Inalienable possession (and lack thereof) in Spanish

3.1 Inalienable possession in Spanish

Inalienable possession between a dative argument and a relational noun in Romance has been analyzed under the scope of Binding and Control (Demonte 1988; Guéron 1983, 1985), predication (Vergnaud and Zubizarreta 1992), and possessor raising (Demonte 1995; Rodrigues 2023; Sánchez López 2007; Suárez-Palma 2019, 2022, 2024).^[4] In her theory of dative arguments in Spanish, Cuervo (2003) observes that dative possessors tend to occur with both stative and dynamic monoeventive predicates (e.g., ver, ‘to see’ and lavar, ‘to wash,’ respectively). Cuervo notes that dative possessors behave syntactically as dative recipients and sources, in that they show the same case, hierarchical position, and word order; besides, they are semantically directly related to the object and not to the predicate. This leads this author to suggest that dative possessors must be projected as low applicative arguments, the type of functional head introducing goals and recipients in double object constructions (henceforth, DOCs), according to Pylkkänen (2002). Pylkkänen’s proposal relies on different semantic notations attributed to low applicative heads to account for the source and goal interpretations of the dative in these configurations.

Dative possessors, however, are understood as the possessor or location of the possessee, never as losing or receiving it; in other words, a relation of transfer of possession cannot be forced in the structures containing them. Therefore, in addition to the two types of low applicative heads in (9), Cuervo adds a third one whose semantics denote a static relation of possession, which can be alienable or inalienable (10), i.e., dative possessors would be instances of DOCs.

The advantage of aligning DOCs with those containing possessor datives is that the latter’s participation in the event is accounted for; this is not always the case in genitive possessor constructions, as first noted by Demonte (1995). Moreover, the affectedness reading that sometimes is present in these sentences is derived from the dative’s interpretation as being the possessor of an affected object.

The possessor dative in (11) is preferred because it is hard to imagine a context in which Luisa’s nose is operated on in her absence. According to Cuervo’s (2003) analysis, a sentence with a dative possessor like (12a) would have the derivation shown in (12b).

(12)

a.

Pablo le	lavó	el	auto	a Valeria.
Pablo 3sg.dat	washed	the	car	Valeria.dat
‘Pablo washed Valeria’s car.’
(Cuervo 2003: 86)

b.

In Cuervo’s (2003) proposal, the dative possessor originates in the specifier of the low applicative head, which relates it to the theme in its complement position, and the whole applicative projection merges as the root’s complement. Cuervo, however, does not delve into how the inalienable possession construal arises in this configuration. One must assume it is due to the nature of the relational noun in Appl’s complement position, since the semantic notation of the low applicative of possession would be identical in alienable and inalienable contexts; moreover, this construal is already present in DPs hosting an internally possessed body-part nominal (e.g., mis manos, ‘my hands’). Nevertheless, all the aforementioned analyses of this phenomenon share the assumption that the possessor originates inside the DP containing the relational noun.^[5]

In fact, Guéron (1985) claims that inalienably possessed nouns take a possessor as an argument, an empty category which can presumably be a null pronominal of some sort inside NP (13).

More recently, Armstrong (2021) develops an analysis of multiple external possession relations in Spanish prepositional double object verbs (e.g., te puse la mano en el hombre, ‘I put my hand on your shoulder’); Armstrong assumes a possessor raising approach whereby this argument originates inside the DP containing the relational noun, where the relation of possession is established. Later, the possessor raises to the specifier of an applicative phrase where it receives a theta role and dative case, as shown in (14).

(14)

a.

Diego	me i	levantó	la	mano i .
Diego	1sg.dat	raised	the	hand
‘Diego raised my hand.’

b.

(Armstrong 2021: 9)

The analysis I am proposing in the next section follows this trend, in that it combines Cuervo’s low applicative of possession and middle (affected) applicative with a possessor raising account in order to explain how inalienable possession originates, or fails to do so, in the structures under consideration.

