AthemAtic PArticiPle choice is Predicted by Argument structure

This article presents two experimental studies that investigate the role of syntax in constraining morphological processes related to past participle forms in Brazilian Portuguese. Analyses within the autonomous approach to morphology have defended that diachronically information Aronoff (1994), ChAgAs (2007). however, the results of two such forms, along the lines of LoBATo (1999) and CALABrese ResUmo Apresentamos dois estudos experimentais com vistas a investigar o papel da estrutura de diferentes análise para as formas curtas e longas. na proposta da morfologia autônoma Aronoff (1994) ChAgAs gramática. em contrapartida, análises como LoBATo (1999) e CALABrese (2014) será sua forma participial. Athematic Participle Choice is Predicted by Argument Structure


introduction
Recently the interface between syntax and morphology has been subject to detailed investigation.within the minimalist program this interface is viewed as a point in the derivation in which the output of syntax is delivered to morphology, which will, based on the properties of the syntactic output, apply operations and mechanisms to guarantee the convergence of the derivation on the PF side.this integrated architecture of the grammar is well developed within the minimalist Program and the distributed morphology framework hALLe & mARANtz (1993), hARLey & NoyeR (2000), embiCK (2010), bobALJiK (2011).within these formalisms, morphology is intrinsically dependent on syntax, and there is only one computational system, which operates merging features at different structural levels, going from a zero level derivation within words to a maximal level derivation within phrases.this integrated view is disputed, however.since ARoNoFF (1994), it has been proposed that morphology can be a completely independent component of the grammar.that is, the rules, operations and mechanisms used by morphology do not take into consideration information from other components -syntax, semantics and/or phonology ARoNoFF (1994).
Research into Romance languages has presented a series of evidence in favor of this autonomous view of morphology.mAideN (2005), for instance, has defended that the subjunctive paradigm is a morphome, an autonomous morpheme.in his analysis, all the cells of the subjunctive paradigm piggyback on the 1sg form of the indicative mood, forming what has been called the 'L-shaped morphome', as shown in table 1: tAbLe 1: the L-shaped morphome for verbs such as ouvir 'to hear': the same stem is shared among the 1sg present indicative and all forms of the subjunctive, to the exclusion of the 2sg/3sg indicative mAideN (2005).Along similar lines, it has been proposed that past participle forms (long vs. short) of the type found in Romance are also based on the 1sg of the indicative, constituting another example of morphome: a paradigm that emerges based solely on a morphological inter-paradigm relation with the 1sg cell of the indicative mood ChAgAs (2007).
the L-shaped paradigm as a productive base of formation for the subjunctive mood was contested by NeviNs & RodRigUes (2013).the authors conducted a series of experiments using wug forms in different Romance language (european Portuguese, spanish and italian) and concluded that Romance speakers have a statistically consideration syntactic/formal features, such as person and mood, rather than arbitrary morphological geometric shapes within paradigms.while L-shaped irregular verbs may have had a diachronic foothold in the language, they are no longer productive, arguably due to the fact that speakers have gradually refused to extend/transmit a pattern with no grounding in syntactic natural classes.
in the present article, we will present two experiments that employ a pseudoword (or 'wug') methodology, investigating past participle forms: long vs. short.the results show that the licensing of short forms depends heavily on the argument structure of the verb in question.the lighter the argument structure of verb, the higher the preference for short-form participles.thus, it follows that the morphological process of forming past participles piggybacks on the syntactic and semantic information contained within the verbal root.this, together with the results of NeviNs & RodRigUes on subjunctive formation, undermines the main evidence that researchers have presented within Romance languages to justify an autonomy of morphology.
the paper is organized as follows: in section 2 we present data from brazilian Portuguese, which displays innovative short-form participles.in section 3, we discuss two different analyses for the phenomenon: Chagas' proposal defending an autonomous morphology based analysis, and LobAto/CALAbRese's analysis, relating participle forms with syntactic structure.in section 4, we present two experiments conducted in brazilian Portuguese, testing the hypothesis that verbal argument structure plays a role in determining the form of participles.in these experiments four verbal classes were considered: unaccusatives, relation between verbal argument structure and past participle forms.Unaccusative verbs have a preference for short-form participles, whereas transitives and ditransitives do not.Psych verbs, on the other hand, prefer long participle forms.A theory based on the autonomy of morphology does not account for these results.therefore, our conclusion (section 5) shows that a theory along the lines proposed by Lobato/Calabrese has a better chance to explain the formation of past participle forms, as morphology seems to be syntactically grounded.

