1 Introduction

The recent growth in formal work on allocutivity has begun to reveal important ways that the syntax of allocutive marking varies across languages. To date, however, little research has considered the formal nature of these cross-linguistic differences or what relationships, if any, exist among them. (See, however, Alok (2021) for some discussion, and Antonov (2015) for an extensive overview from a typological perspective.) In this paper, we summarize results from recent formal work on nine allocutive varieties in an effort to characterize some aspects of this variability. In particular, we develop a model of four kinds of variation across allocutive languages.

The first of these is whether the allocutive morpheme is restricted to root contexts. While some initial models of allocutivity have characterized it as strictly a root clause phenomenon (Miyagawa 2013; 2017; Portner et al. 2019), more recent work has described allocutive marking in embedded domains in some varieties including Magahi (Alok & Baker 2018; Alok 2021), Japanese (Yamada 2019) and Basque (Haddican & Etxeberria 2022). A second kind of variation across allocutive varieties less discussed in the recent literature concerns the kinds of left peripheral elements allocutive morphemes interact with. In one class of allocutive languages, allocutivity interacts with clause typing features (Portner et al. 2019), or C features (Yamada 2019; Miyagawa 2012; 2022), while other varieties show no such interaction. A third kind of variation is the surface position of the allocutive morpheme. In some languages such as Korean, morpheme ordering suggests a high surface position (Pak 2017), while in others, like Japanese, a much lower placement (Yamada 2019). The final kind of variation we consider is the form of the allocutive morpheme. In particular we distinguish two broad kinds of allocutive morphemes—those in which the allocutive morpheme behaves morphologically like a bound pronoun, and those in which the allocutive marker behaves like the exponence of agreement or a particle.

Following Baker (2008), Alok & Baker (2018), Portner et al. (2019) and Alok (2021), we take allocutivity, in all cases, to reflect exponence of addressee features licensed by a silent Addressee DP introduced high in the clausal spine. We propose that this is in fact the sole unifying formal property of allocutive languages. We propose, moreover, that variation in the properties of allocutive morphemes outlined above reflects four distinct loci of variation. The first concerns which elements in the allocutive structure are non-silent: in some languages, the allocutive morpheme is a head exponing agreement with the Addressee DP, or bearing an addressee feature, while in other languages, the allocutive morpheme is a bound pronoun. A second kind of variation is the position in which the silent Addressee DP may be introduced in the clause, as reflected by root/embedded asymmetries. Following Portner et al. (2019), we take root-sensitive allocutive morphemes to involve an Addressee DP merged in a high, utterance-related projection merged only in true root environments. Varieties allowing embedded allocutivity instantiate a lower first merged position of the allocutive morpheme, as suggested also by morpheme order facts and indexical shift. A third locus of variation is the presence vs. absence of an applicative head capable of introducing an allocutive pronoun. Support for an additional, lower structure capable of hosting allocutive forms comes from varieties with multiple allocutive marking within a single domain and/or in non-finite constituents. Fourth and finally, we propose that variation in the way that allocutive morphemes interact with other C-field elements reflects different licensing requirements of the head responsible for introducing the allocutive morpheme.

A principal outcome of the analysis is that allocutivity involves greater formal heterogeneity than has been described previously in the literature. We show that principal aspects of cross-linguistic variation nevertheless can be modeled in terms of a limited set of formal options independently motivated in descriptions of cross-linguistic variability in the behavior of thematic addressee-denoting DPs. That is, from the perspective of a theory that takes allocutive marking to be related to the syntax of addressee DPs more generally (Alok & Baker 2018; Portner et al. 2019), the variability observed across allocutive systems is no more nor less, we suggest, than what might reasonably be expected given well-described cross-linguistic differences in the behavior of thematic addressee DPs.

The discussion is organized as follows. Section 2 deals with empirical facts. It presents an extensive description of four kinds of variation across varieties. Section 3 outlines the formal variation we take these facts to motivate. Section 4 concludes the discussion.

2 Four kinds of variation across allocutive varieties

2.1 Root-embedded asymmetries

The first kind of variation across allocutive languages that we focus on is whether allocutive morphemes are restricted to root contexts. We consider three well-described root-allocutive varieties—Korean, conservative Basque dialects and Punjabi—before turning to a larger but less well-described set of languages that freely permit allocutive markers in embedded clauses.

2.1.1 Varieties with root restrictions

The clearest exemplar of this class of languages is Korean, as described by Pak (2017), Kim (2019) and Portner et al. (2019), where addressee status and/or context formality are marked by one of six rank-ordered allocutive particles on the finite verb.1 In example (1), the allocutive marker -supnita indicates that the addressee of the sentence is older or socially higher than the speaker, the sentence is a declarative, and the context where the sentence is uttered is formal. Moreover, the example shows that these particles are available in root contexts (1a), but sharply out in embeddings (1b).2 These forms are also available in root interrogatives as in (2), where -eyo indicates that the sentence is polite and in root imperatives as in (3), where -ala indicates that the speaker is higher than or equal to the addressee in social status (Pak 2017; Portner et al. 2019). Note, importantly, that the allocutive marker here spells out as a portmanteau form together with a clause-typing morpheme, which varies across declarative, interrogative and imperative contexts, a fact to which we return later, in section 2.2.

    1. (1)
    1. a.
    1.   Inho-ka
    2.   Inho-nom
    1. choysen-ul
    2. best-acc
    1. ta
    2. all
    1. ha-ess-supnita.
    2. do-pst-dec.formal
    1.   ‘Inho did his best.’
    1.  
    1. b.
    1. *Inho-ka
    2.   Inho-nom
    1. sensayngnim-kkey
    2. teacher-to
    1. [choysen-ul
    2. best-acc
    1. ta
    2. all
    1. ha-ess-supnita-ko]
    2. do-pst-dec.formal-comp
    1. malhayss-ta.
    2. said-dec.plain
    1.   ‘Inho told the teacher that he did his best.’
    2.   (Portner et al. 2019)
    1. (2)
    1. ∅you
    2. ∅you
    1. encey
    2. when
    1. ka-si-ess-eyo?
    2. go-hon-pst-int.pol
    1. ‘When did you go?’
    2. (Pak 2017)
    1. (3)
    1. Na-lul
    2. I-acc
    1. ttala-o-ala!
    2. follow-come-imp
    1. ‘Follow me!’
    2. (Portner et al. 2019)

A similar set of facts comes from conservative Basque dialects, where allocutive morphemes are second person clitics on the finite auxiliary. The most general pattern, found in southern dialects, is one where familiar addressees are marked via a clitic that also agrees with addressee’s gender. These dialects are like Korean in that allocutive morphemes are available in all root clause types, but sharply out in embeddings, as shown in (4) (Hualde 2003; Arregi & Nevins 2012; Haddican 2018). (We use the diacritic % here in view of the availability of embedded allocutivity in a subset of southern dialects that we return to shortly, in section 2.1.1.)

    1. (4)
    1. a.
    1. Jon
    2. Jon-erg
    1. etorri-ko
    2. come-fut
    1. d-u-k.
    2. expl-root-2sg.fam.m
    1. ‘Jon will come.’
    1.  
    1. b.
    1. Jon-ek
    2. Jon-erg
    1. esa-n
    2. say-prf
    1. d-i-k
    2. expl-root-2sg.fam.m
    1. etorri-ko
    2. come-fut
    1. d-u-(%k)-ela.
    2. expl-root-2sg.fam.m-c
    1. ‘Jon has said that he will come.’

A third allocutive language of this class is Punjabi, where allocutive markers are possible only in root contexts. In example (5a), the allocutive marker ii encodes a singular addressee while in (5b) the marker je encodes a plural addressee. The marker je is also used for a singular honorific addressee.

    1. (5)
    1. a.
    1. aman
    2. Aman.nom
    1. kitaab
    2. book
    1. paRh
    2. read
    1. reyaa
    2. prog.m.sg
    1. ii
    2. alloc.sg
    1. ‘Aman is reading a book.’                                                       [spoken to a familiar addressee]
    1.  
    1. b.
    1. aman
    2. Aman.nom
    1. kitaab
    2. book
    1. paRh
    2. read
    1. reyaa
    2. prog.m.sg
    1. je
    2. alloc.pl
    1. ‘Aman is reading a book.’       [spoken to a group of familiar addressees/honorific addressee]

Moreover, in embedded contexts, Kaur & Yamada (2019) show that the allocutive marking is restricted to the complement clauses of speech predicates. As shown in (6), je is possible under the speech verb ‘say’ but impossible under the thought predicate ‘think’.

    1. (6)
    1. a.
    1.   karan-ne
    2.   Karan-erg
    1. keyaa
    2. said
    1. [ki
    2. that
    1. miiraa
    2. Mira
    1. kal
    2. tomorrow
    1. aayegii
    2. come.fut
    1. je]
    2. alloc.pl
    1.   ‘Karan said that Mira will come tomorrow.’
    1.  
    1. b.
    1. *karan-ne
    2.   Karan-erg
    1. soceyaa
    2. thought
    1. [ki
    2. that
    1. miraa
    2. Mira
    1. kal
    2. tomorrow
    1. aayegii
    2. come.fut
    1. je]
    2. alloc.pl
    1.   ‘Karan thought that Mira will come tomorrow.’

In addition, Kaur & Yamada (2019) report that outside verbal complements, the allocutive marker may occur in temporal/location-clauses as in (7a) but not in purpose and reason-clauses as in (7b). Moreover, the marker is found only in finite clauses.

    1. (7)
    1. a.
    1.   maiN
    2.   I
    1. tadd
    2. then
    1. jaavaaNgii
    2. go.fut
    1. [jaddoN
    2. when
    1. karan
    2. Karan
    1. vii
    2. also
    1. jaayegaa
    2. go.fut
    1. je]
    2. alloc.pl
    1.   ‘I will go when Karan does too.’
    1.  
    1. b.
    1. *karan
    2.   Karan
    1. bajaar
    2. market
    1. gayaa
    2. go.prf
    1. [kyoNkii
    2. because
    1. o-nuu
    2. 3.sg-dom
    1. ikk
    2. a
    1. kuRii-ne
    2. girl-erg
    1. bulaayaa
    2. call.prf
    1. je]
    2. alloc.pl
    1.   ‘Karan went to the market because a girl called him.’

We follow Kaur & Yamada (2019) and Kaur (2020) in taking this distribution to reflect a root/non-root distinction in the analysis below.

2.1.2 Root-insensitive allocutive morphemes

Contrasting with the above varieties is a class of languages that freely permits allocutive morphemes in embedded contexts. One such variety is Magahi, an Indo-Aryan language spoken in northern and eastern India, principally in the state of Bihar (Verma 1991; 2003; Bhattacharya 2016; Alok 2020; 2021). In example (8a), the morpheme -au indicates that the addressee is nonhonorific to the speaker. The morpheme -o, in (8b), indicates that the addressee is honorific, while in (8c), the morpheme -ain indicates that the addressees is high honorific to the speaker. Examples are taken from Alok (2021).

    1. (8)
    1. a.
    1. Santee-aa
    2. Santee-nh
    1. dauR-l-au.
    2. run-prf-nh.sbj;nh.alloc
    1. ‘Santee ran.’                                                                           [spoken to a friend]
    1.  
    1. b.
    1. Santee-aa
    2. Santee-nh
    1. dauR-l-o.
    2. run-prf-nh.sbj;h.alloc
    1. ‘Santee ran.’                                                                           [spoken to a father]
    1.  
    1. c.
    1. Santee-aa
    2. Santee-nh
    1. dauR-l-ain.
    2. run-prf-nh.sbj;hh.alloc
    1. ‘Santee ran.’                                                                         [spoken to a teacher]

In recent work, Alok & Baker (2018) and Alok (2021) report that Magahi allocutive marking freely appears in finite embedded contexts. Example (9a) illustrates embedding of allocutive agreement under a perceptual predicate, example (9b) shows allocutive agreement in relative clauses, and example (9c) illustrates it in noun complement clauses. All these sentences are spoken to a friend, triggering the -au suffix on the verb.

