Skip to main content

Ezafe, PP and the nature of nominalization


In the paper we argue that the English VP/NP structures in (i) a-d have exact counterparts in the i(ranian)Persian PP/NP structures in (ii) a-d, where P1-P3 are three different classes of iPersian Ps and where -EZ is the so-called “Ezafe” morpheme. (i. a) John [VP destroy the evidence] “Pure VP”; (i. b) John -’s [NP destroying the evidence] Nominalized VP; (i. c) John -’s [NP destroying of the evidence] Nominalized V; (i. d) John -’s [NP destruction of the evidence] Deverbal N; (ii. a) NP [PP P1 NP] “Pure PP”; (ii. b) NP -Ez [NP P2 NP] Nominalized PP; (ii. c) NP -Ez [NP P2 -Ez NP] Nominalized P; (ii. d) NP -Ez [NP P3 -Ez NP] “Deprepositional” N. The notion “nominalization” is thus shown to be relevant to both of the lexical categories - V and P - identified by Chomsky (1974) as [-N]. Our demonstration proceeds in three steps: 1) We establish a common syntactic function for English -’s/of and iPersian -EZ, viz., case-assignment, following Samiian 1994; Karimi and Brame 1986/2012; Larson and Samiian 2020; 2) We argue for a shared cross-categorial structure for VP-PP, developing proposals by Jackendoff 1973; van Riemsdijk 1990; Svenonius 2003. We show that if Jackendoff’s (1977) “scopal nominalization” analysis of gerunds is extended to iPersian PPs, the parallelism in (i) and (ii) is accounted for; 3) We show that the full extension of nominalization to iPersian PPs suggests a more general view of nominalization than has been recognized previously, viz., a “split-feature” view of category specification. This has a variety of implications, which we briefly explore.

This is a preview of subscription content, access via your institution.


  1. 1.

    The three main geographical variants of Modern Persian spoken in Iran, Afghanistan and Tajikistan are largely mutually intelligible, but nonetheless grammatically distinct, including for the phenomena discussed in this paper. We adopt the terminology “iPersian,” “aPersian” and “tPersian” for these variants in order to accommodate the strong desire of all three communities to be identified as speaking “Persian” (Farsi), but also to distinguish them for linguistic purposes.

  2. 2.

    Karimi and Brame (1986/2012) argue that Ezafe occurs between nouns, which requires them to argue that the category N in iPersian embraces a far larger group of expressions than might otherwise be thought. Samvelian (2007) argues that Ezafe is specifically nominal morphology; given that Ezafe can attach to adjectives (7j) and to Partitive quantifiers, this entails the claim that these elements are nominal. Samiian (1994), Larson and Yamakido (2008), Larson and Samiian (2020) adopt the claim in the form given in (12).

  3. 3.

    An anonymous NLLT reviewer notes that some quantifiers display nominal behavior insofar as they occur in argument position and take a plural morpheme (1a-b):

    1. (i)

    This is true of many quantifiers but not all. Specifically, all iPersian quantifiers that are followed by a classifier behave nominally as in (15)a. And some quantifiers without classifiers also occur nominally, e.g., hame ‘all’ in (iia-b):

    1. (ii)

    However, some iPersian quantifiers don’t occur in nominal positions, such as bishtar ‘most’ (iiia-b), aksar ‘most/majority’ (iva-b), aqhlab ‘most/temporal’; even tamam ‘all’ is not acceptable as a nominal for some native speakers (va-b).

    1. (iii)
    1. (iv)
    1. (v)
  4. 4.

    Az has independent use in iPersian as an ablative preposition meaning ‘from’ (ia-b); example (ib) shows both partitive and ablative uses together:

    1. (i)

    iPersian speakers detect an “ablative flavor” with az in some cases in (19)-(22), such as (21d). Note that from/of alternation is also found in English in examples like (iia-b):

    1. (ii)

    Presumably the semantic differences between use of Ezafe versus az follow from the fact that az carries genuine semantic features whereas Ezafe is semantically empty, a pure case-marker, counterpart to the difference between English from and of.

  5. 5.

    The close parallelism between iPersian Ezafe and English of is noted explicitly in Karimi and Brame (1986/2012) and Samiian (1994).

  6. 6.

    When the head noun bears the indefinite suffix in iPersian, Ezafe is excluded (ref.). In this context, az becomes obligatory (i-v):

    1. (i)
    1. (ii)
    1. (iii)
    1. (iv)
    1. (v)

    Note, however, that this distribution does not hold of adjectival modifiers, which never co-occur with az.

    1. (vi)

    We assume that whereas adjectives can co-occur with the pure case-marker -EZ, they cannot occur with the prepositional case-marker az in virtue of residual ablative semantic features inhering in the latter.

  7. 7.

