Causativity in zero and overt nominalizations: An experimental study

Causativity in zero and overt nominalizations: An experimental study Gianina Iordăchioaia (University of Graz) Jeannique Darby (University College Volda) GIANINA IORDĂCHIOAIA — JEANNIQUE DARBY ABSTRACT We experimentally evaluate three patterns of deverbal nominalizations derived by means of zero, -ing and Romance suffixes (i.e., -(at)ion, -ment, and -ance) in English as to whether they can express causativity and anticausativity like their base verbs. We report the results of afirst study which uses native speaker judgments to test the acceptability of these competing nominalizing suffixes (-ing vs. zero and -ing vs. Romance suffixes) in realizing event readings with argument structure inherited from their causative, and respectively, anticausative base verbs. While previous literature claims that zero cannot realize the structurally more complex causative readings, and -ing cannot realize anticausative readings, our current results indicate that all three suffixes may realize both types of readings. This is in line with data attested in natural text corpora and suggests that zero suffixes are not necessarily structurally simpler than overt suffixes as often claimed in previous literature. KEYWORDS Anticausativity, causativity, -ing suffix, nominalization, Romance suffixes, zero suffix DOI https://doi.org/10.14712/18059635.2025.1.3 1 INTRODUCTION Deverbal nominalization in English involves at least three morphological patterns, which may compete for the same verbal base, as illustrated in (1). (1a) to divide (verb) (1b) the divis-ion (ATK-N) (1c) the divid-ing (ING-N) (1d) the divide-Ø (Zero-N) From the verb to divide we may derive division with aRomance suffix, which we will refer to as an ATK-N(ominal) (from ‘-ATion and Kin’, Borer, 2013), dividing involves the suffix -ing (ING-N), and divide is derived by means of zero (Zero-N). Among these suffixes, -ing is known as the most productive one, attaching to any verb except for states (Asher, 1993; Borer, 2013; Iordăchioaia & Werner, 2019), while ATK-suffixes and zero are selectionally more restricted. For this reason, not many verbs show all three patterns as in (1), and in this paper we therefore focus on competing pairs such as -ing vs. zero, ing vs. ATK, and, indirectly, also zero vs. ATK. The analysis of zero forms such as in (1d) is controversial: some authors speak of derivation by means of azero suffix (Marchand, 1969; Kiparsky, 1982; Don, 1993, 2005; OPEN ACCESS 34 LINGUISTICA PRAGENSIA 1/2025 Alexiadou & Grimshaw, 2008; Iordăchioaia & Melloni, 2023a), others argue against azero suffix and refer to this process as conversion, category underspecification and the like (Lieber, 1981, 2016; Farrell, 2001; Plag, 2003; Borer, 2013). Although we do not directly address this debate (but see Bauer & Valera, 2005; Dahl & Fábregas, 2018; Iordăchioaia & Melloni, 2023b for critical overviews), our guiding hypothesis follows the zero-derivation approach, as we aim to directly compare zero with overt suffixes and hypothesize that there is no structural difference between them. We propose an acceptability judgment task, in which we test the ability of Zero-Ns to realize structurally complex readings of their base verbs in comparison with overtly derived ATK-Ns and ING-Ns. There are two questions related to this issue. The first is whether Zero-Ns may inherit argument structure from their base verbs like ATK-Ns and ING-Ns usually do, forming event nominals (see Grimshaw, 1990; Borer, 2013; Lieber, 2016). The second question is whether such zero event nominals with argument structure may express causative events or are limited to anticausative events. Causative events are structurally more complex than anticausative ones (Alexiadou et al., 2015) and implicitly more difficult to be hosted by zero morphology in Zero-Ns, according to previous literature such as Grimshaw (1990), Marantz (1997), Borer (2013) and others. Our project consists of two parts, investigating different aspects of the acceptability of the data. First, in an acceptability judgment study we test the acceptability of pairs of nominalizations built on anticausative and causative verbs in eventive contexts with verbal argument structure realization, and second, in aforced choice interpretation study we aim to check the compatibility of such previously tested nominalizations with contexts that impose causative vs. anticausative readings of alternating verbs such as to crash in (2) below. In this paper we present the first part, namely the acceptability judgment study, and draw conclusions on the basis of these intermediate results, which provide preliminary support for our guiding hypothesis. We start by introducing the empirical issue and the previous claims in the literature on causative/anticausative readings in nominalizations in Section 2. In Section3 we present our experimental design and in Section 4 we summarize our results and interpret them within the background of the claims and predictions in the previous literature. Section 5 concludes on these findings and proposes afollow-up to compensate for some shortcomings of the current study. 