining morpho-syntactically marked focused constructions in five Ghanaian languages, including Akan, Schwarz and Fiedler show that there is formal parallelism between the out-of-focus part of ex-situ (i.e., fronted) focus and narrative clauses. In Fiedler and Schwarz (2005: 136) they write: “[There] are striking similarities on the morpho-syntactic level between the non-focal part of the focus constructions and [Narrative Clauses]”, and in Schwarz and Fiedler (2007: 277) they propose that “a narrative clause constitutes the non-focal part of such ex-situ focus constructions”.

Claiming that lt-na functions as a discourse linking marker in general, we could explain why it is used in both kinds of constructions, namely, to conjoin sentences or mark focus.

4 ht-ná: intensional linking marker

Richard Montague, notably in Montague (1973), defines the denotation of a sentence as a function from possible worlds and moments of time to truth values, calling it an ‘intension’. Accordingly, he argues that the truth-value of a sentence must be determined with respect to a time and world (in addition to other parameters). This is because a certain sentence may be true with respect to a time t but false with respect to some other time t’, or true in a world w, but false in some other possible world w’.

In what follows we will show that ht-na is an intensional anaphora marker, which functions to link times (§4.1) or worlds (§4.2) within its respective discourse.

4.1 Temporal anaphora

The idea of temporal anaphora was first introduced by Partee (1973) and developed further in Partee (1984), where she implements the notion of reference-time introduced by Reichenbach (1947). Her idea, adapted by linguists such as Bonomi (1995), was to deal mainly with the behavior of tenses in discourse.

Partee argues that tenses are like pronouns, such that their (temporal) interpretation may be dependent on an antecedent. Consider one of her examples, depicted in (26a) below, along with one of Bonomi’s (1995) examples, depicted in (26b):

    1. (26)
    1. a.
    1. Sheila had a party last Friday and Sam got drunk. (Partee 1973: ex. 10)
    1.  
    1. b.
    1. When I turned off the stove, Leo turned off the light. (Bonomi 1995: ex. 6)

Determining the temporal interpretation of the sentence in (26a), Partee (1984: 245) writes: “a time is specified in the first clause and the second clause is most naturally understood as referring to the same time” and, therefore, “the past tense can be viewed as an anaphoric element”.

Adapting Partee’s analysis, Bonomi (1995) suggests talking about the anaphoric referential nature of tense in a weaker sense, claiming that its antecedent is a temporal framework suggested by the context (rather than an exact time). Therefore, the when-clause in (26b), for example, “contributes to the temporal location of the event described by the main clause”, but “there is no reference (in the strong sense) to a specific moment” (p. 6). This seems to us to be a more adequate characterization of the phenomenon.

Partee, furthermore, shows that temporal anaphoric relationship may hold also in case of bound tenses. Consider examples (27) and (28) from Partee (1973):

    1. (27)
    1. Whenever Susan comes in, John immediately leaves. (Partee 1973: ex. 17)

Partee suggests (27’) below as the temporal interpretation of (27):

    1. (27’)
    1. Every time Susan comes in, John leaves in the immediate future of that time. (Emphasis added.)

As indicated by the boldfaced expression, the time at which John leaves is to be defined by the subordinated whenever-clause. In (28), the conditional structure alone suggests such temporal anaphora:

    1. (28)
    1. If Susan comes in, John will leave immediately. (Partee 1973: ex. 14)

Partee examines two ways for the temporal interpretation of (28), and sides with the one given under (28’) below:

    1. (28’)
    1. If there is a time when Susan comes in then John will leave in the immediate future from the time at which Susan comes in. (Emphasis added.)

The time of the (possible) event reported in the apodosis (the ‘then’ part) of the conditional is defined with respect to the time of the (possible) event reported in the protasis (the ‘if’ part).

Note that the anaphoric relationship observed by Partee for examples like (27) above is not indicated explicitly, which might suggest that it is based on pragmatic considerations. But such a relationship may be marked explicitly by an adverb such as ‘at the same time’ or a when-clause. We suggest that in Akan, temporal anaphora is marked explicitly by ht-na.

