Grammatical Words and Spreading of Contexts: Evidence from the Spanish Preposition a

Abstract

The paper shows that when grammatical words are involved, context is then the unit of language change. Certain changes consist in an active spreading of a form to new contexts, without changing the category or grammatical status of the form; in these cases, context must be considered the unit of language change. The empirical evidence is the diachrony of the Spanish preposition a ‘to’. Throughout history, this preposition pervasively extended to new and different contexts, but the form a never changed, remaining a grammatical preposition with a basic meaning of ‘directive telicity towards a goal’ (goal maybe locative, temporal, transitivity, finality, discursive, etc.). The paper labels this kind of change as ‘context construction’, and considers it an analogical extension induced by context. Finally, to test whether the diachrony of a is grammaticalization or not, the paper reviews fourteen related theoretical concepts, checking them against the diachronic evidence of the preposition a.

1. Introduction. The Theoretical Problem1

The history of the Spanish preposition a ‘to’ constitutes both a striking case in general syntactic diachrony and a challenge for several theoretical frameworks of language change. It especially presents a challenge for Grammaticalization, because the preposition underwent many syntactic changes without modifying its basic, general meaning throughout the history of Spanish, and without modifying its category status of grammatical preposition. The changes, as a whole, may be characterized as a very active spread of the preposition into new and different distributions and contexts, but the form a never changed, always being a grammatical preposition, and always having a basic meaning of ‘directive telicity towards a goal’, the goal may be locative, temporal, final, inmediate or mediate goal of transitivity, modal goal, discourse goal, the hearer as a goal, etc. It is a case of change of contexts and not a case of change of the status of the form. There is no polysemy of a, there is a great context polifuncionality, and there is also the creation of new constructions with a. I will label this kind of change as a ‘pattern of pervasive spreading of contexts’.

Each new context is an extension of the previous one, via metonymic or metaphorical elaborations, and each new context has gradually moved away from the locative context encoded by the classical Latin preposition ad, which is the etymon of the Spanish a. The progress as regards to the context is, then, the expected one in diachronic syntax. However, the behavior of the preposition is not the expected one in diachronic syntax, because it maintained its grammatical status, and always coded ‘directive telicity towards a goal’; the goal being of different kinds, as I said above, locative, temporal, purposive, modal, imperative, discoursive, etc. (cf. §2 below).

Table 1. Pervasive extension of preposition a ‘to’ into new contexts ^†.

a	→
LOC	LOC	LOC	LOC	LOC	LOC	LOC	LOC	LOC	LOC
	IO	IO	IO	IO	IO	IO	IO	IO	IO
		HDO	HDO	HDO	HDO	HDO	HDO	HDO	HDO
			TEMP	TEMP	TEMP	TEMP	TEMP	TEMP	TEMP
				FUT	FUT	FUT	FUT	FUT	FUT
					PURP	PURP	PURP	PURP	PURP
						IDO	IDO	IDO	IDO
							MOD	MOD	MOD
								IMP	IMP
									DISC

^† LOC = locative; IO = indirect object; HDO = human direct object; TEMP = circumstantial temporal complements; FUT = temporal future in the periphrasis ir a + infinitive; PURP = purpose; IDO = inanimate direct object; MOD = modal contexts; IMP = imperative; DISC = discursive contexts.

(apud Company and Flores 2014, p. 1328) shows schematically the pervasive spreading of contexts undergone by the Spanish preposition a in the history of Spanish. The growing line a → , with a bigger a in each step, does not mean polysemy of a nor changes in the preposition itself, it only means the gradual and constant diachronic spreading of the preposition to new contexts. The lines in the scheme reflect, to a certain degree, the diachronic stages of the advance of a to new contexts, and also reflect the cumulative or stratified effect (Hopper 1991) produced by any syntactic change. For simplicity, the scheme represents only ten contexts, but there are many more; the historical depth of this change, still in progress, is also very old; it begun, at least, in Late Latin.

Pervasive spreading into new and more varied contexts with no change in the category status of the form or construction involved in those contexts is not a type of change that is typically considered within the research framework of grammaticalization. In fact, the well-known classical definitions of grammaticalization in (1) posit that language change is characterized by a change in the grammatical status of a form or construction (e.g., a change from a lexical to a grammatical form, or a change from a grammatical form to a more grammatical one), and that these developments are driven by changes in contextual use, context being only the medium through which forms or constructions acquire a new grammatical status or a new grammatical category. The definitions in (2) do not mention context or take it into account at all. Probably, context is absent from many definitions because it is obvious that forms need context and actual use for changing; that is, context is taken for granted in grammaticalization, although all works point out the local nature of syntactic change, that is, they recognize that change always takes place within specific contexts (Diewald 2002; Heine 2002). In consequence, specialized theoretical literature does not bear in mind changes such as the one analyzed here.

(1)

a. A process by which a lexical form or construction, in specific pragmatic and morphosyntactic contexts, assumes a grammatical function or by which an already-grammatical form or construction acquires an even more grammatical one (Kuryłowicz [1965] 1975, p. 69).2

b. Grammaticalization is the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions or grammatical items develop new grammatical functions (Traugott 2001).

c. Grammaticalization refers to that part of the study of language change that is concerned with such questions as how lexical items and constructions come in certain linguistic contexts to serve grammatical functions or how grammatical items develop new grammatical functions […] grammaticalization refers most especially to the steps whereby items become more grammatical through time […] the change occurs only in a very local context (Hopper and Traugott 2003, pp. 1–2).

