Syntactic structure of Spanish parasynthesis: Towards a split little-v via affectedness

In this paper, I claim that the syntactic structure of Spanish parasynthetic verbs a-...-ar (e.g. ablandar ‘to soften’) and en-...-ar (en-dulzar ‘to sweeten’) provides (further) evidence for a decomposed vP structure. I propose that these verbs are represented with a structure where v.Caus and verbalizer-v are distinct. In particular, I analyze transitive a-/en-...-ar verbs as complex predicates with a v.Caus-headed vP and a PP. These verbs differ in the properties of a projection (ResultP), located between v.CausP and PP, whose head expresses how specific the result state/location for the internal argument is: a-...-ar specifies a (particular) result state/location for the internal argument whereas en-...-ar remains underspecified with regard to a (particular) result state/location. I address the consequences of my proposal for Spanish, and suggest how it can be extended to French and Italian, and how Romance deadjectival parasynthetic verbs are different from English degree achievements, e.g., straighten and soften.

In this paper, I claim that the syntax of Spanish parasynthetic verbs a-…-ar and en-…-ar (e.g.ablandar 'to soften,' en-dulzar 'to sweeten') provides (further) evidence for a decomposed vP structure, in particular, a structure where v.Caus and verbalizer-v are distinct (Hale & Keyser 1993, 1998;Harley 1995Harley , 2009Harley , 2016;;Marantz 1997;Embick 2004;Serratos 2008;Key 2013;Jung 2014).I propose that transitive a-/en-…-ar verbs are complex predicates with a v.Causheaded vP and a PP that makes explicit the relation between the internal argument and the adjectival/nominal base predicate.They differ in the properties of a projection, ResultP, (Ramchand 2008), located between v.CausP and PP, whose head determines how specific the result state/location for the internal argument is: a-...-ar specifies a (particular) result state/location for the internal argument whereas en-...-ar remains underspecified with regard to a (particular) result state/location (Beavers 2011).In addition, based on Montalbettis's (1996) argument that a-/en-attach to an already verbalized base predicate and the fact that, in general, the absence of a-/en-makes the verbs ungrammatical, I claim that a-/en-are the heads of ResultP-which merges with verbalizer-v, which is distinct from v.Caus.Thus, the syntactic structure of parasynthetic verbs I propose looks as follows (Bowers 2002;Alexiadou, Anagnostopoulou & Schäfer 2006;Folli & Harley 2004;Marantz 2008;Pylkkänen 2008;Harley 2009Harley , 2016;;Wurmbrand & Shimamura 2016) (EA = external argument; IA = internal argument; v = verbalizer-v; a/n = adjectival/ nominal base predicate): ( In what follows, I give arguments for the structure in (1).In section 2, I provide a general template that characterizes a-/en-…-ar verbs as complex predicates with a v.Caus-headed vP and a projection that denotes change of state/location, which makes explicit the relation between an Towards a split little-v via affectedness internal argument and a base predicate.This discussion works as a point of departure to pursue refinements to the projection that denotes change.In section 3, taking into account adjectival/nominal base predicates, I argue that the crucial difference between these verbs lies in the fact that a-...-ar specifies a (particular) result state/location for the internal argument, whereas en-...-ar remains underspecified with regard to a (particular) result state/location.This discussion provides the elements that are needed to characterize, in section 4, the changedenoting projection, which would be formed by (at least) a PP and a result state/location phrase (ResultP).In this section, I also make explicit that v.Caus and verbalizer-v are distinct heads due to the fact that ResultP-a projection that specifies the internal argument for result state/location whose morphological spell-outs are a-/en--merges with an already verbalized adjectival/nominal base predicate.In section 5, I address how parasynthetic verbs interact with verbalizers, extend my account to des-…-ar parasynthetic verbs in Spanish, examine the differences and similarities in Spanish, French and Italian parasynthetic verbs, and suggest how Romance deadjectival parasynthetic verbs are different from English degree achievements, e.g., straighten and soften.Finally, in an appendix, I go one step further with regard to base predicates and present some arguments that suggest that they are uncategorized bases-not adjectives or nouns, as traditional grammar suggests.

Common structure: v.Caus + change 1
In this section, I characterize transitive a-/en-…-ar verbs as complex predicates with a v.Causheaded vP and a projection that denotes change (BecomeP).I propose that a-/en-…-ar are complex predicates formed by v.Caus, which introduces a causer external argument.This projection triggers a change (of state/location) that establishes a predicative relation between an internal argument and a base predicate.I propose that a-/en-…-ar verbs involve v.Caus. 2 The presence of v.Caus was first observed by Mendívil (2003), and argued for by Martínez Vera (2014) and Gibert Sotelo & Pujol Payet (2015), who independently reached this conclusion as well.If, following Harley (2003), Folli & Harley (2004), Harley (2013), among many others, inanimacy is a property of v.Caus, the examples below provide evidence for the presence of v.Caus since inanimate EAs (in addition to animate ones) are possible: (2) Juan/el problema a-tontó a Pedro.John/the problem PAR-made.dumbDOM Peter 'John/the problem made Peter feel dumb.' 1 This section puts together the basic elements of the proposal as a departure point.In particular, the projection denoting change will be problematized and enriched in sections 3 and 4 (so the links to proposals such as those by Gumiel, Pérez & Nieto 1999, Mateu 2002, 2012and Fábregas 2015a become apparent).In a way, it recapitulates previous literature, since I am adopting (and expanding to a certain extent) a common structure that has been proposed in the literature (Mendívil 2003;Martínez Vera 2014;Gibert Sotelo & Pujol Payet 2015).Although these inanimate causers, hunger and the storm, are, in fact, ungrammatical in such examples, it is not the case that these verbs (a-delgazar 'to thin,' em-pedrar 'to put stones') cannot take inanimate causers, as the examples in (7) show (for some speakers at least, these inanimate causers are grammatical): Thus, the ungrammaticality of inanimate causers is not as straightforward as the examples in (6) would suggest.I take this to mean that inanimate causers are, in principle, possible.Even if there were examples that cannot take inanimate causers (which, under my proposal, would mean that instead of v.Caus there is v.Do (Cuervo 2003)), the intuition that v.Caus characterizes the general case of a-/en-…-ar verbs would still hold.Thus, although there could be alternations as the ones mentioned, previous literature acknowledges that there is a strong tendency pointing towards the fact that v.Caus is involved in most parasynthetic verbs.
I further suggest that v.Caus triggers a change, which I represent with BecomeP. 4 Both traditional grammar (Serrano-Dolader 1999) and syntactic approaches to parasynthesis (Mendívil 2003;Fernández Alcalde 2011) have captured this intuition by means of the notion of change, which the latter represent with Small Clause (SC) and a Become predicate respectively.I extend this usage to include both changes of state and location, which seems more empirically adequate (e.g., em-botellar 'to put in a bottle,' etc.) (Beavers 2011). 5s Harley (2003), Folli & Harley (2004), among many others argue, if there is a causer EA, change is triggered.Under a decomposed VP approach, provided that the test lo que le ocurrió/sucedió al tema x es 'what happened to theme x is' (my translation of Cruse's 1973 test) determines whether a predicate entails change for the internal argument, actual evidence for a structure where BecomeP is the complement of v.Caus can be given.As the examples below illustrate, when change is entailed, as in (8), the pattern is grammatical, whereas when it is not, as in (9), the result is ungrammatical (Beavers 2011;Martínez Vera 2014).
a. the men destroyed the city a. 'The men destroyed the city.' b.Lo que le ocurrió/sucedió a la ciudad es que los hombres la destruyeron.b. the that DAT happened to the city is that the men ACC destroyed b. 'What happened to the city is that the men destroyed it.'4 I would like to thank an anonymous reviewer for pointing out that BecomeP is a better label than Small Clause (SC) at this point of the discussion, since the former can be thought of denoting change of state/locate (which is the notion I am capturing in this section), while the latter only emphasizes the presence of a result state/location.As anticipated in the introduction in relation to (1), not only do I include ResultP in the final structure I propose, but also PP.In fact, these two are the elements of BecomeP that are minimally needed (within my proposal).I motivate the presence of ResultP and PP in section 3 and formalize it in section 4.
a. the men saw the star a. 'The men saw the star.' b. #Lo que le ocurrió/sucedió a la estrella es que los hombres la vieron.b. the that DAT happened to the star is that the men ACC saw b. #'What happened to the star is that the men saw it.'With regard to a-/en-…-ar, (10-13) provide evidence for the claim that the complement of v.Caus is BecomeP, which denotes change, since EAs trigger a change in the internal arguments, i.e.Peter in (10), Mary in (11), the chickens in ( 12) and the man in ( 13).( 10 to.the man is that the avalanche ACC PAR-buried 'What happened to the man is that the avalanche buried him. Notice that the adjectival/nominal base predicates in (10-13) determine what kind of state/location the internal argument is moving towards through the change it undergoes: dumb in (10), trouble (lit.storm) in (11), fat in (12) and bury (lit.ground) in (13).to those boys is that the gangsters ACC PAR-beat b. 'What happened to the boys is that the gangsters beat them.'The reviewer mentions that perseguir 'to chase' and apalear 'to beat' do not involve change.They seem to involve change though.As Beavers (2011) points out, whenever there is an affected theme in an event, it has undergone some kind of change.Consider perseguir 'to chase.'The dictionary (RAE 2014) shows that, in most entries, the theme is, in fact, affected (e.g., the theme is bothered or instigated, or is escaping from somebody/something), which, in Beavers (2011) terms, indicates that it undergoes a change.The last entry for the verb is a judicial use, e.g., perseguir las infracciones (lit.'to chase the infractions (of the law)').Intuitively, in this case, the theme does not seem to be affected (thus, it does not undergo any change).Interestingly, when the test is applied to this example, the outcome is degraded (at least for the speakers I consulted):