Past Participles: short and long forms
Romance languages in general present regular and irregular past participle forms.As the Portuguese past participles in (1) illustrate, regular forms keep the thematic vowel of the original verb and add the morpheme -d, whereas irregular forms do not.thus, regular forms are often called the long forms and the irregular forms are called 'rhizotonic', 'athematic', or simply short-form participles: (1) secado/seco, limpado/limpo, aceitado/aceito, acendido/aceso, dried, clean, accepted, ignited however, not all verbs accept both forms.in european Portuguese and also in some dialects of brazilian Portuguese, verbs like the ones in (2) accept only long participles, whereas others verbs (3) are compatible with short-form participles only.
(2) matado/*mato, apaixonado/*apaixono, fechado/*fecho killed, impassioned, closed (3) *abrido/aberto, *escrevido/escrito, *dizido/dito opened, written, said Romance speakers in general seem to have a strong preference for long forms.this is clearly observed during the acquisition process of Portuguese, a period in which long forms for the verbs in (3) are readily found.Adults might also have this preference; tUCKeR (2000) shows that italian adult speakers, when introduced to unfamiliar italian verbs, have a strong preference for forming long participles based on standard dialects of brazilian Portuguese (henceforth bP) involves the spreading of short forms for past participles that historically have only long forms.For example, although the short-form participles in (4) are unacceptable in european Portuguese and standard brazilian Portuguese, they are found in many varieties of colloquial bP. in the present discussion, we restrict ourselves to varieties spoken in Rio de Janeiro & minas gerais.
(4) chego, trago, compro arrived, , (cf.chegado, trazido, comprado) the short and long forms seem to be accepted in both passive and past perfect contexts, as shown in (5). the short form, however, doesn't in (5b), the perfect in (6c), and optionally the passive and the perfect in (5d).
(5) Forms like (5b) acender/aceso are largely frozen, though examples of this sort are widely found in italian (e.g.mettere/messo 'to put').diachronic developments from the so-called stems inherited from Latin (see bAChRACh & NeviNs, 2008 for an overview of approaches to this phenomenon).Notably, the irregular/short participles can be called athematic, as they lack the theme vowel (preserved in apagado but lost in aceso). in LobAto (1999), it is suggested that these innovative participles involve an alternative spell-out, without the theme vowel and the participial morpheme (the nature of which has seen various proposals across the Romance languages, e.g.deverbalizer, resultative, etc).
what is largely unique to contemporary bP (although apparently parallels are found in sicilian italian; cf.mAideN, 2013) is the emergence and indeed, some of which are considered substandard or variant, but are nonetheless increasing in attestation.this may be a result of the grammaticalization and generalization of sporadic hypercorrective tendencies that are found in a 'diglossic' situation such as bP, in which the prescriptive norm is rather distant from the colloquial forms and in which speakers across all class levels often feel uncertain about 'correct' usage.given the stigmatized usage of overregularized and 'natural' productions such as % foi instead of foi 'was open', it may be that speakers come to treat athematic short-forms as 'correct', paving the way for extension and innovation of short-forms to etymologically novel contexts.Nonetheless, our central emphasis in this paper is to raise the point that even hypercorrection often submits to grammaticallybased generalizations, and, as a result, the 'sporadic' tendency to produce short-forms may not apply equally across-the-board.
whatever the source of the innovation (which, to our mind, must interestingly, for some verbs, the passive particle favors the shortform while the perfect favors the long form (e.g.imprimir 'to print', with ser impresso, ter imprimido; see sCheR et al. (2014) for detailed discussion).this preference for either the short or the long form might indicate that these forms are near doublets in the sense of KRoCh (1994).the contrast in (6), for example, suggests that they do not have the same syntactic distribution, and might not have the same meaning (7), as discussed in PiRes (1998), LobAto (1999), boeChAt (2008)  'the dishes having become dry, put them on the table' indeed, some of these differences may be akin to embiCK 's (2004) distinction between participial forms in english such as an open door and an opened door, and mAideN (2013: 509) explicitly suggests such an interpretive difference for sicilian short-form vs long-form participles such as apì rtu/graputu 'open/opened'.while these potential syntactic and semantic differences merit further exploration, in this paper we will focus on testing different classes of verbs to verify if argument structure will not pursue them in this paper, the syntactic and semantic differences above might follow from our conclusion that the form of the derived participle depends on the internal argument structure of the original verb.some native speakers have the intuition that (8a) is better than (8b), which in turn is better then (8c).importantly, the main difference between these examples is the argument structure of verbs. in (8a) the participle comes from an unaccusative verb, whereas in (8b) and (8c) it comes from a transitive and a ditransitive verb.thus, it is in principle worthwhile investigating a potential relation between argument structure and participle forms.if this is on the right track, the general conclusion about the architecture of grammar is that there is a close integration between syntax and morphology, contrary to the autonomous morphology proposal.