    1. (9)
    1. a.
    1. Ham
    2. I
    1. dekhl-i-au
    2. saw-1-nh.sbj;nh.alloc
    1. [ki
    2. comp
    1. Santeeaa
    2. Santee
    1. bhag
    2. escape
    1. gel-au].
    2. went-nh.sbj;nh.alloc
    1. ‘I saw that Santee ran away.’                                                                                                    [spoken to a friend]
    1.  
    1. b.
    1. Laikwaa
    2. Boy
    1. [je
    2. rel.pro
    1. uhan
    2. there
    1. khaRaa
    2. stand
    1. h-au]
    2. be-nh.sbj;nh.alloc
    1. hamar
    2. my
    1. bhaai
    2. brother
    1. h-au.
    2. be-nh.sbj;nh.alloc
    1. ‘The boy who is standing there is my brother.’                                                                        [spoken to a friend]
    1.  
    1. c.
    1. Aphawaah
    2. rumor
    1. [ki
    2. comp
    1. Santeeaa
    2. Santee
    1. inaam
    2. prize
    1. jitl-au]
    2. won-nh.sbj;nh.alloc
    1. sahii
    2. true
    1. ha-l-au
    2. be-prf-nh.sbj;nh.alloc
    1. ‘The rumor that Santee won the prize was true.’                                                                      [spoken to a friend]

Cross-linguistically, such a range of embedded contexts do not typically permit embedded root phenomena i.e., instances of lower clauses with root clause syntax (Hooper & Thompson 1973; Heycock 2006; Julien 2009; Wiklund et al. 2009), suggesting that embedding of allocutive marking is an instance of true embedding in Magahi. Additional evidence that these are true embeddings as opposed to embedded root contexts comes from the fact that allocutive marking is possible in contexts both with and without indexical shift. In (10), for example, ham, ‘I’, in the lower clause can be interpreted as either John (shifted interpretation) or the utterance author (unshifted interpretation). From the perspective of the consensus approach that takes the binding operation implicit in indexical shift to be structure sensitive, the availability of the shifted interpretation in (10) suggests that the allocutive morpheme sits in a true embedding.

    1. (10)
    1. John
    2. John
    1. socha
    2. think
    1. h-au
    2. be-nh.alloc
    1. ki
    2. that
    1. ham
    2. I
    1. tej
    2. smart
    1. h-i-au.
    2. be-1.sbj-nh.alloc
    1. ‘John thinks that I (=speaker, =John) am smart.’                         [spoken to a friend]
    2. (Alok & Baker 2018)

Similar facts are described for Tamil by Sundaresan (2018) and McFadden (2020). In the below Tamil example, the morpheme -ŋgæ indicates that the addressee of this sentence is either a group or a person with whom the speaker would use polite forms i.e., an honorific addressee (McFadden 2020).

    1. (11)
    1. Naan
    2. I
    1. dӡaangiri
    2. Jangri
    1. vaang-in-een-ŋgæ.
    2. buy-pst-1sg.sbj-alloc
    1. ‘I bought Jangri.’
    2. (McFadden 2020)

As in Magahi, this allocutive morpheme may appear in embedded domains, and, as in the case of Magahi, evidence for taking these contexts to be bona fide embedded contexts comes from indexical shift. In (12), -ŋgæ appears in a lower clause whose anaphoric subject is interpreted as coreferential with the matrix subject. In such a context, moreover, the allocutive morpheme is interpreted as honorifying the matrix recipient—the hearer of the reported speech event.

    1. (12)
    1. Mayai
    2. Maya.nom
    1. [taani,*j
    2. anaph.nom
    1. pooʈʈi-læ
    2. contest-loc
    1. dӡejkka-ppoo-r-een-ŋgæ-nnŭ]
    2. win-go-prs-1sg-alloc-comp
    1. Seetha-kiʈʈæ
    2. Seetha-loc
    1. so-nn-aa.
    2. say-pst-3.f.sg.sbj
    1. ‘Mayai told Seethaj that shei would win the contest.’ (Plural/polite form to embedded addressee)
    2. (adapted from Sundaresan (2018))

In Galician, the relevant set of morphemes is usually referred to as “solidarity pronouns” or “solidarity clitics” in the Romance literature, and will here be called “allocutive clitics”.3 These forms are accepted by most Galician speakers but are most common in rural dialects and in informal contexts. They have been discusssed in much of the formal and descriptive literature on Galician (Álvarez Blanco 1980; 1994; 1997; Álvarez et al. 1986; Campos 1989; Carbón Riobóo 1995; Uriagereka 1995a; Longa & Lorenzo 2001; Huidobro 2009; 2022; Haddican 2019), but have not previously been discussed extensively in connection to theories of allocutivity or speech act encoding generally. We illustrate such a construction in (13), where the clitic che marks agreement in familiarity with the addressee of the sentence.4

    1. (13)
    1. Non
    2. no
    1. che
    2. 2sg.fam.dat
    1. me
    2. 1sg
    1. dá
    2. give
    1. pena
    2. sorrow
    1. ningunha.
    2. any
    1. ‘It doesn’t make me feel bad at all.’
    2. (Álvarez Blanco 1997: 8)

Importantly, as discussed by Huidobro (2022), these morphemes are available in embedded contexts, as in (14).

    1. (14)
    1. Xurso
    2. George
    1. cre
    2. believes
    1. que
    2. that
    1. este
    2. this
    1. ministro
    2. minister
    1. che
    2. 2sg.fam.dat
    1. é
    2. is
    1. un
    2. a
    1. pouco
    2. little
    1. ladrón.
    2. thief
    1. ‘George believes that that Minister is a little of bit of a thief.’
    2. (Huidobro 2022: 142)

Evidence that these contexts are true embeddings rather than embedded root contexts comes from the fact that these clitics appear in environments that fail standard root clause tests. Galician shares with European Portuguese a set of embedded-root asymmetries for clitic placement called “clitic second”. These constraints hold that, in the general case, root clauses require enclisis except in the presence of negation, a moved wh-phrase and some quantifiers (Raposo & Uriagereka 2005). Subordinate clauses, on the other hand, require proclisis (16).5

    1. (15)
    1. a.
    1. Todo
    2. all
    1. o
    2. the
    1. mundo
    2. world
    1. o
    2. it
    1. viu
    2. saw
    1. /
    2. /
    1. *viu-no
    2.   saw-it
    1. ‘Everyone saw it.’
    1.  
    1. b.
    1. Xan
    2. Xan
    1. Rodriguez
    2. Rodriguez
    1. *o
    2.   it
    1. viu
    2. saw
    1. /
    2. /
    1. viu-no
    2. saw-it
    1. ‘Xan Rodriguez saw it.’
    2. (Adapted from Uriagereka (1995))
    1. (16)
    1. El
    2. he
    1. dixo
    2. said.3sg
    1. que
    2. that
    1. a
    2. her
    1. viu
    2. saw.3sg
    1. /
    2. /
    1. *viu-na
    2.   saw.3sg-her
    1. onte.
    2. yesterday
    1. ‘He said that he saw her yesterday.’6

Importantly, when allocutive clitics appear in embedded contexts, they do so pursuant to the usual morpheme order constraints for embedded clauses, as in (17), suggesting that allocutive clitics are possible in true syntactic embeddings.

    1. (17)
    1. Creo
    2. think.1sg
    1. que
    2. that
    1. *viu-cha(<che+a)
    2.   saw.3sg-2sg.fam.dat.3sg.f
    1. /
    2. /
    1. cha
    2. 2sg.fam.dat.3sg.f
    1. viu.
    2. saw.3sg
    1. ‘I think he/she/it saw her.’

In addition, unlike embedded root phenomena in other languages, allocutive clitics are not sensitive to embedding type (Julien 2009; Wiklund et al. 2009), nor precluded in contexts commonly taken to involve embedded operators (Haegeman 2006; Haegeman & Ürögdi 2010). Example (17) illustrates the possibility of allocutive clitics under ‘think’-type verbs. Example (18)–(20) illustrate the possibility of allocutive clitics in factive embeddings, in relative clauses and in embedded wh-questions.

    1. (18)
    1. Esquecin
    2. forgot.1sg.pst
    1. que
    2. that
    1. vos
    2. 2pl.fam.dat
    1. está
    2. cop
    1. aquí.
    2. here
    1. ‘I forgot that he is here.’
    1. (19)
    1. A
    2. the
    1. rapaza
    2. girl
    1. que
    2. that
    1. che
    2. 2sg.fam.dat
    1. está
    2. cop
    1. aquí.
    2. here
    1. ‘The girl that is here.’
    1. (20)
    1. Non
    2. neg
    1. sei
    2. know.1sg
    1. quen
    2. who
    1. cha(<che+a)
    2. 2sg.fam.dat.3sg.f
    1. viu.
    2. see.3sg.pst
    1. ‘I don’t know who saw it.’

As foreshadowed, some Southern Basque dialects also allow for allocutive morphemes in embedded domains as in (4b), repeated here. We refer to these dialects as ‘Innovative Southern Basque’, in observance of the fact that embedded allocutive marking is commonly taken to be novel (Azkue 1923; Hualde 2003).

    1. (21)
    1. Jon-ek
    2. Jon-erg
    1. esa-n
    2. say-prf
    1. d-i-k
    2. expl-root-2sg.fam.m
    1. etorri-ko
    2. come-fut
    1. d-u-(%k)-ela.
    2. expl-root-2sg.fam.m-c
    1. ‘Jon has said that he will come.’

As described in Haddican & Etxeberria (2022), these morphemes appear in all finite embedding types including those typically inhospitable to embedded root phenomena such as relative clauses, embedded yes/no-questions and temporal clauses (Haegeman & Ürögdi 2010). As noted by Haddican & Etxeberria (2022), allocutive clitics in these innovative varieties are possible in contexts that fail standard root-clause tests. In particular, (22) shows allocutive marking in relative clauses, where “ba-support”—a verb-second-position repair operation that applies only in root contexts—is sharply out (Ortiz de Urbina 1989; Elordieta & Haddican 2018).

    1. (22)
    1. Relatives
    1. (*ba-)z-etorre-(%k)-en
    2. ba-expl-come-2sg.fam.m-comp
    1. ekaitza.
    2. storm.def
    1. ‘The storm that is coming.’
    1. (23)
    1. Embedded yes/no questions
    1. Ez
    2. neg
    1. z-aki-a-t
    2. expl-know-2sg.fam.m-1sg.erg
    1. [(ba)-z-etorre-(%k)-en
    2. epen-expl-come-2sg.fam.m-comp
    1. ala
    2. or
    1. ez].
    2. neg
    1. ‘I don’t know if he’s coming or not.’
    1. (24)
    1. Temporal clauses
    1. Jon
    2. Jon
    1. z-etorre-(%k)-en-ean
    2. expl-come-2sg.fam.m-comp-in
    1. ikusi-ko
    2. see-fut
    1. d-i-a-t.
    2. expl-root-2sg.fam.m-1sg
    1. ‘When John comes, I will see him.’

A fifth language recently reported to exhibit true embedded allocutivity is Japanese. In a series of influential pieces, Miyagawa (2012; 2013; 2017), following a brief discussion by Oyharçabal (1993), draws parallels between the Japanese allocutivity particle -mas-, and allocutivity in Zuberoan Basque. Miyagawa proposes that -mas- is restricted to root contexts, defined a bit more broadly than in Korean and Basque. In particular, Miyagawa reports that -mas- is possible in four main contexts: finite monoclausal contexts (25a), because-clauses (25b), in complement clauses embedded under verbs of saying (25c), and relative clauses (25d). Miyagawa notes that this distribution is parallel to those in which English speaker-oriented adverbs frankly, truthfully and honestly are possible. Following Emonds (1969), Miyagawa (2017) takes these contexts to be root contexts, defined as unselected contexts. Examples are taken from (Miyagawa 2017).