    An interesting implication of Samiian’s analysis is that iPersian adjectives can be directly case-marked like iPersian nouns (a point also noted by Haig 2011). This implication is explored and developed for iPersian and a number of other Iranian languages in Larson (2018).

  8. 8.

    See Appendix for a full list of iPersian prepositions.

  9. 9.

    The P1-P3 division classifies iPersian prepositions with regard to Ezafe, but some iPersian Ps show other options. For example, the prepositions bad/pas ‘after’, gabl/pish ‘before’, and qeyr ‘except’ pattern like P3s, but instead of showing Ezafe before their objects they govern the preposition az (see 21b-c). iPersian also exhibits compound prepositions like bar-asaas-e ‘based on’, banaa bar ‘according to’, bar-mabnaa-ye ‘on the basis of’ and dar-baare-ye ‘about’. See Appendix for a more complete classification.

  10. 10.

    Pantcheva (2008), following Svenonius (2006), analyzes P2s and P3s as “Axial Parts,” a locative category distinct from nouns and distinct from functional prepositions (P1s). To account for the nominal characteristics of P2/P3s she posits an empty place node before P2/P3s. A major argument she presents is the occurrence of a demonstrative determiner before P2/P3s as in (i) which is semantically synonymous with (ii). Since ja means ‘place,’ she assumes a silent place before ru-ye miz in (i).

    1. (i)
    1. (ii)

    However, (i) and (ii) exhibit some distinctions. For example, (i) disallows demonstratives before the final noun (iiia), whereas (ii) allows them (iiib). Note the bare PP form allows final demonstratives (iiic).

    1. (iii)

    Moreover, P3s can occur without a complement with a determiner, but not preceded by inja and unja.

    1. (iv)

    Also, absent from Pantcheva (2008) is any account of the external occurrence of Ezafe before PPs. (The internal occurrence of Ezafe is not accounted for in detail beyond the assumption that Ezafe is a case assigner and an additional projection KP can host the Ezafe morpheme as proposed by Svenonius 2006.) Finally with respect to optionality of Ezafe following P2s, Pantcheva proposes to classify them as P1 when there is no Ezafe following them as in (v) or as Axial Part P3 when there is as in (vi).

    1. (v)

      tu jabe

    1. (vi)

      tu-ye jabe

    This appears to us simply to restate the distributional facts rather than to explain them.

  11. 11.

    We are grateful to an anonymous NLLT reviewer for comments prompting the discussion in this section and to Jonathan Washington (p.c.) for pointing out the relevance of English inside and outside for the discussion of iPersian prepositions.

  12. 12.

    Svenonius (2003, 2012) analyzes what are here identified as relational nouns as of the syntactic category Axial Part, and offers an extensive decompositional analysis of spatial adpositions.

  13. 13.

    This difference in prepositional trajectory for inside and interior plausibly traces to their different origins. According to the OED, inside originated (1504) from an adjective-relational noun combination yn-syde meaning ‘inner side,’ later generalized to ‘interior.’ By contrast interior (1490) derives from a Latin comparative adjective intere meaning ‘situated more within.’

  14. 14.

    This identity of form led Karimi and Brame (1986/2012) to conclude that iPersian P3s simply are nouns. But this cannot be correct, as we have seen, given that prepositions and relational nouns have distinct semantics.

  15. 15.

    We remain neutral in this discussion as to whether incorporation occurs syntactically as in Baker (1988) or whether it is a word formation process as in Rosen (1989). For us the key point is that incorporated nominal material, like jelo in (50b), does not bear case features like argumental nominals.

  16. 16.

    Jackendoff 1973 did not consider sentential complement-taking verbs [VP V CP], which, as later noted by Emonds (1976), are matched by clause-taking prepositions like before, after, while, because and although—so-called “subordinating conjunctions.” An anonymous NLLT reviewer also notes the clause-taking P in (Reading is a skill, in that we need to practice it). The core VP complementation patterns not apparently matched in PP appear to be: double objects ([VP V NP NP]; give John a peach)), double PPs ([VP V PP PP]; talk to John about Max) and object control structures ([VP V NP CP]; urge John to eat).

  17. 17.

    One of Jackendoff’s (1973) striking demonstrations is that expressions like from Kyoto to Tokyo can function as a single constituent (54d′) and not simply as a sequence of two PPs. Jackendoff establishes this with classic constituency tests and examples like (ia-c):

    1. (i)
  18. 18.

    The characterization of of and ez as “optional” in (55b,c) and (56b,c) (resp.) is purely descriptive. As we argue below, the two pairs have different structures.

  19. 19.

