Abstract
This paper addresses the notion of close apposition. It begins by analyzing the quintessential appositive construction the poet Burns, arguing that such strings do not have a stable internal structure. This conclusion can be applied to other so-called close appositions, such as we boys, but not to similar-looking structures like my friend Burns or a friend John in linguistics. Most close appositions are characterized by a strong degree of conventionalization and fixity, and remain relatively fossilized schemas, a fact that contributes to their recognizability as specific constructions. However, not all these seemingly binominal constructions have exactly the same constituency, but are rather relatively distinct symbols with distinct functions. The focus of this paper is thus to show how so-called close appositions do not constitute a fixed form. It will also be suggested—although due to space limitations not fully pursued—that both their similarities and the “taxonomic” and “inheritance” ties they share motivate an analysis of close apposition in general as a constructional “network,” understood in a way which is compatible with various forms of Construction Grammar.
Introduction: Close and Loose Apposition
Linguists have generally referred to the concept of apposition when dealing with structures characterized by intonational detachment, such as in (1). 1
(1) The writer, Tom Lebowski, insisted on prizes being awarded by categories.
Such “loose” structures fall out of the scope of the present paper (for a review, see, e.g., Matthews 1981; Meyer 1992; Acuña-Fariña 1999, 2006; Heringa 2011), where the focus will be on “close,” “restrictive,” or “integrated” constructions like (2) to (10), also widely assumed to be appositions in the literature.
(2) Sister Livy (Poutsma 1904:195)
(3) My friend Smith (Curme 1947:59)
(4) We girls who Bill likes (Delorme & Dougherty 1972:12)
(5) The verb “use” (Huddleston & Pullum 2002:447)
(6) This bloke Mark (Keizer 2007:29)
(7) Holmes the meddler (Keizer 2005:458)
(8) The city of New York (Meyer 1989:148)
(9) The idea that cats can fly (Meyer 1989:148)
(10) Down there on the ground (Taylor 2002:236)
In fact, since reasons of space make a consideration of all these subtypes impossible, I will be concerned with the major subtypes recognized in most of the literature on the topic. These include the structures in (11) to (16).
(11) The poet X.
(12) X the poet.
(13) My friend X.
(14) A friend X.
(15) This bloke X.
(16) We Xs.
Meyer (1992) and Keizer (2005) provide excellent corpus-based studies of these constructions. The basic issues addressed here have to do with: whether (11) to (16) all have the same constituent structure and the same basic function; whether they all have one or two heads; and whether there can be more than one referring nominal expression inside them. The answer to the first and third of these questions is no. The answer to the second question is that the internal constituency of all these constructions is more varied than has often been recognized in the literature; some of these structures have a clear head, but others do not. This does not mean that some close appositions have two heads, but rather that their internal constituency is not fully elaborated. It will be argued that no one-size-fits-all approach to the grammar of the noun phrase can capture the complexity of the general category of close apposition (henceforth CA), since it is characterized by both a unique mixture of idiosyncratic conventionalization and large pockets of idiomaticity.
The structure of this work is as follows. In the next section, I briefly discuss the structure that traditional grammarians have usually taken as representative of CA at large, the the poet Burns type. This is the type that has presented the greatest challenge in the past, as it seems to have motivated the view of apposition as a peculiar non-coordinative nominal structure which has two heads instead of one. In the subsequent sections, the bulk of this contribution, the main subtypes of CA are briefly analyzed. I show that most of these seemingly bipartite or binominal constructions are characterized by a strong degree of conventionalization and fixity, and that not all have exactly the same constituency. The final section contains a few concluding remarks. It must be noted that although a full constructionist account of CA is not possible here, this work assumes basic ideas of constructionist models (see, e.g., Fillmore, Kay & O’Connor 1988; Culicover 1999; Kay & Fillmore 1999; Goldberg 1995, 2006; Goldberg & Jackendoff 2004; Jackendoff 2008). The next section clarifies some of these ideas.
Basic Construction Grammar Tenets
A foundational principle of constructionist models of grammar is that constructions are real, not merely the epiphenomenal result of the application of preexisting rules. By “real” is meant the idea that once the mind generates, say, a ditransitive structure a thousand times (e.g., John gave Mary the candy), then it forms a “template” or “schema” of that kind of structure based on the occurrence of thousands of similar examples in actual usage (so it will not have to generate or recognize the same structure ex novo over and over again). Constructionist models are thus usage-based in assuming that whatever regularities emerge in representation (such as a ditransitive schema) must be anchored in the actual experience of language users with particular linguistic stimuli (hence the connection of this kind of approach to grammar with corpus linguistics). Frequency of occurrence is therefore a formative component of the system, leading to “entrenchement.”
Since whatever is entrenched in the mind must come directly from actual experience, every construction encodes the totality of information that co-occurs when using it (phonological, morphological, semantic, pragmatic, etc.). Constructions are therefore form-meaning correspondences, that is, symbols. Such symbols range from the purely contentful (lexical items) to the more schematic (syntactic templates), with idioms and “prefabs” like the let alone construction or the What’s X doing Y? construction (Kay & Fillmore 1999) occupying a place in the middle. Direct commitment to memory via experience/entrenchment allows constructionist models to accommodate in their accounts the very many structures of language whose composite meaning is not the sum of their parts (e.g., idioms): these are simply listed (or stored) as wholes independently of their varying degrees of internal compositionality.
Another common assumption of constructionist accounts is that constructions form networks: a “constructicon.” In the constructicon every construction is a node that relates to the other nodes in the system either because they share features or because they inherit features from a more schematic construct. For instance, the ditransitive John gave Mary the candy and the prepositional structure in John gave the candy to Mary share obvious semantics. Agreement markings such as those frequent in the Romance languages (say, -o for masculine gender and -a for feminine gender) are instances of shared form in that such markings are applied both to sentient beings and objects in a semantically unmotivated way (for instance, chico ‘boy’ is masculine in Spanish and so is banco ‘bank’ and both take the same determiner el ‘the’ because both are formally masculine; but see below). These shared features are “taxonomic links.” By contrast, “Inheritance links” (of various kinds) typically take place between a parent and a daughter construction. For instance, the idiom kick the bucket inherits a transitive schema that is activated by the semantics of kick (which obviously does not exhaust its idiomatic meaning) and the transitive schema in its turn inherits a subject-predicate schema one step up the hierarchical ladder. These links are motivated (i.e., not arbitrary) in that formally similar constructions are also often semantically similar in various ways. Since each construction exists to perform a specific functional job in actual use, no two constructions are completely alike (the “Principle of no Synonymy of Form”; Goldberg 1995:67): thus, active and passive sentences, for instance, are not seen as coming (or being generated) from an underlying common source, but simply as being two very close (and pragmatically motivated) nodes in the network. The functional specificity of each construction guarantees its survival in the grammar. Constructional networks are therefore “ecological” systems in the sense that categories emerge as distinct symbols to the extent that each new functionally motivated shade of meaning develops minimal formal reflexes, guaranteeing recognition inside the larger network. A major objective of this study will be to show how the different types of CAs behave ecologically, understood in this way.