2 THEORETICAL BACKGROUND 2.1 (ANTI)CAUSATIVITY IN THE VERBAL DOMAIN The causative alternation, one of the most studied topics in lexical semantics, refers to the ability of some verbs such as to crash to exhibit both atransitive causative reading as in (2a) and an intransitive anticausative (essentially, unaccusative) reading as in (2b). Unlike other languages, English does not mark this alternation on either of the two forms: OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 35 (2a) The hijackers crashed the plane. (causative reading) (2b) The plane crashed. (anticausative reading) The causative reading realizes both an external (the hijackers) and an internal argument (the plane) and behaves like causative-only verbs such as to destroy— and unlike unaccusative/anticausative-only verbs such as to disappear— in building passives with an optional by-phrase, as in (3a). The anticausative reading in (3b) behaves like typical unaccusative verbs: it realizes the internal argument of the causative verb as asubject and cannot license external arguments in aby-phrase (Rappaport Hovav & Levin, 1998; Schäfer, 2009; Alexiadou et al., 2015). Following this literature, we consider the causative variant of an alternating verb to pattern with causative-only verbs and the anticausative variant with non-alternating unaccusative verbs. (3a) The plane was crashed/destroyed/*disappeared (by the hijackers). (3b) The plane crashed/*destroyed/disappeared (*by the hijackers). Since the early studies in lexical semantics, the causative variant has been argued to be structurally more complex than the anticausative one independent of the type of (semantic or syntactic) theory adopted (see Dowty, 1979; Rappaport Hovav & Levin, 1998; Ramchand, 2008; Alexiadou et al., 2015; Beavers & Koontz Garboden, 2020 and references therein). In Dowty (1979), Rappaport Hovav & Levin (1998 and later work), and Beavers & Koontz-Garboden (2020), the causative variant involves aCAUSE operator on top of the content of the anticausative one, while in Ramchand (2008) and in Alexiadou et al. (2015), the causative form involves additional syntactic structure. 2.2 DEVERBAL NOMINALIZATIONS The causative alternation itself has not been thoroughly investigated in the domain of deverbal nominalizations. Related claims have been made within the larger discussion on how nominalizations may resemble their base verbs and how much they may preserve from the verbal event structure, especially building on Grimshaw (1990). The main observation that Grimshaw (1990) makes is that (overtly suffixed) deverbal nominalizations may realize what came to be later known as argument structure nominals (ArgStr-Ns; in Grimshaw’s terms complex event nominals), which inherit the full argument and event structure of their base verbs, as in (4a), in contrast with result nominals (Res-Ns), which have lost any connection to the verb’s event structure and behave like lexical nouns, as in (4b). While the ArgStr-N examination of the patients in (4a) requires apredicate of events such as took along time and is incompatible with apredicate of individuals such as was on the table, the Res-N examination in (4b) behaves like the clipped lexical noun exam in this respect. (4a) The examination/*exam of the patients took along time/*was on the table. (4b) The examination/exam was on the table. OPEN ACCESS 36 LINGUISTICA PRAGENSIA 1/2025 Grimshaw’s claim is that ATK-Ns like examination are systematically ambiguous between the two readings, as in (4), while ING-Ns always form ArgStr-Ns.1 She further claims that Zero-Ns never form ArgStr-Ns due to their zero morphology. She consequently argues for asplit between overtly marked nominalizations, which may form ArgStr-Ns, and Zero-Ns, which never do. Borer (2013) further investigates this issue and strengthens Grimshaw’s original claims by contrasting Zero-Ns with either ATK-Ns or ING-Ns built on the same verb, as in (5). In these examples it is clear that the Zero-Ns do not form ArgStr-Ns, and only the overtly suffixed ATK and ING nominals do. (5a) the importation/*import of goods from China in order to bypass ecological regulations (5b) the walking/*walk of the dog for three hours Following such contrasts, Borer argues that Zero-Ns cannot accommodate any compositional event structure from their base verbs and they are idiosyncratic like lexical nouns. She uses these observations in support of abroader thesis against derivational zero morphology, given that, as she claims, it cannot encode acomplex syntax-semantics mapping, as required by ArgStr-Ns. However, more recent developments in the study of Zero-Ns have argued that these can form ArgStr-Ns as well, even if they are admittedly more idiosyncratic in that some do not primarily express events (Newmeyer, 2009; Lieber, 2016; Iordăchioaia, 2020; Iordăchioaia et al., 2020; Iordăchioaia & Melloni, 2023a). For instance, while Iordăchioaia et al. (2020) present Zero-Ns such as cut, crack, and bruise, which are less likely to form ArgStr-Ns, Iordăchioaia & Melloni (2023a) offer examples of ArgStr-Ns as in (6), taken from the Corpus of Contemporary American English: (6a) acontinuation of the drain of capital from the developing world (6b) our surrender of freedom and property under coercion and threat (6c) the spread of disease by travelers These new insights suggest that at least some Zero-Ns behave similarly to INGand ATK-Ns in their ability to inherit the event and argument structure of their base verbs in the shape of ArgStr-N readings. 