    An anonymous NLLT reviewer questions the parallelism between (55a-d) and (56a-d) based on presumed thematic differences. Specifically, in (55a-d) John(’s) is claimed to be an argument of the boldfaced phrases that follow it whereas in (56a-d) the boldfaced phrases constitute adjunct modifiers of the preceding N-ez. While it is true that John(’s) functions as a semantic argument in (55a-d), in at least (55b-d) it is dubious that John’s is in fact a syntactic argument. As discussed in detail by Grimshaw (1991) in all of such cases, the genitive nominal is optional and replaceable by a simple determiner (the destroying/destruction), behavior quite uncharacteristic of a true argument. More plausible, as Grimshaw discusses, is that gerunds and derived nominals contain covert subjects and that the genitive-marked nominal has the status of an adjunct that, when present, is bound to the covert subject. If so, then the fundamental relation between John’s and the boldfaced phrases in (55b-d) and N-ez and the boldfaced phrases in (56b-d) (resp.) is not fundamentally different, presumably interpreted by predicate conjunction in both cases, a symmetric relation. Furthermore we note that the parallelism that truly matters in (55) and (56) is exactly between (55b-d) and (56b-d) since (56b-d) are precisely the cases where presence of Ezafe is unexpected and in need of explanation. In short then, where parallelism in (55)/(56) truly matters, it does indeed appear to obtain, including with respect to thematic relations.

  20. 20.

    Jackendoff (1977) builds on Horn (1975), who was the first (to our knowledge) to propose that -ing can function to nominalize a verbal projection. Horn (1975:363) offers a general structure for verbal gerunds as in (i), utilizing X-bar theory:

    1. (i)
  21. 21.

    As is well known, verbal gerunds also allow aspectual verbs (ia) and negation (ib). In the first case, -ing would be assumed to attach to the corresponding projections above vP, e.g., PerfP. The second case appears to be a negative verbal gerund and raises an interesting word order problem for Jackendoff that we discuss below:

    1. (i)
      1. a.

        John’s [NP having destroyed the evidence] (was problematic).

      2. b.

        John’s [NP not destroying the evidence] (was fortunate).

  22. 22.

    Abney (1987) and Kratzer (1996) adopt variants of Jackendoff’s scopal nominalization; cf. also Borsley and Kornfilt’s (2000) analysis of “mixed categories.” See Sect. 7 for more discussion.

  23. 23.

    As noted in fn. 16, the genitive subjects of gerunds and nominals do not behave as true syntactic subjects in being optional; we follow Grimshaw (1991) in taking these genitives to be adjuncts, bound to a covert pronominal subject (represented in (58a) by “…”) that remains within vP.

  24. 24.

    We are grateful to Nazila Shafiei for assistance with corpus research.

  25. 25.

    In (61c-e), the preposition be, bar and az are optional with Ezafe present. With Ezafe absent, be, bar and az become obligatory. The meaning is the same in all cases.

  26. 26.

    For developments of, and alternatives to, Vendler’s proposals about gerunds, see Hamm and van Lambalgen (2002), Milsark (2005), Grimm and McNally (2015).

  27. 27.

    Our analysis makes the general prediction that Ezafe should be optional before P1Ps. We are aware of only two counter-examples from the literature, viz., (ia) from Samvelian (2007) and (ib) from Khanemouyepour (2014). In addition, the Bijan Khan corpus of 103 items contains one such item (BK#27), reproduced in (ic).

    1. (i)

    Interestingly, the three P1Ps in (i) behave differently than P2Ps and P3Ps despite requiring a preceding Ezafe like the latter. With the P3 kenar ‘beside’ in (iia) and the P2 tu ‘inside’ in (iib), we can have an Ezafe requiring modifier before PP and a possessive after.

    1. (ii)

    However, with goruh-e dar shahr ‘the group in town’ a preceding modifier is acceptable only if the Ezafe before P1P is omitted (iiia,b).

    1. (iii)

    Similarly with sob.ha-ye ba madar ‘mornings with mother’ (iva-c)

    1. (iv)

    Similarly with aks-e dar ganje ‘picture in the closet’ (va-c).

    1. (v)

    We conjecture that the examples in (ia-c) cannot be broken up by other modifiers. If so, the obligatory Ezafe found with them is not the productive Ezafe found elsewhere, and is not a counterexample to the generalization made here.

  28. 28.

    Note that the truth of (68a) presupposes the truth of (68b). Similarly for (69a-b) below.

  29. 29.

    For what it means for an N or V feature to be interpretable see Panagiotidis (2014) for extensive discussion. Regarding valuation, note that in this theory features do not have or assume different values; they simply are valued or not. To be valued is thus simply to be PF-interpretable.

  30. 30.

    Pesetsky and Torrego (2007) generalize the scheme in (70) so that any unvalued feature may probe and agree with a local, c-commanded matching feature. This allows for potential probe-goal relations as in (i), where “⇒” indicates agreement. By contrast, potential probe-goal relations as in (ib) are excluded since they involve a valued probe:

    1. (i)
      1. a.

        iF ⇒ Fv, F ⇒ Fv, F ⇒ Fi, F ⇒ F.

