Right-dislocated constituents Combinations of left-dislocation and preposing

ʋmʋ kʋ nä wälɩklё -bhlogbe -bhlü bhä NH 3: TH on SP 1 matter:part one_single fall: CP PST 1 that day lit. on it something happened.’ [greve 15–17] If they are not coreferential, the first dislocated constituent has a spatial, temporal, or other circumstantial reference with no trace in the following clause, while the second dislocated constituent is coreferential with a core constituent of the subsequent clause and a pronominal trace in that clause. This situation is illustrated in example 65. Note in this example that the first dislocated constituent has a spatial reference ‘on the lagoon’ while the second ‘the canoes that used to transport us’ is coreferential with the non-human pronoun ɩ ‘they’ in the main clause. 65 - mɔɔ nyiee klʋʋ gokpüü LOC lagoon: DEF surface: OBL boat: PL : DEF ‘There on the lagoon, the boats - ɩ tʋ-anyɩ bhlä nä […] kɩ bʋä NHP : REL cross: ICP :1 PO PST 1: LOP SP 1 ASF : NHP be_slow: ICP that used to take us across, […] they are really slow.’ [greve 18, 21] In terms of information processing, the dislocation helps to package the information in such a way that the core structure of the clause has less linguistic material and is thus easier to process. Note that in the above example, the spacer nä marks the end of the relative clause and the main clause consists of only two words.

3.3.2 Right-dislocated constituents

Right-dislocated constituents are rare in Godié. The only occurrences that have been found in the data are coreferential with the clause subject. Thus they are not what Lambrecht 1994:181 calls a right- dislocated anti-topic, which is the only subject constituent of the clause. They either give more precise information about the subject referent, as in 66a, or express an afterthought or a self-correction, such as adding more referents to the subject referent, as in 66b. 66 a. ‘nɩ wa yi yi ŋwnɩä -yä yuë ADD 1 they XFUT come woman: PL ADD 2 child: PL ‘… and they came, women and children.’ [elisabet 72] b. ɔ -kʋ ylä - lɔɔ ɔ -yä ɔ ŋwnɩä sɔ 3 S BE 1 now: LOP there 3 S ADD 2 3 S wife: PL two ‘… he was there, he and his wives’ [sigo 185]