Finally, in many constructional accounts a network architecture for grammar is related to the network structure of the mind/brain. In this context, linguistic links of all kinds are “activation” paths as well, which usage has stored via repeated experience. Activation leads to well-known and pervasive priming effects in language and cognition at large. For instance, it is well-known that at the lexical level polysemous words (e.g., bug) prime all their meanings at once (say, both ‘hidden microphone’ and ‘insect’ for bug) even before context can help rule out the inappropriate ones (Swinney 1979). So too at the syntactic level constructions prime (or activate) other constructions, provided that usage has frozen enough shared links. In what follows I will try to show that the principles briefly outlined in this section (representational existence, syntactic schema, usage-based grammar, entrenchment, symbol, listing, constructicon, inheritance, ecological network, activation, etc.) provide the best possible theoretical framework for capturing the grammar of CA.
“Prototypical” Apposition: The Poet Burns
The constituent structure of a construction such as the poet Burns has never been clear. This has to do with the difficulty in deciding whether the poet Burns patterns more with the poet over there or with the excellent Burns. This is what we need to resolve, but I will argue that resolving it is not feasible. 2 However, before exploring this issue further, it is necessary to clarify what the construction does in discourse. Keizer (2005) offers an analysis of the functions of CAs, observing that the main function of this pattern is an introductory one. 3 Unlike other CAs, the identifying information given in the first nominal does not anchor the name to the previous discourse, but instead provides relevant background information about a newly introduced referent. This means that, from a pragmatic point of view, unlike CAs which provide “descriptionally identifying information” (Keizer 2007:42ff.) that is inferable from the context (as in the pattern my friend Tsonga), in the poet Burns construction the descriptive information is new.
The criteria that one can keep in mind in order to form an analysis of the the poet Burns construction are the following. There is in the first place evidence that points to U1 being the head:
(a) Pluralization: usually speakers prefer to say the painters Van Gogh instead of the painter Van Goghs (or even the painters Van Goghs; except when a contrastive interpretation is intended; Burton-Roberts 1975; Keizer 2007:56ff.).
(b) Agreement: consistent with the pluralization data, formal agreement is established between the verb and the common noun: the painters Van Gogh were so different.
(C) Constituent order: if Burns the poet means the Burns who is a poet and has the theme Burns as head, and a predicative, nonreferential segment the poet as attribute (see below), it seems reasonable to suggest that, on the same thematic grounds, in the construction the poet Burns, poet takes on the role of projecting (i.e., heading) the entire noun phrase, while Burns qualifies it (poet).
(d) Taxonomic ties (Goldberg 1995): this will be discussed more fully below. By considering the meaning of a close analogue of the poet Burns, namely my friend Burns, it soon becomes apparent that we are not talking about *my Burns who is a friend (with U2 as head; compare the poet Burns, where ‘the Burns who is a poet’ is at least a plausible reading). The same can be said of a phrase like this guy Lou (*the Lou that is a/this guy). These proximate “attractors” (Langacker 2006:126-127) have clear U1-as-head constituency; in a constructionist account of grammar, where shared activation paths entail shared representation (and where meaning itself is nothing other than dynamic distributed neural activation), this makes it almost inevitable to extend this kind of constituency to such a similar construction as the poet Burns.
In contrast to the above, there is also evidence suggesting that U2, the proper noun, is the projected constituent:
(a) Phrasal stress: the stress pattern is secondary + primary (Haugen 1953; Keizer 2007:24), the same as in the kitchen table or the large truck.
(b) Selection restrictions: at least in the most natural use of CAs, like the poet Burns, the proper noun seems to impose conditions on the kind of constituents that can co-occur with it, in particular that they be definite and add something to the bare name (Acuña-Fariña 1996). Hence, in normal conditions *the Burns, *A Burns, and *Poets Burns are wrong. Keizer (2007:32ff.) points out that these restrictions on definiteness are not real, in that NPs like an excellent Burns and A Burns came here asking about you, where proper nouns are compatible with indefinite elaborations, are indeed acceptable. 4 However, it is also true that whenever this is the case, other conditions apply. Thus, the name must be recategorized as a common noun, or alternatively the meaning of the construction may change slightly with either an ‘a certain’ interpretation or a contrastive interpretation becoming mandatory. Perhaps the strongest restriction is the one most at issue here: *we saw an excellent tennis player Roger Federer in action yesterday, with the name embedded in a CA, is unlikely without intonational detachment. All this suggests that we are dealing with a related but different construction. Interestingly, when premodifying adjectives occur in the construction, the more clearly the adjective points to the proper noun the more natural is the effect produced: the excellent poet Burns, the great poet Burns, and the famous poet Burns are all acceptable because it is also natural to say the excellent Burns, the great Burns, and the famous.
(c) Elaboration: the first, common noun is severely constrained in its potential for elaboration, that is, in its potential to grow: ?the accused poet Burns, ?/*the poet accused Burns, *the poet of sweet melancholy Burns, *the poet of the lost generation Burns, and *the poet that made us all proud Burns seem increasingly infelicitous if pronounced without intonational boundaries. In this sense, such first units behave like ordinary premodifying nouns (*the cotton of Mali sweater). Notice that in contrast to the limitations suffered by U1, U2 is not too constrained by the construction (although see below); rather, it co-occurs with the determiners and adjectives with which it ordinarily occurs outside close apposition (here the and renowned), as in the Burns that we all know, and the renowned Burns.
The previous summary could be expanded considerably. Yet continuing with the idea that any of the three main views (U1 is the head, U2 is the head, both are heads) might turn out to be the correct one is perhaps a diversion; a simpler possibility is that the head status of this construction is simply not resolved. It has been argued (Acuña-Fariña 2009) that all one can see in these strings is traces of constituency pointing in conflicting directions, and that these traces are inchoate due to a lack of strong functional pressure: in the poet Burns, for instance, whatever the constituency, the reference to the intended individual is never in danger.