    1. (25)
    1. a.
    1. Hanako-wa
    2. Hanako-top
    1. ki-mas-u.
    2. come-alloc-prs
    1. ‘Hanako will come.’                                                       [monoclausal contexts]
    1.  
    1. b.
    1. [Hanako-ga
    2. Hanako-nom
    1. ki-mas-u
    2. come-alloc-prs
    1. kara],
    2. because
    1. ie-ni
    2. home-at
    1. ite-kudasai
    2. be-please
    1. ‘Because Hanako will come, please be at home.’               [‘because’ contexts]
    1.  
    1. c.
    1. Taroo-wa
    2. Taro-top
    1. [Hanako-ga
    2. Hanako-nom
    1. ki-mas-u
    2. come-alloc-prs
    1. to]
    2. comp
    1. itta.
    2. said
    1. ‘Taro said that Hanako will come.’                                                      [under ‘say’]
    1.  
    1. d.
    1. Watasi-wa
    2. I-top
    1. [mizu-tama-moyoo-no
    2. polka.dots
    1. ari-mas-u
    2. exist-alloc-prs
    1. kami-ga]
    2. paper-nom
    1. hosi-i
    2. want
    1. to
    2. comp
    1. omoi-mas-u.
    2. think-mas-prs
    1. ‘I want the paper that has polka dots.’                                        [relative clauses]

Recently, Yamada (2019) has described the availability of -mas- in a broader range of subordinated contexts under the complementizer -koto, as in (26), from (Yamada 2019).

    1. (26)
    1. Karei-mo
    2. he-also
    1. [karei-no
    2. he-gen
    1. musuko-ga
    2. son-nom
    1. kabin-o
    2. vase-acc
    1. kowasi-te
    2. break-cv
    1. simai-masi-ta
    2. mal-hon-pst
    1. koto]-o
    2. comp-acc
    1. wabi-te
    2. apologizing-cvb
    1. ori-mas-u.
    2. prog.hon-hon-prs
    1. ‘Hei is also apologizing for hisi son having broken the vase.’

Similarly, (Miyagawa 2022: 119–127) mentions that in addition to -koto, another complementizer that embeds -mas- is -yooni.

    1. (27)
    1. Isya-wa
    2. doctor-top
    1. kanzya-sani-ni
    2. patient-dat
    1. [ei
    2.  
    1. kusuri-o
    2. medicine-acc
    1. nomi-mas-u
    2. take-mas-prs
    1. yooni]
    2. C
    1. motome-masi-ta.
    2. ask-mas-pst
    1. ‘The doctor asked the patient to take medicine.’

We return to these facts shortly in section 2.2. For the moment, we observe, following Yamada (2019) and Miyagawa (2022), that -mas- is indeed available in some embedded contexts.

2.2 Interactions in the C domain

A second way that allocutive morphemes vary in behavior is in their interactions with other elements in the left periphery of the clause. In particular, allocutive varieties fall into one of two broad classes—those in which exponence of the allocutive morpheme interacts with the local clause typing or the complementizer morpheme and those in which it does not.

In the former class is the Northern Basque Zuberoan dialect, described in Oyharçabal’s (1993) seminal work. In Zuberoan, allocutive marking is only possible in root declaratives and not root interrogatives, as shown in (28) and (29), adapted from (Oyharçabal 1993).7

    1. (28)
    1. a.
    1.   [Manex
    2.   Manex
    1. joan-en
    2. go-fut
    1. d-e-la]
    2. expl-root-comp
    1. uste
    2. think
    1. d-u-k
    2. expl-root-2sg.fam.m
    1.   ‘You think that John will go.’
    1.  
    1. b.
    1. *[Manex
    2.   Manex
    1. joan-en
    2. go-fut
    1. d-u-a-la]
    2. expl-root-2sg.fam.m-comp
    1. uste
    2. think
    1. d-u-k
    2. expl-root-2sg.fam.m
    1.   ‘You think that John will go.’
    1. (29)
    1. a.
    1.   Lan
    2.   work
    1. egiten
    2. do-imperf
    1. d-ui-a
    2. expl-root-q
    1. hire
    2. your
    1. lagunak?
    2. friend
    1.   ‘Does your friend work?’
    1.  
    1. b.
    1. *Lan
    2.   work
    1. egiten
    2. do-imperf
    1. d-i-n-a
    2. expl-root-2sg.fam.m-q
    1. hire
    2. your
    1. lagunak?
    2. friend
    1.   ‘Does your friend work?’

A different kind of interaction is seen in Korean, as noted earlier, where the allocutive marker and clause-typing morpheme spell out as a single head (cf. (1a)). Example (30) shows that exponence of this head varies by clause type, suggesting that the heads responsible for allocutive agreement and clause typing are syntactically adjacent.

    1. (30)
    1. a.
    1. Inho-ka
    2. Inho-nom
    1. choysen-ul
    2. best-acc
    1. ta
    2. all
    1. ha-ess-ta.
    2. do-pst-decl.pln
    1. ‘Inho did his best.’                                                              [declarative]
    1.  
    1. b.
    1. Inho-ka
    2. Inho-nom
    1. choysen-ul
    2. best-acc
    1. ta
    2. all
    1. ha-ess-(nu)nya/ni?
    2. do-pst-int.pln
    1. ‘Did Inho do his best?’                                                    [interrogative]
    1.  
    1. c.
    1. Choysen-ul
    2. best-acc
    1. ta
    2. all
    1. ha-la/ela!
    2. do-imp.pln
    1. ‘Do your best!’                                                                    [imperative]
    2. (Portner et al. 2019)

A third such interaction is seen in Japanese, where the allocutive marker -mas- is possible under only certain complementizers. As noted in Miyagawa (2012; 2022) and Yamada (2019) and discussed in section 2.1.2, the allocutive marker -mas- is possible in embedded clauses with the complementizer -koto and -yooni, but not in clauses headed by the complementizer to, which is restricted to higher predicates of saying and thinking.8

    1. (31)
    1. [Gomeiwaku-o
    2. trouble-acc
    1. o-kake
    2. hon-giving
    1. si-masi-ta-koto]-wa
    2. do-hon-pst-comp-top
    1. zonzi-te
    2. know.hon-cvb
    1. i-mas-u.
    2. prf-hon-prs
    1. ‘I know that I gave you trouble.’
    2. (Yamada 2019: 366)
    1. (32)
    1. Isya-wa
    2. doctor-top
    1. kanzya-sani-ni
    2. patient-dat
    1. [ei
    2. e
    1. kusuri-o
    2. medicine-acc
    1. nomi-mas-u
    2. take-mas-prs
    1. yooni]
    2. comp
    1. motome-masi-ta.
    2. ask-mas-pst
    1. ‘The doctor asked the patient to take medicine.’
    2. (Miyagawa 2022)
    1. (33)
    1. Taroo-wa
    2. Taro-top
    1. [Hanako-ga
    2. [Hanako-nom
    1. kuru/*ki-mas-u
    2. come/come-prs
    1. to]
    2. comp
    1. sinzitei-ru.
    2. believe-prs
    1. ‘Taro believes that Hanako will come.’
    2. (Miyagawa 2012)

In the other varieties in our sample, the allocutive morpheme does not show any interaction in C-domain. For example, in innovative Southern Basque, allocutive morphemes are available in all finite clause, like other clitic forms Haddican & Etxeberria (2022). Similarly, in Magahi and Galician, allocutivity is available across all finite clause types, in imperatives, root and embedded questions (Huidobro 2018; 2022).

2.3 Allocutive morpheme placement

A third kind of variation across allocutive varieties that we consider is the surface position of the allocutive morpheme relative to other clausal material. Much recent work on allocutivity has taken the head implicated in allocutive marking to be merged very high in the clausal sequence, based in part on root-restrictions in some varieties as discussed above (Speas & Tenny 2003; Miyagawa 2013; Portner et al. 2019). Portner et al. (2019), for instance, propose that Korean allocutive particles spell out a head they label “c” merged high in the clause, just above the position of clause-typing morphemes. Their evidence for this claim is of two kinds. First, as noted above, Korean allocutive morphemes are strictly proscribed in embeddings, which Portner et al. (2019) take to indicate a high position not truncated in embedded contexts. Second, Korean allocutive particles spell out portmanteau forms together with the clause-typing morpheme, suggesting the two heads are in some local relationship at spell-out.

Importantly, in several varieties, there is evidence that the allocutive morpheme sits in a lower position than what Portner et al. (2019) propose for Korean. In innovative Basque dialects, for example, the allocutive clitic sits inside the position of the clause-typing complementizer (Haddican & Etxeberria 2022), as shown in (4b), repeated here. Importantly, the -k allomorph in (34) is determined in contexts in which the allocutive morpheme is on the right edge of the morphological word, indicating -k is inside the auxiliary in a constituent distinct from that containing the clause-typing complementizer.9

    1. (34)
    1. Jon-ek
    2. Jon-erg
    1. esa-n
    2. say-prf
    1. d-i-k
    2. expl-root-2sg.fam.m
    1. etorri-ko
    2. come-fut
    1. d-u-(%k)-ela.
    2. expl-root-2sg.fam.m-comp
    1. ‘Jon has said that he will come.’

Similarly, in Magahi two sets of facts suggest a position of the allocutive morpheme just above TP. The first is morpheme order, with the allocutive marker appearing on the right edge of the finite verbal cluster as in (35), suggesting a position above TP material. Second, addressee honorific features and subject honorific features combine for spell-out in Magahi, suggesting the two heads are syntactically adjacent (Alok 2021). We illustrate this in (35), from (Alok 2020; 2021). Example (35a) and (35b) are spoken to a friend, a non-honorified addressee, but differ in the honorific status of the subject. In (35a), in the presence of a non-honorific subject, the combined agreement marker spells out as -au. In (35b), with an honorific subject, the combined spell-out form is -thu(n). Similarly, the addressee of example (35c) and (35d) are the same, a teacher, a high honorific addressee but differ in the honorific status of the subject. In (35c), the verb is marked with the agreement morpheme -ain, while in example (35d), the verb bears the agreement morpheme -thi(n).

    1. (35)
    1. a.
    1. Santeeaa
    2. Santee-nh
    1. dauR-l-au.
    2. run-prf-nh.sbj;nh.alloc
    1. ‘Santee ran.’                                                    [spoken to a friend]
    1.  
    1. b.
    1. Baabaa
    2. grandfather.h
    1. dauR-la-thu(n).
    2. run-prf-h.sbj;nh.alloc
    1. ‘Grandfather ran.’                                             [spoken to a friend]
    1.  
    1. c.
    1. Santeeaa
    2. Santee.nh
    1. dauR-l-ain.
    2. run-prf-nh.sbj;hh.alloc
    1. ‘Santee ran.’                                                  [spoken to a teacher]
    1.  
    1. d.
    1. Baabaa
    2. grandfather.h
    1. dauR-la-thi(n).
    2. run-prf-h.sbj;hh.alloc
    1. ‘Grandfather ran.’                                          [spoken to a teacher]

Punjabi is a third language where the allocutive marker appears to occupy a position below that for allocutive morphemes in Korean as analyzed by Portner et al. (2019). Kaur (2020) notes that the allocutive marker is a free morpheme in Punjabi, which occurs to the right of the auxiliary as in (36). These are the facts as described by Kaur (2020) for past tense contexts. In present tense contexts, the allocutive morpheme alternates with the auxiliary, and appears to the left of the polar question particle kii as in (37). Kaur (2020) takes these facts to indicate that the allocutive head (which Kaur labels “Pers”) is located in a position similar to that described above for Magahi and Basque, just above tense-marked auxiliaries.

    1. (36)
    1. o
    2. 3sg.nom
    1. billii
    2. cat.f.sg
    1. paaldaa
    2. raise.hab.m.sg
    1. sii
    2. be.pst.3sg
    1. je
    2. alloc.pl
    1. ‘He used to raise cats.’
    2. (Kaur 2020)
    1. (37)
    1. karan-ne
    2. Karan-erg
    1. miraa-nuu
    2. Mira-dat
    1. kitaab
    2. book
    1. dittii
    2. give.pfv.f.sg
    1. je/e
    2. alloc.pl/be.prs.3sg
    1. kii
    2. polq
    1. ‘Has Karan given the book to Mira?’
    2. (Kaur 2020)

In Japanese, evidence summarized by Yamada (2019) suggests that allocutive -mas- is introduced even lower, below the position of tense markers, as in (26), repeated here as (38).