Towards a split little-v via affectedness
Based on the previous discussion, I propose that the common structure of transitive a-/en-…-ar verbs is formed by v.Caus, which introduces an EA.Its complement is a change-denoting projection, BecomeP, which explicitly represents the predicative relation between the adjectival/nominal base predicate and the internal argument (IA).The syntactic representation has the same form Spanish verbs with two arguments have (Hale and Keyser 1993,1998; Gumiel,  Pérez & Nieto 1999; Zagona 2002; Cuervo 2003; Mendívil 2003; Mateu 2012): 7,8   (ii) ?*Lo que le ocurrió/sucedió a las infracciones es que el juez las persiguió.?*the that DAT happened to the infractions is that the judge ACC chase ?*'What happened to the infractions is that the judge chased them.'With regard to apalear 'to beat,' the fact that it has been categorized as an activity does not mean that the theme does not undergo a change (Gibert Sotelo & Pujol Payet 2015).For this reason, (ib) is grammatical: the theme the boys transitions in such a way that they are affected (they change from not being beaten to being beaten).Thus, both cases fall under the notion of change.The reviewer also mentions that the test gives a grammatical outcome even though no result state may be entailed, since, in cases like (iii), the theme need not end up dumb or fat with verbs such as a-tontar 'to make sillier' and en-gordar 'to fatten.' (iii) *Juan está a-tontado/en-gordado. *John is PAR-made.sillier/PAR-fattened.*'John is made sillier/fattenend.'As the reviewer points out, this might be related to the nature of the bases, since there are other cases with a-/en-…-ar verbs that are grammatical with estar 'to be' (e.g., John está en-terrado 'John is buried,' etc.).This possibility seems to be a grounded one.In Kennedy & McNally's (2005) theory, adjectives such a fat and dumb do not entail a minimal/maximal degree, which means that the standard of comparison is contextual.In parasyntehtic verbs, this means that there is no minimal/maximal degree for cases like a-tontar 'to make sillier' and en-gordar 'to fatten,' so only a comparative reading for them is available (the theme ends up fatter/sillier).As discussed by the authors, the comparative reading does not entail the standard (e.g., a fat/dumb person) because it is not lexically encoded in such bases, which would account for the examples in (iii)-where no absolute result state can be reached.
Nevertheless, it must be kept in mind that things are more complex than this, as with examples like Juan está en-vejecido 'John is older' show (this example was also provided by the reviewer).The base viejo 'old' seems to be just as fat or dumb, where no minimal/maximal degree is denoted by the base.One very tentative possibility would be that in such cases the event is bounded somehow, so a salient degree is available, and a result state reading becomes possible.The examples in sentences like (iii) have to be treated rather carefully though, because it is not clear what the semantics of estar 'to be' is, as the long-standing debate on the topic illustrates (see García-Pardo & Menon 2016 and references therein for recent discussion on the topic; the literature on the ser/estar contrast is rather vast), so it is not evident what is (ultimately) at issue.Addressing this topic further would take us too far afield.In the end, the label used to represent change is not crucial.What is relevant is that there is an element that makes explicit that there is a predicative relation between the internal argument and the base predicate.In sections 3 and 4 I give content to such projection in terms of PP (Gumiel, Pérez & Nieto 1999;Fábregas 2015a;Mateu 2002Mateu , 2012;;Gibert Sotelo & Pujol Payet 2015) and ResultP (Ramchand 2008).Gabriel Martínez Vera (14) Having discussed the common aspects of a-/en-…-ar, I now turn to what makes these verbs different, which will provide arguments to further characterize the BecomeP in (12) in terms of (at least) ResultP and PP.