Acuña-Fariña (2009) further argues that the construction is a hybrid that emerges from inheritance and taxonomic ties originating in other constructions. This is suggested by the conflicting set of specifications discussed above. It is also suggested by the unique properties that such emergence brings about (Goldberg 1995). First, only nouns denoting occupations and the like are available (?the friend Burns, ?the cousin Burns, ?the man Burns; compare: my friend Burns or we friends and them enemies). Second, there is a “freezing point” that depends on the saturation of the first nominal that makes the Jaguar boss Tom Walkenshaw acceptable, but the Jaguar boss at the time Tom Walkenshaw in need of some structurally demarcating signal that clarifies its constituency (now veering clearly toward a referentially saturated U1), and *the Jaguar boss during all that season Tom Walkenshaw clearly unacceptable without it. These constraints are pertinent in a construction which is made up of two nouns with referential potential occurring right after the definite article the, the quintessential reference activating device. They guarantee that only one referent is projected.
There are further restrictions characteristic of idiomaticity. For instance, *the writer the poet is not acceptable even though there is no a priori obstacle to particularizing generic U1s using more specific U2s in other CAs (my friend Burns can become my friend the poet, and one can also say there in London in a sentence such as I lived there in London). Interestingly, in my friend the poet, competition for headedness is not an issue, since the possessive determiner clarifies which nominal is projected and must occur first: *the poet my friend. Right-branching is problematic too: *the painter Picasso of pain. Thus, not only is the first, common noun incapable of expanding to its right, like ordinary modifying nouns, but even the second, proper noun, which can usually take PP dependents outside this syntactic mold (the Roosevelt of -war fame, I’m the Picasso of pain), seems not to accept such dependents in this configuration. This suggests a strong degree of conventionalization and entrenchment.
Finally, the move from the poet Burns to Burns the poet appears to yield a slightly different construction (because it is associated with meaning change; see next section), but the fact remains that this CA allows such a possibility while most other close structures do not. Thus, the first structure in each pair is possible, while the second is not: that chap Ritchie / *Ritchie that chap; you girls / *girls you; my friend the physicist / *the physicist my friend. In fact, the common belief that apposition involves reversible NPs is no more than a myth, given that only the paradigmatic case under analysis here does so. Note, however, that when the common noun has a plural mark (the biologists Huxley), reversibility is not possible even in this pattern (*Huxley the biologists; see Burton-Roberts 1975:403-404; Keizer 2007:38), yet another sign of constructional rigidity (see also Meyer 1992:130-134).
What is being claimed, then, is that the poet Burns activates a [the + NOMINAL] template whose internal [NOMINAL] part is blurred in representation. 5 The construction blends conflicting features of the [the + Modifier + Proper noun] construction and [the + Head Noun + Modifier] construction, and through its very existence in the representational world of grammar it develops its own competitive profile. Since the phrase starts by constructing “NP by Det” (see Hawkins 2004:32ff.; also below), that is, by a clear NP-introducing signal (the determiner), and since object recognition (one referent for the entire NP) is not compromised, all options (the + poet, the +Burns, the + poet Burns . . .) converge on the same concept; hence, the fine details of a richly articulated componential structure are not needed, the final objective (the activation of one referent, launched by a Det + NOMINAL signal) having already been achieved.
The CA Family
The aim of this section is to present an overview of constructions commonly classed as close appositives, looking at the most typical patterns. A full account of each specific type would require more space than is available here. By concentrating on the most common patterns it will be possible to show both their commonalities and their uniqueness as specific constructions, and also to briefly illustrate their distinct functional roles.
The My Friend Burns / My Friend the Poet Types
The main function of this construction is to provide descriptionally identifying information which allows the hearer to locate the referent of the construction inside his/her knowledge base (Keizer 2007:42ff.). The information provided is directly aimed at preventing a who is s/he? sort of question (Keizer 2005). The first nominal in this type of construction typically represents family terms like cousin, aunt, or brother, which are easy to place in one’s permanent schema of people’s social network of relations. Together with the description provided by the first nominal, the reference of the possessive pronoun itself anchors the reference of the whole phrase unequivocally to the hearer’s knowledge base.
A superficial comparison of two structures like my friend Burns and the poet Burns reveals that one of the traces of constituency that seems evident in the poet Burns (linking the and Burns in order to make sense of a construction that is taken to mean ‘[the Burns] who is a poet’) is not possible for my friend Burns, where clearly there is not even a thread of constituency (any test, symptom, criterion, etc.) that prompts us to understand the phrase as meaning ‘my Burns who is a friend.’ In short, nothing links the determiner and the last noun, and my relates to friend unambiguously.
Does it relate to friend only? Apparently, this is not so clear-cut. Keizer (2007:45ff.), for instance, argues for a structure such as the following for most kinds of CA, including the one with possessive determiners:
Det [[N] Np]]
However, following Burton-Roberts (1975), she wishes to differentiate the features of definiteness and possessiveness. The reason for that is precisely the difference between the poet Burns and my friend Burns in terms of grounding/referentiality. Keizer’s (2007) analysis seeks to reach an intermediate position between those of Burton-Roberts (1975) and Acuña-Fariña (1996). Based on the restrictions imposed by the proper noun, Burton-Roberts (1975) analyzed the poet Burns as a [Mod + Head] structure and, rather surprisingly, retained this analysis for the my friend Burns structure, despite the fact that the possessive does not point to the proper noun in the latter. In order to accomplish that, he made a distinction between “the two features of the determiner, its definiteness on the one hand and its possessiveness on the other” (Keizer 2007: 45-46). In the transformational model used by Burton-Roberts (1975), this could easily be done by invoking different derivational histories and the same surface outcome. Acuña-Fariña’s (1996) account, by contrast, does not rest on derivations, and thus simply assumes that since the possessive does not have scope over both nouns and points only to the common noun, this is the head of the structure. So, on this view: (i) my friend Burns is simply different from the poet Burns; (ii) Burns is some kind of modifier in that configuration (see below). Keizer (2007) concurs with Acuña-Fariña (1996) in that no recourse is made to hidden layers of structure, and she also recognizes that the possessive feature of the determiner does not affect the last noun and that the head of the structure is the first noun. However, perhaps motivated by a desire to apply the same constituent structure to all CAs, she, like Burton-Roberts (1975), restricts the differences between the poet Burns and my friend Burns to a difference in grounding, and, what is more, one that does not affect syntactic constituency.
