Introduction

Successful language processing requires integration of different information sources. While this is well-known for syntactic constraints (e.g., Hagoort et al., 1999; Pearlmutter et al., 1999), world knowledge (e.g., Ferretti et al., 2001; Troyer & Kutas, 2020) and social context (e.g., Münster & Knoeferle. 2018; Nieuwland & Van Berkum, 2006) are also rapidly integrated to resolve ambiguity. Uncovering how context-related information, like formality, affects sentence comprehension can help better understand variability in language processing (e.g., Adger, 2006).

Research questions and hypotheses

Adopting a broad definition of register as “[...] any language variety defined by its situational characteristics [...]” (Biber, 2009, p. 823), we tracked eye-movements during sentence reading to investigate: (i) whether context formality-register congruence is rapidly processed; (ii) whether and how sensitivity to formality-register congruence and to morphosyntactic congruence interact during sentence comprehension. In line with accounts of context and world knowledge effects (e.g., Hagoort et al., 2004; Van Berkum et al., 2008), we expected rapid and incremental integration costs for formality-register mismatches and morphosyntactic mismatches (vs. matches). Additivity or interactivity of these effects might reflect distinct versus shared underlying mechanisms and representations (see Hagoort, 2003). We also explored whether variability in the degree of perceived formality modulates such incremental effects.

Methods

In two eye-tracking pilot studies, we investigated real-time processing of formality-register congruence during sentence reading. Monolingual German adults (aged 18-31) read 40 critical items interleaved with 56 filler items (of which 75% were followed by binary comprehension questions), each comprising two context sentences and a target sentence. Target sentences featured an animate subject (NP1; e.g., Der Polizist, ‘The policeman’), a verb in the past simple (e.g., inhaftierte, ‘detained’, high-register; schnappte, ‘grabbed’, low-register), and a direct object (NP2; e.g., die Aktivistin, ‘the activist’). A pair of context sentences, conveying a formal or informal situation (see examples below), preceded each target sentence, permitting the manipulation of formality-register congruence (match vs. mismatch). Experimental lists were pseudo-randomized according to a Latin square, and blocked by context formality (counterbalanced). In a register-only pilot (N=8), we manipulated formality-register congruence (match vs. mismatch). In a register-by-morphosyntax pilot (N=8), we additionally manipulated subject-verb morphosyntactic congruence (match vs. mismatch), with a target verb in the infinitive (e.g., inhaftieren, schnappen). The first pilot featured two conditions: full match and register mismatch; the second pilot additionally featured morphosyntactic mismatch and full mismatch.

Example critical item:

Formal context:Während der gestrigen Ausschreitungen waren die Einsatzkräfte gnadenlos. Die Polizeidirektorin referierte die Rivalität: (‘During yesterday’s riots, the emergency forces were merciless. The police director lectured the rivalry:’).

Informal context:Bei der Demo gestern war die Stimmung richtig heftig. Die Olle hetzte die Protestler: (‘The atmosphere at the demo yesterday was really intense. The old lady stirred up the protesters:’).

Target sentence:

1. High-register verb, subject-verb morphosyntactic match: Der Polizist inhaftierte die Aktivistin. (‘The policeman detained the activist’).
2. High-register verb, subject-verb morphosyntactic mismatch: Der Polizist inhaftieren die Aktivistin. (‘The policeman *detain the activist’).
3. Low-register verb, subject-verb morphosyntactic match: Der Polizist schnappte die Aktivistin. (‘The policeman grabbed the activist’).
4. Low-register verb, subject-verb morphosyntactic mismatch: Der Polizist schnappen die Aktivistin. (‘The policeman *grab the activist’).

Formality ratings: Through offline Web-based tasks (PCIbex; Zehr & Schwarz, 2018), we assessed the perceived formality of context and target sentences. Monolingual German adults rated stimuli on a scale from 0 (very informal) to 50 (very formal). In the context pretest (N=28), informal sentence pairs received lower ratings (mean=12.4; sd=9.8; median=10) than formal ones (mean=35.5; sd=9; median=36), with an average difference of 23.5 points between conditions. In the target sentence pretest (N=30), low-register target sentences received lower ratings (mean=19.4; sd=11.8; median=18.5) than high-register ones (mean=30.9; sd=10.5; median=32), with an average difference of 11.5 points between conditions. Eye-tracking analysis: Interest regions were defined for NP1, verb, and NP2 in target sentences. Linear mixed-effects models were fitted, using the R package lme4 (Bates et al., 2014), to log-transformed first-pass duration, regression path duration, and total time, in the verb and in the object (NP2) region, where spillover effects might emerge. The predictor was register congruence (sum-coded) in the register-only pilot, and an interaction of register congruence*morphosyntactic congruence (sum-coded) in the register-by-morphosyntax pilot. Average target and context sentence formality ratings served as covariates, and random intercepts were fitted for items and participants.

Results

In the register-only pilot, the analyses corroborated the predicted longer total reading times in the verb region (Figure 1) for register mismatches (vs. matches, t=-2.05, p=.040, d=-.25), and for higher target sentence formality ratings (t=3.22, p=.001, d=.37; post-verbal effects n.s.). In the register-by-morphosyntax pilot, we expected that morphosyntactic congruence effects might modulate those of register congruence, or that the effects might be additive. In the verb region, as predicted, total times (Figure 2) were longer for morphosyntactically mismatching (vs. matching) verbs (t=-1.98, p=.048, d=-.24). Contrary to predictions, total times were longer for register-matching verbs (vs. mismatching, t=3.24, p=.001, d=.39; effects of formality ratings n.s.). In the NP2 region, we observed first-pass effects of subject-verb morphosyntactic congruence (t=2.07, p=.038, d=.25), register congruence (t=-2.50, p=.012, d=-.31), and of their interaction (t=2.00, p=.045, d=.25): nouns following a morphosyntactically matching verb received longer first passes in register-match conditions, whereas in register-mismatch conditions no difference emerged between morphosyntactic matches and mismatches. Finally, longer regression path durations in NP2 regions were observed following morphosyntactically mismatching verbs (t=-2.83, p=.005, d=-.35), register-matching verbs (t=2.55, p=.011, d=.31), and as a function of higher context formality ratings (t=2.45, p=.014, d=.29).

Conclusions: In both pilots, effects of formality-register congruence (and of morphosyntactic congruence, in the second pilot) emerged at a late processing stage in the verb region, and at earlier stages in the object/spillover region (second pilot only). Our current findings, from small samples, do not fully support rapid integration of context effects. Interestingly, when both factors were manipulated, the direction of the register congruence effect was opposite to that predicted. This pattern suggests some interference between social-context and morphosyntactic processing during sentence reading, and their reliance on potentially shared mechanisms. Ongoing replication studies with a larger N (=40) will help further address these questions.