3.2 A hybrid approach to inalienable possession in Spanish

3.2.1 Internal possession

Before delving into how the inalienable possession interpretation between dative possessors and relational nouns comes about, let us explain how this construal originates inside the DP containing a relational noun. To that end, I will adapt Alexiadou et al.’s (2007) proposal for the different positions where possessive elements can merge or be licensed inside the possessee DP, shown in (15). In the present proposal, I assume that possessors originate in Spec,nP in the form of a pronominal or a full DP. Additionally, relational and body-part nouns select for a PRO that merges in their specifier, which is controlled by the possessor in Spec,nP, thus giving rise to the inalienable possession construal. However, if no possessor sits in Spec,nP, the PRO in the specifier of the body-part noun remains uncontrolled, which favors an arbitrary interpretation of the body part’s possessor (cf. (8) above).^[6]

(15)

1. Lexical DP possessives	John’s book	(English)
2. Clitic possessives	Su libro	(Spanish)
3. ‘Weak’ pronoun possessives	El seu llibre	(Catalan)
4. Post-nominal strong possessors	El libro suyo	(Spanish)
5. Alienable possessors
6. Inalienable possessors (PRO)
(Adapted from Alexiadou et al. 2007: 575)

Additionally, I will assume a version of the feature endowment for Spanish possessors proposed by Fábregas (2011); therefore, in addition to its own set of interpretable person and number features, the possessor will contain several uninterpretable phi-features to be checked against the possessee, including a number and a gender feature, although the latter is only morpho-phonologically realized in prenominal position in 1st and 2nd person plural possessive determiners in Spanish (16a). On the other hand, the feature endowment of the determiner is shown in (16b).

(16)

a.	Feature endowment of the possessive: [Person, Num, uNum, uGen]
b.	Feature endowment of the determiner: [D, uNum, uGen]

Thus, a DP containing an inalienably possessed noun like mis cicatrices (‘my scars’) would be derived as in (17).

(17)

a.	Mis cicatrices.
b.

In (17), the relational noun cicatriz takes a PRO in its specifier, which is controlled by the possessor argument generated in Spec,nP. The noun undergoes head movement to Agrº, from where it enters an agreement relation with the c-commanding determiner, therefore valuing its gender and number uninterpretable features: [D, Num_PL, Gen_F]. The possessor, on the other hand, raises to the specifier of AgrP, where it is assigned genitive case by Agrº. As I will show later, this position is also where weak possessors are licensed in languages like Catalan or Italian, and the same one proposed by Ticio (2005) for internal possessor PPs. On its way to Spec,AgrP, the possessor establishes the necessary agreement relations with the nominal: [Gen_F] is valued in Spec,nP against the nominal, and [Num_PL] is valued in Spec,NumP. In languages with possessive determiners, I assume that the possessor further raises to the specifier of the determiner, possibly because this head has an EPP feature that needs checking. Finally, when Lexical Insertion occurs, Spanish resorts to the possessive determiner mis, which is able to spell out the terminal node Dº, as well as the possessor in its specifier; in other words, this would be an instance of Phrasal Spell-Out (Fábregas 2009; Starke 2009).^[7] Alternatively, we could take a morphophonological approach, and interpret this as a case of cumulative exponence (Stump 1998), where several sets of features are materialized by a single lexeme, as suggested by Fábregas (2011).^[8]

In languages with weak possessors, such as Catalan (les meves cicatrius, ‘my scars’), the possessor remains in SpecAgrP, presumably because Dº lacks an EPP feature in these languages. The possessor is lexicalized as the weak possessive meves, independently from the definite determiner, which is realized as les; the latter’s number and gender features are also valued against the relational noun cicatrius in Agrº. The derivation is given in (18).

(18)

a.	Les meves cicatrius	(Catalan)
b.

Additionally, Romance languages such as Spanish and Catalan have strong possessors, which occur after the noun. Fábregas (2011) shows that possessors in the absolute final position of the sentence are interpreted as contrastive foci (19).

(19)

a.

Me	gusta	el	color	de	ojos	suyo,	no	el	tuyo.
1sg.dat	like	the	color	of	eyes	hers	not	the	yours
‘I like HER color of eyes, not yours.’

b.