Whether definiteness and possessiveness are features of a different nature, in the way that Keizer (2007) suggests, is a moot point. So is the idea that my friend Burns is “undetermined for definiteness.” Keizer (2007) wishes to avoid an NP + NP configuration partly because in her view in such a configuration if “a possessive pronoun is a definite determiner, and if in my friend the poet, my friend is a referential NP, then the referent must be identifiable on the basis of the information given in this NP only” (Keizer 2007:26). Since, in both my friend Burns and my friend the poet, both Burns and the poet are necessary to identify the intended referent, Keizer (2007) sees herself obliged to reject the NP + NP structure. This has the advantage of placing all these CAs in an habitual NP mold (a Det + NOMINAL structure, instead of an NP + NP one). Hence the possessive has scope over the first noun only, and the combination of this “semantic” feature with a post-modifying proper noun makes it possible to identify the referent of the entire phrase, even in the absence of “true” or “direct” definiteness.
In the present study, I do not assume a split between definiteness and possessiveness. However, leaving that aside, it is in fact not clear that in an NP + NP configuration where the second NP is some sort of modifier the referent of the entire phrase must be identifiable only through the first NP. This seems to be the case in loose apposition. In the literature there is usually a more solid objection against the double NP analysis of CA, voiced, once again, by Burton-Roberts (1975:396ff.) and it involves a comparison of noun compounds like fighter-bomber and CA. According to Burton-Roberts (1975), truly coreferential NPs in CA, such as *man homo sapiens or *the Bard Shakespeare, are ungrammatical because if “such NPs were acceptable, we should be faced with the odd situation in which, each ‘constituent’ being coreferential with the other, the supposed superordinate entity would be coreferential with each and every one of its constituents, which inherently precludes it from being the superordinate entity and as such it would be totally redundant” (Burton-Roberts 1975:396-397; see also Lekakou & Szendröi 2007). In short, on this view, if both NPs are referents, CA will not be possible when they refer to the same referent.
However, reference to the same individual is not incompatible with a formal construal of the same individual from two parallel, complementary angles. In fact, referential coherence is usually constructed in that way. It is therefore useful to recall the distinction between identical reference and identical denotation. The latter would be sensitive to Burton-Roberts’s (1975) constraints on “NP synthesisation,” and would violate Goldberg’s (1995) “Principle of no Synonymy of Grammatical Forms,” whereas the former would not. The actual restrictions on usage (*the Bard Shakespeare, or *my Bard Shakespeare) thus follow not from the coexistence of two NPs with identical reference, but rather from the existence of two referential NPs. All these obstacles disappear if we abandon both the traditional theory (two equal, referential heads) and Keizer’s (2007) view that in an NP + NP construction, if the head is inside the first NP, this must identify a referent by itself. The most parsimonious account of a structure like my friend the poet, where my refers to friend and not to poet, is simply to assume friend as the referential determinant of the whole phrase, that is, the head. If we then assume that the poet is not a referent in that configuration, it then follows that it must be a modifying phrase. The idea of a modifying NP with a definite determiner without referential powers is not new. In Saul was the prey of obscure emotions and Gareth, the heir to a huge estate, was not a happy soul, such NPs are unproblematic (the first example having a copula that mediates between the two NPs, the second not even that). The claim here, then, is simply that they may occur inside “integrated NPs” (Huddleston & Pullum 2002:1350ff.). Meyer (1989:157) insists that in a CA such as Larry Larkin the cartoonist (see below) there is attribution in that “Larry Larkin is a member of the class of cartoonists.” And, with constructions like my friend the poet in mind, Quirk et al. (1985:1313) point out that “attribution involves predication rather than equivalence.”
A further problem with an analysis like Keizer’s (2007) is that it forces one to recognize some rather strange intermediate nominal projections. Thus, my friend Burns becomes my + friend Burns, although this does not seem to have the same motivation as sports car in my + sports car, nor, importantly, that of architect Norman Foster in the architect Norman Foster (architect Norman Foster can even stand alone as a title). That is, both distributionally and semantically friend Burns makes an odd unit. The situation becomes yet more problematical if we contemplate my + friend the poet. One does not easily conceive of ?a + friend the poet or ?this + friend the poet, or ?two + friends the poet(s). However, one can of course imagine a + sports car, this + sports car, the + sports car, just as easily as a + white table, this + white table, and the + white table. It is true that premodifiers are easier to capture as inherent attributes of the head, but even PPs seem more acceptable in intermediate nominal projections; thus, in a + man in a great suit, this + man in a great suit, or the + man in a great suit, man in a great suit can even be coordinated, yielding, for example, there was a + man [in a great suit and looking very good] (compare ?I saw his [friend [Burns and looking very good]] and ?I saw his [friend [Burns and the poet]]). Overall, then, it seems better to accept that my friend Burns and my friend the poet are somewhat different from the prototypical CA the poet Burns. Further support for this difference can be found in the fact that the possessive determiner is, as already mentioned, completely fixed in order: *Burns my friend / *the poet my friend (compare the poet Burns / Burns the poet). Again, this suggests a clear U1 center with the speaker-oriented possessive determiner claiming preferential (thematic) status by virtue of its superior deictic grounding. Conversely, the reversibility of the poet Burns shows a trace of equality of the two nouns, as noted out above. 6
The Burns the poet Type
There are clear chains of resemblance, or taxonomic links (Goldberg 1995), between a construction like Burns the poet and the paradigmatic appositive type the poet Burns. The very same types of words are involved (definite article, proper noun, a common noun designating an occupation) and a broadly similar meaning is conveyed (in both, Burns is a poet). Also, whatever the omission test may reveal about syntactic structure, the test applies equally to both: one can split a sentence like Tommy the plotter will fool everyone into Tommy will fool everyone and the plotter will fool everyone. Needless to say, this is no guarantee that a phrase such as the plotter will be the same (in terms of grounding/reference, etc.) in the two configurations, just as the black in the black doors are better and the black are better are not the same.