Autonomous morphology or syntactically-grounded morphology?
given the data in (5), one tempting generalization that can be contemplated is that the innovative participle short-forms are 'imitations' of the 1sg.ind forms.while this is not the case for the entrenched forms in (5a-c), ChAgAs (2007) puts forth an analysis that short-forms are the result of an inter-paradigmatic relation: bP innovative shortform participles are formed based on the 1sg of the indicative mood.According to this author, this relation is what licenses the increase in productivity of short-form participles in bP.
this inter-paradigmatic relation, however, is not so easily conceived as the 1sg.ind in particular has no privileged morphosyntactic closeness person/mood combinations.in other words, for ChAgAs' analysis to be correct, short-form participles would have to be morphomesindependent productive pieces of morphology that are truly arbitrarily -resulting from 'purely morphological correspondences', showing no sensitivity to syntactic properties of the original verbs.
in contrast with ChAgAs' proposal, under analyses such as LobAto's (1999), the short-forms are alternative spell-outs of a syntactic structure, and no matter which details of that structure one As such, the syntax of the verbs in question comes to be relevant in understanding the morphophonological form of past participles.
observations such as Lobato's are echoed in formal analyses of related phenomena within distributed morphology (e.g.embiCK, 2004 andCALAbRese, 2014).CALAbRese, for instance, proposes that the absence of the theme vowels in past participles allows for locality between the verbal root and tense, enabling contextual allomorphs such as perso 'lost' in italian perdere) and eleito 'elected' in Portuguese eleger).that is, the theme vowel projects a head that blocks a structural relation between t and the verbal root, and therefore, shortform participles, which lack the theme vowel, are lighter in structure than long forms (which have the extra functional head projected by the conclude that past participle forms are not morphomes at all, as syntax and morphology work together in determining their shape. 2nder a theory in which the relation between short-form participles' spell-out and the word-internal syntax of participle expressions is directly play potential roles in conditioning the acceptability of such forms.thus, it is relevant to mention that of the extant innovative short-forms, very few are found with ditransitives and psych-verbs such as agradar 'to please', 'to annoy', and temer 'to fear'.Nonetheless, this may be related to token frequency in the language or other grammar-external factors.thus, to test the hypothesis that argument structure plays a role in the formation of past participles without having the interference of external variables, wug-testing presents itself as extremely useful, as wugs have a token frequency of zero, allowing participants to rely on the phonological form, the syntactic environment, or neither of the above, in rating the goodness of short-form participles given a particular .