2.3 (ANTI)CAUSATIVITY IN NOMINALIZATIONS Within the debate on the possible structural complexity of deverbal nominalizations and the general observation that causative readings are more complex than anticausatives, the literature has made several observations that are instrumental to our study. In particular, various claims lead to the expectation that ATK-Ns should be compatible with both causative and anticausative readings; that ING-Ns should allow 1 The latter part cannot hold, given ING-Ns such as building, painting and the like, which also have Res-Ns readings, but it is generally acknowledged that ING-Ns— like ATK-Ns— can always realize ArgStr-Ns, even if they may also allow Res-N readings (see Lieber, 2016). OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 37 causative readings but disallow anticausative ones; and that Zero-Ns should only allow anticausative readings, to the extent that they can be eventive at all. Below we summarize these claims. First, ATK-Ns built from causative alternating verbs as in (7c) have been shown to be ambiguous between acausative and an anticausative reading, basically encoding both readings of their alternating base verbs in (7a) and (7b). Adding aby-phrase disambiguates the nominals for a(passive) causative reading in (7c) (see Harley & Noyer, 2000; Alexiadou et al., 2009; Sichel, 2010; Alexiadou et al., 2013). ATK-Ns built from causative-only verbs, such as in (7d), convey only causative meanings, while ATK-Ns built from anticausative/unaccusative-only verbs, as in (7e), derive only anticausative readings, as expected: (7a) John accumulated wealth. (causative) (7b) Wealth accumulated (*by John). (anticausative) (7c) the accumulation of wealth (by John) (causative/anticausative) (7d) the destruction of the vase (by John) (causative-only) (7e) the disappearance of the plane (*by the hijackers) (anticausative-only) Second, acontrast has been observed between ATK-Ns and ING-Ns to the extent that the latter exclude aself-action interpretation in examples such as (8). While in (8a) the agent of the registration event may be identical to the theme the children, (8b) requires an agent different from the theme the children (Kratzer, 2003; Alexiadou et al., 2009). (8a) The registration of the children (was slow). (Ag≠Th; Ag=Th) (8b) The registering of the children (was slow). (Ag≠Th; *Ag=Th) Following Kratzer’s (1996) hypothesis that aVoice projection hosts the external argument of causative active and passive verbs, which in her analysis accounts for the disjoint effect between external and internal arguments, the restriction in (8b) indicates that ING-Ns must project Voice.2 Implicitly, this suggests that anticausative/ unaccusative readings (which lack Voice and external arguments) should not be possible with ING-Ns (Alexiadou et al., 2009, 2013; Iordăchioaia, 2008). Borer (2013, ch. 4) further substantiates the empirical contrast between ATK-Ns and ING-Ns (though in adifferent formal account) and argues that ING-Ns express only causative/transitive events and are excluded with unaccusative (and, implicitly, anticausative) verbs. She offers examples of INGand ATK-N pairs such as in (9), for which she takes the ING-N to be completely ungrammatical. Iordăchioaia & Werner 2 Kratzer (1996) discusses the contrast between eventive and adjectival passives as in (i) below. The eventive passive in (ia) projects Voice and requires an agent disjoint from the theme, while the adjectival passive in (ib) allows also areading in which the children have registered themselves, which indicates that Voice is not present in its structure: (i) a. The children are being registered (by somebody/*themselves). (Ag≠Th; *Ag=Th) b. The children are registered. (Ag≠Th; Ag=Th) OPEN ACCESS 38 LINGUISTICA PRAGENSIA 1/2025 (2019), however, argue that ING-Ns are only slightly less acceptable than the ATK-Ns, due to the latter being more entrenched with Latinate verbs.3, 4 (9a) the arrival/?arriving of the train (9b) the disappearance/?disappearing of the rabbit Although the differences between ATK-Ns and ING-Ns with regard to (anti)causativity may not be as strong as claimed by Borer, there is ageneral agreement in the literature that ING-Ns highlight the process part of acausative event to the exclusion of the result state, while ATK-Ns express both. The presence of an external argument and causativity are thus more prominent in ING-Ns than in