    1. (38)
    1. Karei-mo
    2. he-also
    1. [karei-no
    2. he-gen
    1. musuko-ga
    2. son-nom
    1. kabin-o
    2. vase-acc
    1. kowasi-te
    2. break-cv
    1. simai-masi-ta
    2. mal-hon-pst
    1. koto]-o
    2. comp-acc
    1. wabi-te
    2. apologizing-cvb
    1. ori-mas-u.
    2. prog.hon-hon-prs
    1. ‘Hei is also apologizing for hisi son having broken the vase.’
    2. (Yamada 2019)

Similarly, in Galician, allocutive clitics can appear inside infinitives in modal contexts which lack independent tense values from the finite verb, and those that allow for clitic climbing. If one takes such contexts to reflect a truncated functional structure such that the infinitival constituent lacks CP and TP layers, then such contexts suggest an additional case in which allocutive morphemes may appear TP-internally.

    1. (39)
    1.   Einstein
    2.   Einstein
    1. tivo
    2. have.3sg.pst
    1. que
    2. that
    1. estudar-che
    2. study.inf-2sg.fam.dat
    1. moito.
    2. much
    1.   ‘Einstein had to study a lot.’
    1. (40)
    1. %Maria
    2.   Maria
    1. quere
    2. wants
    1. chegar-che
    2. arrive-2sg
    1. pronto.
    2. soon
    1.   ‘Maria wants to arrive soon.’

In finite contexts, however, Uriagereka (1995a) argues that allocutive clitics are introduced in a left peripheral position, based in part on the order of allocutive clitics in clitic clusters and their participation in second position effects as in (17) from section 2.1.2, repeated here. In such contexts, the clitic always appears to the right of complementizers, suggesting a position below the latter.

    1. (41)
    1. Creo
    2. think.1sg
    1. que
    2. that
    1. *viu-cha(<che+a)
    2. saw.3sg-2sg.fam.dat.3sg.f
    1. /
    2. /
    1. cha
    2. 2sg.3sg.f
    1. viu.
    2. saw.3sg
    1. ‘I think he/she/it saw her.’

The facts for Galician, then, suggest that allocutive clitics, may be introduced in different positions in the clausal sequence. (See Álvarez Blanco (1997) and Huidobro (2022) for discussion.)

The evidence just presented suggests that natural languages allow allocutive morphemes to be introduced in different positions in the clausal spine. An additional kind of evidence to this effect comes from McFadden’s description of Tamil, according to which the allocutive morpheme can appear below or above the question particle, or in both positions simultaneously.

    1. (42)
    1. a.
    1. niiŋgæ
    2. you.pl
    1. saap-ʈ-aačč-aa-ŋgæ?
    2. eat-asp-res-q-alloc
    1. ‘Have you eaten?’
    1.  
    1. b.
    1. niiŋgæ
    2. you.pl
    1. saap-ʈ-aaččŭ-ŋgæɭ-aa?
    2. eat-asp-res-alloc-q
    1. ‘Have you eaten?’
    1.  
    1. c.
    1. niiŋgæ
    2. you.pl
    1. saapʈ-aaččŭ-ŋgæɭ-aa-ŋgæ?
    2. eat-res-alloc-q-alloc
    1. ‘Have you eaten?’
    2. (McFadden 2020)

2.4 Morpheme type

The final kind of difference among allocutive marking languages that we consider is the nature of the allocutive morpheme itself. In particular, allocutive languages are divisible into two classes in this respect: those in which the allocutive marker is a pronoun, and those in which it behaves as an agreeing head—i.e. as an agreement morpheme—or particle. We consider these two groups in turn.

2.4.1 Clitic pronouns

In all formal descriptions that we are aware of, Galician allocutive morphemes are treated as clitic pronouns (Álvarez Blanco 1980; 1994; Uriagereka 1995a; b; Longa & Lorenzo 2001; Raposo & Uriagereka 2005; Huidobro 2018; 2022; Haddican 2019). These morphemes are identical in exponence to thematic dative clitics as summarized in Table 1, with the consequence that strings with second person datives are often ambiguous among allocutive, benefactive and ethical interpretations as in (43) (Huidobro 2022).10 As reflected in the gloss and translation in (43), allocutive clitics also behave like thematic dative clitics in their participation in portmanteau clitic formation rules (44). Finally, as noted in Section 2.1.2, Galician allocutive morphemes also behave like ordinary clitics in their obedience of proclisis/enclisis rules.

Table 1

Dative clitics in Galician.

Person Singular Plural
2 familiar che vos
2 formal lle lles
    1. (43)
    1. Merquei-cha
    2. bought.1sg-2sg.fam.dat.3sg.acc.f
    1. ‘I bought it.’/’I bought you it.’
    1. (44)
    1. che
    2. 2sg.fam.dat
    1. +
    2.  
    1. a
    2. 3sg.acc.f
    1. →
    2.  
    1. cha
    2. 2sg.fam.dat;3sg.acc.f

Similarly, in Basque, allocutive morphemes are near identical in exponence and allomorphy conditions to person morphemes cross-referencing thematic addressees as in (45).

    1. (45)
    1. a.
    1. Hi-ri
    2. 2sg.fam-dat
    1. ema-n
    2. give-prf
    1. d-i-a/na-t
    2. expl-root-2sg.fam.m.dat/f-1sg.erg
    1. ‘I have given it to you.’
    1.  
    1. b.
    1. Hi-k
    2. 2sg.fam-erg
    1. egi-n
    2. do-prf
    1. d-u-k/-n.
    2. expl-root-2sg.erg.fam.m/f
    1. ‘You have done it.’

The consensus in recent Basque formal literature is that person markers on the auxiliary are not the exponence of agreement on a functional head (pace Miyagawa 2013, a.o.), but rather clitic pronouns doubling a possibly silent DP (Laka 1993; Rezac 2008; Arregi & Nevins 2012; Preminger 2009; Rezac et al. 2014). The principal motivation for a clitic analysis of these forms comes from locality in displacement for morphemes marking agreement with internal arguments, as first discussed in Arregi & Nevins (2012). If one takes person markers to reflect agreement on a verbal head, then contexts like (46) seem to require the 1sg.dat morpheme to raise out of the non-finite constituent, skipping over intervening heads such as modals (Travis 1984). If, instead, one takes such morphemes to be clitics, then the movement required seems an unremarkable instance of clitic climbing. We therefore take all person morphemes on the auxiliary—including allocutive morphemes—to be clitics, following Rezac (2006), Arregi & Nevins (2012), Haddican (2018) and Haddican & Etxeberria (2022).

    1. (46)
    1. [Ni-ri
    2. 1sg-dat
    1. eman]
    2. eman.inf
    1. nahi
    2. want
    1. d-i-t.
    2. expl-root-1sg.dat
    1. ‘He/she/it wants to give it to me.’

2.4.2 Agreement morphemes and particles

Contrasting with the above languages are varieties in which allocutive morphemes behave like exponence of agreement on a functional head. As mentioned above in section 2.2, in Magahi, subject honorification and allocutive agreement combines features for spell out, suggesting that, like subject agreement, allocutive markers are agreement morphemes, as in (35), repeated here as (47).

    1. (47)
    1. a.
    1. Santeeaa
    2. Santee-nh
    1. dauR-l-au.
    2. run-prf-nh.sbj;nh.alloc
    1. ‘Santee ran.’                                                    [spoken to a friend]
    1.  
    1. b.
    1. Baabaa
    2. grandfather.h
    1. dauR-la-thu(n).
    2. run-prf-h.sbj;nh.alloc
    1. ‘Grandfather ran.’                                            [spoken to a friend]
    1.  
    1. c.
    1. Santeeaa
    2. Santee.nh
    1. dauR-l-ain.
    2. run-prf-nh.sbj;hh.alloc
    1. ‘Santee ran.’                                                   [spoken to a teacher]
    1.  
    1. d.
    1. Baabaa
    2. grandfather.h
    1. dauR-la-thi(n).
    2. run-prf-h.sbj;hh.alloc
    1. ‘Grandfather ran.’                                          [spoken to a teacher]

Similarly, in Tamil, the allocutive marker -ŋgæ is isomorphic with a general plural marker of the language (McFadden 2020), indicating that the allocutive marker is a true agreement morpheme in Tamil. Consider example (11), repeated here, where the presence of the morpheme -ŋgæ indicates that the sentence is spoken to either a group or an honorific addressee.

    1. (48)
    1. Naan
    2. I
    1. dʒaangiri
    2. Jangri
    1. vaang-in-een-ŋgæ.
    2. buy-pst-1sg.sbj-alloc
    1. ‘I bought Jangri.’

The same morpheme appears in other contexts, as shown in Table 2, taken from McFadden (2020), which presents the regular agreement paradigms for the simple present tense and imperative forms of the verb ooɖǔ ‘run’. We see that the morpheme ŋɡæ appears as a final element on the verb ooɖǔ to mark the 2nd and 3rd person plural agreement.11 Moreover, it is also used to mark second person plural imperatives. An additional context in which -(ŋ)gæ appears as a number marker is as an affix on nominals as in Table 3. (Table taken from McFadden (2020).)

Table 2

Regular verb agreement in Tamil.

Form sg pl
1 ooɖǔ-r-een ooɖǔ-r-oom
2 ooɖǔ-r-æ ooɖǔ-r-iiŋgæ
3f ooɖǔ-r-aa ooɖǔ-r-aaŋgæ
3m ooɖǔ-r-aan ooɖǔ-r-aaŋgæ
3pol ooɖǔ-r-aarǔ ooɖǔ-r-aaŋgæ
3n ooɖǔ-dǔ ooɖǔ-dǔ
imp ooɖǔ ooɖǔŋgæ
Table 3

Number marking in Tamil nominals.

Subject sg pl
1excl naan naaŋgæ
2 nii niiŋgæ
3m avan avaŋgæ
‘girl’ poɳɳǔ poɳɳǔŋgæ
‘tree’ maram maraŋgæ

Korean and Japanese both lack a full paradigm of person agreement on verbs akin to that found in English and French. However, for Japanese, Miyagawa (2017) argues that even though the language lacks agreement in the T-domain, it has an agreement system in the C-domain (also see Yamada (2019)). Portner et al. (2019), for Korean, on the other hand argue that allocutive morphemes are the realization of a head that contains a [status] feature which encodes social aspects of the speaker-addressee relationship. We follow Portner et. al’s proposal that in both languages, allocutive morphemes appear as a particle realizing an addressee-marking head in the clausal spine. We spell out the details of this in section 3.

2.5 An excursus on “optionality”

Zuberoan, Conservative Southern Basque, Korean, Tamil and Innovative Basque have been argued to be languages where allocutive marking must appear if the discourse conditions are met, such as the presence of a specific addressee/interlocutor in a formal situation in Korean (see Portner et al. (2019)). Above in section 2.1.2, in Tamil example (11), we noticed that the allocutive morpheme -ŋgæ is used with a polite addressee. Moreover, in a context where there is a polite addressee and we have to say ‘thanks a lot’, only (49a) is possible. (49b) is ill-formed in such a context. Allocutive marking is obligatory in all these languages.

    1. (49)
    1. a.
    1.   rombaa
    2.   very
    1. thanks-ŋgæ
    2. thanks-alloc
    1.   ‘Thanks a lot’
    1.  
    1. b.
    1. *rombaa
    2.   very
    1. thanks
    2. thanks
    1. (to a polite addressee)
    2.  

In other allocutive varieties, on the other hand, allocutive marking is typically described as “optional”. For example, in example (8), we saw three ways to say ‘Santee runs’ in Magahi depending on the honorificity level of the addressee of the sentence. There is, in addition, a fourth way to say that sentence, with just subject agreement, that is, without the allocutive morpheme present, as shown in (50). The morpheme -ai indicates that the subject is third person and non-honorific to the speaker. (50) can be uttered in the same context where (8a)-(8c) can be spoken.