Viewed more closely, however, there are also obvious differences between the two patterns. In the first place, the connection between the definite article and the proper noun that we argued was visible in the poet Burns is absent in Burns the poet, where the relates to poet only. In the second place, the range of meanings of the construction is sometimes different, either subtly or overtly. In her account of the functional uses of CAs, Keizer (2005:456ff.) attributes to the construction the same descriptionally identifying use that the poet Burns has (among others). In both cases, she adds, the function of the CA is to help the reader or hearer identify the name by linking it to their existing knowledge base. However, a more fine-grained analysis of the examples she adduces reveals interesting nuances. (17) is an example of the poet Burns type and (18) and (19) illustrate the Burns the poet type.
(17) This is really a big home win for Jaguar and indeed for Silverstone because the Jaguar boss Tom Walkenshaw is of course the uh managing director of Silverstone Circuits. (Keizer 2005:454)
(18) Uh, he said that uh his uh feeling was that uh Mr Kook and uh the other guy who was there who uh was Mr Hunt the Finance Director were uh talking about two different things [. . .] (Keizer 2005:457)
(19) [. . .] “I know you, you scoundrel! I have heard of you before. You are Holmes the meddler”. My friend smiled. “Holmes the busybody” His smile broadened. “Holmes the Scotland Yard Jack-in-office!” (Keizer 2005:458)
Example (18) appears to be either a loose apposition, which would therefore require some kind of demarcating use of tonicity (not necessarily pauses) after U1 (Mr Hunt), or a true hybrid falling between CA and loose apposition. Example (19) is a better contrast with (17), and its meaning is noticeably different. In typical journalistic texts, where constructions like (17) occur frequently, the characteristics of the common noun, especially the denotation of one’s professional occupation in life, are used to help identify the individual referred to by name, since, in a community of “disparates” (Meyer 1992:92ff.) with little shared knowledge, the name alone does not usually suffice. By contrast, in (19), as Keizer (2005:458) herself points out, “it may be argued that [. . .] the second element may lose its discourse connection to become more of a general characterisation, serving as a nickname or as a part of the proper noun: Edward the Confessor, William the Conqueror, or Charles the Simple.” In fact, it has been suggested that these CAs ought to be seen as names with no internal constituent structure at all (Acuña-Fariña 1996). The capitalization of the common noun actually suggests that it has acquired a naming or appellative function, and in fact this was the norm until the end of the Middle Ages, when individuals were known by their professions and places of origin (Jack the Smith, Sam the Butcher; see Kjellmer 1986:445-449). Keizer (2005:460-461) observes that the “introductory use” that she attributes to CAs like the poet Burns is not possible for the pattern in (19), as (20)-(21) illustrate. In both of these a community of disparates is invoked, which calls for the use of a construction that specializes in locating a referent in a social context, not a construction that presupposes familiarity with that referent via the use of a nickname.
(20) [on a plaque] */?Burns the poet lived here from 1786 to 1788.
(21) */?An enquiry’s begun into the sinking of a trawler by a nuclear sub off the Scottish coast and Roald Dahl the author has died at the age of seventy-four.
An important characteristic of the Burns the poet pattern seems to be the ability to extend the range of common nouns that can appear in CA. Notice that, in (19) above, the meddler is not a profession in Holmes the meddler, but the construction almost makes it sound as if it were. In this sense, almost any predicative NP can be used to denote the fact that, of all possible predications one can attribute to a given individual, the one chosen defines him or her perfectly. Note that this is not a question of identification of the right individual against one’s knowledge base, but rather one of the best possible definition of that individual. Holmes the busybody and, especially, Holmes the Scotland Yard Jack-in-office, are good examples of this, and it is not easy to imagine one saying the meddler Holmes announced he intends to sue (?), or the busybody Larry could not refrain from taking part (?), using another type of CA. One can look on this extended “type” frequency of the construction as an “emergent” property that developed from the earlier practice of identifying people by profession, one of the most basic constructional meanings of most prototypical CAs. In saying Holmes the meddler, the construction makes it possible to suggest that Holmes’ occupation in life is to meddle. The property is emergent because the other CA types do not have it. It is therefore specific to the Burns the poet type, and must be listed.
Another function of this pattern seems to be to express a contrast between two or more facets of the same individual, as when one says I mean Eve the writer, not Eve the mother. Even though special pronunciation might make this contrastive use possible with, for instance, the the poet Burns type, that function is much less amenable to it: I mean the WRITER Eve, not the MOTHER Eve.
Now consider the structure of this pattern. In the first place, in as much as the construction Holmes the meddler is used to refer to an individual named Holmes and not to somebody we know from context previously identified as the meddler, it makes sense to take the proper noun as head. We can recall that in the case of the construction with the inverted word order (the poet Burns) the head status of the phrase was unspecified (in the sense that poet Burns was all we could identify as a structural center). Thus, the different thematic alignment of both constructions is in synchrony with the different meanings we have identified for each. This form-meaning correspondence is to be expected given constructionist premises: minimal variations, including simple word order rearrangements even inside the NP, cause meaning change (Goldberg 1995:3). In the second place, it follows that the meddler is a dependent phrase. Given that, if we leave out the proper noun, the meddler will have to acquire referential specificity to pick a referent (it is certainly not the same to say Holmes the meddler will be happy with that and the meddler will be happy with that), it also follows that the meddler is not a referent but a predicative phrase inside the CA (Meyer 1989:157). In this sense, the structure [NP + NP] seems evident and is vaguely (and appropriately) reminiscent of the bipartite air of the notion of apposition. However, just as I have said that in the poet Burns there were inchoate traces of constituency leading in (at least two) ways that fail to culminate in a specific nominal structure, so I suggest here that this rather clear asymmetric but nevertheless bipartite structure is concurrent with the fact that the construction as a whole does not have a rich internal structure, and that for this reason it is so easy to regard the whole combination as a composite name or a name plus nickname (see above and Keizer 2005).