(2)

a. Grammaticalization [...] by definition is a process of language change in which an expression moves away from the lexical pole and toward the grammatical pole (Geurts 2000, p. 781).

b. Lexical or less grammaticalized linguistic expressions are pressed into service for the expression of more grammaticalized functions (Heine 2003, p. 578).

Relating the above definitions to the Spanish preposition a, this word neither moved away from the lexical pole towards the grammatical pole, nor was it ‘pressed into service’ for more grammatical functions, because a has always served grammatical functions. Even specialized literature considers it as an element that only establishes relations between two forms or constructions (Di Tullio 1997, pp. 41–42). Indeed, reference grammars often analyze this preposition as an empty form, lacking any meaning (Alcina and Blecua 1975, sct. 6.2.5). Obviously, the preposition a increased its frequency across the history of Spanish, because it appears in many more contexts, but an increase in frequency is not, by itself, a signal of grammaticalization. At best, heightened frequency is a symptom that there may be certain changes in progress.

The theoretical problem of the definitions of grammaticalization above is twofold: first, how is it that a word, which is already very grammatical, becomes more grammaticalized? Second, is it possible to consider context change as a kind of grammaticalization? In this paper I address these questions by analyzing the role of context in language change. In doing so I hope to establish a dialogue between the different approaches and the types of grammaticalization and the Spanish diachronic data of a while determining the extent to which context is responsible for developing the pattern of a → .

This paper contains three proposals. The first is that context may be, in some cases, the ‘unit of change’, besides being simply a ‘locus of change’. Context is understood here neither in a restricted grammatical sense, nor in the traditional meaning of narrative blocks. Instead, context is defined as a form or construction + a specific distribution where the form or construction enters into + optionally, other syntagmatic and/or prosodic units co-occurring with those forms or constructions. This definition, basically, fits with the ‘construction grammar’ framework analysis (Trousdale 2012; Hoffmann and Trousdale 2013; Traugott and Trousdale 2013). The second proposal of this paper is that it is necessary to leave the word level for some types of changes, and take into account both construction and context in order to fully understand some diachronic syntactic processes. Finally, this paper proposes that forms maintain their basic, general meaning for centuries. Forms enter into new contexts but these contexts are always more or less compatible with their general meaning. This stability guarantees that the progression of language change is neither haphazard nor chaotic.

This paper is organized in four sections besides this Introduction. In Section 2, the diachronic data related to preposition a are presented, focusing on the new contexts and constructions in which the preposition appeared, and classifying the types of change corresponding to the new contexts with a. In Section 3, I summarize the various current research frameworks on grammaticalization, and specifically focus on the conceptual theoretical labels that relate to the Spanish diachronic data of preposition a. That is, in this section I try to establish a dialogue between theory and diachronic data in order to inform the diverse views on grammaticalization. In Section 4, I make a specific theoretical proposal for the kind of change consisting in pervasive spreading of contexts. Brief conclusions are in Section 5.

2. Diachronic Data on Preposition a

The basic meaning ‘directive telicity towards a goal’, which is inherent to preposition a ‘to’, is maintained, as I said before, in all contexts, but the ‘goal’, which formally is the term of the preposition, may be of different types, displaying diverse semantic and grammatical nuances, forming different syntactic constructions which enter in very different contexts.

The goal, exemplified in (3–23), may be—among many other constructions—a locative complement (3), taken literally (3a) or metaphorically (3b), a temporal complement (4), a future time (5), a future time without specifying tense nor mood (6), a final purpose (7), an accomplished trajectory (8), a person which is the final locus of movement (9), a personal direct object (do) (10), an inanimate do (11)—do is always the immediate goal of transitivity―, a recipient indirect object (io) (12), which functions as the remote or ultimate goal of transitivity, an autonomous imperative event (13), a nominal construction with modal meaning (14), a nominal modality in construction with a verb (15), an adjective modality in construction with a verb (16), an adverbial modality with adverbs in -as (17), a future event in subordinate relative clause (18), a member acting as the goal of a comparison (19), an interrogative clause (20), a predicative construction close to a discourse marker (21), a fixed expression, an idiom, inherited from Latin (22), which should lack preposition in learned writing, because the Latin idiom is in ablative case, which internally already has a kind of preposition. The example (22) is a conclusive proof of the structural pervasive spreading of a to new contexts, even the most reluctant ones, such as idioms. And the goal may be also a pragmatic one, an abstract location implying abstract movement (23). There are more types of context coding the goal, but (3)– (23) are enough examples to show the diversity of constructions, of contexts, and the diachronic stability of the preposition a. Only (3a) shows the original etymological context coming from Latin.3 The large and great variety of goals displayed in (3b–23) is the consequence of the diachronic spreading of a to more varied contexts along the history of Spanish.4 In the examples, the verb and/or the noun and the prepositional phrase (pp) are in bold type to facilitate the identification of the construction and the context.

