Lecture by Prof. Idan Landau
At NELS
Silent Resumption: A New Test for Ellipsis
(Delivered at the 56th Annual Meeting of the North East Linguistics Society (NELS) at NYU)
Abstract:
Background: Establishing ellipsis, or surface anaphora, as opposed to different kinds of deep anaphora (Null Complement Anaphora=NCA or pro), has proven less straightforward than originally envisioned in Hankamer & Sag's (1976) seminal study. In fact, the analytic survey in Merchant 2013 concludes that only two reliable tests for ellipsis exist: Agreement triggered by an ellipsis-internal element on some ellipsisexternal target, or extraction from the ellipsis site. Unfortunately, these tests are unavailable in many environments suspected of being elliptical, either because the silent category is an island for extraction or it is opaque for agreement. Novel criteria for the existence of silent syntactic structure are therefore needed.
Silent resumption and the Ban on Vacuous Quantification (BVQ): The BVQ mandates that every operator bind a variable at LF (Chomsky 1982). Importantly, this representational condition does not distinguish movement-dependencies from base-generated dependencies insofar as both involve an operator in an Ā-position in need of a variable to bind. Variables might be resumptive pronoun (RP), which, in Hebrew, are found in D-linked wh-questions and in relative clauses. In (1), the genitive RP šelo ‘his’ is required by the BVQ.
(1) Eyze bamayi at lo maskima lir’ot yoter sratim *(šeloi)?
which director you.F not agree.SG.F to.see more movies his
‘Which director do you no longer agree to see movies by?’
We can now ask whether such Operator-RP dependencies are licensed even when the constituent containing the RP is unpronounced. If they are, it means that the silenced constituent has undergone genuine ellipsis, leaving its internal syntactic material intact, and in particular the (silenced) RP, required by the BVQ. If they are not licensed, we can conclude that the constituent is syntactically absent (NCA) or simplex (pro), hosting no internal RP (shading represents ellipsis).
(2) Ellipsis-Internal Resumption (EIR) test
a. ✓ Opi … [TP … V [XP … RPi …] …] ➔ Surface anaphora (ellipsis)
b. * Opi … [TP … V __ / pro …] ➔ Deep anaphora (NCA/pro)
In what follows, we apply the EIR test to various types of XPs; specifically, we use it to establish that Hebrew has genuine nP ellipsis and PP ellipsis.
Resumption inside null nP. DPs often surface with their nominal nucleus missing, sometimes along with its arguments and some adjuncts, leaving only peripheral overt material (demonstratives, possessors, modifiers, numerals, quantifiers, etc.). Rsearch indicates that languages may generate such “incomplete” nominals in two ways, corresponding to deep and surface anaphora : Empty Nouns (EN) or Elided NPs
(ENP), correspondingly (Corver and van Koppen 2011, Merchant 2014, 2023, Saab and Lipták 2016, Saab 2019). Following Merchant and Saab, I assume that both types correspond to the nP portion of the DP; while an EN consists of a bare n, inherently specified for some fatures (e.g., human, feminine), an ENP corresponds to a full nP structure, with a nominal root and its arguments/adjuncts. Crucially, ENs are pragmatically identified, while ENPs require a linguistic antecedent as well as a licensing head, as is standard in other elliptical constructions (Lobeck 1995, Saab 2019).
Because Hebrew makes ample use of ENs in pragmatically supported contexts, it is not obvious that it ever resorts to ENPs. To demonstrate that it does, we can apply the EIR test. First, observe that numerals can occur with pragmatically supported ENs (3a). However, if the EN is required to host an internal RP because an external operator needs a variable to bind (as per the BVQ), the result is ungrammatical (3b). Strikingly, the very same string with an incomplete nominal is grammatical if the linguistic context supplies an antecedent for the missing nP portion (boldfaced in (3c)). This indicates that the missing nP (3c) is an instance of ENP, containing a silent duplicate of the RP, which can then be bound by the local Ā-operator of the relative clause in the second conjunct.
(3) a. [Context: John shows me the first album of “Queens of the Stone Age”,
and then shows me the second one. I say:]
ahavti et ha-šeni.
liked.1SG ACC the-second
‘I liked the second one.’
b. * yeš la’hakot še-ahavti et ha-šeni.
there.exist bands that-liked.1SG ACC the-second
(‘There are bands that I liked the second one.’)
c. yeš la’hakot še-ahavti et ha-albom ha-rišon šela’hen,
there.exist bands that-liked.1SG ACC the-album the-first their
ve-yeš la’hakot še-ahavti et ___ ha-šeni.
and-there.exist bands that-liked.1SG ACC the-second
‘There are bands that I liked their first album,
and there are bands that I liked their second one.’