While one can integrate other CAs into existing nominal paradigms, this is impossible for this type. Thus, one can imagine the red blouse, the bottom chair, the great Burns, and (as an open possibility) the poet Burns as all exhibiting a [Det + Premodifier + Head] schema. Also, we can conceive of his friend the architect, his friend in the choir, his friend here, and his friend Burns to constitute another schema. However, one runs out of schemas for Burns the poet: Burns the poet, *Burns poet, *Burns in the firm, *Burns there. A general, one-size-fits-all X-bar kind of approach to these modified proper nouns is difficult to imagine. The advantage of a constructionist approach to grammar is that one is not obliged to cancel out the pros and cons of a given analysis. That is, in a representational map of the constituent features of a given construction, the features of the construction which point in a given direction coexist with those features of the same construction that may point in a different direction. If we think of these structural biases as “activation paths” (Langacker 2006:141), then the meaning, and the structure, of the construction is quite simply the entire collection of activation paths. Hence, just as in the poet Burns neither noun is as profiled as the entire nominal constituent, so in Burns the poet the otherwise clear formal internal structure of the construction is less profiled—evident—than the composite name meaning that it evokes. 7
The Type A Friend John [who’s in Linguistics]
This pattern, which has a distinctly informal ring, is interesting in that it is the only kind of indefinite CA. In fact, Keizer (2007:32ff.) uses it to reject the standard idea that CA must involve definite grounding mechanisms. Together with the indefinite article a, the pattern also occurs with indefinite this and that (Givón 1993:204ff.). Examples (22) to (24) below come from Keizer’s own corpus (Keizer 2005:448, 461).
(22) I have a friend John who’s in linguistics with me.
(23) He also has to put up with a soppy elder brother Robert who is forever mooning about some girl or other and a sister Ethel who has all the brisk no-nonsense superiority of a true Wodehouse gel.
(24) Oh, I remember I was talking to this bloke Mark some sort of . . . this really old friend of mine.
(25) We had a lecture by that guy Rene Weis over there.
Two conspicuous features of this pattern are worth bearing in mind. The first is that the constructional schema is fairly constrained. This can be seen in two ways: first, it involves a very circumscribed set of common nouns only, nouns “with very low lexical information” (Burton-Roberts 1975:398), such as friend, child, guy, chap, bloke, and several more. One does not usually say ?I used a computer Dell the other day, ?They liked a word ‘stone’ for the commercial, or ?I was talking to a man Peter. 8 Second, reversibility is not possible: *I have John a friend who’s in linguistics, or *He also has to put up with Robert a soppy elder brother.
The second important feature is that, when selection restrictions affect the construction, they affect the first noun only. In the previous examples, for instance, reversibility is impossible because it forces us to construe strings of the form ?I have John, which seem ungrammatical (without elaborate pragmatic remediation). This suggests a clear U1 center. Yet again we see that the notion of a doubly endocentric structure (two heads) is unrealistic. The fact that the second unit in all these appositions is a proper noun by no means guarantees that such a unit has a referential realization (although it of course has a referential potential which may be realized in other constructions).
On closer inspection, it may be argued that the construction branches off into two subtypes, or indeed that we have two distinct constructions. The examples with indefinite this and that appear to be extremely fixed in that the vast majority seem to involve only “indefinite” nouns such as chap, lad, bloke, fellow, and guy only. The specific constructions this guy X, this bloke X, this chap X have a very particular discourse job, presenting an individual by his/her name together with a specific indication that, unlike in the ordinary use of names, the individual in question is not presumed to be well-known by the addressee (Keizer 2005). For this very specific functional role the indefinite article a is not normally accepted either: in *A guy John, what does he think he’s doing?, a would also sound odd instead of this/that, as in 24 and 25 above). The construction is thus completely frozen. A relevant functional feature of this (sub)pattern is that lack of familiarity may be suggested to be deliberate. Thus, in saying this guy Smith, is he nuts or what?, one may code a subtle wish to distance oneself from the individual in question, rather than unfamiliarity in a strict sense. Given its distinctly informal, modern ring, it is tempting to speculate that the construction is the result of a grammaticalization process whereby the familiarity degree and/or the distance that it codes were routinized using the model provided by the nonrestrictive version (this guy-comma-John-comma → this guy John, with no comma or intonational boundary at all).
From a constructionist perspective, one might be inclined to assume that the restrictive phrase is the result of vertical inheritance links from the nonrestrictive construction (Goldberg 1995:72-74). When inheritance occurs, the new construction usually develops a highly specific profile. Thus, whereas loose appositions are open to all manner of expansion (this guy we met the other day, John; that guy over there at the counter, John), the new pattern is much more circumscribed in that even premodifiers appear to be rare, and, even if possible, it would seem better if the whole NP is used as a topic (a weighty predication in its own right), separated from the sentence by means of intonational marks (this stupid guy John// he is so conceited!; that interesting bloke Mark// had you met him before?). 9 PPs seem to be not allowed easily without intonational detachment inside the NP itself: *this guy in the corner John. And the “distancing” effect is, I believe, unique to the construction. A relevant feature of this pattern is that it is functionally well embedded in a very finite set of discourse formulas which specialize in introducing topics, such as you know this guy Pete Doherty, he is the one that got the promotion; and This bloke Ray Prince, who does he think he is anyway?
The second subtype, exemplified by (22) and (23) above, is also compatible with indefinite this (as noted, the converse does not obtain: replacing that with a in (25) above, for instance, would sound odd). The particular niche that it fills in discourse differentiates it from other, historically more established types of CA in that it is possible with names only, and, in fact, it is usually employed with first names (as in the examples shown). Conversely, one would not normally say ?the poet Peter, but would also require a family name to fully identify an individual by his/her profession (the poet Peter Wright). Interestingly, one can also use another CA type and say my friend Peter, without the family name (Keizer 2005). I analyzed this construction as being constituted by an NP containing the ultimate head of the overall structure (friend) followed by a nonreferential (appellative) proper noun (my friend + Peter). I hypothesize that ecologically the pattern under scrutiny developed from this more classic type and extended its appellative function to the maximum. It even admits quantification: well, I have two friends Peter who are in linguistics (compare: *two Peter friends, *two friend Peters). Hence, besides having a clear U1 center, the construction has erased any vestige of referential power of the proper noun; John does not pick out any referent, it simply denotes a characteristic that—like the color of one’s eyes—may be shared by an indefinite set of individuals. This is a stage reached only by this construction, not its “parent” construction; although technically the phrase *my friends Peter might carry a conceivable meaning (‘all my friends who happen to be named Peter’), it does not seem to be a naturally well-formed phrase.
Interestingly, all the examples that Keizer (2005) uses in her account of indefinite appositions occur in object position and are followed by relative clauses (as with 22 and 23 above). It may be that her spoken corpus contains a greater variety of cases, but in fact this combination is reminiscent of a conversational schema (I VERB a friend X who’s . . .). Beginning a sentence with a string like this one seems less likely: ?A friend John who’s in linguistics can do that. This is probably so because the grounding that the parent construction my friend John provides through the possessive determiner must be provided in these indefinite CAs by elements outside the nominal core itself; I have is actually very similar to a possessive determiner (my) in terms of grounding the reference of a friend John in discourse. The fact that somebody has a first name is not enough for that. The relative clause, or any other extra predication, is also normally needed to visualize the relevance of activating a nominal referent out of the blue.