    1. (50)
    1. Santee-aa
    2. Santee-nh
    1. dauR-l-ai.
    2. run-prf-3.nhs
    1. ‘Santee ran.’                                                  [spoken to anybody]

In Magahi, thus, allocutive markers are optional in all contexts with the consequence that allocutive morphemes may appear in either root clauses or lower clauses as in (51). The presence of allocutive marking, however, is said to “involve the listener in facts being related” or ask for solidarity/complicity in the events described (Alok 2021).12

    1. (51)
    1. Santee-aa
    2. Santee-nh
    1. kah-l-ai/(au)
    2. say-prf-nh.sbj/nh.sbj;nh.alloc
    1. ki
    2. comp
    1. [Bantee-aa
    2. Bantee-nh
    1. bhag
    2. escape
    1. ge-l-ai/(au)].
    2. go-prf-nh.sbj/nh.sbj;nh.alloc
    1. ‘Santee said that Bantee ran away.’                                                  (said to a friend)

Like Magahi allocutive agreement morphemes, Galician allocutive clitics are also “optional” as shown in (52). Moreover, like in Magahi, when Galician allocutive clitics are used in narrative contexts, they implicate the listener in facts being related and/or are interpreted as a request for solidarity/empathy (Álvarez et al. 1986; Uriagereka 1995b; Haddican 2019; Huidobro 2022).

    1. (52)
    1. Creo(-vos)
    2. think.1sg(-2pl.fam.dat)
    1. que
    2. that
    1. (vos)
    2. (2pl.fam.dat)
    1. está
    2. cop
    1. aquí.
    2. here
    1. ‘I think that he/she/it is here.’

As Haddican (2019) notes, this empathy-request interpretation appears to be related to allocutive morpheme placement in that speakers prefer to place the allocutive clitic in clauses containing new information. In a context like (53), for example, in which the new information in the response falls in the higher clause, speakers prefer placement of the allocutive clitic in the higher clause.

    1. (53)
    1. Q:
    1. Quén
    2. who
    1. cree
    2. thinks
    1. que
    2. that
    1. vai
    2. it.goes
    1. chover?
    2. rain
    1. ‘Who thinks that it will rain?
    1.  
    1. A:
    1. Xoan
    2. Xoan
    1. cree-(che)
    2. thinks-2sg.fam.dat
    1. que
    2. that
    1. (?che)
    2. 2sg.fam.dat
    1. vai
    2. it.goes
    1. chover.
    2. rain
    1. ‘Xoan thinks that it will rain.’
    2. (Haddican 2019)

Likewise, in a context like (54) in which the new information in the response is in the lower clause, speakers have some preference for placement of the allocutive clitic in the lower clause.

    1. (54)
    1. Q:
    1. Que
    2. what
    1. cree
    2. thinks
    1. Xoan?
    2. Xoan
    1. ‘What does Xoan think?’
    1.  
    1. A:
    1. Xoan
    2. Xoan
    1. cree-(?che)
    2. thinks-2sg.fam.dat
    1. que
    2. that
    1. (che)
    2. 2sg.fam.dat
    1. vai
    2. it.goes
    1. chover.
    2. rain
    1. ‘Xoan thinks that it will rain.’
    2. (Haddican 2019)

Similarly, root structures like e che, ‘it’s that’, which resist a focus interpretation are often reported to be less than fully natural with allocutive clitics.13

    1. (55)
    1. E-(??che)
    2. it.is-2sg.fam.dat
    1. que
    2. that
    1. (che)
    2. 2sg.fam.dat
    1. está
    2. he/she.is
    1. aquí.
    2. here.
    1. ‘It’s that he’s here.’

A different pragmatic effect on allocutive morpheme placement is observed in Japanese by Yamada (2019), who notes that speakers may vary their use of honorifying allocutive morphemes within a single conversational turn to a single addressee. In particular, Yamada (2019), discusses a conversational turn by a fictitious “dissolute student” addressing their teacher in (56).

    1. (56)
    1. a.
    1. Ore
    2. I
    1. zyugyoo-nante
    2. class-top
    1. de-taku
    2. attend-want
    1. nai-yo.
    2. neg-sfp
    1. ‘I do not want to attend the class.’
    1.  
    1. b.
    1. Kagaku-no
    2. chemistry-nom
    1. sensei-no
    2. teacher-gen
    1. hanasi
    2. speech
    1. tumannai-si.
    2. boring-sfp
    1. ‘What the chemistry teacher teaches us is boring.’
    1.  
    1. c.
    1. Geemu
    2. game
    1. si-te
    2. do-cvb
    1. r-u
    2. prog-s
    1. hoo-ga
    2. way-nom
    1. zutto
    2. far
    1. masi.
    2. better
    1. ‘Playing video games is far better.’
    1.  
    1. d.
    1. Ore
    2. I
    1. ie-ni
    2. home-to
    1. kaeri-mas-u.
    2. return-hon-prs
    1. ‘I will go home.’
    2. (adapted from Yamada (2019)

What changes, if the speaker in (56) uses -mas- in a greater number of contexts in which it may appear, is not meaning related to a empathy/solidarity request, but rather, according to Yamada, the overall honorific interpretation, i.e. greater respect shown toward the interlocutor.14 Yamada notes that this is somewhat akin to the enhancement effect achieved by repeating expressive elements like damn as in (57).15

    1. (57)
    1. a.
    1. Damn, I left my keys in the car.
    1.  
    1. b.
    1. Damn, I left my damn keys in the car.
    1.  
    1. c.
    1. Damn, I left my damn keys in the damn car.
    2. (Potts 2007)

Punjabi is another language where the allocutive markers are said to be optional. However, unlike, in Magahi, Galician and Japanese no additional meaning such as a solidarity/empathy request or a cumulative politeness effect has been reported in the literature. Thus the difference between (58a) and (58b) is that there is no reference of the addressee in (58b) (Kaur 2020: 08).

    1. (58)
    1. a.
    1. aman
    2. Aman.nom
    1. kitaab
    2. book
    1. paRh
    2. read
    1. reyaa
    2. prog.m.sg
    1. je
    2. alloc.pl
    1. ‘Aman is reading a book.’
    1.  
    1. b.
    1. aman
    2. Aman.nom
    1. kitaab
    2. book
    1. paRh
    2. read
    1. reyaa
    2. prog.m.sg
    1. e
    2. be.prs.3sg
    1. ‘Aman is reading a book.’

In the sample of allocutive varieties considered here, the presence of allocutive marker in optional allocutive languages depends in part on pragmatic factors that vary across varieties (even though in Punjabi there has been no additional meaning reported). Whether some or all of these kinds of pragmatic meaning should be represented syntactically and whether such structure should be invoked in modeling these pragmatic effects on allocutivity is an issue beyond the scope of this article (see, though, Alok (2021), Yamada (2019) for some discussion). We set these issues aside in the remaining discussion.

3 Ways of licensing addressees

We summarize in Table 4 the data presented in the preceding discussion. The column labeled “Position” in 4, uses labels “C” and “T” for convenience to represent the relative positions of allocutive morphemes described above. In the following discussion, however, where we propose a set of formal options for expressing this variability, we assume a split CP (Rizzi 1997) and the neo-performative hypothesis which takes speaker and addressee speech act roles to be syntactically represented in the clause (Speas & Tenny 2003; Haegeman & Hill 2013).

Table 4

Embedding restrictions, interactions in C, and first-merged position of allocutive morpheme for nine allocutive varieties.

Variety Root-embedded effects Interaction in C-domain Position Morpheme type
Korean Root only Clause type and allocutive features spellout together alloc>c Particle
Zuberoan Root only Under C in declaratives but not in interrogatives c>alloc>t Clitic
Cons. S. Basq. Root only None c>alloc>t Clitic
Inn. S. Basq. Root, embedded None c>alloc>t Clitic
Galician Root, embedded None c>alloc>t>alloc Clitic
Magahi Root, embedded None c>alloc>t Agr
Tamil Root, embedded None alloc>c>alloc Agr
Japanese Root, embedded Under certain Cs in embedded contexts t>alloc Particle

A maximally desirable state of affairs would be one in which all four of the variable properties summarized in Table 4 could be expressed in terms of a single abstract locus of variation. Let us consider, in this light, the first two kinds of variation introduced above—root restrictions and the position of the allocutive morpheme in the clausal spine. From the perspective of the fairly well developed line of research that has modeled embedded/root asymmetries principally as a function of “truncation” of the clausal sequence—i.e. where the functional sequence in embedded domains is topped (Julien 2009; Wiklund et al. 2009; Haegeman & Hill 2013; Elordieta & Haddican 2018)—one possible understanding of the variation in root-restrictions across languages is that it is a consequence of the different first-merged positions of the allocutive morpheme. Specfically, let us suppose that languages without root-restrictions differ from Korean in that the allocutive morpheme is merged in a lower position present in embedded clauses (Alok 2021).16 Without further qualification, however, this approach is insufficient to capture facts from Basque dialects for two reasons. First, as just discussed, allocutive morphemes in conservative dialects are merged in a lower position than Korean, but are restricted to root contexts. A lower merged position, therefore, does not suffice to render these morphemes embeddable. Moreover, innovative and conservative dialects show no difference in the behavior of allocutive morphemes that would suggest different merged positions for these morphemes. These facts, then, suggest that varying the structural position of the allocutive morpheme is insufficient to express differences in allocutive morpheme embeddability across languages. Additionally, Table 4 suggests no relationship between either of these two properties and morpheme type or the kinds of interactions in C the allocutive morpheme participates in. For example, Japanese allocutive particles as described by Yamada may, as in Galician, appear TP-internally but, unlike in Galician, spell out as a particle rather than as a clitic, and are sensitive to complementizer selection.

In light of these facts, we propose that the kinds of variation observed across allocutive varieties motivate a richer set of representational options which we spell out in the remaining discussion. In particular, we propose that the sole formal property unifying allocutive varieties is the overt realization of an agreement relation between an Addressee DP and a head or pronoun in its local domain. Specifically, following Tsoulas & Kural (1999), Baker (2008), Zanuttini (2008), Kratzer (2009), Collins & Postal (2012) and Portner et al. (2019) among others, we assume that representations of utterances contain (usually silent) Addressee and Speaker DPs in the functional sequence of the clause, and that first- and second-person pronouns acquire their speech act interpretation through binding by these, as schematized in (59), for a second person pronoun.

    1. (59)
    1. A second person pronoun
    2. [cP Speaker-DP Addressee-DPk c … [XP DPk X ] ]

Portner et al. (2019) extend this framework to model honorification/politeness marking in Korean. Specifically, Portner et al. (2019) propose that Speaker and (Interlocutor-)Addressee operators are introduced by a head—c—the denotation of which is a function specifying the hierarchical relationship between these two arguments, modelled as feature status. Second-person pronouns bound by the Addressee DP via c acquire both person and status features through operator-variable agreement—an agreement operation parasitic on the binding relation. We schematize this in (60), where a person pronoun acquires politeness and person values via binding by c. (Here, η ϵ {≤, <, =, >, ≥}.)

    1. (60)
    1. Portner et al.’s (2019) structure of politeness

An alternative implementation of Portner et al.’s core idea is that the syntax of politeness interpretation is structurally akin to that of tense/aspect interpretation in a framework that takes the latter to reflect representations of intervals among time-denoting DPs. In particular, Demirdache & Uribe-Etxebarria (2000; 2007; 2014), propose that DPs denoting utterance time, event time and assertion time are introduced as arguments of temporal Ps. Tense and aspect interpretations are intervals denoted by the relations among these DPs determined by different (possibly silent) temporal prepositions present. (61), for example, adapted from Demirdache & Uribe-Etxebarria (2014), shows the temporal ordering of utterance time relative to assertion time.

    1. (61)
    1. Syntax of temporal intervals

From the perspective of Portner et al.’s proposal, a possibility to be considered is that the syntax of utterance anchoring for politeness is parallel to the syntax of temporal anchoring within Demirdache and Uribe-Etxebarria’s framework. That is, Portner et al’s c is a head of category P, which establishes a politeness relation between a Speaker DP, introduced as its specifier and an Addressee DP in its immediate scope. (62), for example, corresponds to politeness interpretations where the Speaker’s status is below, above, or on a par with that of the Addressee, depending on the P morpheme present.17 The proposal to follow will not require us to choose between these two implementations, and we set the issue aside in the remaining discussion.