Distinction between a-/en-…-ar: (under)specification of result state/location
In this section, I claim that a-...-ar are predicates that specify a (particular) result state/location for the internal argument, whereas en-...-ar are predicates that remain underspecified with regard to a (particular) result state/location for the internal argument, i.e., a difference in degree of affectedness (Beavers 2011). 9,10This means that, while both a-/en-…-ar verbs entail a result state/location (Serrano-Dolader 1999; Gibert Sotelo & Pujol Payet 2015), 11 the verbs differ with regard to its specificity.To argue for this distinction, I address adjectival and nominal bases in what follows.
Both a-/en-…-ar allow adjectival bases.I assume that the adjectival bases that appear in these verbal schemes include a scale as part of their denotations (Gumiel, Pérez & Nieto 1999;Beavers 2011).Following Kennedy & McNally (2005:349), I further assume that a scale is a set of ordered degrees.Gradable adjectives "map their arguments onto abstract representations of measurement, or DEGREES, which are formalized as points or intervals partially ordered along the internal argument, just as examples (10-13) do (e.g., Lo que le pasó a Juan es que Pedro lo acuchilló 'What happened to John is that Peter knifed him'); (b) if it were a true instrumental, overt instruments should be ungrammatical, but they are not (e.g., Juan a-cuchilló a Pedro con una navaja/con un pedazo de vidrio con filo 'John knifed Peter with a penknife/with a piece of glass with a blade') (Arad 2003); (c) (b) suggests that what is relevant in bases like cuchillo 'knife' is its manner (i.e., something like a blade-like shape) (in fact, so-called instrumentals and manner seem to pattern together as Erteschik-Shir & Rapoport 2007 point out).Gibert Sotelo & Pujol Payet (2015) assume a similar position with regard to instrumentals.Thus, the brief discussion in this footnote provides arguments for their claim.9 One of the crucial aspects of Beavers' (2011) discussion lies in the presence or absence of a result operator (i.e., the presence or absence of a result state/location) from a semantic point of view.This is the notion I have in mind in this section.10 I do not address (a)telicity of a-/en-…-ar verbs.In general, I assume, as Gibert Sotelo & Pujol Payet (2015) discuss, that these verbs often appear in telic events.Atelic readings are not always excluded, however.Such considerations do not have direct consequences for my proposal, so I leave them aside for future research.
11 Both yield a contradiction with entailment tests (Tenny 1994;Beavers 2011).Consider as examples a-/en-…-ar verbs with the adjectival bases tonto 'silly' and gordo 'fat': #Juan atontó/en-gordó los pollos, pero no hay nada diferente en ellos 'John made sillier/fattened the chickens, but nothing is different about them.'The same holds for nominal bases.Towards a split little-v via affectedness some DIMENSION," such as height, weight, cost, etc.To exemplify this, consider the antonyms empty and full.Both make use of the same dimension (let us call it fullness), but illustrate opposite orderings.While the latter exemplifies a positive ordering (i.e., an entity is fuller than another one if the first one has a higher degree of fullness than the second one), the former exemplifies a negative one (i.e., an entity is emptier than another one if the first one has a lower degree of fullness than the second one). 12Scales vary depending on whether there is or there is not minimal/maximal degrees, i.e., there are scales for which these degrees can be reached and scales for which they cannot be reached. 13If this is the case, three kinds of scales are possible: open scales, closed scales, and partially closed scales.For open scales, no minimal or maximal degrees can be reached; for closed scales, minimal and maximal degrees can be reached; for partially closed scales, only one between minimal or maximal degrees can be reached. 14Examples of these scales, illustrated via antonyms (i.e., they share the relevant dimension), are shown below:15 16), where the scales of length (16a) and width (16b) are involved.The latter is illustrated in (17), where flatness (17a) 12 I illustrate this with comparative forms, because verbal predicates of change with gradable bases resemble comparative uses (Kennedy & Levin 2008) and may include overt comparative morphology (Bobaljik 2012).
13 Kennedy & McNally (2005) assume that degrees are isomorphic to the real numbers between 0 and 1. 0 represents the minimal degree and 1 stands for the maximal degree in a scale.

14
A further specification on whether the scale has a minimal/maximal degree is not relevant for our purposes, since parasynthetic verbs do not seem sensitive to such distinction.See 5.3 for an extension of my proposal to French and Italian where the distinction between low/high bounds (i.e., whether the ordering of the scale is oriented towards the minimal/maximal degree respectively) is relevant (Di Sciullo 1997).
and cleanliness (dirtiness) (17b) are the relevant properties in the scales respectively.Notice that scales that are closed on both ends are not possible, as in ( 18 However, it is not the case that a-/en-…-ar take exactly the same aspects of the scales just discussed.On the one hand, the denotation of a closed end state in a scale seems to be restricted to a-…-ar, as in ( 19).Thus, the bases liso 'straight' in (19a) and plano 'flat' in (19b), which denote a closed end state, are only possible with a-…-ar.This does not mean that scales that include a closed end state are impossible with en-…-ar; what it means is that when an adjective with a scale that is closed on one end appears, it will denote the open end state.Thus, in (20) the scale does include a closed end state, i.e., clean, but the verb with en-…-ar denotes the open end state, i.e., dirty: On the other hand, the general case is that en-…-ar verbs allow idiosyncratic/conventionalized meanings associated with scalar adjectives, whereas this is very restricted in the case of verbs with a-.This is illustrated in (21a) with ancho 'wide' and largo 'long': the sentence is grammatical with en-sanchar 'to widen' but not with a-largar 'to lenghten'-the intended sense is 'to broaden,' which is associated with en-sanchar only.The same takes place in (21b): en-crespar 'to curl' can have an idiosyncratic/conventionalized meaning, i.e., 'to get the nerves,' but in that same context a-lisar 'to straighten' is ungrammatical, i.e., no such idiosyncratic/conventionalized meaning is available.(21c,d) further illustrate that adjectival bases with en-…-ar have developed idiosyncratic/conventionalized meaning: (21c), with en-dulzar 'to sweeten,' exemplifies an idiomatic use and (21d), with en-friar 'to cool,' shows that the object is not restricted to elements that can undergo cooling (in a literal sense): The antonyms gordo 'fat' and delgado 'thin,' which represent the two extremes of the same scale, illustrate a case where a verb with a-can also have an idiosyncratic/conventionalized meaning, as in ( 22): en-gordar 'to fatten' in (22a) is understood as 'become richer,' and adelgazar 'to thin' is understood as 'become poorer.' 17 I am not aware of other examples like adelgazar 'to thin' that show this behavior, which means that there seems to be a strong tendency with regard to the fact that en-…-ar verbs are the ones that develop idiosyncratic/conventionalized meanings the most (Serrano-Dolader 1999).This discussion points towards two aspects: (i) only a-…-ar can denote the closed end state in a scale, i.e., while both verbal schemes can denote an increase of degree within a given scale, only a-…-ar can denote the absolute value (the closed end state) in a scale; (ii) in general, the idiosyncratic meanings that a base has developed can (only) be expressed with en-…-ar.In terms of affectedness and (under)specification of a result state/location, this means that only a-…-ar can specify that a theme reaches the minimal/maximal degree within a scale, 18 but it also means that en-…-ar is less specific-or more flexible-in the sense that it can include conventionalized meaning in what it denotes. 1916 Notice that ??Se calentó el amor de los amantes 'The love of the lovers heated' is degraded.17 I would like to thank an anonymous reviewer for providing the example in (22b).

