The Syntax of European Portuguese Resultatives

This study attempts to propose a syntactic distinction between true resultatives and pseudo-resultatives and to provide an overview of the resultative constructions in European Portuguese (EP). It has been found that true resultatives, which have causative structures and involve Manner Conflation, are not allowed in EP, contrasting to languages such as English. That is because EP (or Romance languages in general) does not allow Manner Conflation. There are some resultative-like sentences in EP, but they should belong to the pseudo-resultative category, which contains an agentive structure and exhibits semantic and syntactic differences from true resultatives. In addition, simple resultatives, which have causative structures formed with light verbs (instead of Manner verbs) are also attested in EP because no Manner Conflation is involved.


Introduction
Resultatives refer to single clause constructions that describe an event of a change of state, involving Manner (the causing eventuality) and Result (the end state), neither one introduced by morphological marker or conjunction, such as John hammered the metal flat (see Simpson 1983, Levin & Rappaport Hovav 1995, Alsina 1996, Washio 1997, Rappaport Hovav & Levin 2001, Kratzer 2005, Williams 2008. Talmy's (1985) typology distinguishes satellite-framed or manner-conflated languages (e.g., Germanic languages) and verb-framed or path-conflated languages (e.g., Romance languages) in how motion events and caused-result events are expressed. For example, English is a satellite-framed language and exhibits resultatives such as John hammered the metal flat (1a), where the Manner is conflated with the main verb, and the Result occurs as "satellites". Contrastingly, such resultatives are not allowed in Romance languages, as shown in (1b-e) (the examples in [1b-d] are from Washio 1997 However, some resultative-alike constructions are observed in EP, as shown in (3).
(3) European Portuguese (Duarte & Oliveira 2010) a. O João pintou a parede de amarelo. the John painted the wall of yellow 'John painted the wall yellow.' b. O arquiteto construiu a cisterna oculta. the architect constructed the cistern hidden 'The architect constructed the cistern hidden.' In this study, we argue that the apparent resultatives in (3) are "pseudoresultatives" (see Carrier & Randall 1992, Levinson 2010, which have a structure distinct from the "true resultatives" such as (1a) and (2a). We maintain that true resultative structure involving both Manner and Result is not allowed in EP. However, EP exhibits "simple resultatives" that involve a causative verb instead of a Mannerdenoting verb, such as (4).