Reconsider the examples given above in (2a) and (2b) from Boadi (2008), repeated below as (29a) and (29b) respectively, with his own glossing of ht-na:

    1. (29)
    1. a.
    1. mè
    2. 1sg
    1. bá-è
    2. come-pst
    1. nó
    2. tp
    1. ná
    2. pst
    1. ɔ̀-rè-dìdí.
    2. 3sg.subj-prog-eat
    1. ‘When I came, he was eating.14
    1.  
    1. b.
    1. (wó
    2. 2sg
    1. bɛ́-bá
    2. fut-come
    1. nó)
    2. tp
    1. ná
    2. fut
    1. Kofi
    2. Kofi
    1. hó
    2. skin
    1. á-tɔ̀
    2. perf-settle
    1. nò.
    2. 3sg
    1. ‘(By the time you come back), Kofi will have enjoyed a state of mental calm.’

As we mentioned in the introduction, Boadi considers ht-na as three homophones, two of which are tenses: past-tense (29a) and future-tense (29b). Boadi’s motivation for considering ht-na to be a tense marker in those sentences seems to be due to the temporal interpretation of its clauses. Both clauses in (29a) are understood to refer to the past. Yet only the verb in the first clause carries a past-tense morpheme. However, Boadi glosses also ht-na as a past-tense marker.

Similarly, both clauses in (29b) are interpreted to refer to the future, but only the first clause has a verb with a future tense morpheme. In this case, Boadi glosses the ht-na appearing in the second clause as a future-tense marker. Boadi’s analysis for ht-na as a tense marker raises several related difficulties.

First, note that Boadi’s analysis suggests that there are two different markers for each tense (a morpheme on the verb and a ht-na), on the one hand, and that the same ht-na particle is ambiguous between two (incompatible) tenses, on the other hand. Ignoring this twofold problem and adopting Boadi’s account, one would still wonder how an Akan speaker would know when to inflect the verb for tense and when to use the ht-na marker. Descriptively speaking, it is always the case that a ht-na clause comes second in a two-clause sentence.15 This descriptive rule raises, in turn, the following question: why is it the case that a ht-na clause cannot be cataphoric?

Our contention is that ht-na is not a tense, so in sentences like (29a) and (29b) the clause in which it appears is tenseless. Instead, in such sentences it functions as a temporal anaphora marker, linking the time of its clause to the time of the previous one. Accordingly, the ht-na clauses get their temporal interpretation via the occurrence of ht-na, which functions to indicate that their reference-time is the same one indicated by the previous clause. Accordingly, we predict that a single clause sentence would always have a tense morpheme on the verb and never the ht-na particle, which is, indeed, the case (cf. discussion on examples 31 and 32 below).

Our analysis may find some support in how studies such as von Stechow (1991), Ogihara (1996) and Kratzer (1998) account for the phenomenon of sequence-of-tenses in languages like English. Consider the English sentence in (30) below:

    1. (30)
    1. John thought that Mary was pregnant.

Temporally speaking, (30) is ambiguous between two readings. On one reading, the time of Mary’s being pregnant depicted in the subordinate clause precedes the time of John’s thinking depicted in the matrix. This interpretation is not problematic, as the past tense is defined to locate a situation in the past of its respective reference-time. On the second reading, however, the time of Mary’s pregnancy overlaps with John’s thinking time. This reading is problematic, as the time indicated by the embedded past-tense clause overlaps the reference-time of the matrix (rather than preceding it). Most semanticists dealing with this problem believe that on its latter (simultaneous) reading, the second clause in (30) is tenseless at the LF (Logical Form) level. Ogihara (1996), for example, suggests that it is deleted at the LF level, and Kratzer (1998) argues that it was not generated in the first place, but added later in the derivation.