ATK-Ns. But the question is whether this sensitivity can really be formulated as arestriction of ING-Ns to causative verbs. Third, Zero-Ns have only been touched upon in this discussion, but the few remarks predict acontrast in the opposite direction from ING-Ns: namely, they are expected to appear with anticausative but not with causative verbs. Smith (1972) contrasts Zero-Ns with ATK-Ns and argues that they are restricted to intransitive (unergative or unaccusative) readings. Cetnarowska (1993), meanwhile, compares zero with -ing and highlights the unergative reading of the Zero-N vs. the causative reading of the ING-N in examples such as (10): (10a) the beat of the heart (10b) the beating of prisoners Although (10) does not contrast acausative with an anticausative reading, the difference is reminiscent of the structural contrast between the two types of verbs: while causative verbs have (complex) bi-eventive structures, unergative verbs (like anticausatives) are simpler, as they have mono-eventive structures. Marantz (1997) addresses the same contrast from the causative/anticausative perspective and argues with data as in (11) that Zero-Ns realize only anticausative and not causative readings, in contrast with ING-Ns: (11a) The elevator’s rise to the top floor (11b) John’s raising/*raise of the glass 2.4 THE CHALLENGE The observations above show interesting differences between the three types of nominalizations, which have found some theoretical explanation in terms of the structural complexity that the different nominalizing suffixes may encode. These observations can be summarized in three points as in (12). 3 The judgments in (9) are taken from Iordăchioaia & Werner (2019). 4 Afurther intervening factor that may be responsible for the slightly reduced acceptability of the ING-N in (9) is the preference of -ing for atelic events, as argued in Borer (2013, ch.4), even though it is not entirely excluded with telic ones, as also indicated in Alexiadou et al. (2010, p.552–553). In our study discussed in Section 4, most nominalizations OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 39 (12) Previous generalizations on (anti)causative events in nominalizations: (12a) ATK-Ns allow both causative and anticausative readings. (12b) ING-Ns allow causative readings but disallow anticausative readings. (12c) Zero-Ns allow anticausative readings but disallow causative readings. The generalizations in (12) create acoherent picture about how the three nominalizing patterns might compete and complement each other, which would be worth testing experimentally in its own right. There is, however, one more reason to test them, namely, the fact that these contrasts have been challenged by empirical counterevidence. For instance, Lieber (2016, p.47–49) provides corpus examples with all three nominalization patterns built on unergative, unaccusative, as well as alternating (anti)causative verbs. (13) lists excerpts from Lieber’s examples, which challenge the generalizations in (12b) and (12c): (13a) and (13b) show ING-Ns with an unaccusative verb and an anticausative reading of an alternating verb (cf. (9b)), while (13c) presents aZero-N with the causative reading of an alternating verb. (13a) you are defining the Resurrection as the disappearing of the body (13b) that portion is the joy that feeds the growing of another life within you (13c) the officer was improper on his stop of this car In particular, we also find various counterexamples to the generalization in (12c), which is central to our discussion. (14a) (from the Corpus of Contemporary American English) challenges the claim about (11b), and (14b) is afurther example of acausative Zero-N (from the corpus News on the Web); see also the examples in (6): (14a) acontinuous raise of salaries (14b) the deliberate crash of apassenger jet 3 ACCEPTABILITY JUDGMENT STUDY Given the controversial picture of the behavior of the three types of deverbal nominalizations with respect to causative and anticausative/unaccusative verbs, we aim to experimentally test the generalizations in (12). As mentioned in Section 1, in this project we plan to investigate different aspects of the acceptability of the data by means of at least two studies. In afirst step we test the acceptability of pairs of nominalizations built on anticausative/unaccusative and causative verbs using native speaker judgments, and in asecond step we will have aforced choice interpretation study to check the compatibility of these nominalizations with contexts that impose causative vs. anticausative readings on alternating verbs. In this paper we present the results of the judgment study. encoded atelic events, and for the few telic ones, ING-Ns did not appear to be less acceptable. If inner aspect played any role, it was negligible for the final results. OPEN ACCESS 40 LINGUISTICA PRAGENSIA 1/2025 3.1 RESEARCH qUESTION AND GUIDING HYPOTHESIS The overarching goal of our research is to find out whether the zero nominalizing suffix is indeed more restricted in accommodating the