Manner Conflation
This study assumes with the general idea of Distributed Morphology (Halle & Marantz 1993, 1994, Marantz 1997, Harley & Hoyer 1999, that the syntax is the only generative system responsible for both word structure and phrase structure, and that "Narrow Lexicon" consists of two classes of units: the "atomic roots" and the "atomic bundles of grammatical features" (Marantz, 1997). Lexical roots are acategorial and contain encyclopedic semantic content; by merging with a categorizing functional head little x, such as v, n, or a, they can get categorized into a verb, noun, or adjective respectively, as illustrated in (5).
Inspired by Folli & Harley's (2005) different flavors of v (i.e., the different nature of events is not determined by the semantic information contained in the lexical roots, but rather, by the different "flavors" of v heads that contain specific eventsemantic content, such as vCAUSE, vBECOME, vDO; see), we propose that true resultatives contain a causative structure involving vCAUSE. For example, in the English true resultative (1a), here repeated as (6a), vCAUSE takes a Small Clause (containing a DP and a Result-denoting root) as its Complement, and a Manner-denoting root ("√") conflates to vCAUSE as an adjunct (see Manner Incorporation/Conflation in Harley 2005 andHaugen 2009; see similar proposal in Mateu 2012 Mateu (2012) proposed that Romance languages do not exhibit such a type of resultatives because the Manner-conflated pattern is possible in Germanic languages but impossible in Romance languages.
Upon the observation that Romance languages allow simple resultatives involving a light verb (see Mateu 2002), Mateu (2012) claims that the Spanish simple resultative in (7a) has the structure in (7b), where the verb poner 'put' moved out of the embedded Small Clause and incorporated into a null verb.  (2005) proposed that instrumental denominal verbs, such as hammer, involve Manner Incorporation applying to the little v, which takes a complement headed by an Eventdenoting Root, as shown in the structure of Sue hammered the metal: This was fine-grained by Haugen (2009), who distinguished between Incorporation and Conflation and claimed that instrumental denominal verbs such as hammer involve Manner Conflation (instead of Incorporation). Other verbs of this type include brush, paddle, string, whistle, saw, anchor, and comb in verb use. Haugen's proposal of Manner Conflation was later applied by Mateu (2012) for resultative sentences such as The boy danced his feet sore.
Pi A nerviosa (Mateu 2012) In contrast to Mateu's proposal, we propose that Romance "simple resultatives" have a causative structure similar to (6b) but without Manner Conflation. For example, the EP simple resultative in (8a) corresponds to the structure in (8b), where vCAUSE embeds a Small Clause. 2 Such a structure is allowed because no Manner Conflation is involvedthe causative verb fazer 'make' is a phonetic realization of vCAUSE. Other light verbs in EP that may occur in the simple resultative structure include tornar 'turn', deixar 'leave, let', etc.
With our account, the possible simple resultatives and impossible true resultatives in EP (or Romance languages in general) can find a straightforward explanation through the inaccessible Manner Conflation. As shown in (9a), the sentence is grammatical with the light verb deixar 'to leave, to let'; however, when the 2 Note that Portuguese also allows secondary predicate constructions like the following (see Baptista 2021): O Pedro considera o João um bom candidato ao cargo the Peter consider the John a good candidate to.the role 'Peter considers John a good candidate to the position.' This is not a resultative construction and therefore is not discussed in this article. However, am embedded Small Clause is also involved.
Manner-denoting verb pontapear 'to kick' occurs, the sentence becomes ungrammatical. That is because the attempted Manner Conflation is prohibited in EP, as illustrated in (9b) Another advantage of our proposal is that the EP resultatives and causatives can be accounted for in a unified manner. We propose that the EP syntactic causatives in (10) (the fazer-Inf construction; see Raposo 1981, Gonçalves 1999, Barbosa & Raposo 2013) also involve the functional head vCAUSE, which is phonetically realized by mandar 'make' in this case. Contrasting to simple resultatives, the embedded constituent in EP syntactic causatives is a VP or a phase instead of a Small Clause, as illustrated in (11a, b) (in light of Pylkkänen's [2002] "selection of CAUSE"). 5 That means, syntactic causatives and simple resultatives in EP both involve a causative structure headed by the functional head vCAUSE. They only differ in the embedded domain: in syntactic causatives, vCAUSE embeds a VP or phase (11a, b), and the caused event is an activity; in simple resultatives, vCAUSE embeds a Small Clause (11c), and the caused event is stative. The adjective aberta 'open' is not intended to form an NP with the noun porta 'door' but to function as a result predicate.

5
The difference between the two examples in (10) is that the verb 'leave' is inflected in (10b) but not in (10a). In (10b), the verb 'leave' is in an inflected infinitive form in agreement with 'the children'; 'the children' is an embedded subject, and thus the sentence contains an embedded phase. In (10a), the verb 'leave' is in a non-inflected infinitive form and is a V.
To summarize, English true resultatives involve a causative structure headed by vCAUSE, which embeds a Small Clause. A Manner-denoting root conflates with this head as an adjunct to denote the causing activity of the event. In contrast, EP (or Romance languages in general) does not allow Manner Conflation and thus can only form simple resultatives where vCAUSE is phonetically realized by a light verb. The English-EP comparison is found in Table 1.