#Me	gusta	su	color	de	ojos,	no	el	tuyo.
1sg.dat	like	her	color	of	eyes	not	the	yours
‘I like her color of eyes, not yours.’
(Fábregas 2011: 22)

Thus, I propose that post-nominal possessors raise to Spec,NumP and remain there to check an uninterpretable [Focus] feature in Numº. Having raised to that position, all the possessor’s uninterpretable features will have been valued and Agrº can still assign it genitive case via Agree. As a result, the possessor is realized as a strong one in both languages: las cicatrices mías (Spanish); les cicatrius meves (Catalan).

(20)

a.	Las cicatrices mías/les cicatrius meves.
b.

When the possessor originates as a full DP or full pronominal, genitive case is realized by means of the preposition de, and the whole PP is adjoined to AgrP, as shown in (21).

(21)

a.	Las cicatrices de él/de Mario.
b.

Finally, when no possessor merges in Spec,nP, the PRO in the specifier of the body-part noun remains uncontrolled, which gives rise to an arbitrary interpretation; however, the body part is still understood to be inalienably possessed by an undetermined possessor, as shown in (8), repeated in (22).

(22)

a.

Esta	depiladora	elimina	el	vello	sin	dolor.
this	epilator	eliminates	the	hair	without	pain
‘This epilator eliminates hair without pain.’

b.

[ DP el [ AgrP vello [ NumP vello [ nP vello [ NP PRO arb [ N’ vell-]]]]]]

Next, let us move on to external possession.

3.2.2 External possession

Having established how a body part/relational noun is internally possessed, let us delve into dative possessors. As noted by Cuervo (2003), these arguments occur in monoeventive configurations containing stative or dynamic predicates when they select a low applicative of possession to be their complement; this functional head would be responsible for relating two entities to convey a static relationship of possession. Unlike Cuervo’s account, my claim is that the relation of possession does not arise inside ApplP; instead, dative possessors, like internal possessors, originate inside the DP containing the possessee. However, such DP is unable to assign genitive case to the possessor in these contexts.^[9] As a result, this argument will need to raise to the specifier of the low applicative head of possession, merging as the complement of the root, to license dative case.^[10] Consequently, Applº will then be realized as a dative clitic, sharing the same phi-features as the possessor DP in its specifier. In other words, low applicatives of possession do not introduce an additional argument to the derivation, but case license an existing possessor argument originating inside the possessum DP. Therefore, in contexts containing a low applicative of possession we will always observe correferenciality between the dative argument and the possessor of the body part.^[11] The derivation of (6) is shown in (23).

(23)

a.

(A mí i )	María me i	miró	las i /*mis i /*tus k cicatrices.
me.dat	María 1sg.dat	looked	the/my/your scars
‘María looked at my scars.’

b.

In (23b), the possessor merges in the specifier of the nP, from where it controls PRO in Spec,NP, and therefore the relationship of inalienable possession is established. The possessor raises to the empty specifier of the low applicative head where it is assigned structural dative case; the applicative head is spelled out as the dative clitic pronoun me, whose person and number features are shared with the possessor DP in its specifier. The possessor in SpecApplP in (23) can be optionally doubled by a dative DP (a mí). The dative clitic incorporates into the verbal root, which head-moves all the way to Tº; finally, the external argument in Spec,Voice (María) will raise to SpecTP to license nominative case and cancel the EPP feature in Tº. This proposal continues to structurally derive the relationship of inalienable possession from inside the DP containing the relational noun. In this sense, I depart from Cuervo’s (2003) analysis that claimed that the possessor originates in SpecApplP, from where the relation of possession – at times, inalienable – arises. The purpose of the applicative head in my story is then to case license a full DP originating inside the possessee. Additionally, the incompatibility between the dative possessor and the possessive determiner is a consequence of the nature of this derivation: Dº cannot be realized as a possessive determiner (mis) because the possessor was unable to value genitive case inside the possessum DP, forcing its exit and raising to Spec,ApplP. In other words, although a lower copy of the possessor is in fact in Spec,DP in the dative possessor configuration, that copy is yet to be case-licensed at that point in the derivation, which renders it ineligible for Lexical Insertion.