past participle forms
we present two experiments designed to test whether short-form participles are productively extended to novel 'wug' verbs, and the extent to which verb class membership can differentially modulate the preference for short-forms.experiment 1 tests four major classes of class of psych-verbs.

experiment 1: verbal classes vs. preference for shortform participles
in this experiment we tested participles formed based in verbs from four lexical semantic structure classes: inherent and non-inherent unaccusatives (9a,b), non-alternating transitives (10), ditransitives (11) and psych verbs (12). 33 For a detailed description of these classes, see LeviN (1993).The play moved the audience 'the play moved the audience' this brings us directly to the hypothesis under evaluation in (13): (13) verbs with less argument structure will display a stronger preference for short-form participles.
this hypothesis predicts then that unaccusative verbs, which have arguably less lexical structure than transitive, ditransitives and psych verbs ChomsKy (1995), hALe & KeyseR (1993) will be more prone to realization via short-form participles, in accordance with the speakers' intuitions presented in (8).

material, design and participants
the use of wugs was necessary to avoid the interference of nonrelated variables, such as token frequency, and the entrenchedness or diachronic 'frozenness' of particular forms.in addition, as sometimes bP speakers are too aware of right vs. wrong in terms of traditional grammar and at times they can be quite sensitive to 'wrong' participle forms, we opted for using wugs in order to minimize these issues.
we did not look for contextual allomorphs, such eleger/eleito 'to elect (inf./participle)',i.e. short-form participles with distinct stem allomorphs, as we concentrated on testing the role of argument structure (L-syntax, in hALe & KeyseR's, form participles differing only in the presence/absence of the thematic vowel and the morpheme, such as chego/chegado 'arrived (short.form/long.form)'.
Participants were exposed to 24 wug forms: 6 wugs for each argument structure class.the wugs were also equally distributed in the three verbal conjugation paradigms, with two wugs per conjugation.this is illustrated below: tAbLe 2: wugs for transitive verbs.the wugs were created taking into consideration the phonotactics of bP.they all had two syllables, and they were created so as to minimize resemblance with existing verbs of bP from the same verb class.A full list of stimuli is included in Appendix 1.
Filler trials, 30 in total, were also wugs and involved unrelated morphophonological alternations such as the diminutive -inho and , nominalizations with vs.
, and other nominally-based productivity-related alternations.
were intended to be transparent with respect to the argument structure pandar the 1sg.ind (e.g.pando) in a later clause.it is not possible to test inherent unaccusatives in a passive context.thus, we elicited the participle forms using the analytic perfect in combination with the auxiliary ter. the wellformedness of the short-form vs. long-form participles were judged by ( 14) eu acho que pandar a casa todos os dias é importante para manter tudo limpo.eu pando a minha casa bem, mas minha mãe, que é mineira, tem um capricho incrível e panda muito bem.No ano novo, depois que ela tinha __________ (pando/pandado delicioso. 'i think that to pandar the house every day is important in order to keep everything clean.i pando my house very well, but my mom, who is from minas, has an incredible knack and panda very well.on New year's eve, after she had ______________ (pando/pandado) the house, it smelled wonderful.'(15) Malir dinheiro para familiares é muito perigoso.eu nunca malo grana male toda a grana que ele tem para os primos.se eu fosse ele, eu não teria _________ (malido/malo) 40,000 Reais para aquele primo esquisito dele que mora na Alemanha.
'to malir money to relatives can be dangerous.i never malo funds to anyone, but poor Fernando might end up penniless.he males all his money to his cousins.if i were him, i wouldn't have ______________ (malido/malo) 40,000 bucks to that weird cousin of his living in germany.' A scale 1-5 was used to indicate relative preference for the form on the left (1 indicating absolute preference for the form on the left; 5 indicating absolute preference for the form on the right; 3 indicating equal preference, and 2 and 4 indicating stronger but not absolute preference for one of the forms).order.
the experiment was conducted with the research platform online (https://www.onlinepesquisa.com),and took around 10 minutes to be completed.186 participants logged in; however, only 98 of them completed the experiment.we did not control for social variance among the participants (e.g.age, educational background and native dialect).
with the independent variables of interest, namely these four classes of verbs.while one particular speaker may overall favor short-form participles more than another speaker, nothing in their extralinguistic demographics per se should cause such a preference to emerge more in some verb classes than in others.

