Evaluating Grammaticalization and Constructional Accounts: the Development of the Inchoative

Evaluating grammaticalization and constructional accounts: the development of the inchoative construction with put verbs in Spanish.

Renata Enghels & Marie Comer, Ghent University

Abstract

This article explores the historical development of the inchoative construction in Spanish with the put verbs poner and meter filling the auxiliary verb slot. Previous studies have focused mainly on the syntactic and semantic processes through which full lexical put verbs come to fulfil an auxiliary role in the inchoative periphrasis. The analysis presented in this article aims at going beyond this traditional,grammaticalization account, and examines whether a constructional approach offers a more nuanced account of the empirical data. The main objective is to verify the extent to which different structural patterns of the inchoative construction have given rise to a general constructional schema.

Keywords

Inchoative construction, put verbs, Spanish, grammaticalization, diachronic constructional studies.

1. Introduction

In order to express inchoative aspect, Spanish speakers dispose of a wide gamut of verbal periphrasis.Different lexical verb classes are usedas auxiliary verb for this purpose. They include lexically ingressive verbs(empezar a/comenzar a bailar ‘to start to dance’), change of state verbs (romper a llorar, lit. ‘to break to cry’, ‘to start to cry’), movement verbs (echarse a reír, lit. ‘to throw oneself to laugh’, ‘to start to laugh’)and put verbs (ponerse/meterse a escribir, lit. ‘to put oneself to write’, ‘to start to write’) (Gómez Manzano, 1992; Gómez Torrego, 1999; Carrasco Gutiérrez et al., 2006; Aparicio et al., 2014). This article focuses on the development of the Spanish inchoative constructionusingthe last of these categories, and its relationship with the intricate semantics and argument structure of its verbal nucleus.

The process by which the above-listed full lexical verbs come to fulfil an auxiliary role in the inchoative periphrasis is particularly suited to studywithin grammaticalization theories (e.g. Lamiroy De Mulder, 2011).Adopting a mainly semasiological perspective, these studies ask how an individual referential expression acquires procedural meaning by means of processes of syntactic reduction and semantic bleaching.ThusVerroens (2011) reconstructs the grammaticalization process which has allowed the French put verb mettretofunction as an auxiliary verb in an inchoative periphrasis. This reconstruction is an expansion ofHeine’s (2002) four-step grammaticalization model[1], and proposes a six-stage scenario:

1. In an initial stage,mettre occurs with its locativesourcemeaning (an agent displaces a theme to another location in space), and has the following argument structure [NP1 + Vmettre + NP2 + LOC] (e.g. je mets le livre dans ma chambre, ‘I put the book in my room’).
2. In an intermediate stage, the theme can be an animate entity without control or willfulnessin being displaced (e.g. je mets Jean dans ma chambre, ‘I put John in my room’).
3. The bridging contextgives rise to a certain ambiguity: the agent causes a theme to move from one place to another, but at the same time, s/he makes him/her start an activity or an event (expressed by the infinitive). Syntactically,mettre simultaneously combines with a locative complement and an infinitive [NP1 + Vmettre + NP2 + LOC + (a + INF)] (e.g. je mets Jean dans ma chambre à faire ses devoirs, ‘I put John in my room to do his homework’).
4. In the next stage, the full lexical locative meaningis gradually substituted by a procedural causative one (an agent incites a theme to do something). The verb then takes the following argument structure [NP1 + Vmettre + NP2 + (a + INF)] (e.g. Je mets Jean à faire ses devoirs, ‘I set John to do his homework’)
5. In the switch context, the verb is used reflexively[NP1 + Vse mettre+ (a + INF)]. The clitic se undergoes reanalysis: it loses its autonomy as an independent argument and becomes an integral part of the verb. This corresponds to the inchoative reading(e.g. Jean se met à faire ses devoirs ‘John starts doing his homework’).
6. In a final stage the construction se mettre à + inf conventionalizes and spreads throughout the linguistic community.

Although this model offers a plausible explanation for the grammaticalization of a lexical put verb into an aspectual marker, it remains as yet a hypothetical design which has not been empirically verified on the basis of diachronic data. Moreover, the model has a number of empirical and theoretical implicationswhich raise doubts about its applicability tocognate Spanish inchoatives.

First, whereas the French inchoative periphrasis is restricted to the put verbmettre, the Spanish equivalent combines with not one, but two put verbs (poner and meter) (Aparicio et al., 2014).Given this difference, it needs to be examined whether generalizations can be made over the historical development of particular instances, and thus whether the inchoative periphrasis with poner has developed in a similar way to that with meter, and by extension to inchoative periphrases with other verbal nuclei.