(3)

a. hallé que abia enbiado a España con çiertas personas çinquenta y çinco mill y seysçientos y ochenta y seys pesos de oro (Letter, 1525, Mexico, cordiam)

“I found that [he] had sent to Spain via certain people fifty-five thousand marcs.”

b. me mandó a la mierda; y me dixo que no quería venir (Legal document, 1816, Mexico, cordiam)

“He sent me to hell and he told me he did not want to come”

(4)

que fue el dicho dia que los mataron a la noche (Chronicle, 1598, United States, cordiam)

“that day they killed them in [lit. to] the night”

(5)

la condenada asma va a darme qué hacer (Emilia Pardo Bazán, Un viaje de novios, 1881, Spain, corde)

“this horrible asthma is going to give me problems.”

(6)

y con ello no tendrá pleyto ni gastos, cino que al comer gastará una gallina, al senar un pollo (Huamán Poma, Crónica, 1615, Peru, cordiam)

“doing it in that way, he’ll never have problems nor spending, on the contrary, for lunch [lit. to the eat], he’ll spend a poultry only, and for dinner [lit. to the have supper], a chicken only”

(7)

el portador desta que es Diego Farias va a ese pueblo a un pleito de unos yndios (Letter, 1583, Colombia, cordiam)

“the person who carries this letter is Diego Farias, who goes to that village to a civil action about some Indians”

(8)

Y llega el rey a por su novia y ve a la negra y le dice … (Anónimo, Cuentos populares españoles, ca. 1920, Spain, corde)

“and here is the king coming for [lit. to for] his fiancée and sees the black woman, and then he tells her …”

(9)

Et lleguemosnos a él et gradescamosle su bien fazer (Anonymous, Calila, 1250, corde)

“and we must come to him and thank him for his well done actions.”

(10)

Y llega el rey a por su novia y ve a la negra y le dice … (Anónimo, Cuentos populares españoles, ca. 1920, Spain, corde)

“and here is the king coming for his fiancée and sees the black woman [lit. sees to the black woman], and then he tells her …”

(11)

los ácidos minerales destruyen al barniz que reviste al hueso del diente (Gazeta de México, 1787, apud Company and Flores 2014)

“mineral acids destroy the varnish [lit. to the varnish] of the teeth.”

(12)

Yo no puedo entender por qué dejó v.s. de enviar luego mi recaudo a el maestro Ávila (Santa Teresa, Cartas, 1560–1580, apud Company 2006)

“I cannot understand why you did not send my news to the master Avila”

(13)

¡Niña, a callar! (Alfonso Zurro, Farsas Maravillosas, 1987, España, crea)

“baby, shut up [lit. to shut up]!”

(14)

¡Con lo fácil que era, pensaba, hacer una tortilla a la francesa! (Camilo José Cela, Esas nubes que pasan, 1945, Spain, corde)

“it is very easy, I thought, to make a plain omelette [lit. an omelette to the French way].”

(15)

Unos jóvenes, andan a cangrejos, metidos en un riachuelo saltarín que brilla en la hondonada (Iñaki Linasazoro, La otra Guipúzcoa, 1969, corde)

“Some young men, seeking crabs [lit. walk to crabs], are into the jumping river.”

(16)

el aire olía mal, a cerrado, a podrido, a miseria (Almudena Grandes, Corazón helado, 2007, apud Company and Flores 2014)

“the air smells bad, closed, rotten, misery [lit. to closed, to rotten, to misery].”

(17)

Cada quien toma veneno a sabiendas, de vez en cuando (José María Arguedas, El zorro de arriba y el zorro de abajo, 1969, Peru, corde)

“everyone takes poison knowingly [lit. to knowingly], from time to time.”

(18)

La categorización de los motivos … constituye también un punto a señalar (José Luis Pinillos, Principios de psicología, 1975, España, crea, apud Company and Flores 2014)

“the analysis of causes is a point that must be considered [lit. a point to consider].”

(19)

Y como a listo (astuto diría yo) nadie le gana … (Jenny E. Hayen, Por la calle de los anhelos, 1993, México, crea, apud Company and Flores 2014)

“as to clever, nobody beats him.”

(20)

Por allí por la Alcarria no veis estas cosas, ¿a que no? (Sánchez Ferlosio, El Jarama, 1956, España, corde, apud Company and Flores 2014)

“there, in the Alcarria, you don’t see those things, that’s right? [lit. to which not?].”

(21)

A ver si vamos haciendo lo que nos corresponde (Spontaneous speech, general Spanish)

“come on [lit. to see], we must do the correct things.”

(22)

La conductancia de un tubo es a grosso modo proporcional al cuadrado del potencial a que están los electrodos (Enrique Mata, La televisión. Fototelegrafía, 1929, Spain, corde)

“Conductivity of the tube is, broadly speaking [lit. to broad mode], proportional to…”

(23)

Si no les pasa, pues bájense y ¡a la chingada! (Gerardo María, Fábrica de conciencias descompuestas, 1980, México, crea)

“If you don’t like it, get down and fuck you [lit. to the chingada].”

As examples (3–23) show, the term-goal of the preposition a may be almost any category, lexical or syntactic: a nominal, noun or pronoun, an adjective, an adverb, a non-inflected verb, a prepositional phrase, a sentence, or a complete autonomous discourse event. Constructions formed with the preposition a may be both argumental and non-argumental; they may be part both of nominal constructions and of verb constructions, may be both a dependent constituent and an autonomous discourse constituent, may appear both in free expressions and in lexicalized ones. That is, the preposition a forms the most diverse constructions and it is, probably, along with preposition de (Rougiest 1980), the most productive preposition in Spanish given the very different syntactic expressions that it forms.