      2. b.

        Fv ⇏ Fi, Fv ⇏ Fv, Fv ⇏ F

    Furthermore, as a condition on legibility at the PF-LF interfaces, Pesetsky and Torrego require each feature to have both an interpretable and a valued instance joined by agreement. Hence feature structures like (iia) will constitute legible objects since they contain instances of both kinds, but those in (iib) will not, because of lack of interpretable F, valued F, or both:

    1. (ii)
      1. a.

        iFv, iF ⇒ Fv, iF ⇒ F ⇒ Fv, iF ⇒ F ⇒ F ⇒ Fv, F ⇒ iF ⇒ Fv

      2. b.

        iF, Fv, iF ⇒ F, F ⇒ Fv, F ⇒ F, iF ⇒ F ⇒ F, F ⇒ F ⇒ Fv, F ⇒ F ⇒ F

  31. 31.

    See Sect. 5.5 below for more about the feature structure of derived nouns like destruction.

  32. 32.

    An anonymous NLLT reviewer inquires whether so-called zero-derived nominals like love in John’s love of chocolate can be handled in this account; e.g., would they be analyzed like nominal gerunds, with [iN] and [Nv] features that are both unrealized phonetically. If so, why is the equivalent of a verbal gerund excluded (*John’s love chocolate), etc. Briefly, we do think this analysis extends to zero-derivation, but that extension involves additional issues of PF feature visibility that carry us outside the scope of this paper. Hence we postpone development to a later occasion.

  33. 33.

    Thus the general labeling convention is that [αP …] will only project the categorial features of a head that is both valued and interpretable. In [destroy -ing], destroy bears interpretable and valued V, but only valued N, hence it is projected as V. By contrast in [ n [destroy -ing]], n bears both interpretable and (by agreement) valued N, hence it is projected as N. Likewise in [ n … [destroy -ing]], all projections up to n will be projected as V since N is not interpretable until that point. Note carefully that in this discussion we are talking about categorial projection in terms of whether features are interpretable and valued, not the lexical items bearing them. This allows for the fully licit projection of elements with no semantic features, such as expletives or a purely grammatical preposition. See below.

  34. 34.

    We assume of to be a categorial P that enters the numeration like other prepositions. It is distinguished only in bearing no inherent semantic features, being a “purely grammatical” P in this sense.

  35. 35.

    An anonymous NLLT reviewer asks about the precise semantics of nominalization that would block (76a). For interesting discussion on this point, see Grimm and McNally (2015).

  36. 36.

    Note that whereas we can resolve projection of the defective [destruct -ion] by adding an interpretable noun feature ([iN]), creating [N n [destruct -ion]], we cannot do so by adding valued verbal feature ([Vv]), creating [V v [destruct -ion]]. Since valued features do not probe in this theory (see fn. 25 above), [Vv] and [iV] could not come into agreement in this configuration. A question arises regarding the residual [iV] feature on destruct-, which does not undergo agreement with any [Vv] feature in the course of derivation and hence isn’t PF-visible—i.e., not fully interpretable. We assume that this is acceptable in the case of roots contained within larger lexical expressions that are fully interpretable heads. This corresponds to our informal intuition that lexemes can have subparts with notional categorial contribution (verbal, nominal, adjectival) even without ever being formally of that category at any stage of derivation.

  37. 37.

    Note in (77)/(78) that since destroying bears only unagreed [Nv] at the point where vP is composed, its feature structure is as in (73a)/(74a); i.e., it is still formally a verb. This entails that it will undergo raising to v.

  38. 38.

    Presumably the marginality that some speakers perceive in examples like (82a-b) can be linked to the marginality of VP as an adverbial attachment site.

  39. 39.

    In this we follow the line of explanation in Carnie (2011), who attributes the occurrence of genitive in certain Irish verbal noun constructions to the absence of a specific accusative case-assigning element, rather than to the presence of nominal vs. verbal projection per se.

  40. 40.

    In the analysis of Panagiotidis (2014), items bearing an interpretable V feature are taken to denote entities extended in time. Being both interpretable and valued destr- would thus be treated as a full fledged verb. This would also presumably link to the fact, noted by an anonymous NLLT reviewer, that derived nominals with a result meaning resist adverbs e.g., I held the translation of the poem (*quickly) in my hand. Since such nominals precisely do not denote entities extended in time, they might be analyzed as bearing a valued V feature, but not an interpretable one.

  41. 41.

    See Fu et al. (2001) for an analysis similar in spirit to (53).

  42. 42.

    We are grateful to an anonymous NLLT reviewer for pointing out the examples discussed in (88)-(93) and the issues they raise.

  43. 43.