Inspired by Kratzer’s theory, we may contend that the second clause of sentences like (29a) and (29b) is generated without tense both in English and Akan. But while in English such a clause is assigned a tense later in its derivation, in Akan it remains tenseless. Therefore, such Akan clauses cannot be given any independent temporal interpretation but only via the anaphoric relationship to their respective previous clause. We argue that this temporal anaphoric relationship is marked by ht-na. ht-na serves as a temporal anaphora marker also in case of bound tenses (cf. the English examples in 27 and 28). Consider sentence (31) below:

    1. (31)
    1. sɛ̀
    2. If
    1. yɛ̀-kyè-è
    2. 3pl-catch-pst
    1. nó
    2. 3sg
    1. áà
    2. cond
    1. ná
    2. ht-na
    1. wó-á-wíá
    2. 2sg-perf-steal
    1. sika
    2. money
    1. ‘If he was caught, then he stole money.’16

Both the protasis (the ‘if’ part) of the conditional and the apodosis (the ‘then’ part) are understood to depict situations in the past, but only the verb of the protasis is inflected for past tense. Our contention is that the apodosis is tenseless and the ht-na appearing in it indicates that its temporal interpretation is to be determined by its “antecedent”, namely the time of the protasis.

    1. (32)
    1. sɛ̀
    2. if
    1. wó-frɛ́
    2. 2sg-call
    1. nò
    2. 3sg
    1. áà
    2. cond
    1. ná
    2. ht-na
    1. ɔ̀-rè-sú
    2. 3sg-prog-cry
    1. ‘Whenever you call him, he will be crying.’

The sentence in (32) reports habitual events whose times overlap. Following Boadi (2008), we consider the protasis verb to be in present tense. The apodosis is tenseless, but due to the presence of ht-na we conclude that its reference-time is the same as that of the protasis.

Our analysis will resolve the difficulties observed above with respect to Boadi’s analysis of sentences like (29a) and (29b). First, as indicated above, we predict that a single clause sentence would always have a tense morpheme on the verb and never the ht-na particle, which is, indeed, the case. Second, we can explain why ht-na cannot appear in the first clause of a two-clause sentence. And third, the peculiar ambiguity indicated by Boadi, namely a morpheme denoting two incompatible semantic values (past and future reference) disappears.17

The property of ht-na to function as a temporal anaphora marker is a subcategory of a more general property, namely an intensional linking marker, which explains its function also as a modal-linking marker, to which we turn now.

4.2 ht-na as a modal-linking marker

As shown in studies of modality in natural language, speakers do not speak only about the actual world, but often make statements about other possible worlds as well (cf. Palmer 2001; Lewis 1973; Kratzer 1981, among others). Compare the sentence in (33) below with the sentences in (34) and (35):

    1. (33)
    1. John lives in Boston.
    1. (34)
    1. a.
    1. I wish John were happy.
    1.  
    1. b.
    1. It is necessarily the case that John is on the beach.
    1.  
    1. c.
    1. Mary may be at the library.
    1.  
    1. d.
    1. If John enters the room, he’ll trip the switch. (Heim 1982: 95, ex. 19)
    1.  
    1. e.
    1. If kangaroos had no tail they would topple over. (Lewis 1973: 1)
    1. (35)
    1. a.
    1. Since her car is here, Mary must have been back.
    1.  
    1. b.
    1. Mary is no longer sick. She must have seen her doctor.
    1.  
    1. c.
    1. Mary must have been out of town.

The sentence in (33) reports some situation the speaker claims to be taking place in the actual world. In contrast, the sentences in (34) are all modal, taking into consideration other possible worlds. Sentence (34a) implies that John is not happy in the actual world and indicates that he is happy in those worlds where the speaker’s wishes come true. Sentence (34b) reports that in all possible worlds, including the actual one, John is on the beach. Sentence (34c) asserts that in some worlds, which may or may not include the actual one, Mary is at the library. The conditionals in (34d) and (34e) imply some relationship between worlds. The indicative conditional in (34d) indicates that the set of worlds where John enters the room restricts the set of worlds where he trips the switch; one of those worlds may or may not be the actual world. The counter-factual conditional in (34e) indicates that the set of worlds in which kangaroos have no tail restricts the set of worlds where kangaroos topple over; none of those worlds is the actual one.

The sentences in (35) illustrate what is referred to in the literature as evidentiality (De Haan 1999; Aikhenvald 2004; among others). Those sentences, too, involve possible worlds, but unlike the sentences in (34), they also report some situation in the actual world that is taken to be evidence for the modal part. For instance, the subordinate clause in (35a) (‘since Mary’s car is here’) presupposes that Mary’s car is here (in the actual world). This presupposition is taken as evidence for the speaker’s conclusion that Mary is back. Similarly, the first sentence in (35b) reports some situation in the actual world (Mary’s being healthy), which is taken by the speaker to be evidence that Mary has seen her doctor. Finally, (35c) is felicitous only if there is some situation understood from the context, which can be taken as evidence for the speaker to conclude that Mary is out of town. In other words, what the speaker seems to be saying is something like “from what is stated/presupposed by the previous clause/sentence/context, it logically follows that (e.g., ‘Mary is back’)”. By that, the speaker seems to be linking some “logical” worlds with the actual world.