Besides the high variety of goals that may constitute the term of a prepositional phrase with a, pp-a, almost any category may be also the governing constituent of a pp-a: a noun (example 24), an adjective (example 25), an adverb (example 26), a verb (example 27), another preposition (example 28), forming a prepositional sequence or a kind of composed preposition with a. The governing constituent and the preposition are in italics and in bold in the examples.

(24)

Iglesias apuntó a las economías asiáticas como un ejemplo a seguir y elogió la forma muy rápida en que salieron de la crisis (El País, 5 Novembre 2002, Uruguay, crea)

“Iglesias pointed out the Asian economies as an example to follow, and he praised…”

(25)

Lo esencial es invisible a los ojos (ABC, 16 February1996, Spain, crea)

“The essential things are invisible to eyes”

(26)

han recibido alguna o varias dosis de tratamiento antibiótico previamente a la obtención de las muestras (Revista Medicina General, 2003, Spain, crea)

“The patients have taken various doses of antibiotic previously to obtaining the samples”

(27)

el fenómeno mundial del fútbol miró a los reporteros gráficos, sonrió (El Tiempo, 12 June 1997, Colombia, crea)

“The football star looked at the journalists [lit. looked to to the journalists] and smiled.”

(28)

produce ciento veinte y nueve mil, y doscientas cuarenta combinaciones de a dos sílabas (Ramón Campos, El don de la palabra en orden a las lenguas, 1804, Spain, corde)

“[chinese language] produces two forty hundred combinations with two syllables [lit. of to two syllables].”

In sum, the preposition a goes through all Spanish grammar and goes through all lexical and syntactic categories. The reason, as I said, is that a progressively entered new contexts diachronically, causing an exponential growth of contexts. With no doubt, two causes might have been working to motivate such spreading to new contexts and such constructional flexibility. On the one hand, the phonological lightness of this preposition, on the other hand, its locative basic meaning, because space, as it is well-known, is a fundamental cognitive grammatical area, prone to be reworked to encode new grammatical domains, such as time, modality, quality, result, etc.

The structural pattern of pervasive spreading of contexts of the preposition a displays six manners of change, some of them exhibiting internal subtypes, but none of these changes affected the grammatical category of the preposition, as I said before. The etymological Latin context remains in Modern Spanish, as I said, because grammatical change is cumulative or stratified (Hopper 1991) that is, old and new distributions and contexts live together for centuries.

2.1. No Preposition > Preposition a ‘to’: Ø > a

This first type of change has two manifestations: a nominal one and a verbal one. The argumental functions do and io exemplify this first type of change well. In this case, the same function existed in the mother tongue, but the construction had the form of a noun phrase (np) not of a pp, which is the structure which that function has in contemporary Spanish. Examples (29a) and (30a) show that do and io, respectively, had in Latin the form of an np. The examples (29b) and (30b) exemplify the same argumental functions in Spanish, with an obligatory preposition a.

(29)

a. Iudaeos       non nocui

acc-do-jewishs     neg past-harm (Baños 2009, p. 195)

“I have not harmed Jewishs.”

b. No he perjudicado a los judíos

   “I have not harmed Jewishs [lit. to the Jewish]”

(30)

a. Nuntiabant   hoc   idem   Curioni

past-announce  do-this  itself   dat-io-propernoun-Curio

(Allen and Greenough 1979, p. 229)

“they announced this fact to Curio.”

b. Anunciaron esto mismo a Curio

“they announced this fact to Curio”

The verbal manifestation of Ø > a is related to periphrastic temporal constructions. In Old Spanish there was a two-verb construction, lacking any preposition linking them, as (31a) shows. From the 14th century onwards, this kind of periphrastic formation obligatorily takes the preposition a, introducing the second verb, as (31b) shows.

(31)

a. Los christianos […] esforçaron se et fueron ferir en los turcos (Anonymous, Gran conquista de Ultramar, 1293, corde)

“Christians tried harder and went to destroy [lit. went destroy] Turkishs.”

b. y éstos dejando quemar sus cosas, […], fueron a herir con gran velocidad en los del Clavero (Alonso de Maldonado, Hechos del maestre de Alcántara, ca. 1492, corde)

“and that people, allowing to burn their things, went to destroy the Clavero’s people quickly.”

2.2. pp Headed by a Preposition Different from a ‘to’ > pp Headed by a ‘to’

This change is as follows: in Old Spanish, the construction had a pp form, but the introductory preposition was another than a, and over the time this last preposition ousted the oldest one. There are both verbal and non-verbal expressions where the preposition a ousted other prepositions. The preposition ousted by a is usually the preposition de ‘of, from’, but other prepositions were ousted too. Examples (32), (33), (34) and (35) display this second type of change.

In (32), the inchoative-verb construction took de ‘of’ in Old Spanish (32a), but from the 15th century onwards, a ousted de. Nowadays, inchoative verbs subcategorize pp-a in Spanish (32b). Example (33) shows that the preposition a also ousted the preposition de even with non-inchoative verbs; (33) is an unusual context for a, because the rule in Spanish is that despedir ‘to say goodbye’ subcategorizes a pp headed by de; however, (33) is a very classic example, although somewhat rare, from the first Spanish text totally written in Castilian, Cantar de mio Cid.5

(32)

a. Quando ellos se vieron fuera, en saluo comenzaron de fablar (Anónimo, Mocedades de Rodri-go, ca. 1400, corde)

“when they were outside, they began to talk [lit. began of talk]”

b. Libado el trago de la amistad, comenzaron a hablar (Gonzalo Zaldumbide, Égloga trágica, 1910, Ecuador, corde)

“Once drunk the glass of friendship, they began to talk”

(33)

Espidiénsse al rey, con esto tornados son (Cid, v. 1914)

“they said goodbye to the king, and came back.”