18
As Beavers (2011) points out, the possibility of denoting a particular degree (specifically, a minimal/maximal one) means that a predicate is more specific with regard to result state than a predicate that cannot do so.In particular, indefinite change (underspecified in my terms) is an endstate with an existentially quantified argument, whereas definite change (specified in my terms) is an endstate with a specific or referential argument (e.g.0/1 for minimal/maximal degree respectively).
19 An anonymous reviewer wonders why it is the case that (quasi-)synonyms such as a-delgazar and en-flaquecer, with bases delgado and flaco 'thin,' respectively, appear in different derivational processes.Although I do not have a good answer for this question, I would like to briefly point out two issues.First, note that flaco can be used with more flexibility than delgado (e.g., thin people could have the nickname flaco, but not delgado).This is consistent with my proposal that en-verbs are underspecified for result state, i.e., bases with more flexible uses appear with it.Second, I would like to suggest something that goes beyond the two examples mentionted above, which deals with a difference between a-…-ar and en-…-ecer.Note that it is a fact that, when the I now turn to nominal bases. 20According to Gibert Sotelo & Pujol Payet (2015), denominal parasynthetic verbs could denote four types of result state/location: property, locatum, locatio and instrument.Property is understood in terms of the internal argument getting a property of the base; thus, la voz 'the voice' ends up stiff in (23a) and Juan 'John' ends up like a roll in (23b). 21Locatum is understood in terms of the base becoming a possession of the internal argument; thus, la ciudad 'the city' ends up fortified in (24a) and María 'Mary' ends up covered in mud in (24b).Locatio is understood in terms of a change of location for the internal argument; thus, el ejército 'the army' ends up quartered in (25a) and el vino 'the wine' ends up bottled in (25b).Instrumentals deal with the use of an object to cause a result state/location; thus, Pedro 'Peter' is hurt with a knife in (26) (no en-…-ar instrumentals are attested as far as I can tell; Gibert Sotelo & Pujol Payet (2015) do not include any).( 23 Gibert Sotelo & Pujol Payet's (2015) classification is helpful to point out two aspects in order to distinguish parasynthetic verbs with a-/en-.On the one hand, while it may be true that both verbal schemes have overlapping meanings, there is a difference in terms of how flexible the meanings denoted by a-/en-…-ar verbs can be-I provide examples in (27).Consider the minimal pair a-vinagrar 'to turn sour' and en-vinagrar 'to put (in) vinegar' with the base vinagre 'vinegar.'Under Gibert Sotelo & Pujol Payet's (2015) classification, the former exemplifies property, and the latter illustrates locatum or locatio.With regard to a-vinagrar 'to turn sour,' there is only one possible meaning, i.e., that of property understood as going wrong by making sour, as in (27a).With regard to en-vinagrar 'to put (in) vinegar,' two possible core meanings are available: locatum (27bi, bii) or locatio (27biii); moreover, two locatum readings are possible.Thus, en-…-ar is more flexible than a-…-ar in that a single verb can denote more suffix -ecer appears with a parasynthetic prefix, it is generally en-, as Serrano-Dolader (1999) points out.Let us (very tentatively) propose that -ecer denotes an event bound.If this is the case, it makes sense that it appears with en-and not with a-, because verbs with a-can be specified with a minimal/maximal degree, i.e., the lower/upper bound of the scale of the base predicate is present, so there is no need for an additional bound.Since no such lower/upper bound is entailed with en-…-ar, the suffix -ecer would be providing it.This is obviously very tentative, since -ecer is not mandatory with verbs with en-, but it can at least provide an idea of why such suffix appears more with en-.See Fábregas (2015a) for an alternative view in terms of selection.20 These bases are non-gradable in general (Morzycki 2009).

21
The verbs with en-…-ar that denote a property, as en-rollar 'to roll,' can be understood as locative, as Gibert Sotelo & Pujol Payet (2015)  It has been claimed, however, that a-…-ar verbs allow such flexibility (Mateu 2002).Consider the verb a-cuchillar 'to knife,' as in (28).There are, in principle, many ways of performing the knifing event of la víctima 'the victim.'Nonetheless, there is something that always remains constant, i.e., the fact that the core meaning of this verb is that of an instrumental.As with a-vinagrar 'to turn sour,' the core meaning of a-cuchillar 'to knife' does not change.Thus, a-…-ar verbs are less flexible than en-…-ar verbs: they can only denote one core meaning associated to the verbal schemes.This flexibility, of course, also depends on the nominal base under consideration.What is relevant, though, is the availability of such flexibility with, mostly, en-…-ar.

23
An anonymous reviewer mentions that a-vinagrar 'to turn sour' appears to have developed an idiosyncratic meaning, as in Su carácter se ha a-vinagrado 'His character has turned sour.'There are two possible ways to address this issue.The first one would be to say that such meaning is not idiosyncratic, but falls under the meaning of property.Thus, sourness is assigned to his character.
The second one is to say that this verb has in fact developed an idiosyncratic meaning; a-vinagrar 'to turn sour' would then be one of those exceptional cases where a-…-ar has developed a conventionalized meaning.Note that the strong tendency mentioned in the text (that, in general, verbs with en-…-ar are the ones that develop idiosyncratic/metaphorical meanings) will still hold.
(32) en-listar 'to enlist (in the army)' base: lista 'list' (33) en-sillar 'to saddle (a horse)' base: silla 'chair' In terms of affectedness and (under)specification, the difference that I stated above with regard to adjectival bases is reinforced when nominal bases are considered: each a-…-ar verb only denotes one core meaning among property, locatum, locatio and instrumental, whereas such restriction does not hold for en-…-ar verbs.This means that a-…-ar verbs are specific with regard to a particular kind of result state/location for the internal argument, while en-…-ar verbs remain underspecified.This (under)specification corresponds to the development of idiosyncratic meanings: they mainly appear with en-…-ar verbs, which is the expected outcome.
In conclusion, my proposal, understood in terms of the (under)specification of result state/location-with a-…-ar specifying a (particular) result state/location for the internal argument and en-…-ar remaining underspecified with regard to a (particular) result state/location for the internal argument-, gives a way of distinguishing a-/en-…-ar verbs despite the fact that they can-and, in fact, seem to have-overlapping meanings.With adjectival bases, (i) open scales cannot (in general) include conventionalized meanings with a-…-ar (e.g.a-largar 'to lengthen') but only with en-…-ar (e.g.en-sanchar 'to widen'); (ii) the denotation of a closed end state is only possible with a-…-ar (e.g.a-lisar 'to straighten') but not with en-…ar (e.g.encrespar 'to curl').With nominal bases, (i) a-…-ar specifies that only one core meaning can be denoted by a given verb (e.g. the transference of a property in a-vinagrar 'to turn sour'), whereas en-…-ar remains underspecified on this regard (e.g. the locatum/locatio ambiguity in envinagrar 'to put (in) vinegar'); (ii) provided that only en-…-ar remains underspecified with regard to the result state/location, the development of idiosynchratic/conventionalized meanings is (for the most part) expected with this verbal scheme (and not with a-…-ar).
I now turn to the syntactic expression of affectedness in terms of result state/location, which includes the split of BecomeP in two projections, i.e., PP and ResultP.

Split little-v via affectedness
In this section, I propose that there is a projection, ResultP, between v.Caus and verbalizer-v, that determines how specific the result state/location for the internal argument is.I further suggest that the morphological spell-outs of the head of ResultP are the prefixes a-/en-.In 4.1, I capture the main points of Ramchand's (2008) proposal that rhematic PPs are located below ResultP, structure that I assume to characterize BecomeP in ( 14); in addition, I address Kratzer's (2004) argument on specification of objects, and Bowers' (2010) argument on the high position of affectees in syntax in order to claim that the specification for result state/location of the internal argument is a projection (ResultP) that is located higher than PP.In 4.2, I argue that the morphological spell-outs of Result are a-/en-.In 4.3, I reinterpret, in syntactic terms, Montalbetti's (1996) argument that a-/en-attach to an already formed verbal base in order to claim that verbalizer-v (a distinct head from v.Caus) is located between ResultP and PP.In 4.4, I reintroduce the syntactic structure of a-/en-…-ar verbs in (1), which makes explicit that v.Caus and verbalizer-v are distinct syntactic heads.

