May 9th, 2020
Georgian is a Kartvelian language spoken by about four million people, mostly in the Republic of Georgia. It is renowned for its flexible word order, threefold casemarking split, and uncommon strategies for marking focus. In this paper, I will analyze two approaches to focus in Georgian Wh-questions, one from Asatiani and Skopeteas (2012) and another from Borise (2019). Asatiani and Skopeteas (2012) provide a survey of information structure in Georgian with some interesting examples of declarative answers with appropriate and inappropriate questions. Borise (2019) gives an overview of Georgian syntax of Wh-phrases and constituents that correspond to Wh-phrases in questions. I will give a brief analysis of the two at the end of this paper.
All six possible word order permutations for the major constituents S, V, and O are possible in Georgian. Deviations from subject-first word order requires a contextual trigger, but all word order permutations are valid and generally occur.
Word order has an effect on focus marking in responses. SOV order appears in responses that license narrow focus on the object (as in 9a), while OSV appears in responses licensing narrow focus on the subject (like in 9b).
(9a) #‘Who bought tomatoes?’ ‘What did Nino buy?’ nino-m i-q’id-a p’amidori Nino-ERG SV-buy-AOR.S.3.SG tomato-NOM “Nino bought tomatoes.” (9b) ‘Who bought tomatoes?’ #‘What did Nino buy?’ p’amidor-i i-q’id-a nino-m tomato-NOM SV-buy-AOR.S.3.SG Nino-ERG “Nino bought tomatoes.”
These focus restrictions are also present in questions, not just responses. For example:
(12) a. vin i-q’id-a p’amidor-i? who.ERG SV-buy-AOR.S.3.SG tomato-NOM “Who bought tomatoes?” b. * vin p’amidor-i i-q’id-a?
Asatiani and Skopeteas (2012) say that there is a requirement for Wh-elements to be adjacent to the finite verb, which also applies to negative and indefinite pronouns, as wells as phrases with only.
Asatiani and Skopeteas (2012) account for the Wh-element verb adjacency restriction through two models. They give the first model, which is a discourse configurational account, as follows:
(13) a. A constituent bearing the feature [foc] moves to spec,FP where this feature is checked. b. If a constituent occupies spec,FP, the head of this projection attracts the finite V. c. The product of these rules is the following syntactic configuration: [_fpα_i^foc [_f V_j [_VP t_i t_j ]]]
In their first model, the focused constituent α moves to the Specifier position of a functional projection fp, and the head f’ attracts the verb V.
The second model assumes the intervening non-focused material β between the focus and V relocates to a peripheral position from its previous place in the thematic layer. The second model is an interface condition account of this phenomenon.
(14) a. Foc-to-V adjacency condition: The output linearization of the syntactic rules has to meet the condition: <..., α^foc , V, ...> b. Whenever non-focused material intervenes between focus and V, this material has to leave the VP. c. The product of this rule is the following syntactic configuration: [_cp β_i [_VP α^foc t_i V]]
β can be realized to the left or the right of the thematic layer, the important thing for the interface condition account is that the non-focused material is in an adjoined position to the core clause.
These two analyses make different predictions about <α,V>. (13) predicts that the operation will only occur if constituent α has a Spec,FP-targeting focus feature. (14) does not use features at all, instead predicting the same linearization in every case where α is part of a focus domain and β is not. The first account predicts the <α,V> to surface in contexts licensing a narrow focused S, while the second account predicts <α,V> to surface anytime O is out of focus. Both accounts predict (9b), but only the second model predicts the same order in SV focus domains, like (15).
(15) ‘There is a box on the table...’ ... q’ut-s k’ac-i a-gd-eb-s. ... box-DAT man-NOM NV-throw-THM-PRS.S.3.SG ‘...a man is throwing the box.’
Georgian allows for postverbal material to be focused due to its mixed OV/VO properties and the fact that verbs can be fronted without a contextual trigger (Asatiani and Skopeteas, 2012). This allows for SVO and OVS sentences to be felicitous in object focus or subject focus contexts and makes postverbal focus free, such as in (16a), where both the object p’amidor-i ‘tomato-NOM’ and adjunct dɣe-s ‘day-DAT’ can be focused.
(16a) nino-m i-q’id-a p’amidor-i dɣe-s Nino-ERG SV-buy-AOR.S.3.SG tomato-NOM day-DAT “Nino bought tomatoes today.” (16b) ‘Who bought tomatoes?’ ‘What did Nino buy?’ nino-m i-q’id-a p’amidori Nino-ERG SV-buy-AOR.S.3.SG tomato-NOM “Nino bought tomatoes.” (16c) ‘Who bought tomatoes?’ ‘What did Nino buy?’ p’amidor-i i-q’id-a nino-m tomato-NOM SV-buy-AOR.S.3.SG Nino-ERG “Nino bought tomatoes.”
If we are to trust Asatiani and Skopeteas (2012)’s two models, there is a rich array of possible preand post-verbal focus positions, as shown in (16b) and (16c). It appears that Georgian’s only restriction regarding focus landing sites is that if the focus is preverbal, it has to be adjacent to the finite verb.
Unlike many Indo-European languages, Georgian has no strict association between focus and pitch accents. Preverbal foci have a left-edge prosodic boundary separating them from nonfocused material and integrating both the verb and the focused material. Postverbal foci are also separated by a prosodic boundary, but unlike preverbal foci, wind up realized in a separate prosodic phrase from the verb.