Non-verbal constructions displayed the same pattern of change, with a ousting other prepositions. For instance, some temporal and locative complements with diverse prepositions in general Spanish, (34a) and (35a), introduced a in the Argentinian dialect of Buenos Aires, and occasionally in some other American dialects, eliminating the old prepositions, as (34b) and (35b) show.

(34)

a. nos vemos por la noche/nos vemos en la noche

“we’ll see you in the night [lit. for/in the night]”

b. nos vemos a la noche

“we’ll see you in the night [lit. to the night]”

(35)

a. vivo en el 1300 de la avenida Insurgentes

“I live in Insurgentes avenue at 1300 [lit. in the 1300]”

b. vivo al 1300 de Corrientes

“I live in Corrientes avenue at 1300 [lit. to the 1300]”

Some set phrases heading a pp also changed the original preposition by a, such as the following cases: de acuerdo con ‘according with’ > de acuerdo a ‘according to’; con base en ‘with basis in’ > en base a ‘in basis to’; en honor de ‘in honor of’ > en honor a ‘in honor to’. Other constructions, lexicalized to some degree, but not so fixed as set phrases, displayed the same pattern in the history of Spanish: amor de la patria ‘love of nation’ > amor a la patria ‘love to nation’; agua del tiempo ‘water not cold’ > agua al tiempo ‘water not cold’. Examples in (36) show non-verbal constructions in which a won the function; (36a) shows the oldest construction, (36b) the newest one, although the first attestations of a are relatively old in these set phrases, as (36c) shows, an example from the year 1561. In many cases, both prepositions continue coexisting to encode the same construction, as in (37), showing the property of cumulative stratification of most syntactic changes.

(36)

a. El dia 4 de acuerdo con algunos traidores, inventaron plaza en San Pedro (Legal document, 1781, Bolivia, cordiam)

“the fourth, according to [lit. according with] some traitors, they put a market in the main square.”

b. en cada signo figuran las características generales de su hijo, de acuerdo a la fecha de nacimiento, que le darán pistas para saber cómo tratarlos (Expreso, 1 October 1991, Peru, crea)

“every star sign displays the traits of your son, according to his/her birthday date, giving you …”

c. tenyendo en si de acuerdo a la calidad de su persona

(Administrative document, 1561, Peru, cordiam)

“having these things according to his qualities as person.”

(37)

Cada estado practica el contraterrorismo de acuerdo con las circunstancias del propio estado y de acuerdo a sus leyes (Cambio 16, 21 May 1990, Spain, crea)

“every government exerts counter terrorism according to [lit. according with] its actual circumstances and according to the laws.”

2.3. Subordinate Clause Headed by Conjunction que > Subordinate Clause Headed by a ‘to’

A subordinate relative clause, whose nominal antecedent is an abstract noun, meaning ‘meeting’, ‘time’, ‘event’, ‘fact’, ‘conference’, etc., may begin the predication with the preposition a when the accomplishment of the event will take place in an immediate future time, such as in (38). Until the 20th century these predicative adjective constructions only took que (38a), but from the 20th century onwards, the preposition a began competing with que as the introductory nexus of the subordinate clause (38b). This pattern is somewhat similar to the change of type 2.2 above, because the preposition a again ousted another grammatical particle, but this time it was a conjunction not a preposition.

(38)

a. El evento, que se hará en el auditorio del Archivo General de la Nación, está dirigido a autoridades universitarias (El Tiempo, 15 September 1996, Colombia, crea)

“The meeting, which will take place at the National General

Archive, is addressed for University authorities.”

b. Unisys de Venezuela ha organizado para esta tarde a partir de las seis un evento a realizase en el salón Naiguatá del Hotel Tamanaco

(El Universal, 21 April 1993, Venezuela, crea)

“Unisys from Venezuela has organized for this afternoon a meeting which will take place [lit. an event to take place] at the Tamanaco hotel.”

2.4. Genuine Spanish Creations with the Preposition a ‘to’ from the Beginning

Many predicative and non-predicative constructions are genuine Spanish formations, displaying a from the very beginnings of the language. Examples above, such as (3b), me mandó a la mierda ‘he sent me to hell’; (6), al comer gastará una gallina ‘for lunch [lit. to the eat] he’ll spend a poultry’; (7), va a ese pueblo a un pleito ‘he goes to that village to a civil action’; (13), niña, ¡a callar! ‘baby, shut up [lit. to shut up]’; (14), una tortilla a la francesa ‘an omelette to the French way’; (15), andan a cangrejos ‘seeking crabs [lit. walk to crabs]’; (16), olía mal, a podrido, a cerrado, a miseria ‘the air smells bad, closed, rotten, misery [lit. to closed, to rotten, to misery]’; (17), toma veneno a sabiendas ‘he takes poison knowingly [lit. to knowingly]’, and (23), y ¡a la chingada! ‘fuck you [lit. to the chingada]’ are good examples of this fourth type of change.