    1. (62)
    1. Syntax of status intervals (politeness)

We propose that the kinds of variation among allocutive varieties introduced in section 2 reflect the following four representational differences.

  1. Different loci of introduction of the Addressee DP. We propose that the Addressee DP of (59), is universally introduced in cP—a projection merged only in root domains, following Portner et al. (2019)—but may also be introduced in a lower position, which we take to be Fin (Alok 2020; 2021). The motivation for this proposal is twofold. First, we take truncation to be the most promising formal avenue for modeling the cross-linguistic difference in root-restrictions, there being no evidence as far as we are aware favoring alternative possibilities including operator intervention (Haegeman & Ürögdi 2010; Haegeman 2013). Second, surface morpheme order facts correlate partially with the presence/absence of root-restrictions: all varieties permitting allocutivity in embedded domains also have the property that the allocutive morpheme sits CP-internally, rather than in the very high position suggested by Korean. (Recall that the relationship is not biconditional, since some varieties with root restrictions, such as conservative Basque, place allocutive morphemes CP-internally.)

  2. Presence/absence of a head introducing an allocutive pronoun. Variation in allocutive morpheme placement reflects not just different positions in which a silent Addressee DP may be introduced, but also the variable presence of an applicative-like projection, here labeled AddrP, where an allocutive pronoun may be introduced. As noted in section 2.4, in Galician and Basque, where the allocutive marker takes the form of clitic pronouns, these forms behave similarly to applicative datives elsewhere.

  3. (Non-)silence of Addressee and/or its targets of agreement. Languages also vary in which of the relevant elements of (59) may be non-silent. In some languages, the signal of allocutive agreement is the exponence of a bound pronoun, realized as a clitic (e.g., Basque, Galician), while in other languages it is the head that is realized as a particle (e.g., Korean, Japanese) or as an exponence of agreement with the Addressee DP of its specifier (e.g., Magahi, Punjabi, Tamil).

  4. Interaction with other C-field material. Sensitivity of allocutive morphology to clause type or complementizer morpheme reflects a dependency relation between the head hosting the allocutive morpheme and other C-field elements. We schematize this interaction as in (63), where Force is taken to spell out the complementizer and to be the locus of clause typing features (Rizzi 1997).

    1. (63)
    1. Force[f] … Addr[+f]

Let us consider now how these assumptions help express the cross-linguistic variation described in section 2. We begin with Korean, where we noted that allocutive markers are found only in root contexts. We adopt Portner et al.’s analysis more or less in toto, as in (64). Here, there is a designated functional projection, which Portner et al. term a “context phrase” (cP) in the left periphery of the clause which hosts a syntactic representation of speaker and addressee and mediates between syntax and discourse. The signal of allocutive agreement in Korean is the exponence of c in the context of determined speaker-hearer relationships, as in (60), above. This head, c, is merged only in root contexts—above the position of the clause-typing morpheme in Force—and it is this property that accounts for the unavailability of allocutive marking in embeddings.

    1. (64)
    1. Korean

Next, we consider Magahi. We observed earlier that (i) allocutive agreement and subject agreement combine features for spellout, (ii) allocutive agreement is associated with finiteness—it occurs only in finite clauses—and (iii) allocutive agreement is possible in all root and embedded finite clauses. We adopt Alok’s (2021) proposal that the Addressee DP may be merged lower in FinP (see Bianchi (2003); Bhadra (2018) and Alok (2021) for a fuller discussion).18

We propose that, unlike in Korean where the allocutive morpheme is spelled out higher in the clause, in Magahi it is relatively low. Specifically, the head Fin bears a [uAddr] feature which is checked against the [iAddr] feature of the addressee DP in the spec of Fin, and it is exponence of this agreement relationship that is the allocutive signal, as shown in (65). Importantly, FinP, unlike cP, is not restricted to root contexts, which makes possible allocutive marking in embedded clauses in Magahi.

    1. (65)
    1. Magahi

In Tamil, like in Magahi, allocutive agreement can be embedded. A notable property of Tamil is that the allocutive marking can be found both above and below the complementizer (cf. (42)). That is, the positional options observed on the one hand in Korean (allocutive marking above C) and, on the other in Magahi (allocutive marking below C) are both instantiated in Tamil. Following Alok (2021), we therefore propose that the agreement relation between the Addressee DP and c and that between Addressee DP and Fin may both be realized as allocutive morphemes. That is, the possibility of having an agreeing addressee DP in two positions is what accounts for multiple allocutive marking in this language.

    1. (66)
    1. Tamil

Next, we discuss Punjabi, for which we noted the following properties. Like Magahi, allocutive agreement morphemes appear in a lower position and only in finite clauses. However, unlike Magahi, it is found only in root contexts. We propose that in Punjabi, like Magahi, the relevant agreement is realized on Fin. Punjabi differs minimally from Magahi, however, in that the feature is only licensed in the presence of c. Specifically, we propose that this variant of Fin must be in the extended projection of V containing c, where extended projection is understood as a sequence of c-selection relations in a single clausal spine the sense of Biberauer et al. (2014).19

We represent this restriction as a binary feature of Fin—[+c]—in structure below. A consequence of this is that the [uAddr] feature only appears on Fin in the presence of c, and therefore only appears in root clauses.20

    1. (67)
    1. Punjabi

So far we have mainly focused on languages where the allocutive morpheme is an exponent of agreement (Korean being an exception). We now turn to languages where this morpheme is a clitic (Galician, Basque) or a particle, like Korean (Japanese). We argue that apart from cP and FinP, there is another way in which languages may introduce allocutive morphemes, namely via an applicative projection. We begin with Galician where the morphological evidence for this proposal is clearest. In section 2.4, we observed that Galician allocutive clitics are identical in exponence and allomorphy rules to thematic datives. Following Huidobro (2022), we take the allocutive clitic to be introduced in an applicative projection, which we label AddrP. As described in section 2.3, allocutive clitics may be introduced in a high TP-external position, or in infinitival contexts, TP-internally. We schematize the first of these possibilities in (68).

    1. (68)
    1. Galician

As noted earlier, in Basque, there is no evidence suggesting any cross-dialectal difference in the position of allocutive morphemes. That is, there is no independent motivation for taking the different distributions of allocutive morphemes to correlate with a distinction in their first merged order. Rather, what we suggest is that the availability of the allocutive morpheme in different contexts is determined by whether the kind of Addr head that introduces it is licensed in the relevant context. In particular, we assume that, in all dialects, the allocutive morpheme is a clitic introduced in the specifier of a TP-external applicative head, Addr, similarly to the above proposal for Galician (Haddican 2018; Haddican & Etxeberria 2022). In Zuberoan, the most restrictive variety, the variant of Addr capable of introducing an allocutive clitic is only licensed in the local domain of c and a declarative force feature, extending the approach just proposed for Punjabi. Specifically, we propose that this variant of Addr must be in an extended projection of V containing {c, Force[Decl]}.

    1. (69)
    1. Zuberoan

Conservative Southern Basque will differ minimally from Zuberoan in that the variant of Addr that introduces the clitic will lack this Force restriction, as in (70). Finally, in Innovative Southern Basque, the Addr head is licensed in all finite contexts, as in (71). The differences in clause type restrictions and root restrictions across these dialects, therefore, do not reflect truncation, or differences in the merged position of the allocutive clitic, but rather agreement relations between the head hosting the allocutive morpheme and other C-field elements, similarly to the above proposal for Punjabi.

    1. (70)
    1. Conservative Southern Basque
    1. (71)
    1. Innovative Southern Basque

In Japanese, following Yamada (2019), we assume the allocutive morpheme reflects a low Addressee head (here labeled Addr). Again, the motivation for a low position of the allocutive morpheme in Japanese comes from the morpheme order relative to tense markers and complementizers as discussed in section 2.3. Moreover, Japanese does not show phi-feature agreement like Korean. We thus treat the allocutive morpheme -mas- as a particle i.e., a realization of a head. Departing somewhat from Yamada’s (2019) proposal, we propose that it is the head of the low AddrP that is realized as -mas-, as shown in (72).

    1. (72)
    1. Japanese

Moreover, the availability of AddrP depends on properties of the complementizer in Japanese. This becomes clearer when we consider embedding contexts in the language. We have seen that embedding of -mas- is possible under the complementizers -koto and -yooni but impossible under the complementizer to. We take this difference to be related to an additional contrast reported between -koto and -yooni on the one hand and to on the other hand in the Japanese literature. In particular, Miyagawa (2022) notes that -koto and -yooni, but not to, can turn a matrix declarative sentence into an imperative.

    1. (73)
    1. a.
    1.   Sono
    2.   that
    1. kusuri-o
    2. medicine-acc
    1. maisyoku-go
    2. each-meal-after
    1. nom-u-yooni.
    2. take-prs-yooni
    1.   ‘Take that medicine after each meal.’
    1.  
    1. b.
    1.   Sono
    2.   that
    1. kusuri-o
    2. medicine-acc
    1. maisyoku-go
    2. each-meal-after
    1. nom-u-koto.
    2. take-prs-koto
    1.   ‘Take that medicine after each meal.’
    1.  
    1. c.
    1. *Sono
    2.   that
    1. kusuri-o
    2. medicine-acc
    1. maisyoku-go
    2. each-meal-after
    1. nom-u-to.
    2. take-prs-to
    1.   ‘Take that medicine after each meal.’

Following Miyagawa (2022), we propose that complementizers -koto and -yooni have an addr (essee) features while to lacks this feature.

    1. (74)
    1. a.
    1. -koto[+addr]
    1.  
    1. b.
    1. -yooni[+addr]
    1.  
    1. c.
    1. to[-addr]

We propose that in embedded contexts, the complementizers -koto and -yooni can license AddrP, similarly to the restrictions proposed above for Galician, Basque and Punjabi. The licensing of AddrP, thereby makes possible the allocutive particle -mas- in embedded contexts as illustrated in (75). In root contexts, this Addr head determining -mas- will be licensed by the higher Addressee DP in Fin. Since to lacks the addr feature, it is unable to license the AddrP, yielding no allocutive particle.

    1. (75)
    1. Licensing of AddrP under -koto/yooni in embedded contexts

4 Conclusion

The tenor of the foregoing discussion has been negative in that it has emphasized the formal non-unity of allocutive systems. In particular, we have argued that some principal facets of cross-linguistic variation in allocutive systems seem unamenable to formal economy, i.e. with multiple surface configurations expressed in terms of one or two abstract loci of variation. Instead, we have suggested that the variation described here reflects the following four kinds of variability.

  1. Variation in positions in which a silent Addressee DP may be introduced.

  2. Variation in positions in which a bound pronoun may be introduced.

  3. Variation in which elements are non-silent. In some languages, the allocutive morpheme is the exponence of agreement between the Addressee DP and its host, while in other languages the allocutive morpheme is a bound clitic pronoun or a particle.

  4. The presence vs. absence of a licensing relation between the head introducing the allocutive morpheme and other left peripheral material including clause-typing features.

It bears noting that the last three of these kinds of variation are also observed across languages for thematic addressee DPs. The extensive literature on applicative constructions suggests considerable variation across languages in the position in which different kinds of event-participant addressees can be introduced, and different syntactic environments in which applicative heads are licensed (Marantz 1997; Cuervo 2003; Pylkkänen 2008). Similarly, variation in the (non-)silence of thematic DPs and/or a host head under agreement has been described in an extensive line of literature dating back at least to the advent of Government and Binding theory (Chomsky 1981; Jaeggli & Safir 1989; Duguine 2013; 2017). Despite, then, the considerable variation in allocutive systems described here and elsewhere in the expanding literature on this topic, we see no reason to view allocutive morphemes as more nor less varied in their distribution than what one might expect given prior work on thematic addressee DPs.