    An anonymous NLLT reviewer observes that the ungrammaticality of (90a) cannot simply be ascribed to the lack of expression of participial morphology associated with have since “do-support” does not save the example (i):

    1. (i)

      *John’s have done destroying the evidence (is problematic).

  44. 44.

    Note here again that the ungrammaticality of (92b) cannot simply be ascribed to the lack of expression of participial morphology associated with be since “do-support” does not save the example (i):

    1. (i)

      *John have was been doing silly.

  45. 45.

    Note in (80b) and (81b) above that the items not and intentionally (respectively) are not potential bearers of -ing ([Nv]), and hence are not barriers to the indicated probe-goal relations under Minimality.

  46. 46.

    As noted by an anonymous NLLT reviewer, an important difference between the prepositional and verbal forms is that the latter derive morphologically from a verbal root and hence any verb with appropriate meaning can occur with each class. By contrast, the iPersian P2 and P3 forms have their nominal content lexically, derived only in a historical sense.

  47. 47.

    See 6.3 below for additions. [P] is treated here as a basic category for convenience. Nothing hangs on this. “P” could as easily be analyzed as a combination of [-v,-n], as in Chomsky (1974) or the more elaborate system in Jackendoff (1977). Similarly for the treatment of [V] and [N] in the text.

  48. 48.

    There is an interesting historical correlate to the hypothesized difference between the “defective” P feature of P3s ([iP]) versus the interpretable, valued feature of P2s ([iPv]). Middle Persian, the historical antecedent of iPersian, exhibits P1s and P3s in examples like (1a-b):

    1. (i)

    By contrast, P2s appear to be absent from Middle Persian and to represent a recent development. Resuming earlier discussion, we propose below in 6.2 that P2s and P3s derive by incorporating a relational noun into a covert preposition. A natural conclusion from these historical facts is that this incorporation process took place only recently with P2s. This appears compatible with the idea that P would have a more independent, less root-like nature in P2s versus P3s, given their more recent development.

  49. 49.

    This view is conceptually similar to Distributed Morphology (Halle and Marantz 1993) wherein category-less roots receive their syntactic category by combination with “categorizers.”

  50. 50.

    In an interesting extension of Borsley and Kornfilt (2000), Cole and Hermon (2011) analyze nominative-genitive alternation in the subject marking of Quechua relative clauses as reflecting scope of nominalization. We believe our approach can be extended directly to Cole and Hermon’s data, along the lines of our analysis of gerunds and hope to pursue these ideas in separate work.


  1. Abney, Steven. 1987. The English noun phrase in its sentential aspect. PhD diss., MIT.

  2. Alexiadou, Artemis. 2001. Functional structure in nominals: Nominalizations and ergativity. Amsterdam: Benjamins.

    Google Scholar 

  3. Baker, Mark. 1988. Incorporation. Chicago: University of Chicago Press.

    Google Scholar 

  4. Baker, Mark. 2011. Degrees of nominalization: Clause-like constituents in Sakha. Lingua 121: 1164–1193.

    Google Scholar 

  5. Borsley, Robert, and Jaklin Kornfilt. 2000. Mixed extended projections. In Syntax and semantics 22: The nature and function of syntactic categories, ed. Robert Borsley, 101–131. San Diego: Academic Press.

    Google Scholar 

  6. Carnie, Andrew. 2011. Mixed categories in Irish. Lingua 121: 1207–1224.

    Google Scholar 

  7. Chomsky, Noam. 1970. Remarks on nominalization. In Readings in English transformational grammar, eds. Roderick Jacobs and Peter Rosenbaum, 184–221. Boston: Ginn.

    Google Scholar 

  8. Chomsky, Noam. 1974. The Amherst lectures. Ms., MIT.

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

    Google Scholar 

  10. Chomsky, Noam. 1994. Bare phrase structure. MIT occasional papers in linguistics. Cambridge: MIT Department of Linguistics and Philosophy, MITWPL.

    Google Scholar 

  11. Chomsky, Noam. 1995. The minimalist program. Cambridge: MIT Press.

    Google Scholar 

  12. Cole, Peter, and Gabriella Hermon. 2011. Nominalization and case assignment in Quechua. Lingua 121: 1225–1251.

    Google Scholar 

  13. Comrie, Bernard, and Sandra Thompson. 2007. Lexical nominalization, 2nd edn. In Language typology and syntactic description. Vol. 3, Grammatical categories and the lexicon, ed. Timothy Shopen, 334–381. Cambridge: Cambridge University Press.

    Google Scholar 

  14. Emonds, Joseph. 1976. A transformational approach to English syntax: Root, structure-preserving, and local transformations. New York: Academic Press.

    Google Scholar 

  15. Fu, Jingqi, Thomas Roeper, and Hagit Borer. 2001. The VP within process nominals: Evidence from adverbs and the VP anaphor do-so. Natural Language and Linguistic Theory 19: 549–582.