Specification of result state/location
In this subsection, I suggest that BecomeP in ( 14) is decomposed into a projection that denotes the specification of result state/location (ResultP) and PP, where the latter is located below the former in the structure.The proposal is based on Ramchand (2008), who claims that ResultP is located above PP.Further support for the location of ResultP above PP comes from an extension of Kratzer's (2004) argument of object specification higher than VP and from Bowers' (2010) claim that affectees are higher than themes in syntax in terms of an affectedness specification (for result state/location) of the internal argument.
Ramchand (2008) proposes that dynamic events (those that denote change, as the ones I am discussing) are formed by subevents, in particular, an initiation subevent, a process subevent and a result subevent, each of which correspond to a syntactic projection. 24,25She proposes that events could incorporate rhemes, i.e., material that does not determine its own particular subevent, but acts as a modifier or descriptor of the subevent under consideration.Thus, rhematic material "further describe[s] the properties of the relevant subevent" (Ramchand 2008:54).In particular, there are three aspects of her treatment of rhemes that are relevant for my discussion: (i) rhematic material can add a description to the result subevent, (ii) rhematic material includes PPs, where the figure and the ground are introduced-the former is the specifier of PP and the latter is its complement-, and (iii) rhematic material is located in the complement position of the projection of which it adds a description (in particular, PP rhemes are structurally lower than ResultP).
Ramchand's theory is helpful to make explicit what characterizes BecomeP in ( 14), the projection that denotes change of state/location, in a more appropriate way.Recall that in section 3 I distinguished a-…-ar from en-…-ar in terms of (under)specification of result state/location.This would correspond to Ramchand's result projection (ResultP). 26Recall as well that, following Gibert Sotelo & Pujol Payet (2015), I introduced four subtypes of parasynthetic verbs with a-/en-, i.e., property, locatio, locatum and instrument.As previous literature has shown (Gumiel Pérez & Nieto 1999;Mateu 2002;Fernández Alcalde 2011;Fábregas 2015a; among others), parasynthetic verbs include a prepositional projection (PP). 27Such PPs could be understood either concretely or abstractly (Gumiel, Pérez & Nieto 1999).When PP is concrete, the meanings involved are those of locatio, locatum and instrument-this would include the verbs with nominal bases such as em-botellar 'to bottle,' a-murallar 'to wall' and a-cuchillar 'to knife' (as in (24-26)); when PP is abstract (an abstract location in Gumiel, Pérez & Nieto's 1999 terms), it denotes the transference of a property-this would include the verbs with nominal bases such as a-cartonar 'to make stiff' (as in ( 23)) and the adjectival bases such as en-dulzar 'to sweeten.'28What is relevant for my proposal is that this means that BecomeP (in parasynthetic 24 Ramchand (2008) has an extensive discussion on the possible combinations of these subeventsand, also, an extensive discussion on when some of them are not present, e.g., the result subevent may not be present, as in push.I only focus on the cases where the result subevent is in fact present.

25
The initiation subevent (roughly) corresponds to v.Caus in my proposal.Gabriel Martínez Vera verbs) is decomposed into ResultP and PP, where PP is lower than ResultP in the structure. 29The theme (the figure) would be in Spec,PP and the base predicate (the ground) would be the complement of PP.
I now turn to the structural position of affectees.On the one hand, Kratzer (2004) claims that there are operators in the event structure located outside the thematic domain (i.e.VP).She discusses the relation between telicity and accusative case in German (and English), for telicity specifies the object in an event.Kratzer proposes that, since the object is relevant to determine the telicity of an event, it cannot be part of the verb, so it moves out of the VP domain; she argues for a telic operator [acc] that establishes a relation with the object, where the first one specifies the second one (MH = Matterhorn): (34) What is relevant for my proposal is that the specification of the internal argument in terms of result state/location is realized in the structure in a position that is distinct from the one where it is base-generated.Assuming Ramchand's proposal, such specification takes place in Spec,ResultP, the landing site of the theme.
On the other hand, since the specification of result state/location is a form of affectedness specification, it is relevant to ask what the position of affectees in syntax is.Bowers (2010) claims that affectees are located in a high position in the syntactic structure.He proposes that, within the Universal Order of Merge (UOM), affectee > theme.Among the many arguments he presents, he discusses the case of Russian, where "there is no double-object construction of the sort found in English and raising of the Th-DP to a position higher than the AffP is also impossible" (Bowers 2010:103).Thus, Russian affectees, which can only be inherently marked with DAT, and themes, marked with ACC, can only appear as the UOM specifies.To test this claim, Bowers applies Barss & Lasnik's (1986) tests, where the order affectee > theme is borne out, because only the affectee (DAT) can c-command the theme (ACC) (the test of reciprocals is illustrated in (35) Mateu (2002Mateu ( , 2012) ) and Fábregas (2015a) for discussion on different structures of PPs in parasynthetic verbs and for relevant distinctions that need to be captured (which in many respects correspond to Gibert Sotelo & Pujol Payet's 2015 characterization of the possible meanings that appear in parasynthetic verbs).See also Gibert Sotelo & Pujol Payet (2015) for a characterization of a-…-ar as having a P meaning to and of en-…-ar verbs as having a P meaning into/onto.

29
In Ramchand's theory, BecomeP would also include the process projection (ProcessP), which I leave aside for not being strictly relevant for my purposes.
[acc] Based on Bowers' discussion of the relative position of affectees in syntax, I propose that the internal argument moves to a higher position where it can be specified for result state/location, specifically, it moves from Spec,PP to Spec,ResultP, where it is specified for result state/location, i.e., the theme/figure is specified for affectedness.
To sum up, BecomeP consists of (at least) two projections: ResultP and PP, with the latter located below the former in the structure (Ramchand 2008).The theme is specified for result/state location via movement from Spec,PP to Spec,ResultP, which is grounded in Kratzer's (2004) discussion regarding the specification of the internal argument and Bowers' (2010) proposal that affectees are higher than themes in the syntactic structure.

Morphological spell-outs of affectedness
In this section, I suggest that the morphological spell-outs of the head of the projection that specifies the result state/location (Result) are the prefixes a-/en-.Provided that each morpheme corresponds to a syntactic head (but not necessarily vice versa) (Baker 1985; 30  Bobaljik & Thráinsson 1998;Harley 2016), as is common practice in syntactocentric views of morphology, a-/en-are the spell-outs of a syntactic head.In particular, I propose that the prefixes are the morphological spell-outs of the Result.If this is correct, there are two interesting outcomes.
On the one hand, if a-/en-are the morphological spell-outs of Result, there could be a PP that merges with either a-or en-with a change of meaning of (at least) the result state/location (in addition to the change of state/location involved).Such issue has not been addressed in the past; instead, the previous syntactic approaches to parasynthetic verbs with a-/en-in Spanish (Gumiel, Pérez & Nieto 1999;Fernández Alcalde 2011) treat a-/en-as meaning the same. 31As discussed in section 2, my proposal points otherwise, which is a positive outcome, since the data show that a-/en-do not have the same meaning as the examples below illustrate: a-botellar means 'to get the shape of a bottle' (property in Gibert Sotelo & Pujol Payet's 2015 classification) and em-botellar means 'to bottle' in (37), and a-vinagrar means 'to make sour' and en-vinagrar means 'to put (in) vinegar)' in (38) (as mentioned with regard to ( 27)): (37) a. Juan a-botelló el vino.base: botella 'bottle' a. John PAR-got.the.shape.of.a.bottle the wine a. 'John gave the wine the shape of a bottle.' 30 I am referring to the Mirror Principle.

31
There is some literarture on this in other Romance languages, in particular, in Catalan.As Acedo-Matellán (2006) points out, a-/en-in Catalan do not mean the same: while the meaning of both correspond to abstract prepositions, a-denotes a change of state and en-denotes the locative meaning inside.On the other hand, although more research is needed, if such specification is required in a-/en-…-ar verbs, the strong claim is that the absence of such projection, i.e. the absence of a-/en-, yields ungrammatical results.If this were so, there would be an explanation to the unsolved puzzle of why it is the case that whenever the prefixes are absent the verbal forms are ungrammatical.In principle, this claim is borne out: 32

Verbalizer-v
In this subsection, I claim that verbalizer-v is a distinct head from v.Caus.To do so, I reinterpret, in syntactic terms, Montalbetti's (1996) morphological argument that, in a-/en-…-ar verbs, a-/en-attach to an already formed verbal base.Thus, the morphological derivation is as in ( 41) and not as in (42) (the base predicates are chico 'small' and gordo 'fat'): An anonymous reviewer wonders about the status of verbs that involve a result state/location and are telic but do not appear with parasynthetic prefixes, e.g., parquear, estacionar vs. a-parcar 'to park' or blanquear 'to whiten,' amarillear 'to turn yellow.'I do not think that the only way to convey the meaning of result state/location in Spanish is by means of a parasynthetic prefix, although the presence of a null prefix (Gumiel, Pérez & Nieto 1999) has been proposed in the literature.Perhaps such null prefix is needed for pairs such as tibiar, en-tibiar 'to warm,' that do not seem to convey different meanings.Another option would be that result states/locations are specified directly by the lexical items (Beavers 2011).Yet another option, as the reviewer points out, would deal with different properties of gradable adjectives.I leave this topic for further research.