An anonymous reviewer points out that if dative possessors are indeed generated inside the possessum DP, just like internal possessors are, we should expect them to block extraction of other material from inside such DP, as Ticio (2005) found was the case for possessor PPs. This linguist develops an analysis of Spanish DPs that mirrors the structure of CP, and explains the different patterns of extraction out of these constituents observed in this language. Specifically, Ticio arrives at the conclusion that possessor and agent PPs block the extraction of argument and argument-like PPs,^[12] since the former occupy a higher position that interferes with the path of the latter in their extraction out of the DP. In (24b), the possessor PP de Ana prevents the extraction of the argument-like PP de cuero out of the DP; however, extraction of the possessor PP is perfectly grammatical, as evidenced in (24c).

(24)

a.

Sujeté	[ DP varios	bolsos	[de	cuero] Arg	[de	Ana] Poss ].
I-held	several	bags	of	leather	of	Ana
‘I held several of Ana’s leather bags.’

b.

*¿[De	qué	(material)] i	sujetaste	[ DP  varios	bolsos	t i  [de	Ana] Poss ]?
of	what	material	you-held	several	bags	of	Ana

c.

¿[De	quién] i	sujetaste	[ DP varios	bolsos	[de	cuero] Arg ]?
of	whom	you-held	several	bags	of	leather
‘Whose leather bags did you hold?’

If my proposal is on the right track, we should expect the same extraction pattern in dative possessor configurations. On the one hand, these possessors also originate in Spec,nP within the possessum DP and pass through the same intermediate positions before reaching Spec,ApplP as internal possessors do. On the other hand, argument and argument-like PPs, originate in a lower position inside the NP, and extraction is therefore thwarted. In (25), I show that this is indeed the case. The argument-like PP de cuero cannot be extracted from inside the possessum DP because the dative possessor a María blocks its extraction, even when this argument does not surface inside the possessum DP (25b). As expected, wh-fronting of the dative possessor is grammatical (25c); these data support the idea that external possessors do originate inside their possessa.

(25)

a.

Le i	sujeté [varios	bolsos	[de	cuero]] i	a María i
3sg.dat	I-held several	bags	of	leather	María.dat
‘I held several of María’s leather bags.’

b.

*¿[De	qué	(material)] k	le i	sujetaste	[varios bolsos t k ]	a María i ?
of	what	material	3sg.dat	you-held	several bags	María.dat

c.

¿[A quién] i	le i	sujetaste	[varios	bolsos [de cuero]] t i ?
whom.dat	3sg.dat	you-held	several	bags of leather
‘For whom did you hold several of their leather bags?’

The sentences in (25) are instances of alienable possession between the dative possessor and the verb’s internal argument. However, the same effects are observed in contexts where the relationship of possession between the two entities is inalienable, i.e., when the possessee is a body-part noun, as in (26).

(26)

a.

Le i	vi	[varias	heridas	[de	arma	blanca]] i	a Antonio i
3sg.dat	I-saw	several	wounds	of	weapon	white	Antonio.dat
‘I saw several stab wounds on Antonio.’

b.

*¿[De	qué	(tipo)] k le i	viste	[varias	heridas	t k ]	a Antonio i ?
of	what	type	3sg.dat	you-saw	several	wounds	Antonio.dat

c.

¿[A quién] i	le i	viste	[varias	heridas	[de	arma	blanca]] t i ?
who.dat	3sg.dat	you-saw	several	wounds	of	weapon	white
‘On whom did you see several stab wounds?’

Therefore, we can conclude that dative possessors originate inside the possessum DP, like internal possessors do. In the next section, I discuss cases of inalienable possession (and lack thereof) in sentences containing change-of-state predicates.

