In Italian active sentences, the gerund PRO corefers with the matrix subject (Rizzi 1982:163) with a few exceptions (e.g. clausal control, see Egerland 2007; 2008). On the contrary, in passive sentences the gerund may yield ambiguous co-indexing. For example, in sentence (1), the gerund (underlined) can be controlled by either the patient 1 (the derived subject) or by the agent 2 (the by-phrase): (1) 1Lucia è interrotta da 2Anna 1|2urlando al telefono Lucia is interrupted by Anna yelling at the phone The phenomenon has been traditionally dealt with under the theory of control. Lonzi (1998) hypothesized that, depending on configurational (Argument-related) and/or on thematic factors, control may be agentive (bearing no overt argument), argumental (syntactic) or clausal/ arbitrary. Other accounts suggested that the choice of the controller of PRO can be influenced by an economy principle (e.g. the Minimal Link Condition), by a processing principle (e.g. Minimal Attachment), by the kind of event structure (symmetry of predicates, Lonzi et al 1994; Winter 2018) and finally by discourse (continuity/discontinuity of the Aboutness-topic, Frascarelli & Hinterholz 2007, Giandoso 2014). In this experiment, we investigated the factors that may affect the choice of either agent or patient control of PRO by Italian native speakers’ (NS). One-hundred-twenty adult (ages ranged between 18 and 69) Italian NS – all with typical language development and various educational backgrounds – participated in an untimed, forced choice experiment. Participants read 20 passive sentences (plus 20 fillers), all displaying animate, feminine agent and patient NPs, such as (2) (2) Anna è sorpresa da Lucia rientrando dalle vacanze Anna is surprised by Lucia coming back from the holidays Participants were asked to choose the controller of PRO (e.g. ‘who is coming back from the holidays?’) between Anna or Lucia in an untimed paper and pencil task. Sentences were pseudorandomized across two different sets – so that each participant encountered a lexical entry only in one condition – and were balanced across the following experimental conditions: (A) ‘position’ (whether the gerund occurs before or after the matrix clause); (B) ‘event’ (whether the event is symmetric such as in giocare ‘play’ or asymmetric such as in notare ‘notice’) and (C) ‘gerund type’ (causal, temporal, modal). The absolute frequency of predicates in different spoken and written corpora of contemporary Italian was controlled with SketchEngine tool. Non-parametric statistics (Kruskal-Wallis) and different significance of independence tests (Chi-square, log-likelihood and Fisher exact) showed that: (1) a significant percentage (almost 23%) of participants chose the agent as PRO controller instead of the grammatical subject; (B) the vicinity of the gerund (after the matrix clause) significantly predicted the choice of agent control, but not the choice of patient control; (C) the presence of multiple values of the gerund strongly predicted agent control whereas the temporal value predicted the choice of the patient; (D) the presence of a symmetric predicates strongly predicted agent control (E) Frequency of predicates was not a factor. These results suggest that (i) the order of NPS in linearization (the former/latter mentioned NP) might be relevant for control; (ii) A-topic (dis)continuity is unlikely a factor and (iii) the temporal value of the gerund can favor the association between patienthood and PRO control.
Nelle frasi attive in italiano, il gerundio PRO si coindicizza con il soggetto della frase matrice (Rizzi 1982:163) con alcune eccezioni (controllo frasale, vedi Egerland 2007; 2008). Al contrario, nelle frasi passive la coindicizzazione del gerundio può risultare ambigua. Per esempio, nella frase (1), il gerundio può essere controllato o dal paziente 1 (il soggetto derivato) o dall'agente 2 (la by-phrase): (1) 1Lucia è interrotta da 2Anna 1|2urlando al telefono Il fenomeno è stato tradizionalmente studiato nell'ambito della Teoria del Controllo. Lonzi (1998) ha ipotizzato che, a seconda della configurazione e/o di fattori tematici, il controllo può essere agentivo, argomentale (sintattico) o frasale/arbitrario. Altri hanno ipotizzato che la scelta del controllore di PRO può essere influenzata da un principio di economia (Minimal Link Condition), da un principio di processing (Minimal Attachment), dal tipo di struttura eventiva (predicati simmetrici, Lonzi et al 1994; Winter 2018) e infine dal discorso (continuità/discontinuità dell' Aboutness-topic, Frascarelli & Hinterholz 2007, Giandoso 2014). In questo esperimento, abbiamo indagato quali fattori possano influenzare la scelta del controllo del paziente o dell'agente nei parlanti nativi di italiano (NS). 