Cross-linguistically, it has been shown that languages may mark evidentiality by grammatical means: affixes, clitics, particles, and the like (see, Palmer 2001: 8, 9, etc. for discussion and examples from a variety of languages; Hammer 1983: 231, 232 for evidential means in German, among many others). The sentences in (35) above show that English uses the modal ‘must’ for that purpose. In Akan, we argue, it is the job of ht-na. Reconsider Boadi’s example (2c), reproduced below under (36):

    1. (36)
    1. sɛ̀
    2. if
    1. wó-hù-ù
    2. 2sg-see-PAST
    1. nò
    2. 3sg
    1. áà
    2. cond
    1. (ɛ́ńdèɛ̀)
    2. (in.that.case)
    1. ná
    2. ht-na
    1. ɔ̀-bà-è.
    2. 3sg.subj-visit-PAST
    1. ‘If you saw him, then it logically follows that he visited.’

Taking the addressee’s word that s/he (the addressee) saw him, the speaker concludes that (in that case) he visited (otherwise the addressee could not have seen him). The linking of the actual world with the “logical” worlds is marked by ht-na. We argue that the ht-na is used for this purpose because it is in general an intensional linking marker.

5 Summary and conclusions

In this paper, we have proposed a toneless super-category na in Akan, which we call Root-na.

Drawing from theories within the approach of truth-conditional semantics and the semantics/pragmatics interface, we argued that this abstract toneless morpheme is used in Akan to link the na-constituent with something in the common ground. An Akan speaker would never pronounce such toneless particle, but add a Low or High tone, generating what we called a lt-na or a ht-na, respectively, according to the kind of linking the speaker wishes to indicate. To appreciate the abstract toneless morpheme Root-na in Akan, compare it to the abstract tenseless morpheme woll suggested by Abusch (1997) for English. As analyzed by Abusch, the tenseless (modal) morpheme woll is never pronounced by English speakers and is realized as ‘will’ or ‘would’, depending on the tense of the said clause.

lt-na has been shown in the Akan literature to be used in conjunction or focus constructions. Adopting Asher’s (1993, etc.) SDRT for discourse coherence relations and Rooth’s (1985, etc.) alternatives semantics account for focus, we have shown that what looks like two different uses of the lt-na are actually two sub-uses of one super function, namely relating the proposition of the lt-na clause to a proposition uttered by a previous sentence or presupposed by the context.

ht-na has been shown to be anaphoric within an intensional discourse. Adopting Montague’s (1973) intensional semantics and theories developed within its approach, we have argued that ht-na is used to link its clause to some intensional expression. Accordingly, it may function as a temporal anaphoric expression, similar to the function Partee (1973; 1984) attributes to tenses in English, or as a modal anaphoric expression that links the actual world with some other possible (“logical”) world(s).

Abbreviations

1 = first person, 2 = second person, 3 = third person, compl = completive, cond = conditional, def = definite, fm = focus marker, fut = future, ht = High tone, imp = imperative, ind = independent, int = intensifier, lt = Low tone, mm = modal marker, neg = negation, perf = perfective, pl = plural, poss = possessive, pst = past, prog = progressive, red = reduplication, sg = singular, subj =subject, top = topic, tp = terminal particle.

Acknowledgements

We would like to thank Mrs. Joscelyn Essegbey and Mrs. Josephine Richter, native speaker consultants, and Felix Ameka for discussions on tone and focus. We are also indebted to four Glossa readers for their insightful comments, which we used to improve our manuscript.

Competing interests

The authors have no competing interests to declare.