meanings and structures of their base verbs compared to overt suffixes, as often claimed in the literature (see Grimshaw, 1990 and Borer, 2013 as representative for the zero nominalizing suffix in English). In this project we focus on causative vs. anticausative/unaccusative verbs against the background that causative verbs are structurally more complex than the anticausative/unaccusative ones. Following several studies on Zero-Ns in English and other languages, which challenge the claims about the inability of zero to host complex syntax-semantics (see Meinschäfer, 2005; Newmeyer, 2009; Fábregas, 2014; Lieber, 2016; Iordăchioaia et al., 2020; Iordăchioaia, 2020; Iordăchioaia & Melloni, 2023a), our guiding hypothesis is that Zero-Ns can express both causative and anticausative/unaccusative readings, i.e., zero morphology is not restricted to simple structures. In this respect, we expect Zero-Ns to behave like ATK-Ns. This means that we aim to test (and challenge) the generalization in (12c). Ideally, to this aim we should test Zeroand ATK-Ns built on the same verbs, and leave ING-Ns out. However, zero and ATK have strong selectional restrictions with respect to their bases and often find themselves in complementary distribution: cf. increase vs. outdated increasement, or archaic amaze, meanwhile replaced by amazement. This makes it difficult to find well-balanced pairs of such nouns. Moreover, -ing is the most productive nominalizing suffix in English, which means that an ING-N would also be in principle available for every verb with both Zero and ATK nouns, which could influence the speakers’ intuitions about their relative acceptability. For these reasons, we are testing verbs with INGand Zero-Ns as well as verbs with INGand ATK-Ns, thus aiming at acomparison of Zero-Ns and ATK-Ns mediated by ING-Ns. 3.2 STIMULI AND DESIGN The study included three types of base verbs: causative-only verbs (CAUS), anticausative/unnaccusative-only verbs (ANTI), and verbs which alternate between these two interpretations (ALT). Verbs were collected from VerbNet (https://verbs.colorado.edu/verbnet/, Kipper-Schuler, 2006), adatabase of English verbs organized into lexical semantic classes based on Levin (1993). Atotal of 24 verbs were chosen for the study: 8 CAUS, 8 ANTI, and 8 ALT. Each of these verbs formed pairs of nominals suffixed with either ATK and ING (4 verbs for each Verb Type) or Zero and ING (4 verbs for each Verb Type), for atotal of 48 separate nominalizations. Nominalizations derived from ALT verbs were included in this task so as to provide baseline acceptability ratings in eventive sentences, for use in our later investigations with explicitly causative and anticausative contexts (see Section 5). However, as these verbs are in principle ambiguous between causative and anticausative readings in the absence of acontext, their nominals may be as well, and it will therefore not be clear which reading is being judged in an isolated sentence. In what follows we thus limit our focus to the CAUS and ANTI verbs, as the only eventive reading their nominals should have is that of the base verb. OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 41 The total number of verbs eligible for inclusion was relatively limited for anumber of reasons: i. Although productive, ING-Ns may be quite infrequent, and not all corpora disambiguate them from other uses of -ing (e.g., the participle, the adjective, etc.). Verbs were therefore required to have their respective ING-N separately attested in the online Oxford English Dictionary (OED, www.oed.com) and assigned to frequency band 2 or higher (at least ~0,009 occurrences per million words). ii. The Zero-Ns had to be attested with event readings and argument structure, rather than solely result readings (based on their OED definitions and attestations in corpora such as COCA, NOW or GloWbE in Davies 2008-, 2013, 2016-; see the dataset in Iordăchioaia & Melloni, 2022). iii. ATK-Ns could not involve affixation onto suffixed bases (e.g., equal+iz+ation), as comparable cases of Zero affixation do not occur; similarly, we also did not include Zero-Ns derived from particle verbs (e.g., bail+out), as such verbs do not form nominals with ATK suffixes. Focusing only on the CAUS and ANTI verbs, there were therefore 32 nominalizations across 8 conditions, based on Verb Type (CAUS vs. ANTI), Pair Type (pairs with ATKNs vs. Zero-Ns) and Suffix (ING vs. Other). Base verbs were matched in frequency across the two Pair Types (ATK: 43,78 per million; Zero: 53,07; F1,30 = 0,151, p = 0,70).5 These 32 nominalizations are presented in Table 1 below. Mean frequencies of the verbs and nominalizations in each condition can be found in the Appendix. Verb Type ATK Verbs Zero Verbs ATK ING Zero ING CAUS correction nourishment reduction suppression correcting nourishing reducing suppressing massacre upload murder raise massacring uploading murdering raising ANTI emergence disappearance expiration putrefaction emerging disappearing expiring putrefying rise decay fall plunge