Pseudo-resultatives
In the previous section, we showed that EP does not allow true resultatives. There are some apparent counter-examples, which were presented earlier in (3) and are repeated below in (12). They show surface similarities to true resultatives since both Manner and Result meanings are overtly expressed. However, there are quite a few semantic differences between true resultatives and the sentences in (12). First of all, while true resultatives involve two subevents (one of causing activity and one of result), these sentences refer to only one event. Taking (12a) as an example, 'painting the wall' and 'the wall being yellow' refer to the same activity: 'yellow' just denotes how the wall was painted. Similarly, in (12b), 'hidden' specifies how the cistern was constructed. Rather than indicating the Result, the second predicate seems to specify the main activity by indicating how it is conducted. We suggest that the examples in (12) are "pseudo-resultatives" (see Carrier & Randall 1992, Levinson 2010, similar to Washio's (1997) "spurious resultatives" or Kratzer's (2005) "adverbial resultatives".
We propose that pseudo-resultatives contain an agentive structure involving vDO. According to Folli & Harley (2005), the difference between vDO and vCAUSE is that, vDO is an agentive v, requires an animate Agent subject, can take a straightforward We thus propose that in Portuguese pseudo-resultatives, the Manner root Merges with vDO to become the main verb, which then takes a DP in its internal argument position and an XP in the Complement position as its modifier. For example, the pseudo-resultative in (12a) corresponds to the structure in (14). The Mannerdenoting root first Merges with vDO to turn into the main verb pintar 'to paint'. Then, this verb takes a DP (a parede 'the wall') in the internal argument position and a PP (de amarelo 'of yellow') as its modifier. The simplified structures of true resultatives and pseudo-resultatives are compared in (15). The main differences are: 1) Pseudo-resultatives involve an agentive structure headed by vDO but true resultatives involve a causative structure headed by vCAUSE. 2) In pseudo-resultatives, the NP in the Object position is an internal argument of the main verb and receives the role of Theme; in true resultatives, it is generated inside a Small Clause and takes the role of Causee. 3) In pseudo-resultatives, the Result-like predicate is in the Complement position of the main verb, functioning as its modifier; in true resultatives, the Result is generated inside a Small Clause.
(15) a. Pseudo-resultative b. True resultative vDOP vCAUSEP We argue that the structural differences can account for the semantic and syntactic differences between true resultatives and pseudo-resultatives. We will illustrate them one by one in the following section.

Strong or weak resultative meanings
According to Washio (1997), resultative expressions may show semantic differences in terms of "strong" or "weak" resultative meanings. Strong resultatives are those in which "the meaning of the verb and the meaning of the adjective are completely independent of each other" (1997: 7)the resulted state of the Causee cannot be predicted from the semantics of the verb denoting the causing eventuality, such as (16a). For weak resultatives, the verb responsible for the causing eventuality, though not necessarily implicating or entailing a certain change, has "a disposition toward a certain result without lexically implying such a result" (Washio 1997: 16). In other words, weak resultatives are those in which the result predicate denotes either the purpose or the conventional result of the activity encoded in the main verb, as in (16b).
(16) English (Washio 1997) a. The lion's roar scared him stiff. b. He wiped the table clean/dry. Following Washio's distinction between strong and weak resultative meanings, we find that true resultatives but not pseudo-resultatives may produce "strong resultative" meanings.
We follow Mateu (2012), who suggests that the reason why English true resultatives may express strong resultative meanings is because there is no structural relation between the conflated Manner root and the Result root, as shown in (15b) above.
In contrast, the EP pseudo-resultatives, including those in (12) above and (17) below, all express weak resultative meanings since the Result-like predicate expresses the purpose or the conventional result of the main activity. We suggest that it can be explained by the syntactic structure that we proposed in (15a): in pseudo-resultatives, the Result-like predicate is a Complement of the main verb and functions as the modifier of the main activity by specifying its purpose or the way how it is conducted. Therefore, the Result-like constituent in pseudo-resultatives always has a relatively close semantic relation with the main verb and thus expresses weak resultative meanings.
(17) European Portuguese a. O João cortou a carne em fatias. the John cut the meat in slices 'John cut the meat in slices.' b. O pintor pintou a paisagem esfumada.
the painter painted the scenery smoke 'The painter painted the scenery smoky.'

Thematic role of the Subject
Differences are also observed in terms of the thematic role of the Subject. The pseudoresultatives require that the Subject is an Agent, as in (12) and (17). However, this is not necessary with true resultatives. For example, in the true resultative The sea ate the beach away, "the sea" is not an Agent of the action "eat", but a Causer of the resulted event. However, the Subject of a true resultative is not required to be an external causer either, as in John hammered the metal flat. The main point is that, pseudo-resultatives, but not true resultatives, show an agentivity constraint on the Subject. That can be explained by the syntactic structures presented in (15). In pseudoresultatives (15a), the Manner root is turned into a main verb by Merging with vDO, which heads an agentive structure. Consequently, the DP at the external argument position receives the role of an Agent. However, true resultatives contain a causative structure headed by vCAUSE (15b), and thus the external argument takes the role of a Causer. In true resultatives, the NP at the Subject position is required to be a possible initiator of the whole caused-result event, but not necessary an Agent of the Manner verb.