    Google Scholar 

  16. Gagnon, Michael. 2013. Part and parcel of eliding partitives. Semantic and Linguistic Theory (SALT) 23: 316–335.

    Google Scholar 

  17. Girbau, Núria Martí. 2010. The syntax of partitives. PhD diss., Universitat Autònoma de Barcelona.

  18. Grimm, Scott, and Louise McNally. 2015. The -ing dynasty: Rebuilding the semantics of nominalizations. In Semantic and linguistic theory (SALT) 25.

    Google Scholar 

  19. Grimshaw, Jane. 1991. Argument structure. Cambridge: MIT Press.

    Google Scholar 

  20. Grimshaw, Jane. 2005. Extended projection. In Words and structure, ed. Jane Grimshaw, 1–74. Stanford: CSLI.

    Google Scholar 

  21. Haig, Geoffrey. 2011. Linker, relativizer, nominalizer, tenseparticle: On the Ezafe in West Iranian. In Nominalization in Asian languages: Diachronic and typological perspectives, eds. Foong Ha Yap, Karen Grunow-Hårsta, and Janick Wrona, 363–390. Amsterdam: John Benjamins.

    Google Scholar 

  22. Halle, M., and A. Marantz. 1993. Distributed morphology and the pieces of inflection. In The View from Building 20, eds. K. Hale and S. J. Keyser, 111–176. Cambridge, MA: MIT Press.

    Google Scholar 

  23. Hamm, Fritz, and Michiel van Lambalgen. 2002. Formal foundations for semantic theories of nominalisation. ZAS Papers in Linguistics 27: 1–21.

    Google Scholar 

  24. Harley, Heidi. 2009. The morphology of nominalizations and the syntax of VP. In Quantification, definiteness, and nominalization, eds. Anastasia Giannakidou and Monica Rathert, 321–343. Oxford: Oxford University Press.

    Google Scholar 

  25. Hazout, Ilan. 2001. Predicate formation: The case of participial relatives. The Linguistic Review 18: 97–123.

    Google Scholar 

  26. Horn, George. 1975. On the non-sentential nature of the poss-ing construction. Linguistic Analysis 1: 333–387.

    Google Scholar 

  27. Ionin, Tania, Ora Matushansky, and Eddy G. Ruys. 2006. Parts of speech: Toward a unified semantics for partitives. In North East linguistic society (NELS) 36, eds. Christopher Davis, Amy Rose Deal, and Youri Zabbal, 357–370. Amherst: GLSA.

    Google Scholar 

  28. Jackendoff, Ray. 1973. The base rules for prepositional phrases. In A festschrift for Morris Halle, eds. Stephen R. Anderson and Paul Kiparsky, 345–356. New York: Holt, Rinehart & Winston.

    Google Scholar 

  29. Jackendoff, Ray. 1977. X-bar syntax. Cambridge: MIT Press.

    Google Scholar 

  30. Karimi, Simin, and Michael Brame. 1986/2012. A generalization concerning the Ezafe construction in Persian. Linguistic Analysis 38: 111–143.

  31. Kahnemuyipour, Arsalan. 2014. Revisiting the Persian Ezafe construction: A roll-up movement analysis. Lingua 150: 1–24.

    Google Scholar 

  32. König, Ekkehard, and Bernd Kortmann. 1991. On the reanalysis of verbs as prepositions. In Approaches to prepositions, ed. Gisa Rauh, 109–125. Tübingen: Narr.

    Google Scholar 

  33. Kornfilt, Jaklin, and John Whitman. 2011a. Introduction: Nominalizations in syntactic theory. Lingua 121: 1160–1163.

    Google Scholar 

  34. Kornfilt, Jaklin, and John Whitman. 2011b. Afterword: Nominalizations in syntactic theory. Lingua 121: 1297–1313.

    Google Scholar 

  35. Kortmann, Bernd, and Ekkehard König. 1992. Categorial reanalysis: The case of deverbal prepositions. Linguistics 30: 671–697.

    Google Scholar 

  36. Kratzer, Angelica. 1996. Severing the external argument for its verb. In Phrase structure and the lexicon, eds. Johan Rooryck and Laurie Zaring, 109–137. Dordrecht: Kluwer.

    Google Scholar 

  37. Krause, Cornelia. 2001. On reduced relatives with genitive subjects. PhD diss., MIT.

  38. Larson, Richard. 1988. On the double object construction. Linguistic Inquiry 19: 335–391.

    Google Scholar 

  39. Larson, Richard. 2018. Zazaki ‘double Ezafe’ as double case-marking. Manuscript, Stony Brook University.

  40. Larson, Richard, and Vida Samiian. 2020, to appear. The Ezafe construction revisited. In Advances in Iranian linguistics I, eds. Richard Larson, Sedigheh Moradi, and Vida Samiian. Amsterdam: Benjamins.