results
the results (n=98) were statistically analyzed using the open-source software r. the analysis conducted, in a within-subjects design, tested the effects of type of argument structure on short vs. long participle forms.the graphic in Figure 1 crosses the types of argument structure considered in experiment 1 with the relative preference for shortform participles, where a higher rating indicates a greater tolerance for short-form participles.Pairwise comparisons were conducted with two-tailed paired t-tests between each comparison of interest.As participle forms when compared with transitives, ditransitive and psych verbs.the comparison between transitives and ditransitives resulted in unaccusatives, transitives and ditransitives.
FigURe 1 experiment 1 results with means by verb class (n=98), on a scale of 1 to 5, where higher numbers represent a greater preference for the athematic participle (e.g.pando) as opposed to the long-form (e.g.pandado).Comparisons among means are represented by p-values or by

discussion
these results demonstrate a clear sensitivity to syntactic aspects of the lexical structure of the verbs from which participles are formed.speakers pay attention to phrasal aspects of the verb in question in determining whether a theme vowel and the of the corresponding participle form are omissible or not.
A straightforward interpretation of these facts emerges rather easily, the hypothesis that unaccusatives have no little v, (di)transitives have a little v, and psych verbs little v plus Causer (e.g.PesetsKy 1995), the ranking of the three levels of omissibility in short-forms directly corresponds to the amount of structure: those verbs with the least amount of heads tolerate short-forms the most, while those with a richer structure prefer to expone it using the long form.suppressability of thus corresponds to amount of structure, and this parallels a great additional syntactic structure building, in a way potentially related to the monotonicity hypothesis of KooNtz-gARbodeN (2008). 4 4 Koontz-garboden's hypothesis asserts that "word formation operations do not remove operators from lexical semantic representations".this makes the prediction that a verb of non-causative changes of state (e.g.unaccusatives) cannot be derived from a verb of causative changes of state (e.g.transitive), because that would involve a morphological operation to delete the cause operator from the lexical representation of the verb.this is indeed visible in Quechua, where less morphology is equal to intransitivization and more is equal to transitivization: (i) hatun-ya hatun-ya-chi languages (e.g.Pima, spanish) suggest the opposite.KooNtz-gARbodeN presents, however, a series of evidence that in these languages the causer operator is not deleted from the lexical conceptual structure of the verb in its intransitive version.our research on athematic participles goes in the same basic direction, showing that, similarly to the data in (i), the short vs. long forms of past participles in Portuguese suggest differences in argument structure.Under the assumption that lexical semantic representations are mapped onto syntactic structure (e.g.hALe and KeyseR, 1993), one can entertain the hypothesis that less syntactic structure requires less morphology.
the relevance of structure may not be limited to simply counting the number of arguments (although this works surprisingly well); in addition, event-structural properties of the verbs (as diagnosed by the relative admissibility of telic-oriented adverbs such as in two hours and get-passives) pair along with the well-formedness of short-form passives in the set below.John feared the situation *in/for two hours b. *John got feared ditransitives patterned similarly to transitives in bP, as shown in Figure 1. this may be related to the fact that bP ditransitives are all of the put-type (involving a direct object NP and a goal PP), and none are of the send-type (involving a double object structure with a higher goal NP) that would involve an applicative structure mARANtz (1993) and thereby an additional argument-introducing functional head.to be necessary as different classes of psych-verbs behave different syntactically.For example, while verbs that place the experiencer in the external argument position [spec, vP] (subjexp verbs), such as fear in (19), are unable to form , verbs that maps the experiencer onto the internal argument position (objexp verbs), such as worry and excite, are able to do so: (20) a. i got worried with Karl's decision b. i got excited by the news in addition, while subjexp verbs do not exhibit an intransitive alternation in terms of middle formation (21a), while objexp verbs do, as shown in (21b): Notice, however, that although objexp verbs can display a middle alternation, they are incompatible with a fully unaccusative frame (22).this might be related to the fact that objexp verbs must project an extra layer of argument structure, which is related to the external participant of the event -a Causer in PesetsKy's (1995) terms.objexps are permitted in middle structures, however, given that the latter involve a mapping of the external argument onto the syntactic structure hoeKstRA & RobeRts (1993), KeyseR & RoePeR (1992), stRoiK (1992)  At any rate, although all psych verbs might display an equally complex argument structure, they do display syntactic differences.thus, it is in principle possible that they pattern differently with respect to past participle formation.in experiment 1 we did not control for subjexp vs. objexp psych verbs.therefore, we decided to launch a second experiment to verify if there is any difference between these two classes of verbs with respect to the formation of short-form participles.