Syntactic structure of Spanish parasynthesis:
Isogloss 2016, 2/2 81 Towards a split little-v via affectedness Since there is no Spanish rule through which a-/en-attach to an adjectival base (e.g.*achico, *en-gordo 33 ) (Montalbetti 1996:331), Montalbetti focuses on noun-verb pairs, for it is not transparent which the base is, as the following examples show: In order to claim that a-/en-attach to verbs, Montalbetti disambiguates the examples applying Myers' (1984) effect: when there are homophone noun-verb pairs, only the base form allows further derivation.The following example, where the noun honor is the base, illustrates the effect: [honor] V !*honor + -al / -ant / -ive (Montalbetti 1996: 333) Since verbal derivation is grammatical (and productive) but nominal derivation is not in a-/en-…-ar verbs, Montalbetti argues that the base forms are verbal, not nominal.As the tables show, the nominal forms, where suffixes are attached to nominal bases, are all ungrammatical, whereas the verbal forms, where suffixes are attached to verbal bases, are possible or existent.Thus, a-/en-attach to verbal (not nominal) bases (ungrammatical = *; existent = "): My claim that verbalizer-v is a distinct head from v.Caus is as follows: (i) provided that the result state/location projection (ResultP) is higher than PP but lower than v.Caus, (ii) provided that a-/en-are the morphological spell-outs of Result, and (iii) provided that the base predicates are adjectives or nouns (crucially, they are not verbs), if the prefixes attach to verbal forms (as Montalbetti claims), ResultP merges with a PP that contains a base predicate that has already been verbalized by means of verbalizer-v (that is distinct from v.Caus).

Syntactic representation of split little-v
In this subsection, I reintroduce the syntactic structure of a-/en-…-ar verbs presented in (1), which makes explicit that v.Caus and verbalizer-v are distinct syntactic heads (Hale & Keyser 1993;Harley 1995Harley , 2009Harley , 2016;;Marantz 1997;Embick 2004;Serratos 2008;Key 2013;Jung 2014). 34As the previous discussion in this section has made explicit, the syntactic structure in ( 14) is enriched with two more projections: (i) a projection that specifies the result state/location of the internal argument (ResultP) and (ii) a verbalizer-v projection (Marantz 2013).ResultP has a feature Result State/Location [S R ], which specifies how specific the result state/location for the internal argument is.Thus, the value of [S R ] could be positive [+S R ]-i.e., a (particular) result state/location is specified-, as in a-…-ar, or negative [-S R ]-i.e., a (particular) result state/location remains underspecified-, as in en-…-ar.The prefixes a-/en-are the morphological spell-outs of the head of ResultP with [+S R ]/[-S R ] respectively.Since the internal argument is specified for result state/location, it moves from Spec,PP to Spec,ResultP.This movement is justified by the presence of an intermediate projection, verbalizer-vP, which blocks a possible relation between a-/en-and Spec,PP.Having addressed the main arguments for the structure introduced in (1), I rewrite it as (50) (Bowers 2002;Alexiadou, Anagnostopoulou & Schäfer 2006;Folli & Harley 2004;Marantz 2008;Pylkkänen 2008;Harley 2009Harley , 2016;;Wurmbrand & Shimamura 2016): 35,36   34   There are two key points in the overall argumentation of these proposals that are relevant for mine.The first one deals with overt evidence (e.g., morphological exponents, argument structure alternations) that suggest the split of the verbal domain (Larson 1988(Larson , 1990;;Hale & Keyser 1993;Chomsky 1995;Kratzer 1996).For my purposes, what is relevant is that there could be more than one verbal (not voice) projection stacked one over the other.For example, Serratos (2008:240) shows that in Chemehuevi, for example, v.Be and v.Caus can appear together (with the latter taking the former as its complement, and the former verbalizing the root).Thus, the verbal composition of verbs involves the presence of more than one verbal projection.The second one deals with the featural composition of projections in the verbal domain (Harley 1995(Harley , 2013(Harley , 2016;;Marantz 1997).One recent proposal is, for example, Key (2013), whose discussion is based on Turkish.He specifically address that verbalizer-v and v.Caus correspond to two different features ([v], [Caus]), which can appear as a bundle (i.e., [v, Caus]) or not (i.e., [v] and [Caus] appear in different projections) depending on the morphological exponents under consideration.These proposals point towards the fact that the verbal domain is formed by different verbal projections and that verbalizer-v and v.Caus can be dissociated.

35
The final verb forms a-/en-…-ar are composed via head movement (Roberts 2010).

36
An anonymous reviewer mentions that verbs such as a-bofetear 'to slap,' a-palear and a-porrear 'to beat' would not have the structure in (50) because they are atelic activities.Recall that in footnote 6 I discussed that case of a-palear 'to beat' and claimed that there is an affected theme (and thus change).The same seems to hold for a-bofetear 'to slap' and a-porrear 'to beat,' which are grammatical with the lo que le ocurrió/sucedió al tema x es 'what happened to theme x is' test.Recall as well that in footnote 10 I mentioned that (a)telicity gives mixed results with parasynthetic verbs.For example, a-lisar el pelo en 15 minutos/durante 15 minutos 'to straightened the hair in 15 minutes/for 15 minutes' is grammatical; a-lisar 'to straighten' does Syntactic structure of Spanish parasynthesis: Isogloss 2016, 2/2 83 Towards a split little-v via affectedness (50) The alternative hypothesis would claim that a-/en-are low heads located in PP (Gumiel, Pérez & Nieto 1999) or are the morphological spell-outs of some kind of Become predicate (Fernández Alcalde 2011).However, these alternatives face certain challenges.On the one hand, they imply that the prefixes attach to the base predicate prior to verbalization, which predicts that word forms such as a-tonto 'PAR-dumb,' a-cuchillo 'PAR-knife,' en-gordo 'PAR-fat' and embotella 'PAR-bottle 37 could be grammatical, but this does not seem empirically correct.My proposal overcomes this challenge by making explicit that a-/en-merge with already verbalized base predicates.On the other hand, the (under)specification for result state/location, if present, would take place in a low position, which contradicts the literature that claims that object specification takes place higher than VP and that affectees are projected high in the structure (see 4.1 above) (in fact, the alternatives mentioned do not capture the affectedness distinction discussed, so the evidence I showed would have to be set aside). 38 The structure in (1,50) is appropriate to explain the ambiguity with adjunction of adverbial expressions, such as casi 'almost' (Gumiel, Pérez & Nieto 1999) and otra vez 'again' (von Stechow 1996;Beck & Johnson 2004).As von Stechow (1996) and Beck & Johnson (2004) discuss, the adjunction of adverbial expressions such as again provides evidence for a decomposed syntactic structure where the adjunct can take scope over different parts of the structure yielding repetitive or restitutive readings for a sentence.Specifically for a-/en-…-ar verbs, this claim means that adverbs such as casi 'almost' and otra vez 'again' should yield two include a change and a result state, but allows both a telic and an atelic reading.Thus, being atelic does not mean that no change is involved.This discussion would suggest that the structure in ( 50) is in fact present with these verbs as well-this does not necessarily mean that all verbs with -ear necessarily have such structure (but only the ones with a-/en-, which, as discussed, seem to have it); I leave this topic for future research.37 These forms are not the conjugations of the verbs, which are existent words.38 Gumiel, Pérez & Nieto (1999)'s proposal faces another problem, because the authors claim that the structure of a-/en-…-ar parallels resultative structures of English, which does not seem right: as Snyder (1995Snyder ( , 2001) ) and Mendívil (2003) discuss, Spanish does not allow resultatives such as those in (i), which means that Spanish has a restriction for combining two complex events (water and flat in (ii)): (i) The To sum up, I claim that Spanish a-/en-…-ar syntactic structure provides evidence for a split little-v, where v.Caus is distinct from verbalizer-v.Given that v.Caus must be higher in the structure than ResultP and PP, and that ResultP is higher than PP, but the base predicate must be verbalized, I have posited that v.Caus and verbalizer-v are different projections.