rising decaying falling plunging table 1.List of items organized by Verb Type (CAUS, ANTI), Pair Type (ATK, Zero) and Suffix (ATK vs. ING, Zero vs. ING) 5 We also attempted to balance the various groups based on anumber of other factors, but were unable to do so due to the limited numbers of possible verbs and their distributional behavior. For example, as noted above, the ING-Ns are consistently much less frequent than their ATK-N or Zero-N partners. Additional factors include etymological origin (where ATK verbs are largely of Romance origin, and Zero verbs more balanced between Romance and Germanic), and morphological complexity of the base (where ATK verbs are largely complex, involving bound Latinate morphemes, while Zero verbs are largely simple). We attempt to account for these issues in our analysis and interpretation (see Section 4). OPEN ACCESS 48 LINGUISTICA PRAGENSIA 1/2025 Nominalization Mean SE ATK verbs, ING, ANTI 0,008 0,165 ATK verbs, ING, CAUS –0,134 0,165 Zero verbs, ING, ANTI 0,275 0,165 Zero verbs, ING, CAUS 0,158 0,165 table 5.Means for ING-Ns 5 CONCLUSIONS AND FUTURE WORK In this study we collected acceptability judgments on English nominalizations derived from causative and anticausative/unaccusative verbs in order to test previous theoretical claims according to which, first: zero nominalizers should not be able to realize causative event readings, and second: the suffix -ing should show arestriction with respect to anticausative readings and allow only causative ones. Both suffixes are expected to contrast with the ATK Romance suffixes, which are compatible with both types of verbs. The results of this study support the observation about the flexibility of ATK-Ns in accepting both causative and anticausative readings. However, they disconfirm the other two expectations, in that no difference was found between the three different types of nominalizations: all three of them are equally possible with both causative and anticausative verbs. While some numerical differences do show up in our data, further investigations are necessary to verify whether they may become significant in agreater pool of data. In conclusion, our results indicate that Zero, ATK and ING are very similar in their possible interpretation with respect to their derivation from causative and anticausative base verbs. All three suffixes may realize both the simpler anticausative/unaccusative and the more complex causative structures of their base verbs. Theoretically, figure 4.Fitted values for ING-Ns by Pair and Verb Type OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 49 this speaks in favor of treating Zero as asuffix, like the overt ones. However, this does not mean that these suffixes are semantically identical, as there may be other differences between them (e.g., -ing cannot denote states, while zero and ATK can; see Iordăchioaia & Werner, 2019). There are afew shortcomings of this study, which we aim to overcome in our follow-up research. First, although the results support our guiding hypothesis, this is the null hypothesis, and in view of the small number of verbs tested, the absence of any statistically significant patterns may simply be the result of too little statistical power. In order to further assess these results, two additional studies may prove helpful. First, we will run alarger acceptability judgment study, aiming to investigate whether Zero-Ns which appear to inherit argument structure are consistently (relatively) acceptable in eventive contexts, regardless of whether they encode argument structure. We will do this by comparing these Zero-Ns to those which form Res-Ns, embedding them in sentences with the predicates tested here (occur, take place, happen). In principle, the Res-Ns should be significantly less acceptable in these contexts; thus, if there is aclear difference, then this would support the idea that some ZeroNs can indeed form ArgStr-Ns, and that their relative acceptability is not purely the result of coercion or repair. Second, we aim to run aforced-choice study in which we will provide causative vs. anticausative contexts (including alternating verbs) and ask participants to select one of the two competing nominalizations (which in the current study have been rated as acceptable). If anominalization pattern has aclear preference for one reading or the other, this task should reveal it, especially as forced choice tasks have been shown to have greater statistical power than Likert-scale tasks (Sprouse & Almeida, 2017). This future work may therefore help to more clearly reveal any differences between types of nominalizations and the contexts in which they can appear. Acknowledgements We thank our student assistants Anna-Lena Feichter, Mariya Kavaldzhieva and Nina Richter for help in the preparation of the experiment, and Iordăchioaia acknowledges the Spanish Ministry of Science, Innovation and Universities’ project PID2020-119851GB-I00 based in the University of Granada. REFERENCES Alexiadou, A., Anagnostopoulou, E. & Schäfer, F. (2009). PP