Abbreviations

Glosses in this article follow the Leipzig Glossing Rules. 1 = first person, 2 = second person, 3 = third person, acc = accusative case, alloc = allocutive, anaph = anaphor, asp = aspect, comp = complementizer, cop = copula, cvb = converb affix, dat = dative case, dec = declarative, dom = differential object marker, emp = emphatic, erg = ergative case, expl = expletive, f = feminine, fam = familiar, fut = future tense, gen = genitive case, hon=honorific, hh=high honorific, imp = imperative, int = interrogative, inf = infinitival, loc = locative, m = masculine, mal = malfactive, neg = negation, nom = nominative case, nh = nonhonorific, pl = plural, pln = plain, pol = polite, polq = polar question, prf = perfect aspect, prog = progressive, pro = pronoun, prs = present, pst = past tense, q = question particle, rel = relative, res = resultative, sbj = subject agreement, sfp = sentence final particle, sg = singular, top = topic.

Notes

  1. These are usually referred to as “honorific” particles in the literature. [^]
  2. A reviewer points out that (1b) becomes grammatical under complementizer -rako/-hago with a direct quote reading. Our focus here is on allocutive embeddings with indirect speech readings. [^]
  3. Uriagereka (1995a) calls them “colloquial clitics”. [^]
  4. Importantly, such clitics differ from ethical dative clitics in that the addressee is not interpreted as an “affectee”, nor indeed as having any relationship to the event described beyond being an audience to its description (Álvarez Blanco 1997; Uriagereka 1995a). They also differ from true ethical clitics in their (non-)participation in clitic doubling and the fact that multiple allocutive forms can appear in a single clitic cluster (Álvarez Blanco 1980; Carbón Riobóo 1995; Uriagereka 1995a; Haddican 2019; Huidobro 2022). [^]
  5. Preverbal embedded topics, however, trigger embedded proclisis. This need not concern us. See Uriagereka (2005) for discussion. [^]
  6. Unless otherwise sourced, data reported are from native speaker consultations, including, in the case of Magahi, native speaker intuitions by one of the authors. [^]
  7. It is not possible to test whether imperative clause type in itself interacts with allocutivity. Basque imperatives appear in both non-finite (i) and finite (ii) forms.
      1. (i)
      1. J-oa-n!
      2. epen-go.inf
      1. ‘Go!’
      1. (ii)
      1. H-oa!
      2. 2sg.fam-go
      1. ‘Go!’
    In the former case, allocutive marking is not possible, presumably for the independent reason that allocutive clitics, like all person clitics, are not possible in non-finite constituents. In the latter case, it is blocked presumably for the independent reason that allocutive clitics are unavailable in the presence of a second person argumental clitic, here h-. [^]
  8. Here, for convenience, both forms are glossed as “C”, however as Yamada (2019:346) notes, to and -koto are likely not of the same category since they can co-occur in certain contexts. [^]
  9. This is the case for central southern dialects. In some eastern dialects the -a/-na allomorphs appear. [^]
  10. Galician also has teista dialects in which the second person singular dative form is te (Álvarez Blanco 1994). [^]
  11. Agreement for 3rd person neuter has no number distinction. [^]
  12. Similar facts have been described for Lebanese Arabic by Haddad (2013; 2014), where a class of first or second non-thematic clitics are exponed as datives, as in (i). Haddad calls these “attitude datives” and describes them as marking an evaluation of events relative to the speaker or hearer’s evaluative position. The behavior of Levantine Arabic attitude datives in terms of root/embedded asymmetries are not well described in the literature as far as we are aware and we set them aside in the remaining discussion.
      1. (i)
      1. Ziya:d
      2. Ziad
      1. biʔadʕdʕi:-li/lak
      2. spend-1sg.dat/2sg.dat
      1. kil
      2. all
      1. waʔt-o
      2. time-his
      1. ne:yim
      2. sleeping
      1. ‘Ziad spends all his time sleeping.’
      2. (Haddad 2014)
    [^]
  13. Note that the allocutive clitic here precedes the finite verb, suggesting that the second clause is a true embedding. [^]
  14. Importantly, Yamada (2019) notes that, in the case of allocutive particles, honorific interpretation does not always increase with the addition of further allocutive morphemes. In (i), for example, an additional allocutive morpheme, -desi- is determined and the utterance is not interpreted as belonging to a higher honorific level than one with only one morpheme. See Yamada (2019), for discussion.
      1. (i)
      1. Hasiri-mas-en
      2. run-hona-neg
      1. desi-ta.
      2. hona-pst
      1. ‘(I) did not run.’
    [^]
  15. Similarly, in Thai, Iwasaki & Ingkaphirom Horie (2000) discuss “adaptive speech level” contexts where conversational participants may vary their use of allocutive particles for sociolinguistic reasons. In particular Iwasaki & Ingkaphirom Horie (2000) discuss the example in (i), taken from natural conversation in which a student is discussing her recollection of an Earthquake in Thailand, to a more senior interlocutor, to whom showing deference is expected. The speaker starts off using the intermediate formality marker hà, switches to zero marking for several turns and then finally switches to the formal marker khá, for discourse strategic reasons in Iwasaki & Ingkaphirom Horie’s (2000) analysis.
      1. (i)
      1. a.
      1. hĕn
      2. see
      1. tùk
      2. building
      1. toŋnía
      2. right.here
      1. hà
      2. hon
      1. ‘(I) saw the building right here.’                                                  [mid-level marking]
      1.  
      1.  
      1. […]
      1.  
      1. b.
      1. tùk
      2. building
      1. yày
      2. big
      1. mày
      2. new
      1. yùu
      2. exist
      1. lǝǝy
      2. emp
      1. ‘the big building was still new.’                                                           [zero marking]
      1.  
      1.  
      1. […]
      1.  
      1. c.
      1. man
      2. it
      1. tùk
      2. building
      1. yày
      2. big
      1. ŋay
      2. PP
      1. khá
      2. hon
      1. mŭan
      2. same
      1. bὲp
      2. like
      1. ‘It’s big, you know. It was like … ‘                                                      [formal marking]
      2. (adapted from (Iwasaki & Ingkaphirom Horie 2000: 538))
    [^]
  16. Alok (2021) was unaware of variation among different dialects of Basque. His analysis was based on the variety discussed in Oyharçabal (1993). [^]
  17. A question that arises is whether there is any direct morphological evidence for such a morpheme of category P, as there is in the case of Basque progressives or English a-prefixing (Demirdache & Uribe-Etxebarria 2000). From the perspective of P-incorporation approaches to have.e alternations, one possibility is that incorporation of a P head in (62) is responsible for the be>have switch in Basque auxiliary roots in the presence of Basque allocutive clitics in (i) (Rebuschi 1981; Oyharçabal 1993; Albizu 1997; Haddican 2018).
      1. (i)
      1. a.
      1. Irun-darra
      2. Irun-er
      1. na-iz.
      2. 1sg-be
      1. ‘I’m from Irun.’          (No allocutive clitic present)
      1.  
      1. b.
      1. Irun-darra
      2. Irun-er
      1. na-u-k.
      2. 1sg-have-2.sg.fam.msc
      1. ‘I’m from Irun.’              (Allocutive clitic present)
    [^]
  18. The representation of the speaker and addressee coordinates in FinP have different motivations than their representation in cP (Portner et al. 2019) or SAP (Speas & Tenny 2003). Their presence in FinP is related to finiteness: the speaker and addressee coordinates of FinP are present only in finite clauses. Moreover, the speaker and addressee coordinates are co-referential with the speaker and addressee coordinates found in cP/SAP in general cases. However, they can also be contra-indexed in special cases (see Bhadra’s (2018) discussion on evidentiality in Bangla). [^]
  19. With spine defined, following Biberauer et al. (2014) as such:
      1. (i)
      1. A sequence of nodes Σ = (α1, …, αn) is a spine if and only if:
        • i.   αn is a lexical category and an X0;
        • ii.   for all αi<n in Σ, either:
        •       αi is a projection of and immediately dominates αi+1, or
        •       αi is an X0 and the sister of αi+1.
    [^]
  20. We take temporal clauses and and complements of verbs of saying which show allocutive marking to be embedded root contexts with a cP layer. [^]

Acknowledgements

We thank three anonymous reviewers for detailed and thoughtrful comments that improved the paper considerably. We also thank audiences at the 2021 LSA annual meeting and the 9th Cambridge Comparative Syntax conference (CamCoS 9) where this work was presented. We are grateful to the editors of this volume, Paul Portner, Miok Pak and Raffaella Zanuttini, whose comments and guidance were invaluable. Many thanks also to Sam Al Khatib, Rosario Álvarez, Xabier Azkue, Francisco Dubert, Maia Duguine, Irantzu Epelde, Ricardo Etxepare, Youssef Haddad, Aritz Irurtzun, Ghada Khattab, Beñat Oyharçabal, Xulio Sousa, Alina Tigau, Juan Uriagereka, Alba Aguete, Daniel Amarelo, Maria Beatriz Domínguez Oroña, Manuel Dopazo, Clara Lago Caamaño, Reyes Rodríguez Rodríguez, the Instituto da Lingua Galega and IKER UMR 5478-CNRS. All errors are ours.

Funding information

This work is supported by the Spanish Ministry of Science, Innovation and Technology (PGC2018-096380-B-100), through the Bas&Be (Basque and Beyond) project.

Competing interests

The authors have no competing interests to declare.

References

Albizu, Pablo. 1997. Generalized Person-Case Constraint: A case for a syntax-driven inflectional morphology. In Mendikoetxea, Amaya & Uribe-Etxebarria, Myriam (eds.), Theoretical issues at the morphology-syntax interface, 1–34. Euskal Herriko Unibertsitatea.

Alok, Deepak. 2020. Speaker and Addressee in natural language: Honorificity, indexicality and their interaction in Magahi: Rutgers dissertation.

Alok, Deepak. 2021. The morphosyntax of Magahi addressee agreement. Syntax 24(3). 263–296. DOI:  http://doi.org/10.1111/synt.12213

Alok, Deepak & Baker, Mark. 2018. On the mechanics (syntax) of indexical shift: Evidence from allocutive agreement in Magahi, MS. Rutgers.

Álvarez Blanco, Rosario. 1980. O pronome persoal en galego: Universidade de Santiago de Compostela dissertation.

Álvarez Blanco, Rosario. 1994. Secuencias de pronomes átonos en galego moderno. In Lorenzo, R. (ed.), Actas do xix congreso internacional de lingüística e filoloxía románicas, santiago de compostela 1989, vol. VI, 247–265. A Coruña: Fundación Pedro Barrié de la Maza, Conde de Fenosa.

Álvarez Blanco, Rosario. 1997. O complemento de solidariedade: A complicidade entre os interlocutores. In Fernaández Salgado, Benígno (ed.), Actas do IV Congreso Internacional de Estudios Galegos, 37–53. Centre for Galician Studies.

Álvarez, Rosario & Regueira, Xosé Luís & Monteagudo, Henrique. 1986. Gramática Galega. Galaxia.

Antonov, Anton. 2015. Verbal allocutivity in a crosslinguistic perspective. Linguistic Typology 19(1). 55–85. DOI:  http://doi.org/10.1515/lingty-2015-0002

Arregi, Karlos & Nevins, Andrew. 2012. Morphotactics: Basque auxiliaries and the structure of spellout, vol. 86. Dordrecht: Springer. DOI:  http://doi.org/10.1007/978-94-007-3889-8_1

Azkue, María Resurrección. 1923. Morfología vasca (Gramática básica dialectal del Euskera). Bilbao: Editorial La Gran Enciclopedia Vasca 2nd edn.

Baker, Mark C. 2008. The syntax of agreement and concord, vol. 115. Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511619830

Bhadra, Diti. 2018. Evidentials are syntax-sensitive: The view from Bangla. Glossa: a journal of general linguistics 3(1). DOI:  http://doi.org/10.5334/gjgl.449

Bianchi, Valentina. 2003. On finiteness as logophoric anchoring. In Guéron, Jacqueline & Tasmovski, Liliane (eds.), Temps et point de vue/tense and point of view, 213–246. Nanterre: Université Paris X.

Biberauer, Theresa & Holmberg, Anders & Roberts, Ian. 2014. A syntactic universal and its consequences. Linguistic Inquiry 45. 169–225. DOI:  http://doi.org/10.1162/LING_a_00153

Campos, Héctor. 1989. Clitic position in modern Gallegan. Lingua 77(1). 13–36. DOI:  http://doi.org/10.1016/0024-3841(89)90037-5

Carbón Riobóo, María de la Soledad. 1995. Ethical datives and solidarity clitics in Galician. MA Thesis, University of Kansas.

Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.

Collins, Chris & Postal, Paul Martin. 2012. Imposters: A study of pronominal agreement. MIT Press. DOI:  http://doi.org/10.7551/mitpress/9780262016889.001.0001

Cuervo, María Cristina. 2003. Datives at large: Massachusetts Institute of Technology dissertation.

Demirdache, Hamida & Uribe-Etxebarria, Myriam. 2000. The primitives of temporal relations. In Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 157–186. Cambridge, Massachusetts: MIT Press.

Demirdache, Hamida & Uribe-Etxebarria, Myriam. 2007. The syntax of time arguments. Lingua 117(2). 330–366. DOI:  http://doi.org/10.1016/j.lingua.2005.07.006

Demirdache, Hamida & Uribe-Etxebarria, Myriam. 2014. Aspect and temporal anaphora. Natural Language & Linguistic Theory 32(3). 855–895. DOI:  http://doi.org/10.1007/s11049-014-9231-2

Duguine, Maia. 2013. Pro-drop and linguistic variation: A minimalist analysis: University of the Basque Country (UPV/EHU) & University of Nantes dissertation.

Duguine, Maia. 2017. Reversing the approach to null subjects: A perspective from language acquisition. Frontiers in Psychology 8. 27. DOI:  http://doi.org/10.3389/fpsyg.2017.00027

Elordieta, Arantzazu & Haddican, Bill. 2018. Truncation feeds intervention: Two clausetype effects in Basque. Natural Language & Linguistic Theory 36(2). 403–443. DOI:  http://doi.org/10.1007/s11049-017-9381-0

Emonds, Joseph Embley. 1969. Root and structure-preserving transformations.: Massachusetts Institute of Technology dissertation.

Haddad, Youssef A. 2013. Pronouns and intersubjectivity in Lebanese Arabic gossip. Journal of Pragmatics 49(1). 57–77. DOI:  http://doi.org/10.1016/j.pragma.2012.12.006

Haddad, Youssef A. 2014. Attitude datives in Lebanese Arabic and the interplay of syntax and pragmatics. Lingua 145. 65–103. DOI:  http://doi.org/10.1016/j.lingua.2014.03.006

Haddican, Bill. 2018. The syntax of Basque allocutive clitics. Glossa 3(1). 1. DOI:  http://doi.org/10.5334/gjgl.471

Haddican, Bill. 2019. Toward a unified syntax of Addressee in C: Evidence from Galician solidarity datives. Revue Roumaine de Linguistique 64(4). 373–386.

Haddican, Bill & Etxeberria, Urtzi. 2022. Embedded allocutivity in Basque. Syntax. DOI:  http://doi.org/10.1111/synt.12238

Haegeman, Liliane. 2006. Conditionals, factives and the left periphery. Lingua 116(10). 1651–1669. DOI:  http://doi.org/10.1016/j.lingua.2005.03.014

Haegeman, Liliane. 2013. The syntax of adverbial clauses. In Anderson, Stephen & Moeschler, Jacques & Reboul, Fabienne (eds.), The language-cognition interface, vol. 45 (Langues et Cultures), 135–165. Droz.

Haegeman, Liliane & Hill, Virginia. 2013. The syntactization of discourse. In Folli, Raffaella & Sevdali, Christina & Truswell, Robert (eds.), Syntax and its limits, 370–390. Oxford University Press. DOI:

Haegeman, Liliane & Ürögdi, Barbara. 2010. Referential CPs and DPs: An operator movement account. Theoretical Linguistics 36(2–3). 111–152. DOI:  http://doi.org/10.1515/thli.2010.008

Heycock, Caroline. 2006. Embedded root phenomena. In Everaert, Martin & Van Riemsdijk, Henk (eds.), The blackwell companion to syntax, 174–209. Oxford: Blackwell. DOI:  http://doi.org/10.1002/9780470996591.ch23

Hooper, Joan B. & Thompson, Sandra A. 1973. On the applicability of root transformations. Linguistic Inquiry 4(4). 465–497.

Hualde, José Ignacio. 2003. Finite forms. In Hualde, José Ignacio & Ortiz de Urbina, Jon (eds.), A grammar of Basque, 211–252. Berlin: Walter de Gruyter. DOI:  http://doi.org/10.1515/9783110895285

Huidobro, Susana. 2009. Datives as “concordia” case: evidence from Spanish and Galician. Paper presented at the 19th Colloquium on Generative Grammar, April 2009, Vitoria-Gasteiz.

Huidobro, Susana. 2018. Allocutive clitics in Galician. MS. City Univerisity of New York.

Huidobro, Susana. 2022. Non-argumental clitics in Spanish and Galician: A case study of the distribution of solidarity and ethical clitics: City Univerisity of New York dissertation.

Iwasaki, Shoichi & Ingkaphirom Horie, Preeya. 2000. Creating speech register in Thai conversation. Language in Society 29(4). 519–554. DOI:  http://doi.org/10.1017/S0047404500004024

Jaeggli, Osvaldo & Safir, Kenneth J. 1989. The null subject parameter and parametric theory. In The null subject parameter, 1–44. Springer. DOI:  http://doi.org/10.1007/978-94-009-2540-3_1

Julien, Marit. 2009. Embedded V2 in Norwegian and Swedish. Working Papers in Scandinavian Syntax 80. 103–161.

Kaur, Gurmeet. 2020. On the syntax of addressee in imperatives: insights from allocutivity. Glossa: a journal of general linguistics 5(1). 1–44. DOI:  http://doi.org/10.5334/gjgl.1083

Kaur, Gurmeet & Yamada, Akitaka. 2019. Embedded allocutivity and its reference. Handout at Person and Perspective workshop, May 4, 2019.

Kim, Sanghee. 2019. Allocutive agreement in korean under cyclic agree. Proceedings of the Linguistic Society of America 4(1). 56–1. DOI:  http://doi.org/10.3765/plsa.v4i1.4543

Kratzer, Angelika. 2009. Making a pronoun: Fake indexicals as windows into the properties of pronouns. Linguistic Inquiry 40(2). 187–237. DOI:  http://doi.org/10.1162/ling.2009.40.2.187

Laka, Itziar. 1993. The structure of inflection: A case study in X0 syntax. In Hualde, José Ignacio & de Urbina, Jon Ortiz (eds.), Generative studies in Basque linguistics, 21–70. Amsterdam/Philadelphia: John Benjamins. DOI:  http://doi.org/10.1075/cilt.105.02lak

Longa, Victor M. & Lorenzo, Guillermo. 2001. Universal constraints on ‘superfluous’ elements: The case of Galician ‘Arb Che’. Current Issues in Spanish Syntax and Semantics 53. 175. DOI:  http://doi.org/10.1515/9783110850536.175

Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. University of Pennsylvania working papers in linguistics 4(2). 201–225.

McFadden, Thomas. 2020. The morphosyntax of allocutive agreement in Tamil. In Smith, Peter W. & Mursell, Johannes & Hartmann, Katharina (eds.), Agree to agree: Agreement in the minimalist programme. 391–424. Berlin: Language Science Press.

Miyagawa, Shigeru. 2012. Agreements that occur mainly in the main clause. In Aelbrecht, Lobke & Haegeman, Liliane & Nye, Rachel (eds.), Main clause phenomena: New horizons, 79–112. Amsterdam: Benjamins. DOI:  http://doi.org/10.1075/la.190.04miy

Miyagawa, Shigeru. 2013. Surprising agreements at T and C. M.s. MIT.

Miyagawa, Shigeru. 2017. Agreement beyond phi. Cambridge, Mass: MIT Press. DOI:  http://doi.org/10.7551/mitpress/10958.001.0001

Miyagawa, Shigeru. 2022. Syntax in the treetops. Cambridge, Mass: MIT Press. DOI:  http://doi.org/10.7551/mitpress/14421.001.0001

Ortiz de Urbina, Jon. 1989. Some parameters in the grammar of Basque. Dordrecht: Foris Publications. DOI:  http://doi.org/10.1515/9783110876741

Oyharçabal, Bernard. 1993. Verb agreement with non arguments: On allocutive agreement. In Hualde, José Ignacio & Ortiz de Urbina, Jon (eds.), Generative studies in basque linguistics, 89–89. Philadelphia/Amsterdam: John Benjamins. DOI:  http://doi.org/10.1075/cilt.105.04oyh

Pak, Miok D. 2017. Towards understanding the syntactic representation of honorifics in Korean. Korean Linguistics 17(2). 132–166. DOI:  http://doi.org/10.1075/kl.17.2.01pak

Portner, Paul & Pak, Miok & Zanuttini, Raffaella. 2019. The addressee at the syntaxsemantics interface: Evidence from politeness and speech style. Language 95. 1–36. DOI:  http://doi.org/10.1353/lan.2019.0008

Potts, Christopher. 2007. The expressive dimension. Theoretical linguistics 33(2). 165–198. DOI:  http://doi.org/10.1515/TL.2007.011

Preminger, Omer. 2009. Breaking agreements: Distinguishing agreement and clitic doubling by their failures. Linguistic Inquiry 40(4). 619–666. DOI:  http://doi.org/10.1162/ling.2009.40.4.619

Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, Mass.: MIT press. DOI:  http://doi.org/10.7551/mitpress/9780262162548.001.0001

Raposo, Eduardo P. & Uriagereka, Juan. 2005. Clitic placement in Western Iberian: A minimalist view. In The Oxford handbook of comparative syntax.

Rebuschi, Georges. 1981. Autour des formes allocutives du Basque. In Euskalarien nazioarteko jardunaldiak, 307–321. Euskaltzaindia.

Rezac, Milan. 2006. Agreement displacement in Basque: Derivational principles and lexical parameters. Ms., University of the Basque Country, Vitoria-Gasteiz.

Rezac, Milan. 2008. Phi-agree and theta-related case. In Harbour, Daniel & Adger, Daniel & Béjar, Susana (eds.), Phi theory: Phi-features across interfaces and modules, 83–129. Oxford: Oxford University Press.

Rezac, Milan & Albizu, Pablo & Etxepare, Ricardo. 2014. The structural ergative of Basque and the theory of case. Natural Language & Linguistic Theory 32(4). 1273–1330. DOI:  http://doi.org/10.1007/s11049-014-9239-7

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Haegeman, Liliane (ed.), Elements of grammar, 281–337. Dordrecht: Kluwer Academic Publishers. DOI:  http://doi.org/10.1007/978-94-011-5420-8_7

Speas, Peggy & Tenny, Carol. 2003. Configurational properties of point of view roles. In Di Sciullo, Anne Marie (ed.), Asymmetry in grammar, vol. 1, 315–345. Amsterdam/Philadelphia: John Benjamins. DOI:  http://doi.org/10.1075/la.57.15spe

Sundaresan, Sandhya. 2018. An alternative model of indexical shift: Variation and selection without context-overwriting, Ms. University of Leipzig.

Travis, Lisa. 1984. Parameters and effects of word order variation.: MIT dissertation.

Tsoulas, George & Kural, Murat. 1999. Indexical pronouns as bound variables. In Proceedings of WCCFL, vol. 18. 545–57.

Uriagereka, Juan. 1995a. Aspects of the syntax of clitic placement in Western Romance. Linguistic Inquiry 26(1). 79–123.

Uriagereka, Juan. 1995b. An F position in Western Romance. Discourse configurational languages. 153–175.

Wiklund, Anna-Lena & Bentzen, Kristine & Hrafnbjargarson, Gunnar Hrafn & Hróarsdóttir, Þorbjörg. 2009. On the distribution and illocution of V2 in Scandinavian thatclauses. Lingua 119(12). 1914–1938. DOI:  http://doi.org/10.1016/j.lingua.2009.03.006

Yamada, Akitaka. 2019. The syntax, semantics and pragmatics of japanese addresseehonorific markers: Georgetown University dissertation.

Zanuttini, Raffaella. 2008. Encoding the addressee in the syntax: Evidence from English imperative subjects. Natural Language and Linguistic Theory 26(1). 185–218. DOI:  http://doi.org/10.1007/s11049-007-9029-6