Discontinuity

id	Language	Extraction and/or extraposition	Raising	Circumfixes	Infixes
4	Bantawa	Extraposition and extraction are not allowed in Bantawa (M. Doornenbal, personal communication, September 18, 2013).	There is no raising construction in Bantawa. In example (7), it is clear that the first person singular is the subject of the main clause, whereas Syam is the subject of the embedded clause, as Syam has an ergative case marker. However, in (8), Syam is the Undergoer, functioning as the object of the main clause or the embedded clause, but seeing that word order is the same, it is unlikely that Syam belongs to the main clause rather than to the embedded clause. Of course, this is not conclusive evidence, as word order in Bantawa is strongly based on pragmatics. However, word order does not seem to have any pragmatic effects here. Therefore, I will assume that Bantawa has no argument raising. Doornenbal (2009: 202)(1) ɨŋka Syam-ʔa kələm pɨ-Ø-ʔo kha-Ø-ŋ. 1SG S.-ERG pen give-PST-NMLZ see-PST-1SG ‘I saw that Syam gave a pen.’(2) ɨŋka Syam kələm ɨ-pu-ʔa-ʔo kha-Ø-ŋ. 1SG S. pen 3.A-give-PST-NMLZ see-PST-1SG ‘I saw someone give Syam a pen.’	Bantawa has circumfixes, e.g. the predicate markers tɨ-…-ci ‘2DU.NPST’, tɨ-…-in ‘2PL.NPST’, and tɨ-…-aci ‘2DU.PST’ (Doornenbal 2009: 145).	Bantawa has what Doornenbal (2009: 70) after Bickel and Nichols (2006) calls interpositions: “formatives placed between the two parts of a bipartite stem” (Doornenbal 2009: 70). An example is given in (9).Doornenbal (2009: 71)(9) ɨŋ-mɨk-mɨwa my-eye-hair ‘my eyebrow’Interpositions should be seen as a subtype of infixes, that are special as they are inserted into morphological units rather than phonological ones. The difference between interpositions and infixes is thus analogous to that between prepositions and prefixes, and postpositions and suffixes. To me, the difference between an interposition and an interclitic is unclear, but I will not address this matter further, since it is not relevant for the analysis of transparency. There is also a regular infix in Bantawa. <-sa> ~ <-so> is a pronominal marker that is inserted between certain third person pronouns and ergative or genitive case endings, e.g. o-ci ‘this-PL’, o-sa-ʔa ‘this-PRN-ERG’ (Doornenbal 2009: 101ff.).
5	Bininj Gun-Wok	There is extraposition in Bininj Gun-Wok, e.g. in example (8). Semantically speaking, the head bininj and the modifying relative clause ‘belong together’. They are, however, not morphosyntactically adjacent. Note furthermore that the existence of a relativiser is an additional argument for actual subordination in Bininj Gun-Wok. Evans (2003: 643) (8) na-mege bininj ga-m-re, na-wu gogok bi-yame-ng I-DEM man 3-hither-go.NPST I-REL brother 3>3.PST-spear-PST.PFV ‘The man is coming, whom your brother speared.’	I found no examples of argument raising in Bininj Gun-Wok. Since complementation is highly restricted anyway (cf. Section 3.1.4), I assume that Bininj Gun-Wok does not allow argument raising.	There are no circumfixes in Bininj Gun-Wok (cf. Evans 2003: 1).	There are no infixes in Bininj Gun-Wok (cf. Evans 2003: 1).