120 adulti (di età compresa tra i 18 e i 69 anni) parlanti nativi di italiano, con sviluppo tipico del linguaggio e di varia formazione scolastica, hanno partecipato ad un esperimento forced choice non temporizzato. I partecipanti leggevano 20 frasi (più 20 frasi filler) che presentavano sempre agente e paziente animati e di genere femminile, come in (2): (2) Anna è sorpresa da Lucia rientrando dalle vacanze Ai partecipanti è stato chiesto di scegliere il controllore di PRO ("chi rientra dalle vacanze?") tra Anna e Lucia in un questionario carta e penna. Le frasi sono state pseudorandomizzate attraverso due set, in modo tale che ogni partecipante incontrasse un'entrata lessicale in una sola condizione, e sono stati bilanciati attraverso le seguenti condizioni sperimentali: (A) posizione (gerundiva prima o dopo la frase matrice); (B) evento (se si tratta di un evento simmetrico come "giocare" o asimmetrico come "notare"); (C) tipo di gerundio (causale, temporale, modale). Su SketchEngine è stata calcolata la frequenza assoluta dei predicati in corpora di italiano scritto e parlato. Diverse analisi statistiche (Kruskal-Wallis; Chi-square, log-likelihood e Fisher exact) hanno mostrato che: (1) una percentuale significativa (quasi il 23%) dei partecipanti ha scelto l'agente come controllore di PRO; (B) la vicinanza del gerundio (dopo la frase matrice) predice fortemente la scelta del controllo dell'agente, ma non di quello del paziente; (D) la presenza di predicati simmetrici predice fortemente il controllo dell'agente; (E) la frequenza dei predicati non è un fattore che influenza la scelta. Questi risultati suggeriscono che (i) l'ordine degli NP nella linearizzazione può essere rilevante per il controllo; (ii) la discontinuità di A-topic non è un fattore plausibile e (iii) il valore temporale del gerundio può favorire l'associazione tra patienthood e il controllo di PRO.
Controllo argomentale e controllo tematico del gerundio nella frase passiva in italiano: uno studio sperimentale sui tempi di reazione con forced choice task
SANSOTERA, SARA
2018/2019
Abstract
In Italian active sentences, the gerund PRO corefers with the matrix subject (Rizzi 1982:163) with a few exceptions (e.g. clausal control, see Egerland 2007; 2008). On the contrary, in passive sentences the gerund may yield ambiguous co-indexing. For example, in sentence (1), the gerund (underlined) can be controlled by either the patient 1 (the derived subject) or by the agent 2 (the by-phrase): (1) 1Lucia è interrotta da 2Anna 1|2urlando al telefono Lucia is interrupted by Anna yelling at the phone The phenomenon has been traditionally dealt with under the theory of control. Lonzi (1998) hypothesized that, depending on configurational (Argument-related) and/or on thematic factors, control may be agentive (bearing no overt argument), argumental (syntactic) or clausal/ arbitrary. Other accounts suggested that the choice of the controller of PRO can be influenced by an economy principle (e.g. the Minimal Link Condition), by a processing principle (e.g. Minimal Attachment), by the kind of event structure (symmetry of predicates, Lonzi et al 1994; Winter 2018) and finally by discourse (continuity/discontinuity of the Aboutness-topic, Frascarelli & Hinterholz 2007, Giandoso 2014). In this experiment, we investigated the factors that may affect the choice of either agent or patient control of PRO by Italian native speakers’ (NS). One-hundred-twenty adult (ages ranged between 18 and 69) Italian NS – all with typical language development and various educational backgrounds – participated in an untimed, forced choice experiment. Participants read 20 passive sentences (plus 20 fillers), all displaying animate, feminine agent and patient NPs, such as (2) (2) Anna è sorpresa da Lucia rientrando dalle vacanze Anna is surprised by Lucia coming back from the holidays Participants were asked to choose the controller of PRO (e.g. ‘who is coming back from the holidays?’) between Anna or Lucia in an untimed paper and pencil task. Sentences were pseudorandomized across two different sets – so that each participant encountered a lexical entry only in one condition – and were balanced across the following experimental conditions: (A) ‘position’ (whether the gerund occurs before or after the matrix clause); (B) ‘event’ (whether the event is symmetric such as in giocare ‘play’ or asymmetric such as in notare ‘notice’) and (C) ‘gerund type’ (causal, temporal, modal). The absolute frequency of predicates in different spoken and written corpora of contemporary Italian was controlled with SketchEngine tool. Non-parametric statistics (Kruskal-Wallis) and different significance of independence tests (Chi-square, log-likelihood and Fisher exact) showed that: (1) a significant percentage (almost 23%) of participants chose the agent as PRO controller instead of the grammatical subject; (B) the vicinity of the gerund (after the matrix clause) significantly predicted the choice of agent control, but not the choice of patient control; (C) the presence of multiple values of the gerund strongly predicted agent control whereas the temporal value predicted the choice of the patient; (D) the presence of a symmetric predicates strongly predicted agent control (E) Frequency of predicates was not a factor. These results suggest that (i) the order of NPS in linearization (the former/latter mentioned NP) might be relevant for control; (ii) A-topic (dis)continuity is unlikely a factor and (iii) the temporal value of the gerund can favor the association between patienthood and PRO control.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/9401