Like Asatiani and Skopeteas (2012), Borise (2019) identifies the position immediately before the verb as the place where narrow focus and Wh-phrases appear. She gives the following example as evidence:
(1) a. Gušin dila-s bebia ra-s a-lag-eb-d-a? yesterday morning-DAT grandma.NOM what-DAT VER-clean-SF-SM-IPFV.3SG ‘What did grandma clean yesterday morning?’ b. Gušin dila-s bebia samzareulo-s a-lag-eb-d-a. yesterday morning-DAT grandma.NOM kitchen-DAT VER-clean-SF-SM-IPFV.3SG ‘Grandma cleaned *the kitchen* yesterday morning.’
Georgian also allows for postverbal narrow focus, which is less common in languages with such a flexible word order as Georgian.
Borise (2019) follows Rooth’s Alternative Semantics, which means that focus is understood to indicate “the presence of alternatives that are relevant for the interpretation of linguistic expressions” (Krifka, 2008, p. 247). This means that new information foci, contrastive foci, and those modified by focus-inducing particles *even* and *only* are included in the definition, as well as Wh-expressions.
According to Borise (2019), there are two main analytical approaches to focus in Georgian. The first is the Spec-Head approach, where the focal/Wh element undergoes A-bar movement to Spec,XP as the verb moves to X^0. The second is the displacement of intervening material. Borise (2019) seems to favor the former over the latter for some operations, while still noting that neither approach is mutually exclusive. She says that both strategies are required for a full account of all the preverbal facts.
Wh-words are required to be immediately preverbal in Georgian, as shown by (27) and (28). Only negation is allowed to intervene between the Wh-expression and the verb. If a Wh-expression appears in a postverbal position, it is only felicitous with an echo interpretation. Multiple Wh-expressions are grouped together in an unbreakable cluster, which is required to be immediately preverbal. Negation cannot intervene between elements of the Wh-expression block.
(27) a. Bebi-a ra-s a-lag-eb-d-a?
grandma.NOM what.DAT VER-clean-SF-SM-IPFV.3sg
'What did grandma clean?'
b. *Ra-s bebia a-lag-eb-d-a?
what-DAT grandma.NOM VER-clean-SF-SM-IPFV.3sg
('What did grandma clean?')
(28) *Bebia a-lag-eb-d-a ra-s?
grandma.NOM VER-clean-SF-SM-IPFV.3sg what-DAT
('What did grandma clean?)
Borise (2019) considers three types of narrow focus. First on the list is constituents in responses that correspond to Wh-phrases. Second are corrections, and third are constituents modified by only or even. I will focus mainly on Wh-phrases and their corresponding constituents in responses.
Preverbal narrowly focused constituents that correspond to a Wh-word must be placed immediately before the verb, or else they become infelicitous, as shown by (10):
(10) Q: ‘What did grandma clean yesterday morning?’ a. Gušin dila-s bebia samzareulo-s a-lag-eb-d-a. yesterday morning-DAT grandma.NOM kitchen-DAT VER-clean-SF-SM-IPFV.3SG ‘Grandma cleaned *the kitchen* yesterday morning.’ b. * Gušin dilas samzareulos bebia alagebda.
All types of narrow foci in responses can appear in the immediately postverbal position. There is a strong preference for the focused material to be the only material in the postverbal domain.
(11) Q: ‘What did grandma clean yesterday morning?’ a. Gušin dila-s bebia a-lag-eb-d-a samzareulo-s. yesterday morning-DAT grandma.NOM VER-clean-SF-SM-IPFV.3SG kitchen-DAT ‘Grandma cleaned *the kitchen* yesterday morning.’ (12) ??? Gušin dilas alagebda samzareulos bebia.
(11) and (12) together show that Georgian expects postverbal focus to be immediately post verbal and clause-final. There are no interpretational differences between preverbal and postverbal focus position.
Both Asatiani and Skopeteas (2012) and Borise (2019) propose two models for handling focus in Georgian. I believe they generally agree with each other regarding the general structure of the two models. They appear to disagree slightly on which is most important for narrow focus in Georgian. Asatiani and Skopeteas (2012) seem to lean more on the interface condition account (i.e. non-focused material vacates the thematic layer and moves away from the immedately pre- or postverbal position) than Borise (2019) does, though both authors are not really in favor of the Spec,XP account. Asatiani and Skopeteas (2012) differs from Borise (2019) in the discourse configurational account in that they posit a focus feature, while Borise (2019) just has the element in focus undergo A-bar movement to Spec,XP.
I think these two articles should be read together to get a full understanding of focus in questions in Georgian. Asatiani and Skopeteas (2012) sets up some of the basics of information structure and how Georgian syntax works, while Borise (2019) goes into much more detail regarding focus and Wh-questions.
Asatiani, Rusudan and Stavros Skopeteas (2012).
In: *The expression of information structure.* Chap. The information structure of Georgian.
Borise, Lena (2019). “The syntax of wh-phrases and narrow foci in Georgian.” Hosted by Magyar Tudományos Akadémia Nyelvtudományi Intézet.
http://www.nytud.hu/oszt/elmnyelv/elvonal_urali/publ/Borise2019_MS_wh_narrow_focus.pdf
Krifka, Manfred (2008). “Basic notions of information structure.” In: *Acta Linguistica Hungarica* 55.3-4, pp. 243–276.