Second, the process of reanalysis is presented as a straightforward unidirectional model which takes only tangential account ofthe availability of different linguistic choices to the speaker. In fact, historically, the Spanish inchoative periphrasis included not only the preposition a but also en (e.g. se metieron a/en comprar una casa ‘they started to look into buying a house’), and different word orders with or without intercalation of a lexical element between the put verb and the subordinated infinitive are attested (e.g. se pone a leer en su cuarto vs. se mete en su cuarto a leer, ‘he starts to read in his room’). To put this another way, grammaticalization models set off from the idea that changes pertain to an individual linguistic item, and not always pay fullattention to the features of the immediate contexts in which the item is used.

This paper explores the extent to which these two drawbackscan beovercomeby integrating findings from grammaticalization research into the framework of construction grammar.Instead of studying the development of individual lexical items, diachronic constructional studies explore the history of syntactic patterns, including how new constructions come into being and gradually develop new functions(Noël, 2007; BergsDiewald, 2008; Hilpert, 2008, 2013;Trousdale, 2008; Traugott & Trousdale 2013; Coussé, in press; among others).

The central idea of the constructional approach is that different expressions (such as different inchoative periphrases) with analogous meanings and formal properties are instances of a more general abstract schematic structure or construction. The language system is defined as “a network of conventionalized pairings of form and meaning, in which specific micro-constructions inherit properties from more general schemas” (Trousdale, 2014, p. 557).Meaning is carried by constructions as a whole, and abstracts away from the input of individual lexical items. In this approach, the paradigm of inchoative periphrases canbe defined as a conventional form/meaning pairing for which the use of the term ‘construction’ seems appropriate (Goldberg, 1995; Croft, 2001). Formally, both reflexive verbs ponerse and meterse (as well as other verbs such as comenzar, empezar, echarse etc.)function as phonologically substantive elements and occupy a fixed auxiliary slot [V]. They are accompanied by threeschematic positions or ‘open’ slots: a subject [NP1], a preposition [PREP] and an infinitive slot [INF]. Semantically the subject [NP1] of the inchoative construction forces or incites him/her or itself to carry out the event expressed by the infinitive complement in (1).

(1)[NP1 + Vrefl + (PREP + INF)] ↔ agent starts the activity Z

a. ponerse a bailar, ‘to start to dance’

b. meterse a dormir, ‘to fall asleep’

c. comenzar a trabajar, ‘to start to work’; echarse a reír, ‘to start to laugh’; etc.

The constructional approach relies on the idea that the taxonomic network consists of inheritance relations between constructions defined at different levels of schematicity. Ranging from the highest degree of abstraction to the highest degree of concreteness, linguistic schemas are instantiated by subschemas, and at lower levels by micro-constructions and constructs (that is, the empirically attested tokens) (Goldberg, 1995; Traugott Trousdale 2013).[2]Properties of lower level nodes can be inherited froma superordinateschema, or can be specific to that node.Clearly, the inchoative constructionis sanctioned by the higher level caused-motion schema proposed by Goldberg (1995, p. 52) andshown in Figure 1, and more concretely its instantiation through the caused-motion put construction. According to this basic schema, an agent (SUBJ) causes a theme (OBJ) to move to another place (namely, the goal, expressed by a locative complement (OBL)).

Figure 1: Caused-motion schema applied to put (Goldberg, 1995: p. 52)

Indeed, as prototypical put verbs, poner and meter basically express a caused-motion event involving an object or theme’s change of location to a new location in space. Syntactically, this event is rendered by a four-slot locative construction (2) (Cifuentes Honrubia, 1999; Ibarretxe, 2012).

(2) [NP1 + V + NP2 + LOCATIVE COMPL] ↔ agent causes patientto move to a new location inspace

a. poner el mantel en la mesa, ‘to put the table cloth on the table’

b.meterse un chupete en la boca, ‘to put a pacifier in one’s mouth’

As has been shown in a previous study (Comer & Enghels, in press), this caused-motion put construction can be considered the historical source construction of the inchoative construction. Its current parent is to be identified as the ‘periphrastic verb construction’, combining a conjugated auxiliary verb form with a non-finite (mostly infinitive) form. This construction expresses several aspectual (such as the beginning, ending or development of the event) and modal (such as obligation or necessity) meanings.[3]

Returning to the aforementioned drawbacksof the grammaticalization account, a diachronic constructional approach can, first,allow us to formulate conclusions on both the functional and formal development of the inchoative construction at a more general level, abstracting away from the specific import of individual lexical items. Secondly, a particularly appealing implication of the constructional view is the recognition of relationships between patterns and structural variantswithin a paradigm, so thatthe usage of one particular structural variant cannot be seen as independent of other related variants.