The Type We Women, You Men
This is another classic type of so-called CA in the literature, whose main function is to locate a deictically defined topic (we, you, them) inside a social group or subgroup (we doctors, them nurses). Together with the appositive analysis, grammarians such as Postal (1966), Burton-Roberts (1975), and Huddleston (1984:233ff.) have also defended the thesis that the first constituent in these structures is some sort of an article (Huddleston & Pullum 2002:422). Postal (1966) in particular provides an elaborate defense of such a view, which rests strongly on the fact that, superficially at least, personal pronouns such as we, you, and us do have “the same privileges of occurrence” (Hockett 1955:99ff.) as ordinary determiners in some configurations. Thus, just as one can say you girls are always saying that, one can also say the/these/those girls are always saying that. Postal (1966:218) also suggests that the pattern cannot be appositive because it freely occurs in environments such as questions and negatives, where “apposition” is prohibited, but he is obviously referring to loose appositives, not to the kinds of structures under analysis in the present study, which have no such restrictions. The fact is that in all the examples in (26) to (28), which are completely ungrammatical, replacing the pronoun by the article the turns them into perfectly acceptable strings.
(26) (a) *I doctor (b) *you doctor
(c) *he doctor (d) *they doctors
(27) *I architect who Sue prefers.
(28) *She architect who Sue prefers. Compare: We/you architects who Sue prefers.
There is, hence, nothing determiner-like in these U1s. It is clear that the initial position in all the constructions above must be followed by whole noun phrases, not just nouns. This is where the notion of apposition becomes useful. If U1 is not a determiner in disguise, then it must be what it appears to be, a pronoun, and since there is no disputing the fact that, after pronouns, plural men, girls, or doctors are NPs (not just nouns; clear singular nouns are prohibited, as in *him doctor, *she architect), then this construction is made up of two NPs that form another integrated NP, that is, an apposition (Spinillo 2003 views these first units as pronouns but not in apposition). As Aarts (2007:125) points out, these forms show a case contrast and, if analyzed as ordinary determinatives, they would be “the only kind of determinative in English that can occur in the objective case.” Delorme and Dougherty (1972) made a strong case for this view, which sits well with the traditional idea of apposition (see, for instance, Quirk et al. 1985:352). In Delorme and Dougherty’s (1972) opinion, the relationship between examples like you, girls, and you girls “provide[s] evidence that stress, intonation, pauses, etc. are not necessarily revealing in specifying underlying constituent structure or semantic interpretation” (Delorme & Dougherty 1972:11). The problem with such an idea is that the greater part of the paradigm in (29) proves that it is ill-founded (Acuña-Fariña 1996).
(29) (a) *I the girl (g) we the girls
(b) *you the girl (h) you the girls 10
(c) *she the girl (i) *they the girls
(d) *I a girl (j) we girls
(e) *you a girl (k) you girls
(f) *she a girl (l) *they girls
There is, indeed, something strange about this paradigm. In principle, one can see why the singular forms are wrong: if the U2s are used to further specify who is meant by the U1s, then the speaker and the hearer are already given and therefore in need of no specification, which makes the “apposition” inessential. However, there seems to be no strong reason why the third person singular cannot be further specified, since one can distinguish among several third parties in a given context. In addition, it is not easy to explain why the third person plural is also wrong, especially since in the objective case we do find such forms in certain varieties of informal spoken English (e.g., I don’t like them doctors). 11
Inserting intonational boundaries between U1 and U2 salvages most, if not all, of these forms (that is, transforming them into loose appositions; see Huddleston & Pullum 2002:374). Of course, we can always stipulate ways (rules) out of the problem, but the fact remains that these constructions only admit first and second person plural personal pronouns and the accusative third person plural, and that that extreme form of choosiness is incompatible with both the determiner view and the CA view, or with any other rule system we might be inclined to contemplate. The construction must be learned and stored—that is, listed—as such. What is to be learned, though, does not amount to much: four fixed personal pronouns (we, you, us, them) and a fixed set of nouns (names of professions and a few generic nouns like men, women, boys, and girls). Within those limited specifications, productivity is not a problem (we boys / girls / doctors / engineers / nurses / runners, etc). The fact that the indexical power of the pronoun is superior to that of the noun in U2 position suggests that this pronoun is the head; after all, in the context of an office with both men and women, saying we men are lazy does not entail that men in general are lazy but entails that part of we is necessarily lazy, at least the speaker. However, Huddleston and Pullum (2002:374) take the opposite view and contend that we, you, and us are “personal determinatives [. . .] exactly parallel to other definite determiners such as demonstratives and the definite article in, for example, permitting the universal quantifier as a predeterminer.” Thus, one can truly say all we supporters of a federal Europe, but not *all we will win the argument. Whether this is enough to confirm the determiner thesis is a moot point, especially given the restrictions shown in (26) to (28).
In a constructionist framework, the determiner thesis and the appositive thesis can simply be seen as reflecting simultaneous features of the same construction, with neither of the two managing to coerce it into a pristine necessary-and-sufficient kind of superordinate form of an X-bar type. Whatever the internal structure, however, the recognizability of the whole construction is never in great danger. Its neat formal specifications make for easy categorization, and categorization is all that is needed to achieve the whole package, which includes the meaning (that is, the final destination). In other words, once again their internal structure need not be resolved. In that sense, the construction’s rigid formal make-up makes it similar to other “syntactic nuts” (Culicover 1999) like the let alone construction (Fillmore, Kay & O’Connor 1988), which have a peculiar syntax, usually a mixture of closed morphology and the possibility of open variables which may be filled productively (Jackendoff 2008:15). This grants them a private space inside the CA network of constructions.
Summary: A Sketched Network
In the preceding pages I have claimed that the pattern the poet Burns shows an underspecified nominal node with no clear center. I have also made the point that the constituency of constructions such as Burns the poet is clear but shallow in that the entire phrase projects a name at the top which obscures the internal structure of the construction and highlights the top itself. Likewise, in the we girls pattern, the idea of a rich internal constituency is incompatible with the insurmountable difficulties in deciding which N is the projected head. Even though the focus of this paper has thus been to show that the typical homogeneous analysis of all CAs in the literature is unwarranted, the “category” CA still exists because most of these subconstructions do share important properties.