3.3 Inalienable possession (and lack thereof) in change-of-state contexts

As shown in (7), repeated as (27), dative possessors are also possible in sentences containing verbs denoting a change of state; when the verb’s theme is a relational noun, a relationship of inalienable possession may arise between the former and the dative. However, this reading appears to be optional, unlike what happens with the structures examined thus far.

Notice that the dative in (27) can have two different interpretations: (i) affected possessor of the body-part DP – headed by the definite determiner las – undergoing the change of state; and (ii) benefactive/malefactive of someone else’s body-part’s resulting state, in which case the latter is internally possessed via the possessive determiner tus. Although less transparent, the relational noun could also be interpreted to be inalienably possessed by an arbitrary possessor, in which case it surfaces headed by the definite determiner since no possessor is projected inside the possessee, and the dative argument is still interpreted as a benefactive.

Cuervo (2003) argues that change-of-state predicates involve a bi-eventive configuration comprising an activity subevent (v_DO) and a stative one (v_BE). Evidence for this claim comes from the fact that the modifier casi (‘almost’) can have scope over the whole event (vP_DO), or only over the stative projection (vP_BE) (Cuervo 2014).

Cuervo proposes that when an additional dative argument appears in these structures, it is introduced in the specifier of an affected applicative head (Affected-APPL), a middle applicative, that relates the dative argument to the event or to the end state of the theme. This functional head merges sandwiched between the two subevents, giving rise to the affected interpretation of this argument. An example of how this derivation would come about is given in (29).

(29)

a.

Pedro	le i	rompió	el	teléfono	a Luis i .
Pedro	3sg.dat	broke	the	phone	Luis.dat
‘Pedro broke Luis’s phone.’

b.

In (29b), a bi-eventive structure is projected by combining a subevent of activity (v_DO) – on top of which a Voice projection introduces the causer in its specifier – and a stative one (v_BE), denoting the resulting state of the verb’s internal argument; the latter, merges as the root’s complement, inside the stative subevent. A middle applicative head, an Affected-APPL one, sits between the two verbalizing heads, and introduces an additional argument marked with inherent dative case in its specifier. Applº applies this argument to the theme’s resulting state, the vP_BE, and it is therefore interpreted as affected by it. Regarding the additional possessor interpretation of the dative, Cuervo explains that “when the dative is understood as the possessor of the theme object, to assume possession is a way to (pragmatically) accounting for the affectedness of the dative as a result of the change of state of the theme object” (2003: 126). While I will adopt her notion of affected applicatives here, I will show that my proposal is able to structurally account for the possessor interpretation of the dative, not just pragmatically. I will begin by showing in (30) the derivation of (27) where the dative is interpreted as the benefactive, and the relational noun is internally possessed.

(30)

(A mí) María me quemó tus verrugas

‘María burned your warts off for me.’

In (30), we see a causative configuration comprising a subevent of activity (v_DO) and a stative one (v_BE); a Voice projection merges on top of the former and introduces the external argument María in its specifier. Merged between the two subevents we find the affected applicative head, in whose specifier is merged a non-selected dative argument pro, bearing inherent dative case, whose first-person singular features are spelled out by the dative clitic pronoun me. The position where the applicative head merges determines this argument’s affected interpretation, for it is as applied to the stative vP_BE, i.e., the resulting state of the verb’s internal argument. The internally possessed relational noun verrugas is headed by the possessive determiner tus; the derivation of this DP would follow the same steps as the one shown in (17) above. In sum, the role of the affected applicative in this configuration is the one described by Cuervo (2003): to introduce and license an additional argument with an affected interpretation in the structure.

When it comes to the interpretation in which the dative argument is the possessor of the relational noun in (27), I propose that an affected middle applicative with an empty specifier is the functional head responsible for licensing the possessor originating inside the possessum DP, in Spec,nP. As in the case of low applicatives of possession (Low-APPL-AT), this functional head does not introduce an additional argument in its specifier, which is available for the possessor to raise to and be case licensed. In essence, low and middle applicatives of possession serve the same purpose, the only difference being the position they occupy in the derivation, which determines the possible interpretations of the possessor they license in each case.