experiment 2: subject experiencer vs. object experiencer psych verbs.
beLLetti & Rizzi (1998) observed the existence of three natural classes of psych verbs in italian: temere 'to fear', preoccupare 'to worry' and piacere 'to please'.Although according to these authors, preoccupare and piacere are different structurally,5 they have in common the fact that both of them map the experiencer onto a position within the vP domain, whereas temere maps the experiencer onto a position outside the vP domain.For that reason, verbs of the class of temere are called subject experiencer (subjexp) verbs, and verbs of the class of preoccupare and piacere are called object experiencer (objexp) verbs.
having granted the beLLetti & Rizzi classes of two psychverbs, we designed a second experiment to investigate the following question: (23) do subjexp and objexp psych verbs behave differently with respect to the licensing of short-form participles?

material, design and participants
similarly to experiment 1, in experiment 2 we used wug psych-verbs to probe participants' preference for short-form participles.suppletive participle forms (e.g.confundir/confuso 'confuse.inf/confused.participle') were not considered.there were 12 target wug forms: 6 wugs for each class of psych-verbs, equally distributed within the three verbal conjugation paradigms (2 for each conjugation).in table 3, we include the subjexp verbal wugs and in table 4 the objexp verbal wugs that were tested.As shown, all wugs respected the phonotactic constraints of bP and they all had three syllables.Filler trials (24 in total) were also wugs involving noun formation with alternations such as the diminutive -inho and , nominalizations with vs.
, and other nominally-based productivity-related alternations.
intended to make the counter-hypothesis (i.e.short forms are formed based on 1sg -autonomous morphology) stronger.the participle forms were elicited in the context of 3sg perfect tense in combination with the auxiliary.the judgments about short vs. long form were elicited using a equally counterbalanced within the experiment.
(24) eu a todas as mães.mas eu sempre fui assim, pois eu muito meus pais também.mas minha irmã era bem diferente de mim.eu acho que ela nunca tinha __________ ( ) ninguém antes de ter tido um 'i really my son.i think that to a child is inherent to all mothers.but i was always like this, as i my parents a lot as well.my sister was very different from me.i think she had never ______________ ( anyone before having a child.' eu nunca alguém da minha família com meus próprios problemas.
'i never members of my family with my own problems.to someone with your problems is not right.the other day, i noticed that i my father by talking about my emotional issues and he got really sad.this is even worse because my sister had already ______________ ( him with the same type of problem.' A complete list of the frames presented in experiment 2 is included in Appendix 2, with the accompanying extant psych-verbs presented along side the intended frame/wug combination.the experiment was again conducted with the research platform (https://www.onlinepesquisa.com),and it took around 10 minutes to be completed.52 participants logged in, but only 36 completed the experiment.All participants saw all items, with a randomized order of presentation.