Notes

  1. (a) Tone is not marked in Akan orthography. However, because it is crucial for the discussion (as we show in section 2), we mark tone on all examples, including those taken from other sources that do not have tone marking. We have also modified the glossing of cited examples (where possible) to conform with the Leipzig glossing rules. (b) Note that not all those who write on Akan agree on what constitutes tense and aspect in the language. For the purposes of this paper, we adopt the tense, aspect and mood categories used in most analyses of Akan and summarized in Boadi’s (2008: 13) Table 1. [^]
  2. Yɛ occurs on final verbs that take the past tense morpheme and occur without a following complement in the Asante dialect. These include intransitive verbs, transitive verbs whose complements have been fronted and, as in (1b), instances where the complement is a null object. For recent analyses of the phenomenon see Ofori (2018) and Kandybowicz (2015). [^]
  3. The word nó functions in different ways: as a tp, top, or def, depending on the sentence. [^]
  4. Akan distinguishes between the third person singular subject pronoun and the object pronoun. We capture this in the glossing by using 3sg.subj for the former and 3sg.obj for the latter. [^]
  5. The conditional which Boadi writes as áà is homophonous with the relativizer which Saah (2010) writes in the same way (see discussion in the next section). [^]
  6. Our analysis is based on data from the literature, and consultation with native speakers. [^]
  7. There is evidence that when a vowel segment with a high tone is deleted before the relativizer, the falling tone is maintained. Consider the example below:
      1. (i)
      1. Awùráá
      2. Lady
      1. n(ó)
      2. def
      1. áà
      2. REL
      1. ɔ̀bá-à
      2. 3sg:subj-come-pst
      1. há
      2. here
      1. nó
      2. tp
      1. dè
      2. be_called
      1. Mary
      2. Mary
      1. ‘The lady who came here is called Mary.’
    Even though the vowel ó of the definite determiner, which carries a high tone, has been elided (as occurs in fast speech), the falling tone of the relativizer remains intact. [^]
  8. We will only discuss here what is relevant for this paper. A concise summary of SDRT may be found in Caudal (2012) and other studies adopting this theory. [^]
  9. Note that some studies argue that focused phrases in Akan are always fronted (Boadi 1974; Saah 1988; among others), while others disagree (Ameka 2010; Schwarz & Fiedler 2007). Bearth (1999b, etc.) tends to agree with the former studies, acknowledging, however, that sometimes the focused element remains in its original position. He explains the difference between in-situ and ex-situ focus in terms of counter-value, where one element is asserted at the expense of the other (p. 261). Since word order does not seem to affect our analysis of na, we will not discuss the different views in this paper. [^]
  10. Jackendoff (1972) calls it Presupposition, but as Rooth (2008: 279) notes, this term may result in confusion with the notion of presupposition in natural language, which is why he avoids using it and suggests the term ‘focus skeleton’ instead. [^]
  11. A contrary position is Erteschik-Shir (1986), who argues explicitly that the wh-phrase in a wh-question does not function as the focus of the question. [^]
  12. Recall that indexes, 6 in our example, are to indicate the relationship between the two constituents that share the same index. [^]
  13. Note: we gloss nà as lt-na (unlike Titov). Note also that Saah’s original examples gloss it as FOCUS. [^]
  14. Kandybowicz (2015) provides an extensive discussion of ht-na use in past tense. Sentence (29b) shows that ht-na occurs in sentences that refer to the future as well. [^]
  15. Note, however, that ht-na can occur in monoclauses like the one below:
      1. (i)
      1. ná
      2. ht-na
      1. m-à-dá
      2. 1sg-perf-sleep
      1. ‘I was asleep.’
    But this is not incompatible with what is described here, as in a sentence like this, ht-na always takes temporal reference from an immediately preceding utterance. [^]
  16. As noted in FN. 5 above, the morpheme áà which is glossed here as ‘conditional’, is homophonous with the relativizer (see discussion of Saah 2010 above). [^]
  17. A Glossa reviewer lends some more support to our analysis, showing that it is in accord with Clements (1982: 8), who points out that “the function of this particle is to ‘key’ the temporal interpretation of the clause it occurs into a time established in the preceding discourse context.” As we do not have access to Clements’ manuscript, we cannot comment on it any further. [^]

References

Abusch, Dorit. 1997. Sequence of Tense and Temporal De Re. Linguistics and Philosophy 20. 1–50. DOI:  http://doi.org/10.1023/A:1005331423820