  41. Larson, Richard, and Hiroko Yamakido. 2008. Ezafe and the deep position of nominal modifiers. In Adjectives and adverbs: Syntax, semantics and discourse, eds. Louise McNally and Chris Kennedy, 43–70. Oxford: Oxford University Press.

    Google Scholar 

  42. Libert, Alan. 2013. Adpositions and other parts of speech. Bern: Peter Lang.

    Google Scholar 

  43. Lieber, Rochelle. 2016. English nouns: The ecology of nominalization. Cambridge: Cambridge University Press.

    Google Scholar 

  44. Lieber, Rochelle. 2018. Nominalization: General overview and theoretical issues. In Oxford research encyclopedia, linguistics. Oxford: Oxford University Press.

    Google Scholar 

  45. Manzini, Rita. 2010. The structure and interpretation of (romance) complementizers. In The complementiser phase: Subjects and operators, ed. Phoevos Panagiotidis, 167–199. New York: Oxford University Press.

    Google Scholar 

  46. Marvin, Tatjiana. 2003. Past participles in reduced relatives: A cross-linguistic perspective. Linguistica 47: 141–160.

    Google Scholar 

  47. Matthewson, Lisa. 2001. Quantification and the nature of crosslinguistic variation. Natural Language Semantics 9: 145–189.

    Google Scholar 

  48. Milsark, Gary. 2005. Gerundive nominalizations. In The Blackwell companion to syntax, eds. Martin Everaert and Henk van Riemsdijk, 436–458. New York: Blackwell.

    Google Scholar 

  49. Ogawa, Yoshiki. 2014. Grammaticalization of near from adjective to preposition via head-movement, gradability declination and structural reanalysis. Interdisciplinary Information Sciences 20(2): 189–215.

    Google Scholar 

  50. Panagiotidis, Phoevos. 2014. Categorial features. Cambridge: Cambridge University Press.

    Google Scholar 

  51. Pantcheva, Marina. 2008. The place of place in Persian. In Syntax and semantics of spatial P, eds. Anna Asbury, Jakub Dotlačil, Berit Gehrke, and Rick Nouwen, 305–330. Amsterdam: Benjamins.

    Google Scholar 

  52. Pesetsky, David, and Esther Torrego. 2007. The syntax of valuation and the interpretability of features. In Phrasal and clausal architecture, eds. Simin Karimi, Vida Samiian, and Wendy Wilkins, 262–294. Amsterdam: Benjamins.

    Google Scholar 

  53. Quirk, Randolph, and John Mulholland. 1964. Complex prepositions and related sequences. English Studies 45(Suppl.): 64–73.

    Google Scholar 

  54. Reuland, Eric. 2011. What’s nominal in nominalization? Lingua 121: 1283–1296.

    Google Scholar 

  55. Sarah Thomas Rosen. 1989. Two types of noun incorporation: A lexical analysis. Language 65: 294–317.

    Google Scholar 

  56. Roy, Isabelle, and Elena Soare. 2011. Nominalizations: New insights and theoretical implications. Recherches Linguistiques de Vincennes 40: 7–23.

    Google Scholar 

  57. Samiian, Vida. 1983. Structure of phrasal categories in Persian, an X-bar analysis. PhD diss., UCLA.

  58. Samiian, Vida. 1994. The Ezafe construction: Some implications for the theory of X-bar syntax. In Persian studies in North America, ed. Mehdi Marashi, 17–41. Bethesda: Iranbooks.

    Google Scholar 

  59. Samvelian, Pollet. 2007. A (phrasal) affix analysis of the Persian Ezafe. Journal of Linguistics 43: 605–645.

    Google Scholar 

  60. Sauerland, Uli, and Kazuko Yatsushiro. 2004. A silent noun in partitives. In North East linguistic society (NELS) 34, Vol. 2, eds. Keir Moulton and Matthew Wolf, 505–516. Amherst: Graduate Linguistic Student Association.

    Google Scholar 

  61. Siloni, Tal. 1995. On participial relatives and complementizer do: A case study in Hebrew and French. Natural Language and Linguistic Theory 13: 445–487.

    Google Scholar 

  62. Simpson, Andrew, and Zoe Wu. 2001. The grammaticalization of formal nouns and nominalizers in Chinese, Japanese and Korean. In Language change in East Asia, ed. Thomas McAuley, 250–283. London: Routledge Curzon.

    Google Scholar 

  63. Sleeman, Petra. 2019. Participial relative clauses. In Oxford research encyclopedia, linguistics. Oxford: Oxford University Press.

    Google Scholar 

  64. Stickney, Helen. 2009. The emergence of DP in the partitive structure. PhD diss., University of Massachusetts, Amherst.