A parallel argument can be constructed with missing DPs, however, due to space limitations, I omit the data (see Landau 2018, 2021, 2025 for extensive evidence for DP elipsis in Hebrew).
Resumption inside null PP. Turning to missing PPs, the EIR test is particularly useful. Argumental PPs are very often optional, which is another way of saying – licensed as deep anaphors. The extensive literature on Argument Ellipsis focuses mostly on DP and CP arguments, providing little compelling evidence that PPs too are subject to ellipsis. Indeed, pragmatically supported missing PPs are easily found in Hebrew (4a). However, the same pragmatic context cannot salvage a null PP occuring inside a relative clause, whose operator needs a variable to bind (4b), precisely because the deep anaphor PP (probably a case of NCA) has no internal syntactic structure. Yet once an explicit linguistic antecedent is provided for the missing PP (including the internal RP), it can undergo ellipsis, introducing the necessary variable in the relative clause (4c).
(4) a. [Context: My friend and I approach the city square,
where a number of NGOs have lined up their petition booths. My friend tells me:]
ani lo xotemet ___.
I not sign.SG.F
‘I don’t sign.’
b. * yeš irgunim še-ani lo xotemet ___.
there.is organizations that-I not sign.SG.F
(‘There are organizations that I don’t sign.’)
c. yeš irgunim še-ani xotemet al ha-acumot šelahem,
there.is organizations that-I sign.SG.F on the-petitions their
ve-yeš irgunim še-ani lo xotemet ___.
and-there.is organizations that-I not sign.SG.F
‘There are organizations that I sign their petitions
and there are organizations that I don’t.’
Broader implications. The EIR test provides us with a poweful, novel tool of diagnosing ellipsis. Notice that its force is independent of whether one analyzes RPs as base-generated or as residues of movement (Sichel (2014) motivates the latter analysis for Hebrew on the basis of reconstruction; see Guilliot and Malkawi 2007 and especially Hewett 2023 for non-movement accounts of the reconstruction data in Semitic). All that matters is that the EIR test is applicable in environments blocking Ā-dependencies terminating in a gap. Time permitting, I will also discuss some puzzling cases of “mixed anaphors” (Hauser et al. 2007, Thompson 2014, Houser 2010, Heynen 2022) and how the EIR test can help decide their status as deep or or surface anaphors.
Selected References. Guilliot, N., and Malkawi, N. 2007. Reconstruction without Movement. In Coreference, Modality, and Focus, ed. by Luis Eguren and Olga Fernández-Soriano, 111-131. Amsterdam: John Benjamins. Hauser, M., Mikkelsen, L., and Toosanvardani, M. 2007. Verb Phrase Pronominalization in Danish: Deep or Surface Anaphora? In Proceedings of the 34 Western Conference on Linguistics, ed. by Erin Brainbridge and Brian Agbayani, 183-195. Fresno, CA: Department of Linguistics, California State University, Fresno. Hewett, M. 2023. Types of Resumptive Ā-dependencies. PhD dissertation, University of Chicago. Landau, I. 2018. Missing Objects in Hebrew: Argument Ellipsis, not VP Ellipsis. Glossa 3(1), 76:1-37. Landau, I. 2025. Type-restricted Argument Ellipsis and Generalized Quantifiers. Linguistic Inquiry 56:97-129. Merchant, J. 2013. Diagnosing Ellipsis. In Diagnosing Syntax, ed. by Lisa Lai-Shen Cheng and Nobert Corver, 537-542. Oxford: Oxford University Press. Merchant, J. 2023. Nominal Ellipsis, n/N Heads, and Lexical Nominalization. Handout of a talk given in CGG 32, University of the Basque Country, Vitoria/Gasteiz. Saab, A. 2019. Nominal Ellipsis. In The Oxford Handbook of Ellipsis, ed. by Jeoren Van Craenenbroeck and Tanja Temmerman, 526-561. Oxford: Oxford University Press. Sichel, I. 2014. Resumptive Pronouns and Competition. Linguistic Inquiry 45:655-693. Thompson, A. 2014. Beyond Deep and Surface: Exploarations in the Typology of Anaphora. PhD dissertation, UCSC.
Congratulations!