The first is that they are all semantically restrictive (my friend Burns, for instance, is not just any friend). The second is that most of them are short, fixed NP formulas which require little recognition and processing time. The third and probably most important characteristic is that, adopting Hawkins’s (2004:32ff.) terminology somewhat loosely, they all “construct NP by Det.” More narrowly, they start with a classic determiner which projects NP at the top of the structure, thus facilitating the constituent recognition domain) by providing clear early immediate constituency: a Det constituent is almost invariably followed by a nominal phrase. Coupled with the form, there is no doubt that, regardless of often fuzzy referential constituency within, all these constructions synthesize and project one unambiguous referent at the top (not two or more). This is a general requirement of the structure of nominal phrases, not just of apposition. Clearly, the overall structure of all these strings is “NP,” and the key is the initial instruction provided by the Det constituent. Since “NP by Det” is open to no ambiguity, a lack of rich internal elaboration is no problem at all. In line with these observations, it makes sense that the grammar of English does not allow such potential referential indeterminacies as *the friend the poet.
In addition to the overarching N + N schema, the semantic restrictiveness, the reduced size, and the clear “NP by Det” projection they all share, what in fact gives the category of CA more conceptual, formal, and even ontological reality is the massive interconnectivity they all share, that is, the existence of numerous taxonomic ties (Goldberg 1995) that connect all the constructions together and define a region in grammatical space through sheer “tie density.” The region actually emerges as a result of taxonomic tie density (Langacker 2006:120, 141), since the same density is not apparent when one compares any of the individual constructions with other constructions outside the present network. Thus, for instance, the constructions the poet Burns, Burns the poet, and we poets share the fact that they all express occupations. If instead of my friend Burns we choose to say my friend the poet, then the possessive construction also codes occupations. The constructions my friend Burns, Burns the poet, and a/his guy Ritchie all contain proper nouns. The constructions the poet Burns and Burns the poet are reversible, and thus practically identical in formal terms. The constructions my friend Burns and a friend Burns who . . . very often share that very same U1. The construction we poets is linked to we the poets and this in turn to Burns the poet. Finally, for present purposes, the constructions my friend Burns, Burns the poet, and a/this friend Burns all share something which may very well stand for the prototypical schematic structure of CA: a [NP + NP] schema where the first NP is the referential center and profile determinant of the whole phrase. It is not that the constructions the poet Burns and we poets cannot be covered by this schematic form, but rather that such a schema is not so well rooted in them. The [NP + NP] template preserves the binominal semblance that is reminiscent of the venerable concept of apposition, but it renounces the idea that the two NPs are truly equifunctional (contra Lekakou & Szendröi 2007). Viewing meaning as a collection of activation paths helps explain the connection between one CA and the rest; the differences between the various constructions are as real as their commonalities, which are to be seen simply as shared (parts of) activation paths. The only limit to a diffuse internal structure or an unprofiled one is that the overall meaning of the construction be reached safely anyway.
The grammar of English makes sure that, despite poor internal elaboration, and even despite conflicting attracting schemas exerting their influence on the same form (as in the poet Burns construction and the we girls construction), all the different kinds of CAs remain unique patterns with unique functional jobs. The rigidity of each pattern is correspondingly unique. Thus, the pattern the poet Burns can only be used with U1s that designate professions and the like, and typically involves full names or family names as U2s. It is reversible, but reversibility creates a different functional range of application (no strong discourse introductory potential) and an extended type frequency (almost any activity can be conceived as a “job”; consider Holmes the meddler). The pattern my friend Burns may be used with some professions, but even if this is the case, a CA does not sound easily acceptable, or at least as acceptable as a loose apposition (compare ?my doctor Peter Thomson / my doctor, Peter Thomson). Strings such as ?my poet Joe Stevenson, ?your writer Tom Harald, and ?his philosopher Jane Barkin are odd (compare my favourite poet, Joe Stevenson). The pattern only applies naturally to family terms and a few generic nouns such as friend, colleague, and flat-mate. It must have a possessive determiner (*the friend Sue) and is therefore nonreversible (with speaker deixis imposing preferential thematic prominence: *Sue my friend).
The (sub)pattern a friend John who . . . might appear to be just the indefinite counterpart of my friend John, but that impression is deceptive. In the first place, although friend can exist in the two patterns, the range of nouns allowed by each construction is different: a guy Louis who . . . / *my guy Louis; a chap Carlos-something / *his chap Carlos. In the second place, the a friend John who . . . pattern occurs naturally in informal English with two or three fixed introductory formulas: I know a guy X who . . . , I have a friend X who. . . . The pattern is expandable by definition, since the extra predication that must attach to the proper noun is the one giving it discourse grounding. The (sub)type this guy David Green involves a very small number of nouns in U1 position: chap, lad, bloke, fellow, and guy are perhaps the most frequent. It therefore has an extremely reduced type frequency, which contributes to its frozen form. These last two (sub)types present names with no referential power at all, but only this-marked phrases achieve a distancing effect, and only they appear embedded in larger topic-introductory constructional schemas (you know this/*a chap Billy Ray, well apparently he’s the one in charge of acquisitions now). Finally, the we girls construction only admits four N1s: we, you, us, and them. It is therefore entirely frozen, with remarkably low type frequency. Except for the the poet Burns and Burns the poet types, none of the other constructions are reversible, a clear index of absence of true functional equivalence.
There is thus a delicate balance between a high degree of interrelatedness and individual existence in a shared ecological niche. Given the minimal formal variations at stake (e.g., the vs. my in my/the lawyer Martin Cook), the grammar of CAs thus shows us how formal minimality and actual, discrete symbolizations not only interact, but do so in various ways. With both function and representation in mind, all that matters is that the formal minimality be associated with a conventional interpretation. It also matters that one recognizes that the grammar of the noun phrase cannot all be captured within an all-encompassing, fixed, one-size-fits-all type of X-bar schema. In sum, CA helps us realize that there are “syntactic nuts” inside the structure of the NP, and that these self-organize and relate to their more standard neighbors like any other nuts in the system.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The present study was supported by a research grant from the Spanish Ministry of Science and Technology (grant number PSI2012-32834) and another research grant from the Autonomous Government of Galicia (GRC2015/006).