Relation between the Object and the main verb
Pseudo-resultatives require that the Object is an internal argument of the main verb. as in (12) and (17). However, that is not the case with true resultatives. For example, in the true resultative He drank the pub dry, 'the pub' is not an internal argument of 'drink'. This difference can also find an explanation from our proposal. As shown in (15a), pseudo-resultatives involve an agentive structure, where the main verb (derived by the Merge of the Manner root with vDO) takes an NP at its internal argument position. Therefore, this NP, which surfaces as the Object in the sentence, always has a thematic relation with the main verb.
However, in true resultatives, the Manner root adjoints to the vCAUSE head as an adjunct, and the NP in the surface Object position is generated inside a Small Clause embedded by vCAUSE. Consequently, this NP is structurally independent from the Manner and does not have to have any thematic relation with the Manner.

Optionality of secondary predicate
If the result predicate is removed, the pseudo-resultative sentences will remain grammatical (19) but the true resultatives may not (18) That is due to the structural differences between the two constructions. In true resultatives, the Result is part of an embedded Small Clause and therefore, its occurrence is obligatory. If the Result-denoting constituent is removed from a true resultative, the structure will become unacceptable (18). However, note that if the embedded NP happens to be a legit Theme of the Manner verb, the sentence may remain grammatical (e.g., John hammered the metal (flat)), but in this case, the sentence corresponds to a different structure (i.e., an Agentive structure with vDO).
Contrastingly, in pseudo-resultatives, since the Result-like predicate is just a modifier of the main verb, the sentence would remain acceptable if it is absent (19).
The implication of this difference is that, in a resultative-like construction, if the secondary predicate cannot drop, the construction is a true resultative; if the sentence remains grammatical when the secondary predicate drops, other tests should be applied.

The "how" question
Since the Result part in pseudo-resultatives is adverbial-like, it usually can serve as a proper answer to the "How" question (see Kratzer 2005, Marcelino 2014), as shown in (20). However, it is not the case with true resultatives, as shown in (21) This response may be considered infelicitous by some speakers since an answer involving instruments or means is usually expected from questions with como 'how'. However, this test is still meaningful since the infectiousness of (21b) is definitely higher than that of (20b).

Conclusions
True resultatives, which involve both Manner and Result, are attested in English, but not in EP. We follow Mateu (2012) and maintain that the contrast between EP (and Romance in general) and English (and other satellite-framed languages) exists in the possibility of Manner Conflation to the causative head vCAUSE. When there is no Manner Conflation, the same causative structure could work, with vCAUSE being phonetically realized by a light verb. In this case, the outcome is a simple resultative. With our approach, the simple resultatives, true resultatives, and causatives are accounted for in a unified manner: they all involve a causative structure headed by vCAUSE, and what differs is the embedded domaina VP, phase, or a Small Clause.
Although there are some resultative-like constructions in EP, we hold that they belong to the pseudo category and exhibit properties distinct from true resultatives. We proposed that while true resultatives contain a causative structure headed by vCAUSE, the pseudo-resultatives contain an agentive structure headed by vDO. Due to the structural differences, true resultatives and pseudo-resultatives exhibit differences both semantically and syntactically. Therefore, with true resultative structure inaccessible, EP exhibits two types of resultative constructions: the pseudo-resultatives with an agentive structure headed by vDO and the simple resultatives with a causative structure headed by vCAUSE, which is phonetically realized by a light verb.
In all, this study offers a systematic view of EP resultative constructions and shows how true resultatives, pseudo-resultatives, and simple resultatives are structurally different in EP. The analysis can be potentially extended to other Romance languages and may contribute to a broader typological study on causatives and resultatives.
Due to the specific focus and approach of this study, result-encoding verbs such as partir 'break', achatar 'flatten', and entrar 'enter' are not discussed. Our preliminary view is that language variation in result-encoding verbs stems from the language's semantic depository of roots (see Yao 2022). Portuguese exhibits plenty of "result roots" (terminology from Beavers et al. 2017); when turned into verbs by merging with vDO, they come to encode resultative meanings and thus correspond to the verb-framed pattern. However, for languages where result roots are less available (e.g., Chinese), the satellite-framed pattern would be needed. It would be interesting to analyze the result-encoding verbs along with the resultative constructions in future studies. Recently, Folli & Harley (2020) proposed that head movement, more specifically the Res-to-v movement, is involved in the verb-framed pattern (with result-encoding verbs), and that such a movement is "on" in verb-framed languages (e.g., Romance), but "off" in satellite-framed languages (e.g., English). It is an alternative approach to account for cross-linguistic resultative variation. It would be interesting to investigate a wider set of languages in future studies to find out the feasibility of both approaches.