licensing in nominalizations. In A.Schardl, M.Walkow & M.Abdurrahman (Eds.), Proceedings of NELS 38 (pp.38–52). Amherst, MA: GLSA. Alexiadou, A., Anagnostopoulou, E. & Schäfer, F. (2015). External arguments in transitivity alternations: Alayering approach. Oxford: Oxford University Press. Alexiadou, A. & Grimshaw, J. (2008). Verbs, nouns, and affixation. In F.Schäfer (Ed.), Working Papers of the SFB 732 Incremental Specification in Context Vol.1 (pp.1–16). Stuttgart: Universität Stuttgart. Retrieved from http://dx.doi.org/10.18419/ opus-5695 Alexiadou, A., Iordăchioaia, G. & Soare, E. (2010). Number/Aspect Interactions in OPEN ACCESS 50 LINGUISTICA PRAGENSIA 1/2025 the Syntax of Nominalizations. Journal of Linguistics, 46(3), 537–574. Alexiadou, A., Iordăchioaia, G., Cano, M., Martin, F. & Schäfer, F. (2013). The realization of external arguments in nominalizations. Journal of Comparative Germanic Linguistics, 16, 73–95. Anwyl-Irvine, A.L., Massonnié, J., Flitton, A., Kirkham, N.Z., & Evershed, J.K. (2020). Gorilla in our midst: an online behavioural experiment builder. Behavior Research Methods, 52, 388–407. Asher, N. (1993). Reference to Abstract Objects in Discourse. Dordrecht: Kluwer. Bates, D., Kliegel, R., Vasishth, S. & Baayen, H. (2015). Parsimonious mixed models. Retrieved from http://arxiv.org/ abs/1506.04967,ms.,arXiv Bates, M.M., Bolker, B. & Walker, S. (2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1–48. Retrieved from https://dx.doi.org/10.18637/ jss.v067.i01 Bauer, L. & Valera, S. (2005). ‘Conversion or zero-derivation: An introduction’. In L.Bauer & S.Valera (Eds.), Approaches to conversion/ zero-derivation (pp.7–17). Münster: Waxmann. Beavers, J. & Koontz-Garboden, A. (2020). The Roots of Verbal Meaning. Oxford: Oxford University Press. Borer, H. (2013). Structuring sense: Volume III: Taking form. Oxford: Oxford University Press. Cetnarowska, B. (1993). The Syntax, Semantics and Derivation of Bare Nominalisations in English. Katowice: Wydawnictwo Uniwersytetu Śląskiego. Dahl, E. & Fábregas, A. (2018). Zero morphemes. Oxford research encyclopedia of linguistics. Oxford: Oxford University Press. Retrieved from https://doi.org/10.1093/ acrefore/9780199384655.013.592 Davies, M. (2008-). The Corpus of Contemporary American English (COCA). Retrieved from https://www.englishcorpora. org/coca/ Davies, M. (2013). Corpus of Global Web-based English (GloWbE). Retrieved from https://www. englishcorpora.org/glowbe/ Davies, M. (2016-). Corpus of News on the Web (NOW). Retrieved from https://www.Englishcorpora.org/now/ Don, J. (1993). Morphological conversion [Doctoral dissertation, Utrecht University]. Don, J. (2005). On conversion, relisting and zero-derivation. SKASE Journal of Theoretical Linguistics, 2(2), 2–16. Dowty, D. (1979). Word Meaning and Montague Grammar. Dordrecht: Kluwer. Fábregas, A. (2014). Argument structure and morphologically underived nouns in Spanish and English. Lingua, 141, 97–120. Farrell, P. (2001). Functional shift as category underspecification. English Language and Linguistics, 5(1), 109  130. Grafmiller, J. (2013). The Semantics of Syntactic Choice: An Analysis of English Emotion Verbs [Doctoral dissertation, Stanford University]. Grimshaw, J. (1990). Argument structure. Cambridge, MA: MIT Press. Harley, H. & Noyer, R. (2000). Formal versus encyclopedic properties of vocabulary: Evidence from nominalisations. In B.Peeters (Ed.), The Lexicon-Encyclopedia Interface (pp.349–374). Amsterdam: Elsevier. Huang, N., Almeida, D., & Sprouse, J. (in press). Anearly-exhaustive experimental investigation of bridge effects in English. To appear in Language. Retrieved from https:// www.jonsprouse.com/papers/Huang%20 et%20al.%20-%20bridge.pdf Iordăchioaia, G. (2008). External argument PPs in Romanian nominalizations. In F.Schäfer (Ed.), Working Papers of the SFB 732 Incremental Specification in Context Vol.1 (pp.71–84). Stuttgart: OPUS. Retrieved from http://dx.doi. org/10.18419/opus-5695 Iordăchioaia, G. (2020). Categorization and nominalization in zero nominals. In A.Alexiadou & H.Borer (Eds.), Nominalization: 50 Years on from Chomsky’s Remarks (pp.231–253). Oxford: Oxford University Press. Iordăchioaia, G. & Melloni, C. (2022). Zeroderived nouns and deverbal nominalization: Databases for English and Italian (1.0) OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 51 [Dataset]. Zenodo. Retrieved from https://doi. org/10.5281/zenodo.6384134 Iordăchioaia, G. & Melloni, C. (2023a). The zero suffix in English and Italian deverbal nouns. Zeitschrift für Sprachwissenschaft— Journal of the Linguistic Society of Germany, 42(1), 109–32. Iordăchioaia, G. & Melloni, C. (2023b). Zero affixes in derivational morphology: Introduction. Zeitschrift für Sprachwissenschaft— Journal of the Linguistic Society of Germany, 42(1), 1–11. Iordăchioaia, G., Schweitzer, S., Svyryda, Y. & Buitrago Cabrera, C.M. (2020). Deverbal zero-nominalization and verb classes: Insights from adatabase. Zeitschrift für Wortbildung/ Journal of Word Formation, 4(2), 