  65. Stowell, Timothy. 1981. Origins of phrase structure. PhD diss., MIT.

  66. Svenonius, Peter. 2003. Limits on P: Filling in holes vs. falling in holes. Nordlyd: Tromsø Working Papers on Language and Linguistics 31(2): 431–445. Proceedings of the 19th Scandinavian Conference of Linguistics.

    Google Scholar 

  67. Svenonius, Peter. 2006. The emergence of axial parts. Nordlyd: Tromsø Working Papers on Language and Linguistics 33(1): 49–77. Special Issue on Adpositions, eds. Peter Svenonius and Marina Pantcheva, Tromsø: University of Tromsø.

    Google Scholar 

  68. Svenonius, Peter. 2012. Structural decomposition of spatial adpositions. Ms., CASTL, University of Tromsø.

  69. van Riemsdijk, Henk. 1983. The case of German adjectives. In Linguistic categories: Auxiliaries and related puzzles, eds. Frank Heny and Barry Richards, 223–252. Dordrecht: Springer.

    Google Scholar 

  70. van Riemsdijk, Henk. 1990. Functional prepositions. In Unity in diversity, eds. Harm Pinkster and Inge Genée, 229–241. Dordrecht: Foris.

    Google Scholar 

  71. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca, New York: Cornell University Press.

    Google Scholar 

  72. Waters, Cathleen. 2009. The preposition cycle in English. In Cyclical change, ed. Elly van Gelderen, 287–300. Amsterdam: Benjamins.

    Google Scholar 

Download references


For helpful comments and clarifying questions we thank audiences at CLS54, Tehran University and Stony Brook University, where earlier versions of this material were presented. We are also grateful to Xuhui Freddy Hu, Anna Maria Di Sciullo, Jim Wood and four anonymous NLLT reviewers for comments and suggestions that have substantially improved this manuscript.

Author information



Corresponding author

Correspondence to Richard K. Larson.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix:  A partial list of iPersian prepositions

Appendix:  A partial list of iPersian prepositions

Forbids Int. ez
Allows Ext ez
Allows Int. ez
Requires Ext ez
Requires Int. ez
Requires Ext. ez
Requires Int. az
Requires Ext. ez
(Compound Ps): Allows Int. ez if
ending in [+N]; Allows Ext. ez
[az] ‘from’
[dar] ‘in’
[bar] ‘on/at’
[bâ] ‘with’
[be] ‘to’
[bi] ‘without’
[tâ] ‘until’
[barâye/bare] ‘for’
[bâlâ] above
‘on top of’
[tu] inside
[ru] on top of
[jelo] in front of
[pahlu] ‘next to’
[atrâf] ‘around’
[aleyh] ‘against’
[baqal] ‘next to’
[beyn] ‘between
[birun] ‘outside
[dâxel] ‘inside’
[darun] ‘inside’
[dowr] ‘around’
[kenar] ‘next to/by’
[miyân] ‘between/among’
[mesl] ‘similar to’
[nazd] ‘at/near’
[nazdik] ‘near’
[posht] ‘behind’
[pâin] ‘below’
[pâ] ‘foot of’
[pish] ‘beside/at’
[sar] ‘head of/at’
[taraf] ‘side of’
[tavasot] ‘by (agent/instrument)’
[zir] ‘under’
[vasat] ‘between/middle
[taraf] ‘in the vicinity of (temporal)]
[mâbeyn] ‘among’
[hamrâh] ‘along with’
[bad az] after
[pas az] after
[pish az] before
[qabl az] before
[qeyr az] except for
[râje-be] ‘about’
[nesbat-be] ‘w. respect to’
[ru-be] ‘facing’
[banâ-bar] ‘according to’
[az-taraf-e] ‘from/on behalf of’
[az bahr-e] ‘for, for the sake of’
[alâ.raqm-e] ‘despite’
[ba-vojud-e] ‘in spite of’
[be-joz] ‘except for’
[be-raqm-e] ‘according to’
[be-jâ-ye] ‘instead of’
[be-taraf-e] ‘towards’
[be-su-ye] ‘in the direction of’
[bar-aks-e] ‘opposite to’
[bar-zed-e] ‘against’
[bar-mabnâ-ye] ‘on the basis of’
[bar-asâs-e] ‘based on’
[dar-bâre-ye] ‘about’
[dar-moqâbel-e] ‘against’
[dar-pey-e] ‘following’
[dar-zemn-e] ‘in the midst’
[dar-hin-e] ‘at the time of’
[dar-tul-e] ‘during’

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Larson, R.K., Samiian, V. Ezafe, PP and the nature of nominalization. Nat Lang Linguist Theory 39, 157–213 (2021).

Download citation


  • Ezafe
  • Prepositional phrases
  • PP
  • Case
  • Nominalization
  • Iranian languages