2.5. Recodification of Prepositional Constructions via Insertion of the Preposition a ‘to’

Constructions with movement verbs + preposition por ‘for’ introduced the preposition a, forming a prepositional sequence a por ‘to for’ in co-occurrence with movement verbs, both with directive ones, such as ir ‘go’ or venir ‘come’, and with non-directive ones, such as pasar ‘pass through’ (Company and Flores 2018).

This change took place in Peninsular Spanish only, never in American Spanish, but it is another proof of the invasion of a. Example (39a) exhibits the oldest and more generalized construction, (39b) shows the new construction with the insertion of a. The last sentences in the two examples of (39), y lo trujeron en unas andas ‘and they brought the dead body in a stretcher’ in (39a), and Y fueron y llegaron al chozo ‘and they went and arrived to the hut’ in (39b), confirm the meaning of the accomplished goal which these expressions have.

(39)

a. Y que sabiendo los naturales dél de la muerte de dicho cazique, fueron por el cuerpo difunto y lo truxeron en unas andas (Administrative document, 1694, Mexico, cordiam)

“the people knowing the cacique’s death, went for dead body and brought it on a stretcher.”

b. Y dijeron: Pos vamos a por lo nuestro. Y fueron y llegaron al chozo (Marciano Curiel Merchán, Cuentos extremeños, 1944, Spain, corde)

“and they said: go for [lit. go to for] our properties, and they went and arrived at the hut.”

2.6. Insertion of a ‘to’ in Learned Idioms which Must go with no Preposition

That is the case of example (22) above: grosso modo > a grosso modo ‘broadly speaking [lit. to broad mode]’.

Summing up, the examples and types of changes show a systematic pattern of spreading to new and more varied contexts with four basic mechanisms: (a) the preposition a invaded Latin non-prepositional functions; (b) the preposition appeared ex novo in genuine Romance constructions; (c) the preposition ousted other prepositions in certain constructions, and (d) the preposition joined other prepositions, forming a prepositional sequence or a composed preposition. And summing up again, via this pervasive spreading, the preposition a won eighteen new contexts, as compared to its presence in Latin.

As far as I know, the preposition a only lost one context, exemplified in (40). The compulsory future modal periphrasis with haber ‘to have’ plus an infinitive took a in Medieval Spanish, as (40a) shows, but from very early times the preposition de ‘of, from’ ousted a in this construction, as (40b) shows.

(40)

a. toda obra que por mano de omne se aya a fazer (Alfonso X, General estoria. Primera parte, apud Company and Flores 2014)

“every deed of good men which they must have to do.”

b. por los serbiçios que nos a fecho en las dichas yslas y esperamos que ha de fazer (Letter, Dominican Republic, 1499, cordiam)

“for the services that he has done us in the mentioned islands and we hope he will do [lit. has of do] in the future.”

Theoretically, the changes themselves and the six types of pattern listed above are heterogeneous. Some extensions of a involved reanalysis of the preposition when it appears in new contexts, and, as it is well known, reanalysis is the basic mechanism of grammaticalization.6 Such are the cases of the encoding of do and io functions, because the preposition took on dative and accusative case-marking. In this case, it might be said that it was a case of tradicional grammaticalization because a case-marking is arguably more grammaticalized than a grammatical preposition. Other changes did not involve reanalysis at all, seeming more such as analogical extensions via a metonymic or metaphorical basis: physical space > metaphoric space > time > attitude > quality > result, etc., or also nominal goal > event goal > discourse goal, the preposition preserving its status of a very grammatical preposition. In some other extensions, the preposition is subcategorized by the verb, and there is no choice between a and other preposition, but in other cases, a forms an oblique complement, having certain syntactic freedom. Thus, it might be said that a subcategorized preposition is somewhat more grammaticalized than a preposition heading a non-argumental complement. And yet in some other extensions, the context progress is completely gradual following the expected channel from more favorable contexts > less favorable ones, as it is the case for io and do. In other cases, the change seems to be so abrupt, or so sudden, that it is difficult to establish gradual phases, as it is the case of voy por agua > voy a por agua ‘I go for some water’ > ‘I go to for some water’, whose first attestation, diffusion and standardization took place in less than one hundred years (Company and Flores 2017).

In sum, from a theoretical point of view, internal heterogeneity is the main feature of the pervasive spreading of a to new contexts, but, with no doubt, there is a salient feature of homogeneity in all changes, which together construe a whole unique pattern: a pervasive spreading to new and different contexts—even, a systematic invasion in many cases—and the creation of new constructions with a.

3. Theory and Data in Dialogue

In this section I will briefly revise both classical and recent definitions of grammaticalization and related theoretical concepts of language change. I frame the review as a check list, discussing whether or not the spreading of contexts and creation of new constructions without modifying the category status of the form fit with each theoretical definition and framework.

3.1. Traditional Grammaticalization → no/yes

The classical definition, in (1) above, of change from a lexical form to a grammatical form or from a grammatical form to a more grammatical one (Hopper and Traugott 2003, p. 2, and many other authors) does not fit with the spreading of a to new and more varied contexts, because this word did not move away from the lexical pole to the grammatical pole, and it was not pressed into service for more grammatical functions.