120–142. Iordăchioaia, G. & Werner, M. (2019). Categorial shift via aspect and gender change in deverbal nouns. Language Sciences, 73, 62–76. Kiparsky, P. (1982). From cyclic phonology to lexical phonology. In H. van der Hulst & N.Smith (Eds.), The structure of phonological representations (pp.131–175). Dordrecht: Foris. Kipper-Schuler, K. (2006). VerbNet: ABroad-coverage, Comprehensive Verb Lexicon [Doctoral dissertation, University of Pennsylvania]. Kratzer, A. (2003). The Event Argument and the Semantics of Verbs [Manuscript]. University of Massachusetts at Amherst. Retrieved from https://works.bepress.com/angelika_ kratzer/5/ Kratzer, A. (1996). Severing the external argument from its verb. In J.Rooryck & L.Zaring (Eds.), Phrase Structure and the Lexicon (pp.109–137). Dordrecht: Kluwer. Lenth, R. (2024). emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 1.10.4. Retrieved from https:// CRAN.R-project.org/package=emmeans. Levin, B. (1993). English Verb Classes and Alternations. Chicago, IL: University of Chicago Press. Lieber, R. (1981). On the organization of the lexicon [Doctoral dissertation, University of New Hampshire]. Lieber, R. (2016). English Nouns. The Ecology of Nominalizations. Cambridge: Cambridge University Press. Marantz, A. (1997). No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In University of Pennsylvania Working Papers in Linguistics Vol.4.2 (pp.201–225). Marchand, H. (1969). The categories and types of present-day English word formation. München: C.H.Beck. Marty, P., Chemla, E. & Sprouse, J. (2020). The effect of three basic task features on the sensitivity of acceptability judgment tasks. Glossa, 5, 1–23. Meinschäfer, J. (2005). Deverbale Nominalisierungen im Französischen und Spanischen. Ein Modell der Schnittstelle von Syntax und Semantik [Habilitation thesis, University of Konstanz]. Newmeyer, F. (2009). Current challenges to the lexicalist hypothesis: An overview and acritique. In W.Lewis, S.Karimi, H.Harley, & S.Farrar (Eds.), Time and Again: Theoretical Perspectives on Formal Linguistics (pp.91–117). Amsterdam, Philadelphia: John Benjamins. Plag, I. (2003). Word-formation in English. Cambridge: Cambridge University Press. Ramchand, G. (2008). Verb Meaning and the Lexicon. AFirst Phase Syntax. London, New York: Cambridge University Press. Rappaport Hovav, M. & Levin, B. (1998). Building Verb Meanings. In M.Butt & W.Geuder (Eds.), The Projection of Arguments: Lexical and Compositional Factors (pp.97–134). Stanford, CA: CSLI Publications. Schäfer, F. (2009). The causative alternation. Language and Linguistics Compass, 3(2), 641–681. Sichel, I. (2010). Event structure constraints in nominalization. In A.Alexiadou & M.Rathert (Eds.), The Syntax of Nominalizations across Languages and Frameworks (pp.151–190). Berlin: Mouton De Gruyter. Smith, C. (1972). On causative verbs and derived nominals in English. Linguistic Inquiry, 3, 36–38. OPEN ACCESS 52 LINGUISTICA PRAGENSIA 1/2025 Sprouse, J., & Almeida, D. (2017). Design sensitivity and statistical power in acceptability judgment experiments. Glossa, 2, 14. Sprouse, J., Caponigro, I., Greco, C. & Cecchetto, C. (2016). Experimental syntax and the variation of island effects in English and Italian. Natural Language and Linguistic Theory, 34, 307–344. Vendler, Z. (1968). Adjectives and Nominalizations. The Hague: Mouton. Gianina Iordăchioaia Department of Linguistics, Faculty of Humanities, University of Graz Merangasse 70, 8010 Graz, Austria ORCID: 0000-0002-8847-6307 [email protected] Jeannique Darby Institute of Language and Literature, Faculty of Humanities and Teacher Training, Volda University College Joplassvegen 11, 6103 Volda (Norway) ORCID: 0009-0006-3299-227X [email protected] APPENDIX A. FREqUENCIES (OED, # PER MILLION OCCURRENCES) A.1 BASE VERBS Pair Type Verb Type Mean SD ATK Verbs CAUS 61,25 92,77 ANTI 26,30 28,63 Zero Verbs CAUS 27,88 48,28 ANTI 78,25 92,36 A.2 NOMINALIZATIONS Pair Type Verb Type Suffix Mean SD ATK Verbs CAUS ATK 22,75 25,91 ING 0,05 0,03 ANTI ATK 7,58 8,61 ING 0,04 0,04 Zero Verbs CAUS Zero 9,55 13,77 ING 0,26 0,43 ANTI Zero 20,25 17,90 ING 1,30 1,45 OPEN ACCESS GIANINA IORDăCHIOAIA — JEANNIqUE DARBY 53 B. OUTPUT OF LINEAR MIXED EFFECTS MODEL B.1 MODEL STATISTICS Effect Estimate SE t Intercept 0,008 0,165 0,05 Pair Type[Zero verbs] 0,267 0,231 1,16 Verb Type[CAUS] –0,142 0,231 –0,61 Suffix [Other] 0,269 0,231 1,16 Pair Type [Zero verbs]:Verb Type[CAUS] 0,025 0,326 0,08 Pair Type[Zero verbs]:Suffix[Other] 0,013 0,326 0,04 Verb Type[CAUS]:Suffix[Other] 0,118 0,326 0,36 Pair Type[Zero verbs]: Verb Type[CAUS]:Suffix[Other] –0,269 0,461 –0,58 Random effects Variance SD Subject (intercept) 0,025 0,16 Item (intercept) 0,088 0,30 Residual 0,435 0,66 Note: Contrasts were treatment coded with ATK verbs, ANTI, and ING as the reference levels. B.2 LEAST-SQUARES MEANS Pair Type Verb Type Suffix Mean SE ATK Verbs CAUS ATK 0,253 0,165 ING –0,134 0,165 ANTI ATK 0,276 0,165 ING 0,008 0,165 Zero Verbs CAUS Zero 0,289 0,165 ING 0,158 0,165 ANTI Zero 0,557 0,165 ING 0,275 0,165 OPEN ACCESS