Aikhenvald, Aleksandra Y. 2004. Evidentiality. Oxford: Oxford University Press. DOI:  http://doi.org/10.1093/oso/9780199263882.001.0001

Ameka, Felix K. 2010. Information packaging constructions in Kwa: micro variation and typology. In Aboh, Enoch O., & Essegbey, James (eds.), Topics in Kwa Syntax, 141–76. London/New York: Springer. DOI:  http://doi.org/10.1007/978-90-481-3189-1_7

Amfo, Nana Aba A. 2007. Clausal conjunction in Akan. Lingua 117. 666–84. DOI:  http://doi.org/10.1016/j.lingua.2006.04.005

Asher, Nicholas. 1993. Reference to Abstract Objects in Discourse. Dordrecht and Boston: Kluwer Academic Publishers. DOI:  http://doi.org/10.1007/978-94-011-1715-9

Asher, Nicholas. 1998. Discourse and the focus/background distinction. In Bosch, Peter & van der Sandt, Rob (eds.), Focus: Linguistic, cognitive, and computational perspective, 247–267. Cambridge: Cambridge University Press.

Asher, Nicholas & Lascarides, Alex. 2003. Logics of conversation. Cambridge University Press.

Bearth, Thomas. 1999a. The contribution of African linguistics towards a general theory of focus: update and critical review. JALL 20. 121–156. DOI:  http://doi.org/10.1515/jall.1999.20.2.121

Bearth, Thomas. 1999b. The Inferential Gap Condition. Pragmatics 9(2). 249–80. DOI:  http://doi.org/10.1075/prag.9.2.03bea

Bearth, Thomas. 2002. Fokus im Akan – Outline. Handout of a presentation at the Linguistic Colloquium of the Seminar of African Studies at the Humboldt University.

Beck, Sigrid. 2006. Intervention Effects Follow from Focus Interpretation. Natural Language Semantics 14. 1–56. DOI:  http://doi.org/10.1007/s11050-005-4532-y

Boadi, Lawrence A. 1974. Focus-Marking in Akan. Linguistics 140. 5–57. DOI:  http://doi.org/10.1515/ling.1974.12.140.5

Boadi, Lawrence A. 2005. Three major syntactic structures in Akan: interrogatives, complementation and relativisation. Accra: Black Mask Limited.

Boadi, Lawrence A. 2008. Tense, Aspect and Mood in Akan. In Ameka, Felix & Dakubu, Mary Esther Kropp (eds.), Aspect and modality in Kwa languages, 9–68. Amsterdam: John Benjamins. DOI:  http://doi.org/10.1075/slcs.100.04boa

Bonomi, Andrea. 1995. Riportiamo qui di seguito il saggio di Andrea Bonomi pubblicato su Lingua e Stile, a.XXX, n.3, settembre 1995 (483–500). Ogni riferimento va fatto citando la suddetta rivista. http://www.sifa.unige.it/SIFA94/BONOMI.pdf

Caudal, Patrick. 2012. Pragmatics. In Binnick, Robert I. (ed.), The Oxford Handbook of Tense and Aspect, 269–305. Oxford & New York: Oxford University Press.

Clements, George N. 1982. Temporal Switch-reference in Akan. Unpublished Harvard University manuscript (draft version).

Culicover, Peter & Rochemont, Michael. 1983. Stress and Focus in English. Language 59. 123–165. DOI:  http://doi.org/10.2307/414063

De Haan, Ferdinand. 1999. Evidentiality and epistemic modality: Setting boundaries. Southwest Journal of Linguistics 18. 83–101. (Online: www.u.arizona.edu/~fdehaan/papers/SWJL99.pdf)

Dolphyne, Florence A. 1988. The Akan (Twi-Fante) language: its sound systems and tonal structure. Accra: Ghana Universities Press.

Dolphyne, Florence A. & Dakubu, Mary Esther Kropp. 1988. The Volta-Comoe languages. In Dakubu, Mary Esther Kropp (ed.), The languages of Ghana, 50–90. London: Kegan Paul International.

Eckardt, Regine. 2007. Inherent Focus on Wh-phrases. In Puig-Waldmüller, Estela (ed.), Proceedings of Sinn und Bedeutung 11. 209–228. Barcelona: Universitat Pompeu Fabra.