However, if we look at categories as not internally homogeneous, even the most grammatical ones, as in the case of a grammatical preposition, it would be possible to say that a underwent some kind of grammaticalization in passing from a preposition to a dative and accusative case-marking. Elaborating on the possibility that the preposition would have undergone grammaticalization, it is convenient to consider the following fact: It is not by chance that in formal frameworks, a (along with de ‘of, from’) is the grammatical preposition that can most easily govern its term without specifying a thematic role, although in some cases the preposition a can also behave as a true preposition assigning a thematic role. In sum, the formal analysis of the preposition would seem to mean that a is more grammaticalized in Spanish than in Latin.

3.2. Grammaticalization as “Emergent Grammar” → yes

If “there is no grammar but grammaticalization” (Hopper 1987, 1998), any grammatical change, including analogy, is obviously grammaticalization. Hopper’s position is epistemologically very interesting because all changes are a recreation of previous lexical or grammatical material. Nevertheless, definitions such as this are so vague and so broad that they lack any operative specificity to work with.

3.3. Grammaticalization as “the Emergence of Grammatical Systems” → yes

This definition (Frajzyngier 2010; Hurford 2012, chp. 7) has the same problem as the preceding one: it lacks operative specificity.

3.4. Grammaticalization as “the Emergence of Language” → yes

This definition (Hurford 2012, chp. 7; Smith 2011, p. 144) has the same problem as the two preceding ones, and is worse because it refers to the evolution of human language and not to the diachrony of specific grammatical systems or specific grammatical items.

3.5. Secondary Grammaticalization → yes/no

This subtype of grammaticalization is defined in three ways: (a) “Cases of grammaticalization affecting elements that already have grammatical function and proceed to develop a new grammatical function” (Breban and Kranich 2014); (b) “later stage processes in grammaticalization” (Breban and Kranich 2014), and (c) “The reanalysis of markers of one syntactic category into another one” (Givón 1991). None of these definitions cover the pervasive spreading of a grammatical word to new and more diverse contexts. On the one hand, definitions of secondary grammaticalization do not differ, essentially, from the second section of the definition of traditional grammaticalization: a grammatical form developing more grammatical functions; in that sense, the path preposition > case-marking could also be a secondary grammaticalization. On the other hand, the diachrony of a does not correspond necessarily to “later stage processes in grammaticalization”, because many context extensions are very old, and because the spreading is not a case of grammaticalization. Finally, most extensions of the preposition a, even in cases where they involve new constructions, do not involve reanalysis. Thus, the changes described in Section 2 are incompatible with Givón’s definitions (1991).

3.6. Connecting Grammaticalization → yes/no

Grammaticalization: chains of grammations, regrammations [transition from one grammatical status to another] and degrammations are seen as one connected process: change a is a precondition for b which again is a precondition for c and d (Nørgård-Sørensen et al. 2011, p. 5).

The pervasive spreading of a grammatical word to new and more diverse contexts is not strictly covered by this definition. However, this definition says that everything in a language system is connected, and in that sense, some contexts of the preposition a are a conditioning step for advancing to a new context. Nevertheless, the changes presented in Section 2 above are not as linear as this definition postulates, because there is a strong overlap in creating new constructions via the progression of a to new contexts (Company and Flores 2014). In addition, this definition lacks explanatory force, in my opinion, because almost any change would be a connecting grammaticalization, following the Saussurean principle of language as “a système où tout se tient” (Saussure 1916).

3.7. Exaptation 1 → no

Lass (1990) considers exaptation a specific type of language change, not necessarily related to grammaticalization, although this proposal implies reanalysis: “Junk or garbage morphemes acquire a new function”. The pervasive spreading of a to new contexts never implied the unemployment of the preposition; that is, a never was a junk particle. All changes happened without previous existence of phonetic or grammatical garbage. Actually, no change undergone by a emerged from prepositional residue.

3.8. Exaptation 2 → yes

“Conceptual invention, […] the model itself is what’s new” (Lass 1997, 318 ss.). The pervasive spreading of a to new contexts fits in with this second definition of exaptation. It creates a new model of changing, having context as the protagonist and the locus of change.7 However, one must be careful, because exaptation seems to be a fairly polysemous concept, losing its descriptive force.

3.9. Exaptation 3 → no

Morphemes may end up encoding a meaning that is unrelated, or distantly related to what they used to express. Such a functional leap of an existing form can be referred to as linguistic exaptation (Van de Velde and Norde 2016, p. 9).

Every expression with a preserves the basic abstract meaning of ‘directive telicity towards a goal’. There never was a functional leap in the diachrony of a.

3.10. Functional Renewal → no

An old form reappears in the grammar with a new meaning (Brinton and Stein 1995). That was never the case of a. This preposition has maintained the same basic abstract meaning for centuries, as described above, winning new distributions and entering to new contexts. The constructional and distributional outcomes of a are, in fact, new, the contexts where the pp-a appears are new also, but the category status of a never changed.

3.11. Regrammaticalization → no

The preposition a is not “a desemanticized item found only in a few lexical forms […] reinterpreted in a new function” (Greenberg 1991). Actually, as I said before, a is probably the most productive preposition in Spanish. It was already very productive in Latin (Baños 2009; Glare 2012, s.v. ad), and it became more productive.

3.12. Capitalization → yes/no

The historical process by which a linguistic feature which already exists in a language comes to be substantially exploited for wider purposes (Pountain 1997).