Against this background, this article aims to empirically evaluate whether, by applying the constructional approach, we are indeed capable of offering a more nuanced model of the development of inchoative periphrases with put verbs. This goalhas significance for one of the main research questions running through Hilpert’s (2013) case studies on constructional change, namely to what extent have different structural patterns given rise to a general constructional schema?[4] Do patterns dissimilate over time or do they undergo functional and structural leveling? The different patterns would then include different auxiliary verbs (poner vs. meter), or variable structural properties (such as preposition selection or variable word order - see Section 2.2). As claimed by Hilpert (2013, p. 157) “a leveling process would indicate that an abstract constructional schema is forming, allowingspeakers to produce new tokens that mix and match characteristics fromdifferent source constructions.” This process would then point towards analogical matching, also identified as an important mechanism of change inthe constructionalization model(see Traugott Trousdale, 2013, p. 21 and footnote 4). If, on the contrary, growing divergence is observed, this would point towards a looser connection between individual variants of constructions, and a lack of schema-forming. In line with these considerations, this article pursues a double goal:

1. On a more theoretical level, it assesses the application of a constructional approach to inchoative periphrases (with put verbs). Is the inchoative ‘subschema’ a valid concept, or is the inchoative meaning mainly due to the grammaticalization of an individual lexical item?
2. As concerns the concrete case study, it offers insights into the linguistic parameters and contextual factors which have promoted the development of the inchoative periphrasis in Spanish.

The remainder of this paper is organized as follows. Section 2 provides details of the data collection, and sets out the parameters that define the structural variation within the inchoative construction. Section 3 gives the results of the corpus analysis, and zooms in on the development of the relative token frequency of the inchoative construction with poner and meter (3.1). In addition, it discusses the evolution of its structural variants defined in terms of the selected preposition (3.2), the internal word order (3.3),andthe different semantic types that can be expressed by the inchoative construction (3.4).Finally, Section 4compares the benefits of the constructional approach with those of existing grammaticalization accounts.

2. Materials and method

2.1 Data collection

In order to achieve the above-mentioned goals, this study outlines the development of different patterns within the inchoative construction family between the 13th and 21st century. The analysis relies on a large historical sample of the two Spanish put verbs. This samplewas compiled from the two reference corpora designed by the Real Academia Española: CORDE (Corpus Diacrónico del Español) and CORPES XXI (Corpus of the 21st century). From these were extracted all inchoativeperiphrases of poner and meter as defined in (1)above from peninsular Spanish only, from the 13th,15th, 17th, 19th, and 21st centuries.[5]

The compiled corpus chronologically covers each of the five phases which traditionally divide the history of Spanish language: Medieval Spanish (13th, 15th c.), Classical Spanish (17th c.), Modern Spanish (19th c.) and Contemporary Spanish (21st c.) (Cano Aguilar, 1992).This method of investigating every other century instead of consecutive ones iscommonly applied in other diachronic studies of Spanish (see for instanceTorres Cacoullos, 2000; Bybee & Torres Cacoullos, 2009; Vergara Wilson, 2009, p. 278,p. 286; Jansegers, 2015, pp. 219-220), and has the particular advantage of revealing important tendencies in the data by coveringlonger spans of time (Vergara Wilson, 2009, p. 286). Another, more practical, advantage is that it allows us to cover larger time spans with quantitatively more restricted data sets.[6] In total, 4765 tokens were retrieved of the inchoativeconstruction with poner, and 378 with meter (see Table 1). The inchoative construction with poner thus clearly outnumbers that with meter(see Section 3.1).

Table 1. Absolute frequencies of the inchoative construction per time period[7]

inchoativeponer / inchoativemeter / Total
13th c. / 7 / 43 / 50
15 th c. / 115 / 33 / 148
17 th c. / 648 / 86 / 734
19 th c. / 1066 / 144 / 1210
21st c. / 2929 / 72 / 3001
Total / 4765 / 378 / 5143

2.2 Parameters of constructional variation in inchoative periphrases

Constructions are defined as “multifaceted linguistic units that maydisplay simultaneous change at different levels of description – frequency,structure, meaning, and distribution in the linguistic community” (Hilpert, 2013, p. 21). Consequently, in order to comprehend the development of the inchoative construction, we need to trace relevant changes in terms of its form, function, frequency of use, and distribution. Therefore, the starting point for the empirical analysis is that, in order to understand how a construction has developed, it is crucial to observe possible changes in the distribution of different structural and functional variables within a more general schema. These observations can point towards changes in the configuration of a network, such as the creation or decline of particular structures, or an internal reorganization of a network.