Beyond a-/en-…-ar 39
In this section, I address the consequences of my proposal.In section 5.1, I address how my proposal with regard to a-/en-…-ar interacts with verbalizers.In 5.2, I suggest how my proposal can be extended to des-…-ar parasynthetic verbs in Spanish.In 5.3, I address the differences among Spanish, French and Italian in terms of the feature composition of Result.In 5.4, I suggest that Romance deadjectival parasynthesis differs from English degree achievements such as straighten and soften in terms of the feature composition of Result.

5.1.a-/en-…-ar and verbalizers in Spanish
In this subsection, I address the fact that my proposal predicts that verbalizer-v in Spanish could have different 'flavors.'Since my account posits that verbalizer-v is a different projection from v.Caus, I predict that, in principle, the former could have different morphological spell-outs.
To address this prediction, two elements are needed: there should be at least one overt verbalizer and this verbalizer should be able to combine with a-/en-.Regarding the first issue, Spanish has more than one overt verbalizer (the last one is the covert verbalizer present in a-/en-…-ar, as well as in many other verbs without a-/en-): (53) -e(ar) blanqu-e(ar) 'to whiten' 39 I would like to thank an anonymous reviewer for suggesting the possible extensions of my proposal that are addressed in 5.2, 5.3 and 5.4.
Although many details still remain to be worked out, I think my proposal is in principle compatible with his, because I am adding a projection, distinct from v.Caus, where these suffixes could appear.

41
An anonymous reviewer wonders about the contribution to the composition of the structure of the suffixes in (53-57) (since I split v.Caus from verbalizer-v).As mentioned in footnote 40, it seems that the presence of more verbal projections could be a positive outcome, precisely because these suffixes appear in structures that seem to correspond to different argument structures (e.g., the case of -iz(ar) that Fábregas 2015b discusses).I would like to suggest two aspects that are worth considering with regard to this issue.The first one deals with the featural composition of such suffixes.All of them would have a feature, let us call it [v] (see footnote 34), that verbalizes a (categorized) base (or a root if one assumes a Distributed Morpohology (DM) approach where roots are categorized in the syntax), and, in addition, they could have some other features depending on their contribution to the structure.One example could be the very tentative proposal that -ec(er) denotes a bound.According to my proposal, [Caus] would be in v.Caus.The second issue to be worked out is how features are materialized in an approach such as DM.Take two Spanish examples: if Fábregas (2015b) is correct with regard to -iz(ar) in the sense that this suffix appears in different argument structures (linked to predicational and/or prepositional projections), this would suggest that this suffix is the exponent of different feature bundles (recall as well that [Caus], in his approach, is not part of the feature bundle that -iz(ar) materializes).He suggests that -e(ar) shows the same kind of varied behavior (with regard to manner or an attributive element), so it would also seem that it can materialize different feature bundles.This kind of evidence, then, suggests that the verbal domain in Spanish is actually split and my proposal provides further evidence for it.

42
Note that this topic is also relevant with regard to the discussion in section 1 where I mentioned cases such as a-terr-iz(ar) ' The question is, thus, whether the bases are verbalized prior to merging with ResultP, for which Montalbetti's (1996) test can be applied.To do so, I make use of the verbs des-huesar 'to bone' and des-plumar 'to pluck,' which have the nominal counterparts des-huese and des-plume, respectively.The evidence points towards the fact that verbalization takes place first: (62) Thus, des-…-ar seem to have the same structure I have proposed for a-/en-…-ar: the base predicate combines with P, and is verbalized prior to merging with Result.The crucial difference between a-/en-/des-thus lies in the feature composition of Result: [+S R ], [-S R ] and [¬P], respectively. 44,45  43   My proposal does seem to be extensible to cases where the base predicate is an adjective, e.g., des-bravar 'to tame,' but I limit my attention to the cases that they address, which are thoroughly studied in their proposal.

44
These verbs would also differ in the particular meaning of P. Following Gibert Sotelo & Pujol Payet, a-…-ar verbs have a P with meaning to, en-…-ar verbs have a P with meaning into/onto and des-…-ar verbs have a P with meaning from.
Since I am only focusing on the counterparts of Spanish a-/en-, this discussion is rather tentative, in particular, when the Italian prefixes are considered, which are many, e.g., s-, ri-and dis-, in addition to a-/in- (Di Sciullo 1997:66).Furthermore, this discussion is only centered on adjectival bases.

Contrasting Romance deadjectival parasynthesis and English degree achievements
This section follows up on the previous one: I distinguish Romance deadjectival parasynthetic verbs with a-/en-from English (morphologically derived) degree achievements. 48I propose that English ResultP does not have a feature specification as the one discussed in the previous section for Romance.
In particular, to derive degree achievements, English derivational morphology includes several affixes, e.g., en-(en-large), -en (straight-en), -∅ (cool-∅) (Hay, Kennedy & Levin 1999).Crucially, it is worth noting that there is no relevant difference in derivational morphology between the presence/absence of a minimal/maximal degree in the base predicate nor a distinction between low/high bounds.Thus,examples (68a,b) show that there is no distinction between low/high bounds with -en/en-(in-); (68c) shows that, in addition to such distinction, -∅ can take bases that denote a minimal/maximal degree and bases that do not denote such degree; and (68d) provides examples where -en takes bases that denote a minimal/maximal degree.
(68) a. hard-en soft-en b. en-rich im-poverish c. clean-∅ dirty-∅ d. straight-en, flatt-en As mentioned in the previous subsection, these distinctions are relevant in Romance.This means that, while it can be maintained that both Romance and English include ResultP (following Ramchand 2008:102), Result in English does not show a feature specification as in Romance (in terms of [±S R ,±B L ]).Thus, English morphology is less restricted than Romance regarding what counts as a possible adjectival base to derive a degree achievement.48 English allows the derivation of locatio/locatum verbs via a very productive process of zeroderivation.In Ramchand's (2008) theory, this means that the rheme of these verbs is a PP.It is worth pointing out that Romance brings together deadjectival and denominal verbs in parasynthesis, whereas English maintains a split between them (in particular, in terms of derivational morphology, i.e., degree achievements show the affixes en-/-en/-∅, whereas location/locatum verbs are derived via zero-derivation).Ramchand (2008) addresses these alternations explicitly in her section 4.4.