3.3.3 Combinations of left-dislocation and preposing

Combinations of left-dislocation and preposing are very frequent in Godié, with the left-dislocated constituents preceding the preposed one, as in examples 67a and 67b. In example 67a, the subject ‘that same beer’ is left-dislocated and the adjective ‘much’ is preposed. In example 67b, the subject ‘our fathers and grandfathers’ is left-dislocated and kaz ɩɩ ‘raffia washcloths’ is preposed. 67 a. ɛmɛ biee nä kpänyɩ ɛɛ kpä tɛ yɩ - pɛɛ NH 2: TH beer SP 1 much NH 2: GEN price be_strong now hence ‘… that same beer, its price is very high at present.’ [radio 85] b. tɛɛtɛɛ -a bäsɩ -yä -a tɩtɩä formerly 1 P father: PL ADD 2 1 P grandfather: PL ‘In former times, our fathers and grandfathers, kazɩɩ wa budo wä raffia_washcloth: PL : OBL 3 P wash: ICP PST 3 it was with raffia washcloths that they used to wash themselves.’ [kazo 18–19] Example 68 has two left-dislocated constituents ‘second ones’ and ‘their group’ and one preposed one the locative pronoun -m ɔɔ. The referent of the first dislocated constituent has its pronominal trace in the second left-dislocated constituent, and the preposed locative pronoun -m ɔɔ is the pronominal trace of the second left-dislocated constituent in the clause. 68 sɔna - lʋä nä waa kpaa mnö - mɔɔ n - kʋmʋ bhlä second DPF : PL SP 1 3 P : GEN group: DEF inside LOC 1 S BE 1: CP : COM PST 1: LOP ‘The second group lit. the second ones, in their group, it was there that I was also …’ [greve 82– 84] 34 4 Information management The need for speakers to manage information arises out of several constraints that the construction and interpretation of a discourse impose on both the speaker and the hearer. 1 All speakers in any language have the same tasks in the management of information. It is only in the strategies they use to carry out these tasks that they differ according to the language they speak. One fundamental constraint on the speaker is what Brown and Yule 1983 call the linearisation problem. Simply put, it consists “first of all in the constraint on the speaker that he can produce only one word at a time” Brown and Yule 1983:125. A speaker who wants to produce a text therefore “has to choose a beginning point” and “this point will influence the hearerreader’s interpretation of everything that follows in the discourse” ibid.. After choosing the beginning point, the speaker who is about to produce a narrative text must decide on how he will bring into sequence the events he intends to talk about. In doing so he needs to take into account the inferences the hearer will draw from the sequence in which the events are presented. One of these inferences consists in assuming, unless the speaker gives a cue to do otherwise, that the first-mentioned event also happened first. Similarly, the hearer will make the assumption, again unless told otherwise, that the second-mentioned event is related to the first. Information is not primarily processed by analysing word meanings but by constructing mental representations, or models, of entities and referents see Brown and Yule 1983:125ff and Lambrecht 1994. It is these mental representations that are stored in the mind. As a discourse progresses, existing mental representations are activated, and possibly changed, in the hearer’s mind and new mental representations are added. In order to enable the hearer to draw the desired inferences and to construct a coherent mental representation of the information he is given to process, the speaker needs to comply with the strategies for information management that are provided by the speaker and hearer’s common language. The purpose of any discourse is to give information. Most clauses in a discourse are a combination of old and new, or established and non-established, information. This guarantees a balanced information rate, where the “established information provides the grounding and the framework within which we process and store the newly asserted information” Runge 2010:188. Established information is information that the hearer knows and of which he has a mental representation. It includes information that was previously established in the same discourse, information that the hearer inferred from what was previously said, and information that is part of the hearer’s general knowledge about the world. Non-established information, on the other hand, is information that is newly established in the clause. What is non-established information in one clause in a discourse will be established information in the following clause and so forth through the whole discourse. It is therefore crucial for the study of information management to determine for each clause the knowledge state of the information conveyed by the constituents of that clause. One typical cross-linguistic strategy to manage established and non-established information in a text is the “principle of natural information flow” Comrie 1989:127–128. It stipulates that speakers generally move from what is most known to their hearers to what is least known. In other words, the normal, or unmarked, way of managing information is for speakers to present established information before non-established information. The knowledge state of information whether it is established or non-established needs to be distinguished from its activation state, which concerns its consciousness in the hearer’s mind or the availability of a given piece of information at a given time. Indeed, while huge amounts of knowledge are stored in a person’s mind, only a very small portion of this knowledge can be focused on at any one time and is therefore available, or active, in the person’s consciousness. On the basis of this fact, Chafe 1987:22ff distinguishes between three different activation states, or degrees of accessibility of information in the hearer’s consciousness at a given time and in a given 1 These observations also apply to the writer and the reader but for the sake of simplicity we will refer only to the speaker and the hearer. context: active, semi-active, and inactive. Active information is information that is “currently lit up” in the hearer’s consciousness. Inactive information is information that is in the person’s long-term memory and not active. At the time the utterance is processed, inactive information either has not been activated at all yet in the course of the discourse or, if it was activated earlier on in the discourse, it has been de- activated and fallen out of a person’s consciousness. Finally, semi-active information is information that automatically enters the hearer’s consciousness together with other information, because it is associated to that information in the hearer’s mind. This last category is therefore also referred to as coactivated or accessible information. Finally, from the speaker’s point of view, not all information given to the hearer is of equal importance or weight in the narrative. Indeed, in order to enable the hearer to construct for himself the mental representation of the narrative that the speaker intends, the speaker gives him cues about which information has more weight than other information. As Carozzi 2010:246 rightly states, “Writers and speakers have more than one strategy at their disposal to enable the audience to appreciate the relative weight of information in the discourse they are producing, and in fact they may manipulate a variety of linguistic items to this end.” Like probably all languages, Godié has linguistic means to show that some entities events, participants, etc. have more relative importance than others at the current point in the narrative see §s 4.1.2 and 4.3.1 below. However, as will be shown in §4.3.2, a Godié speaker also has means to let the hearer know which information is of lesser importance at the current point in the narrative. Information management in language has mostly been studied at the clause level and is often referred to as information structure. Chafe 1976 referred to it as “information packaging” and Prince