In concrete, the inchoative construction comprises a set of constructional variants (or “allostructions”, in Cappelle’s 2006terms) which can be defined along four dimensions: (1) the semantics of the put verb (poner vs. meter); (2) the semantics of the preposition (a vs. en); (3) the internal word order, in particular the placement of adverbial complements; (4) the semantic event type expressed by the construction.Each of these four variables is discussed below.

(1) Lexical verb: poner vs. meter.Throughout the analysis, extensive attention is paid to the effect of the lexical source verbsponer and meter on the evolution of individual micro-constructions.The two verbs radically differ with regard to their etymological source. Whereas poner derives from Latin pōnĕre, meaning ‘to place’, meter comes from Latin mĭttĕremeaning ‘to send, to throw’. In their etymological dictionary of Spanish, CorominasPascual (1980-1991) mention thatin the Middle Ages,the semantic field of meterextended to encroach on that of locative poner. In present-day Spanish, the two verbs are near-synonymous but have semantically specialized in different domains, even if, at first glance, they share a similar locative meaning(Cifuentes, 1999; Comer, Enghels Vanderschueren, 2015).The verb poneris far more widely used in discourse than meter, and is the less semantically specific. It refers to placements both on a horizontal (e.g., poner/*meter los platos sobre la mesa, ‘to put the plates on the table’) and vertical surface (e.g., poner/*meter el cuadro en la pared, ‘to put the painting on the wall’). The verbmeter, on the other hand, is preferred for internally directed (concrete or abstract) movements (e.g., meterse/*ponerse en la vida de alguien, lit. ‘to put oneself in somebody’s life). Moreover, compared to meter,poner extends more frequently toward other, more grammaticalized domains (such as the pseudo-copulative, e.g. ponerse nervioso ‘to get nervous’). Consequently,the following research question permeates the empirical analysis: do particular structural variables occur more frequently with one of the two lexical verbs, or does the variability abstract away from the unique semantic profile of the verb and isthus situated at a more abstract constructional level?

(2) Variation in the prepositional slot.A second source of structural variability is the prepositional slot, which can be filled not only by the preposition a (‘to’) but also by the preposition with locative meaning en (‘in’). To what extent do the two prepositions alternate freely throughout the history of the inchoative construction, and do we observe a gradual fixation of form of the construction? (Section 3.2)

(3) Internal word order and placement of adverbials.This variable involves the relative word order within the inchoative construction and, more concretely,the intercalation of lexical elements between the put verb slot and the INF slot. Two structural variants are to be distinguished: an adverbial complement can be placed in between the put verb and the infinitive (NP1 + V + ADV + (PREP + INF)), e.g., ponerse en su cuarto a leer ‘(lit.) to put oneself in one’s room to read’) or after the infinitive (NP1 + V + (PREP + INF) + ADV) e.g., ponerse a leer en su cuarto ‘to start to read in one’sroom’). This variable relates to the degree of syntactic incorporation and unithood of the inchoative construction, which is, of course, higher when no lexical element appears between the two verbs. It also links up with the degree of compositionality of the construction, defined by Traugott Trousdale (2013) as the relationship between the meaning of individual items and the meaning of the construction as a whole. A higher degree of unithood within a construction coincides with a reduced degree of compositionality, as the import of the meaning of individual elements becomes less transparent (Section 3.3).[8]

(4)Semantic types. The fourth and final variable relates to the semantic development of the inchoative construction. Different variants are distinguished in terms of the event type expressed by the construction, for instance an animate participant commencing an activity (se pone a leer, ‘he starts to read’) versus the commencement of a non-volitional event (se pone a llover, ‘it starts to rain’). As opposed to the formal featuresexplained in (2) and (3) above, where the development of the inchoative construction is expected to go hand in hand with reduced structural variability, this semantic parameter is expected to point towards more variation as the inchoative construction develops. Indeed, a higher degree of constructional vitality coincides with a higher probability of the constructional schema generating new instances(Hilpert, 2013, p. 116ff;Traugott Trousdale, 2013; Trousdale, 2014). Thiscan be seen as a correlate of an increasing degree of schematicity of a construction. When constructions are used more frequently, speakers generalize over the individual types and develop more abstract schemas. In its turn, this abstract schema attracts new types, which again contributes to the productivity at the schema level (Barðdal, 2008)(Section 3.4).