Conclusion
In this paper, I have claimed that v.Caus and verbalizer-v are different syntactic heads by analyzing a-/en-…-ar verbs in Spanish.Their syntactic structure provides (further) evidence for a decomposed vP structure: I claim that these verbs are complex predicates with a v.Caus-headed vP and a PP.The two verb types only differ in the properties of ResultP, located between vP and PP, whose head specifies whether the internal argument reaches a result state: a-...-ar specifies a (particular) result state/location for it but en-...-ar remains underspecified with regard to the result state/location.Since a-/en-attach to an already verbalized base predicate and that the absence of a-/en-makes the verbs ungrammatical, I claim that a-/en-are the heads of ResultP, which merges with verbalizer-v which is distinct from v.Caus.This proposal can be extended to des-…-ar verbs in Spanish and to French and Italian parasynthetic verbs.Romance differs from English degree achievements with regard to the feature composition of ResultP: Romance shows features in that projection that English lacks.

Appendix. On uncategorized bases in Spanish parasynthetic verbs
In this appendix, I suggest that the bases that appear in a-/en-…-ar predicates are not categorized, but are instead uncategorized (Hale & Keyser 1993, 1998;Marantz 1997Marantz , 2013;;Embick 2004;Bobaljik 2012;Harley 2013Harley , 2016;;among others). 49An argument that points towards this direction is the fact that derived bases are not possible in parasynthetic verbs.Thus, (69) shows that bases with the adjectival suffix -oso and with the nominalizer suffix -ura, respectively, yield ungrammatical results: (69) a. *a-deseos-ar / *en-deseos-ar 'to make eager' base: dese-oso 'keen, eager' b. *a-pintur-ar / *en-pintur-ar 'to be painting like' base: pint-ura '(a) painting' There are two interesting outcomes that follow from this.First, allowing for uncategorized bases correctly accounts for the fact that the bases that appear in a-/en-…-ar, when categorized, belong to very different grammatical categories: 'to make turbid' base: turbio 'turbid' (adjective) c. en-simismar 'to be absorbed in oneself' base: sí mismo 'self' (anaphoric pronoun) (6) a. *El hambre a-delgazó (a) los pollos.a. *the hunger PAR-thinned DOM the chickens a. *'Hunger thinned the chickens.'b. *El temporal em-pedró el camino.b. *the storm PAR-put.stones the road b. *'The storm formed the road with stones.'

(
21) a. Los alumnos *a-largaron/ en-sancharon su conocimiento.a. the students PAR-lengthened/PAR-widened their knowledge a. 'The students *lengthen/broadened their knowledge.'b.Una falla en Facebook *a-lisó/ en-crespó los nervios (de los usuarios).b.A failure in Facebook PAR-straightened/PAR-curled the nerves of the users) b. 'A failure on Facebook got on the users' nerves.'Syntactic structure of Spanish parasynthesis: Isogloss 2016, 2/2 73 Towards a split little-v via affectedness c.Juan le en-dulzó la píldora a María.c. John DAT PAR-sweetened the pill DOM Mary c. 'John softened the news for Mary.' d.Se en-frió el amor de los amantes.d.SE PAR-cooled the love of the lovers d. 'The love of the lovers cooled down.' 16 (22) a.El presidente corrupto en-gordó sus bolsillos durante su gobierno.a. the president corrupto PAR-fattened his pockets during his government a. 'The corrupt president fattened his pockets during his government.'b.Han a-delgazado las arcas públicas.b. have PAR-thinned the treasury b. 'The treasury has been thinned.' point out.Syntactic structure of Spanish parasynthesis: Isogloss 2016, 2/2 75 Towards a split little-v via affectedness than one core meaning. 22(27) a.El mal clima a-vinagró la mayonesa.a. the bad weather PAR-turned.sour the mayonnaise a. 'The bad weather turned the mayonnaise sour.' b.Juan en-vinagró el pollo.b.John PAR-put.vinegar the chicken b. (i) 'John put vinegar on the chicken.' (locatum) b. (ii) 'John put vinegar inside the chicken.' (locatum) b. (iii) 'John put the chicken in vinegar.' (locatio)

( 28 )
El ladrón a-cuchilló a la víctima.the thief PAR-knifed DOM the victim 'The thief knifed the victim.'Another interesting difference between a-/en-…-ar verbs with nominal bases in terms of flexibility deals with the development of idiosyncratic/conventionalized meanings-which was mentioned as well for adjectival bases.As descriptive work has pointed out (Serrano-Dolader 1999), such meanings are only found, in general terms, in en-…-ar verbs, in particular, with nominal bases: 23 (29) en-arbolar 'to hoist, to raise' base: árbol 'tree' (30) en-raizar 'to take root' base: raíz 'root' (31) em-plazar 'to place' base: plaza 'square' 22

26I
am thankful to an anonymous reviewer for pointing this out.

27I
am thankful to two anonymous reviewers for pointing out some advantages of including PP in the structure I propose.
b. Juan em-botelló el vino.b.John PAR-bottled the wine b. 'John bottled the wine.' (38) a. Juan a-vinagró la crema de leche.base: vinagre 'vinegar' a. Juan PAR-turned.sour the cream of milk a. 'Juan made the whipping cream sour.' b.Juan en-vinagró la ensalada.b.Juan PAR-put.vinegar the salad b. 'Juan put vinegar in the salad.' (39) a. a-grandar -*grandar 'to enlarge, to make big' b. a-calorar -*calorar 'to get hot' (40) a. en-cajonar -*cajonar 'to box' b. en-lodar -*lodar 'to make muddy' and Italian appear to have the same feature composition for Result, i.e., a contrast of [±S R ], where a-has [+S R ] and en-/in-has [-S R ], while French, in addition to [±S R ], would have another feature, let us call it [B L ] for low bound.When the feature is [+B L ], then there is a low bound, and when the feature is [-B L ], there is not a low bound but a high one.Therefore, ain French has the features [+S R ,+B L ] and en-has the features [-S R ,-B L ]. Thus, French parasynthetic prefixes are more specified-in terms of the number of features-than their Spanish and Italian counterparts.

Table of Contents
) Lo que le ocurrió/sucedió a Pedro es que el problema lo a-tontó.the that DAT happened to Peter is that the problem ACC PAR-made.dumb'What happened to Peter is that the problem made him feel dumb.' (12) Lo que le ocurrió/sucedió a los pollos es que Juan los en-gordó.the that DAT happened to the chicken is that John ACC PAR-fattened 'What happened to the chickens is that John fattened them.' and the test of bound variables is illustrated in (36)): Sotelo and Pujol Payet's (2015)-…-arIn this subsection, I propose that des-…-ar verbs have the same structure as the ones with a-/en-…-ar, but differ with regard to the feature specification of Result.To do so, I follow GibertSotelo and Pujol Payet's (2015)analysis (which is limited to nominal bases 43 ).They suggest that these verbs denote a change of state characterized in terms of a 'detachment from.'Thus, the theme changes from having a particular property/entity to not having it, e.g., des-cabezar means 'to behead' (the base predicate is cabeza 'head'), where the theme transitions from having a head to not having it, or des-terrar 'to exile' (the base predicate is tierra 'land'), where the theme transitions from being in the land to being exiled from it.They capture the detachment from-part of the meaning via a prepositional component with the meaning from.As it can be readily seen, the basic elements of des-…-ar verbs resemble what I have proposed for a-/en-…-ar in (50): there is a prepositional component (a PP with P from) and a result state (basically, the negation of the base predicate, which can be formalized with a feature [¬P], where P is the base predicate).
to land' (among other examples): since there are a reduced number of cases where v.Caus alternates with other flavors of v (e.g., v.Become, v.Do), the proposal in (1,50) provides enough projections for v.Caus/v.Become/v.Do, for different verbalizers and for parasynthetic prefixes.