Erteschik-Shir, Nomi. 1986. Wh-questions and focus. Linguistics and Philosophy 9(2). 117–49. DOI:  http://doi.org/10.1007/BF00635608

Fiedler, Ines & Schwarz, Anne. 2004. Focus or Narrative Constructions? Morphosyntactically Marked Focus Constructions in some Gur and Kwa Languages. Paper presented at a Conference on Topic and Focus: Information Structure and Grammar in African Languages. University of Amsterdam, December 3–4.

Fiedler, Ines & Schwarz, Anne. 2005. Out of focus encoding in Gur and Kwa. Interdisciplinary Study on Information Structure 3. 111–42.

Halliday, Michael A. K. 1967. Notes on transitivity and theme in English. Journal of Linguistics 3. 199–244. DOI:  http://doi.org/10.1017/S0022226700016613

Hamblin, Charles L. 1973. Questions in Montague English. Foundations of Language 10. 41–53.

Hammer, Alfred E. 1983. German grammar and Usage. (Reprinted edition with corrections and supplement) London: Edward Arnold.

Harbour, Daniel. 2008. Klivaj Predika, or Predicate Clefts in Haitian. Lingua 118. 853–871. DOI:  http://doi.org/10.1016/j.lingua.2007.11.010

Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. PhD dissertation, Amherst: University of Massachusetts.

Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press.

Kandybowicz, Jason. 2015. On prosodic vacuity and verbal resumption in Asante Twi. Linguistic Inquiry 46(2). 243–272. DOI:  http://doi.org/10.1162/LING_a_00181

Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1. 3–44. DOI:  http://doi.org/10.1007/BF00351935

Kiss, Katalin É. 1998. Identificational Focus and Information Focus. Language 74(2). 245–273. DOI:  http://doi.org/10.2307/417867

Koranteng, E. Owusu. 1991 [1974]. Guasohantan. Accra: Bureau of Ghanaian Languages.

Kratzer, Angelika. 1981. The Notional Category of Modality. In Eikmeyer, Hans J. & Reiser, Hannes (ed.), Words, Worlds and Contexts. New approaches in Word Semantics. Berlin and New-York: de Gruyter. Reprinted 2002 in Portner, Paul & Partee, Barbara (eds.), Formal semantics: The essential readings. Oxford: Blackwell, 289–323; and in Gutiérrez-Rexach (ed.), Semantics critical concepts, Vol. iv. London: Routledge. 365–403.

Kratzer, Angelika. 1998. More structural analogies between pronouns and tenses. In Strolovitch, Devon & Lawson, Aaron (eds.), Proceedings of SALT VIII, 92–110. DOI:  http://doi.org/10.3765/salt.v8i0.2808

Kratzer, Angelika & Shimoyama, Junko. 2002. Indeterminate Pronouns. The view from Japanese. In Otsu, Yukiko (ed.), Proceedings of the Third Tokyo Conference on Psycholinguistics, 1–25. Tokyo: Hituzi Syobo. Longer version electronically available at Semantics Archive, URochester.

Lascarides, Alex & Asher, Nicholas. 1993. Temporal Interpretation, Discourse Relations and Commonsense Entailments. Linguistics and Philosophy 16(5). 437–493. DOI:  http://doi.org/10.1007/BF00986208

Lewis, David. 1973. Counterfactuals. Cambridge, Mass.: Harvard University Press.

Montague, Richard. 1973. The proper treatment of quantification in ordinary English (PTQ). In Hintikka, Kaarlo Jaakko Juhani & Moravcsik, Julius & Suppes, Patrick (eds.), Approaches to Natural Language (Synthese Library 49), 221–242. Dordrecht: Reidel, Reprinted in Portner and Partee (eds.), 2002. Formal Semantics: The Essential Readings. 17–35. Oxford: Blackwell. DOI:  http://doi.org/10.1007/978-94-010-2506-5_10

Ofori, Seth Atwi. 2011. On the Basic Focus Marker and the Basic Focus Sentence in Akan (Twi). Nordic Journal of African Studies 20(3). 241–262.