(31)

(A mí) María me quemó las verrugas

‘Maria burned my warts off.’

The derivation in (31) is also a causative one in which an affected applicative head with an empty specifier merges sandwiched between the two subevents. The verb’s internal argument is a DP containing a relational noun that is inalienably possessed by a null pronominal generated in the specifier of the nP. Because this argument is unable to be assigned genitive case inside the possessum DP, it vacates it and raises to the specifier of the middle applicative head, where it receives structural dative case. As a result, the possessee is headed by a definite determiner agreeing in gender and number with the relational noun. Additionally, because of the position the applicative projection occupies in the structure, the possessor also acquires the interpretation of being affected by the resulting state of its possessee. In the next section, I will discuss the scope of the current proposal when it comes to accounting for other external possession configurations containing body-part nouns.

3.4 Scope of the proposal

The analysis developed here shows potential when it comes to accounting for the inalienable possession effects present in certain prepositional configurations in Spanish, like the one in (5c) above, repeated in (32).

Harley (2002) provides an analysis of double object constructions in English in which the theme and the goal/source/possessor argument are related by means of a null preposition P_HAVE, as in (33).

(33)

a.	John gave Mary a book.
b.

I propose that the same type of projection is at play in (32), the only difference being that, in Spanish, P_HAVE is not phonologically null. The derivation of (32) is shown in (34).

(34)

In (34), the DP Manuel originates in Spec,nP, where it controls the PRO in the specifier of the NP headed by the relational noun mano, whereby the relationship of inalienable possession is established. As explained before, this DP needs case, and undergoes XP-movement to the specifier of the functional projection embedding the whole possessum DP, in this case PP_HAVE, which merges as the root’s complement. Because the preposition assigns structural oblique case to the DP containing the relational noun, it must be v (or Voice) that values the possessor’s uninterpretable case feature, assigning it accusative case; notice that if the possessor were replaced by a clitic pronoun, it would surface as an accusative one in non-leísta dialects of Spanish: María [lo_ACC]_i golpeó en [la mano]_i (‘María hit him on the arm’).

This analysis may seem rather unconventional for it assumes that verbs like golpear can take a PP as an argument, when they normally take DPs, and because I am proposing that possessor raising can occur outside of a PP (when the possessor surfaces as a clitic), which tend to be analyzed as adjuncts and therefore subjected to island constraints. However, Landau’s (1999) analysis of possessor raising in Hebrew argues that locative, source and instrumental PPs are argumental in that language, since they allow possessor extraction, whereas typical adjuncts expressing cause, purpose, opposition, etc. do not. Furthermore, Sánchez López (2007) elaborates an analysis of sentences like (32) along the same lines, where possessor and possessee are related inside a locative PP, whose head is the preposition en. The possessor sits in its specifier, where it receives accusative case from the verb, while the body-part DP occupies its complement position, where it is assigned oblique case by the preposition; as in (34), the locative PP merges as the verb’s complement. What differentiates her proposal from mine is the fact that Sánchez López assumes that the external possessor does not originate inside the possessee but in the specifier of the PP; in other words, her analysis does not involve possessor raising.

An anonymous reviewer brings up the striking similarities between the sentence in (32) and its English counterpart (Maria hit Manuel in the hand) – a rare instance of external possession in this language –, and wonders whether the analysis in (34) could be applied to English. At this point, I do not see why this would not be the case since, in all respects, the structural properties of the English construction are identical to those of the Spanish one, including the fact that body-part nouns in these configurations allow a token-reading (35a), which is impossible in other structures containing this type of nominals (35c) (Vergnaud and Zubizarreta 1992).

The advantage of the proposal outlined here is that it can structurally account for the inalienable possession effects (and lack thereof) arising between relational nouns and dative arguments in Spanish, as well as in other related constructions involving external possession. It seems sensible to assume that the inalienable possession construal arises inside the DP containing the relational noun, since it is also available when the latter surfaces internally possessed via a possessive determiner. Additionally, this analysis is able to account for the fact that possessor datives and possessive determiners are in complementary distribution.