6	Chukchi	Extraposition or extraction is not allowed in Chukchi. M. J. Dunn (personal communication, January 25, 2013) points at example (4), which could be argued to contain an extraposed relative clause ‘which were torn’. Dunn (1999: 268) (4) … pucʔe-t təni-tku-jw-ə-ninet cimirʔet-ə-lʔ-ə-t sleeve-ABS.3PL mend-ITER-COLL-ə-3SG>3PL tear-ə-PTCP-ə-ABS.3PL ‘She mended the sleeves which were torn’ However, Dunn (idem) states that this relative clause might as well be seen as an independent nominal constituent functioning as a repeated object. Then, a better translation would be ‘She mended the sleeves, the torn ones.’ In this scenario, there is no extraposition, since the constituent ‘the torn ones’ is independent and does not form a whole with ‘sleeves’. Since there is no pressing argument to assume extraposition in Chukchi, I will take it that it is not allowed.	There is no argument raising in Chukchi. In fact, Chukchi does not have syntactically dependent clauses (cf. Section 4.1.4) so that raising is impossible – the feature does not apply. M. J. Dunn (personal communication, January 25, 2013) confirms this.	Chukchi has circumfixes, e.g. the negating circumfixes luŋ-…-(t)e and e-…-ke (Dunn 1999: 325).	There are no infixes in Chukchi.
7	Dutch	Extraposition is allowed in Dutch, as illustrated by example (4). (4) heb je die man gezien [die door rood reed]? have you that man seen that through red drove ‘Did you see that man that drove through the red traffic light?’	Dutch also exhibits argument raising, as illustrated by example (5). (5) a. het lijk-t dat de paard-en wit zijn 3SG.N seem-3SG COMP the.PL horse-PL white be.PRS.PL ‘It seems that the horses are white.’ b. de paard-en lijk-en wit (te zijn) the.PL horse-PL seem-PL white INF be.INF ‘The horses seem (to be) white.’	Circumfixation is present in Dutch, e.g. ge-maak-t ‘PTCP.PST-make-PTCP.PST’. Another circumfix, ge-N-te, turns nouns into collective nouns, e.g. een steen ‘a stone’ vs. gesteente ‘stone’.	There are no infixes in Dutch.
8	Egyptian Arabic	Egyptian Arabic allows for extraposition of a part of a constituent away from its head. An example is given in (5), in which the long relative clause (between square brackets) is placed after the verb, instead of adjacent to its head, ‘man’. Gary & Gamal-Eldin (1981: 55) – bracketing mine (5) ʔir-rˤaagil da ʔabuu-ja [l[ā]bis ʔig-galabijja ʔid-dabalaan DEF-man that father-POSS.1SG wearing DEF-galabeya DEF-linen ʔil-miχatˤ[tˤ]a[tˤ]a ʕala ʔazraʔ bit[ā]ʕit ʔaχuu-ja s[ā]mi] DEF-striped on blue belonging_to brother-POSS.1SG S. ‘That man who is wearing the blue striped linen galabeya belonging to my brother Sāmi is my father.’	There is no true raising in Egyptian Arabic, as illustrated in example (6). The subject of the embedded clause, ‘I’, is the subject of the main clause in (6b). However, it is not raised out of the embedded clause, since it is still present at that location; rather, the subject of the embedded clause is copied to the main clause. The embedded clause is not discontinuous, and therefore, I do not count this as argument raising. Wise (1975: 70) (6) a. miʃ mumkin a-ʕmil kida NEG possible 1SG-do like_that ‘It is not possible that I do so.’ b. ana miʃ mumkin a-ʕmil kida 1SG NEG possible 1SG-do like_that ‘It is not possible that I do so.’ (Lit.: “Me it is not possible that I do that.”)	There are circumfixes in Egyptian Arabic, e.g. several person and number markers and the negation marker ma-V-ʃ ‘NEG’ that attaches to perfective and imperative verb forms.	Lexical stems in Arabic colloquial varieties consist of three consonants (called roots or radicals), that are integrated in a CV-pattern (e.g. CaCaC ‘PFV’) that determines the parts-of-speech properties and the exact meaning of the word. For example, KTB ‘write, book’ combined with the perfective pattern CaCaC gives katab-Ø ‘write.PST.PFV-3SG, he wrote’ (perhaps a more adequate representation is k< a > t < a >b ‘write’), but with the imperfective pattern CCiC we get ji-ktib ‘3SG-write.PRS.IPFV, he is writing’ (or: ji-kt< i>b, ‘3SG-write’), and the pattern CiCaaC results in kitaab ‘write.NMLZ’ or k< i>tb ‘write’ (Gary & Gamal-Eldin 1981: 115). The described CV-patterns have been analysed in the morphological literature as a type of infixes called transfixes (e.g. Haspelmath 2002: 23), i.e. inflectional affixes that are integrated into the root consonants to form stems. These patterns create discontinuity in their hosts, since they prevent the consonantal roots from being adjacent, which is the reason to count them here under infixation.