This definition fits in well with the increase of constructions in new contexts of a. However, there are two problems with the concept of capitalization. Firstly, it does not imply a pattern of progressive spreading, which is the main trait of the diachrony of a. Secondly, Pountain’s main example for capitalization, the Spanish auxiliary verb estar ‘to be, to stand’ that grows at the expense of the verb ser ‘to be, to exist’, implies category weakening of this last verb, which is not the case with the preposition a nor with the prepositions ousted by a. All of them continued to be prepositions in Spanish.

3.13. Refunctionalization → no

“A process whereby a form loses its original function and takes on a new function” (Smith 2006). The preposition a never lost its etymological function of introducing a locative complement, and never lost its original relating prepositional function.

3.14. Adfunctionalization → no

“A process in which a form assumes a new function alongside or in addition to its original function” (Smith 2006). The preposition a has always been a grammatical preposition, it never changed its category.

Some gaps and some problems are present in the definitions above. First of all, the role of context has been ignored in grammaticalization and in the related frameworks. In consequence, changes consisting in a pervasive spreading of a grammatical form to new contexts without modifying the grammatical status of the form have not been taken into account. Moreover, many definitions overlap with one another and thus lose their explanatory value and descriptive usefulness. For instance, exaptation, secondary grammaticalization, refunctionalization, functional renewal, regrammaticalization, and, in some cases, grammaticalization, are very similar to one another, and similarly, adfunctionalization and capitalization also share many common characteristics. Likewise, some concepts are polysemous, thus losing their descriptive force, although all proposals and types of changes enhance our understanding of language change, with no doubt. In sum, the concept of grammaticalization, which has so powerfully explained many changes and has connected apparently unrelated diachronic processes, “risks to become the victim of its own success”, as Breban et al. (2012, p. 2) have noted.

4. A Theoretical Proposal

It seems to be necessary to give its own theoretical status to changes such as a → , as it does not readily conform to previously described frameworks. It is not ‘traditional grammaticalization’; it only partially conforms to ‘capitalization’ or ‘connecting grammaticalization’; and it fits with ‘exaptation’, but only in its revised second definition. Obviously, it is ‘emergent grammar’ and it is characteristic of emergence in grammatical systems, because any language change implies recreation of distribution and recreation of encoding. In a general sense, every language change exists in emergent grammar and in emergent language, as variation, change and malleability are inherent to the everyday language functioning.

In sum, it is not clear that a → would be a completely different change from the ones revised in the preceding section. On the one hand, the pp-a shifted to new functions, in a similar way as predicted by most of the definitions above, but it always preserved its category status (except for the reanalysis as a case-marking) and the change was a pervasive extension to new contexts, two facts not considered by the concepts revised in the previous section.

The main difference and the novelty of changes regarding a analyzed in this paper consist, then, in the argument that context, not the form itself, is the protagonist of linguistic change, where context acts in specific distributions and specific constructions. Logically, context goes hand in hand with ‘constructionalization’, in the sense of ‘construction grammar’ (Trousdale 2012, pp. 171–72; Traugott and Trousdale 2013, chp. 1), as a lexeme plus morphosyntactic properties plus semantic properties. Syntagmaticity is inherent to human language. This paper adds to the literature by providing a framework for change that can be applied and extended analogically to other contexts.

In sum, the spreading of a form to new contexts with no changes in the category status of the form lacked a term until now. My proposal is that changes such as a → are ‘analogical extensions induced by context’, where context is both the unit of change and the locus where the change takes place. The analogical model is not a proportional equation, as traditional analogy is, but an extension motivated by the preceding context and by the stability of the general basic abstract meaning of linguistic forms. Certain contexts emerge because they are motivated or induced by other syntactically and semantically close contexts. The change thus can be described as context addition.

Other changes in Spanish involving grammatical words, such as clitics and particles, follow the same diachronic pattern. That is the case of le ‘him’ ‘her’, se ‘-self’, or of some prepositions, but to a lesser extent than the preposition a. They continue to have the same category and the same form, and only new contexts appear (Company 2006; Company and Sobrevilla 2014).

My proposal is also that, in many cases, it is necessary to leave the word level in diachrony and consider context as the unit of change. The type of change analyzed here must be included both in general grammatical change and in constructionalization. It would be labeled as ‘context construction’, a type of construction not listed in the inventory of Hoffman & Trousdale in their handbook on Construction grammar (Hoffmann and Trousdale 2013, p. 2).

5. Conclusions

This paper has shown five theoretical points: first, that, under certain conditions, context is the unit of language change, when some grammatical words are at work. Context, in addition, is always the locus of change. Second, that certain changes consist in a very active spreading of a form to new contexts, without changing the category status of that form. Third, that in certain cases of language change, it is necessary to leave the word level. Fourth, that the spreading to new contexts is an analogical extension. Finally, that this type of change may be termed as spreading of context, a type of construction not included in ‘construction grammar’ nor in ‘constructionalization’ works.

The paper has also shown that spreading of context is not grammaticalization, demonstrating that grammaticalization has neglected the essential role of context in language change, at least in changes involving various grammatical words.

The empirical evidence to make the five proposals above has been the history of the Spanish preposition a, characterized by a very active diachrony, but, at the same time, by a high stability, both in category status and in meaning.

Finally, we have linked theory and diachronic data reviewing fourteen theoretical concepts related to grammaticalization, and checking them against the diachronic evidence of the preposition a.