Ofori, Seth Atwi. 2018. Vowel Doubling before the Verbal Unit -YƐ [jɛ] in Asante Twai. Ghana Journal of Linguistics 7(2). 71–101. DOI:  http://doi.org/10.4314/gjl.v7i2.4

Ogihara, Toshiyuki. 1996. Tense, Attitudes, and Scope. Kluwer Academic Publishers. DOI:  http://doi.org/10.1007/978-94-015-8609-2

Osam, Kweku. 1994. Aspects of Akan grammar: a functional perspective. PhD thesis. University of Oregon, Eugene.

Palmer, Frank Robert. 2001. Mood and Modality. 2nd ed. Cambridge, UK, New York, and Melbourne Australia: Cambridge University press.

Partee, Barbara. 1973. Some analogies between tenses and pronouns in English. Journal of Philosophy 70. 601–9. DOI:  http://doi.org/10.2307/2025024

Partee, Barbara. 1984. Nominal and temporal anaphora. Linguistics and Philosophy 7. 243–286. DOI:  http://doi.org/10.1007/BF00627707

Pesetsky, David. 1987. Wh- in situ: Movement and unselective binding. In Reuland, Eric & ter Meulen, Alice (eds.), The representation of (in)definiteness. 98–129. Cambridge, Mass: MIT Press.

Ramchand, Gillian. 1997. Questions, polarity and alternative semantics. Proceedings of NELS 27. 383–396. GLSA, University of Massachusetts at Amherst.

Reichenbach, Hans. 1947. Elements of symbolic logic. New York: The Free Press & London: Collier-Macmillan.

Rochemont, Michael S. 1978. A Theory of stylistic rules in English. Unpublished PhD dissertation, University of Massachusetts, Amherst.

Rooth, Mats. 1985. Association with Focus, Diss., University of Massachusetts, Amherst.

Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics 1. 75–116. DOI:  http://doi.org/10.1007/BF02342617

Rooth, Mats. 2008 [2007]. Notions of focus anaphoricity. Acta Linguistica Hungarica 55(3–4). 277–85. Appeared also in Fery, Caroline & Fanselow, Gisbert & Krifka, Manfred (eds.), The notions of information structure. Interdisciplinary studies on information structure 6. 57–67. DOI:  http://doi.org/10.1556/ALing.55.2008.3-4.3

Rullmann, Hotze & Beck, Sigrid. 1998. Presupposition projection and the interpretation of which-questions. In Strolovich, Devon & Lawson, Aaron (eds.), Proceedings of SALT 8. 215–232. Cornell: CLC Publications. DOI:  http://doi.org/10.3765/salt.v8i0.2811

Saah, Kofi. 1988. Wh-questions in Akan. Journal of West African Languages l8(I). 17–28.

Saah, Kofi. 1994. Studies in Akan syntax, acquisition and sentence processing. PhD Thesis, University of Ottawa.

Saah, Kofi. 2010. Relative clauses in Akan. In Aboh, Enoch O. & Essegbey, James (eds.), Topics in Kwa Syntax, 91–107. London/New York: Springer. DOI:  http://doi.org/10.1007/978-90-481-3189-1_5

Sandler, Wendy & Lillo-Martin, Diane. 2006. Sign Language and Linguistic Universals. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9781139163910

Schwarz, Anne & Fiedler, Ines. 2007. Narrative strategies in Gur and Kwa. In Aboh, Enoch O. & Hartmann, Katharina & Zimermann, Malte (eds.), Focus strategies in African languages: The interaction of focus and grammar in Niger-Congo and Afro-Asiatic, 139–160. Berlin: Mouton de Gruyter. DOI:  http://doi.org/10.1515/9783110199093.3.139

Schachter, Paul. 1973. Focus and relativization. Language 49(1). 19–46. DOI:  http://doi.org/10.2307/412101

Titov, Elena. 2019. Morphosyntactic encoding of information structure in Akan. Glossa 4(1). 1–36. DOI:  http://doi.org/10.5334/gjgl.576

Von Stechow, Armin. 1991. Current Issues in the Theory of Focus. In von Stechow, Armin & Wunderlich, Dieter (eds.), Semantics; An International Handbook of Contemporary Research, 804–825. Berlin: Mouton de Gruyter. DOI:  http://doi.org/10.1515/9783110126969.10.804