9	Fongbe	E. Aboh (personal communication, February 25, 2012) indicates that extraposition is not easily allowed in Fongbe. However, he does give example (5) from Gungbe, which is in this respect again similar to Fongbe. Gungbe – E. Aboh (personal communication, February 25, 2012) (5) ùn ɖɔ̀ xó ná súrù [ɖɔ̀ ví étɔ̀n ná wá égbè] I speak word to S. that child his will come today ‘I told Suru that his child will return today.’ The object xó ‘word’ and the modifying relative clause arguably ‘belong together’ semantically, but the relative clause is extraposed, possibly because of its weight. I will assume that extraposition is possible in Fongbe.	Fongbe exhibits argument raising (Lefebvre & Brousseau 2002: 277). In example (6), all argument are in the clause where they semantically belong, while in (7), an Undergoer argument is raised out of the embedded clause to function as the subject of the main clause. Lefebvre & Brousseau (2002: 278) (6) é hwɛ̀ [j]ɛ̀ ɖɔ̀ núsúnû ɔ́ mɛ̀ it lack salt at soup DET in ‘It lacks salt in the soup.’ (7) [j]ɛ̀ hwɛ̀ ɖɔ̀ núsúnû ɔ́ mɛ̀ salt lack at soup DET in ‘Salt is lacking in the soup.’	There are no circumfixes in Fongbe (Lefebvre & Brousseau 2002: 187).	There are no infixes in Fongbe (Lefebvre & Brousseau 2002: 187).
10	Georgian	Extraction or extraposition of a part of a constituent out of that constituent is allowed in Georgian (B. G. Hewitt, personal communication, November 27, 2012).	Harris claims that there is argument raising in Georgian, and gives examples like (7). Supposedly, the object of the embedded clause in (7a), ‘good examples’, is raised to be the subject of the main clause in (7b). Harris (1981: 54) (7) a. čem-tvis ʒnel-i=a, k’arg-i magalit-eb-is mo-ʒebn-a me-for hard-NOM=COP.3SG good-ATTR example-PL-GEN PFV-find-NOM ‘It is hard for me to find good examples.’ b. k’arg-i magalit-eb-i ʒnel-i=a good-ATTR example-PL-NOM hard-NOM=COP.3SG mo-sa-ʒebn-ad čem-tvis PFV-FUT.PTCP-find-ADV me-for ‘Good examples are hard for me to find.’ However, I disagree with this analysis: ‘good examples’ cannot be the subject of the main clause in (7b), as it would trigger plural agreement on the predicate ‘hard’, which it does not. Rather, I think that ‘good examples’ is in a non-default position, perhaps for pragmatic reasons, which is perfectly possible in Georgian since word order is relatively flexible. I assume, then, that raising is not allowed in Georgian.	Georgian has circumfixes, e.g. the pronominal verb marker g-…-t ‘2PL’ (Hewitt 1995: 128).	There are no infixes in Georgian.