results
the results (n=36) were statistically analyzed using the open-source software r. in a within-subjects design, we tested the effects of type of argument structure (subjexp vs. objexp) on short vs. long participle forms.As shown in Figure 2 and measured by a two-tailed paired t-test, psych-verbs with respect to preference for short participle forms.
FigURe 2: experiment 2 results with means by verb class (n=36), on a scale of 1 to 5, where higher numbers represent a greater preference for the athematic participle (e.g.pando) as opposed to the long-form (e.g.pandado).the difference between subject experiencer verbs (left) and

discussion
the result suggests that speakers do not consider the difference in argument structure between subjexp and objexp while computing past participle forms.these results are expected if the argument structure of both subjexp and objexp involves an extra head responsible for projection of the external argument, as proposed by PesetsKy (1995).that is, both types of verbs possess a heavy lexical structure, in which the vP layer is dominated by another argument taking head.therefore, the results are in accordance with our initial hypothesis.
there is, however, an issue that has to be considered.First, it is unclear that every sentence designed to contain a wug psych-verb indeed receives a psych interpretation.ARAd (1998) has argued that a sentence containing one of beLLetti & Rizzi's objexp psych verbs can receive two different readings: agentive and stative.the agentive reading, she argues, does not deliver a psych-reading, and only stative readings are truly psychological readings.All the syntactic properties attributed by beLLetti & Rizzi to objexp psych-verbs are observed in stative readings, but not in agentive readings, which behave syntactically like regular transitive agentive predicates.From Arad's argumentation, we can thus conclude that psych-verbs are better tested using stative predicates.this was not controlled in the experiments we conducted.
A priori it might be then that beLLetti & Rizzi's dual of our current investigation. in fact, the inner aspect of the event described by the verb (state vs. event) might emerge as an important issue to be considered in general, as it seems that habitual readings favor short participles.For example, according to some speakers' intuitions, in the examples below, the habitually-framed sentence (a) sounds a bit better than episodic sentence (b): (26) a. eu tenho chego atrasada todos os dias I have arrived late all the days 'i have arrived late every day' b. % eu tinha chego atrasada na festa I had arrived late in.theparty 'i had arrived late at the party' the past participle forms of the verb.this should inform all future work on this topic.
For now, let us emphasize that our current results show that given beLLetti and Rizzi's categorization of psych-verbs, there experiencer verbs with respect to their preference for short past participle forms.given that the same methodology, employed in experiments 1 and 2, was sensitive enough to detect differences in the former, we contend that the null result in the latter may stem from either a 'ceiling effect' whereby both subjexp and objexp verbs, despite their differences, are both too syntactically heavy to tolerate short-form of these verb types, whereby factors such as stativeness and habitualness verbs truly distinct.

general conclusion
the present results point to the fact that morphological realization can make reference to the amount of hierarchical syntactic structure underlying it (pace models such as ANdeRsoN, 1992). in sum, suppression of the theme vowel and in the short-form participles is not simply a morphomic generalization based on analogical reasoning stemming from Latin stems (acendido/aceso).instead, speakers reliably show tolerance for short-forms when they correspond to smaller amounts of argument structure.while the growing emergence of these innovative participles in bP may be the result of the complex 'diglossic' situation of hypercorrection and metalinguistic awareness of 'correct' short forms in the suppletive cases, our central point in this paper is to show that reference to 'morphology by itself ' is not enough to explain structure and event structure) on the dependent variables, namely the acceptability of athematic participles.the sporadic omission of the theme vowel and found in bP (and rarely found to this productive extent in other Romance languages) certainly may result from social and educational factors, but the differential modulation of their application by verb classes implicates a undeniable role for the syntax.
arrived late for the meeting' b. ?* eu tenho pago as contas todo mês I 'i have paid the bills every month' c. * eu tenho apresento o João para o pessoal 'i have introduced João to the folks' (9) a.As crianças chegaram atrasadas the children arrived late 'the children arrived late' b. o vaso quebrou ' the vase broke' (10) As mulheres esfregaram as roupas 'the women scrubbed the clothes' (11) os ricos doaram computadores aos órfãos 'the rich people donated computers to the orphans' (12) A peça comoveu o público document to mary in/?for two hours b. the document got sent to mary (19) Psycha.
respects easily (cf.John respects the truth) b.Children bore easily (cf.your story bored the children) bored with your story b. * John annoyed with your behavior