11	Huallaga Quechua	As Weber (1989: 250) explains, phrases may be discontinuous in Quechua, but if they are, each separate part gets a case marker of its own. Therefore, the separate units are in fact independent phrases themselves, and we should speak of nominal apposition rather than discontinuity. This is illustrated by example (5). In (5a), there is an NP with a relative clause containing a nominalisation. In (5b), this relative clause is separate from its head, but since it is marked for accusative case, it can be seen as an independent argument. The translations reflect this difference. Weber (1989: 250) (5) a. maqa-sha-n runa-ta rika-: hit-NMLZ-3.POSS man-ACC see-1 ‘I see the man who hit him.’ b. runa-ta rika-: maqa-sha-n-ta man-ACC see-1 hit-NMLZ-3.POSS-ACC ‘I see the man, the hitting one.’	There is raising in Huallaga Quechua, as demonstrated by (6). Example (6a) contains a main clause with a subordinated infinitive, marked for a first person Object. In (6b), the main clause is also marked for a first person Object, while this does not belong at that position semantically. Strictly speaking, the object is not raised out of the subordinate clause, since it is still present in the subordinate clause as well. Rather, the object argument is copied to a semantically inappropriate position. Weber (1989: 237) (6) a. maqa-ma-y-ta muna-n hit-1.OBJ-INF-ACC want-3 ‘He wants to hit me.’ b. maqa-(ma)-y-ta muna-ma-n hit-1.OBJ-INF-ACC want-1.OBJ-3 ‘He wants to hit me.’ (Lit.: “He wants me to hit me.”)	Weber (1989: 9) states that all affixation is suffixing in Huallaga Quechua. This indicates that there are no circumfixes.	Weber (1989: 9) states that all affixation is suffixing in Huallaga Quechua. This indicates that there are no infixes.
12	Japanese	S. Iwasaki (personal communication, October 17, 2013) states that extraposition and extraction are not allowed in Japanese, and provides example (3). S. Iwasaki (personal communication, October 17, 2013) (3) a. akai seetaa o kita otokonoko ni kinoo atta red sweater ACC wear boy DAT yesterday met ‘I met that boy, who was wearing a red sweater, yesterday. b. *akai seetaa o kita kinoo otokonoko ni atta red sweater ACC wear yesterday boy DAT met ‘I met that boy yesterday, who was wearing a red sweater.’	I have not seen examples of argument raising in Japanese. S. Iwasaki (personal communication, October 17, 2013) confirms that raising does not occur.	There are no circumfixes in Japanese; Iwasaki (2002: 45) states that affixes are either suffixes or prefixes.	There are no infixes in Japanese; Iwasaki (2002: 45) states that affixes are either suffixes or prefixes.
13	Kayardild	Evans (1995: 234) states that a Kayardild NP must be realised as one contiguous element, thus disallowing discontinuous NPs, “except under special discourse conditions”. Some modifying elements may be found separated from their heads, as in (7), but since all units are marked for case, they can be seen as independent NPs. Under that analysis, (7) displays apposition rather than extraposition. Evans (1995: 332) (7) niya dangka-na kaba-tharra jalji-nurru-na NOM.3SG person-MODC.ABL find-PST shade-ASSOC-MODC.ABL yiiwi-n-kina sleep-NMLZ-MODC.ABL ‘He found the person in the shade, the sleeping one.’ I have not found examples of the special discourse conditions that Evans mentions, but I assume that if those conditions cause any dislocation of NP modifiers, they are still cases of apposition rather than extraction or extraposition. Hence, Kayardild is transparent with respect to this feature.	There is argument raising in Kayardild, as shown in example (8). In this sentence, the third person singular pronoun is syntactically an argument of the main clause, which is proven by the fact that it has modal case, as all arguments in the VP (cf. Section 11.1.2), while semantically speaking, it is the Actor argument of the embedded clause. This is a typical case of subject-to-object raising. Evans (1995: 502) (8) ngada kurri-ja niwan-ji, NOM.1SG see-REAL 3SG-MODC.LOC natha-wurrka dana-thurrk camp-MODC.LOC.COMPC.OBL leave-IMMED.COMPC.OBL ‘I saw him leaving the camp.’	The only type of affixes in Kayardild is suffixes (Evans 1995: 1). Hence, there are no circumfixes.	The only type of affixes in Kayardild is suffixes (Evans 1995: 1). Hence, there are no infixes.