Embed Size (px)
Citation preview
Dear Author, Please, note that changes made to the HTML content will be added to the article before publication, but are not reflected in this PDF. Note also that this file should not be used for submitting corrections.
1
3 Sequential processing during noun phrase production
4
5
6 Audrey Bürki a,b,⇑, Jasmin Sadat c,b, Anne-Sophie Dubarry b, F.-Xavier Alario b
7 a Faculté de Psychologie et des Sciences de l’Education, Université de Genève, Bd du Pont d’Arve 42, 1205 Genève, Switzerland8 bAix Marseille Université, CNRS, LPC UMR 7290, 3, Place Victor Hugo, 13331 Marseille, France9 cAix Marseille Université, CNRS, LPL UMR 7309, 5, Avenue Pasteur, 13100 Aix-en-Provence, France
1011
1 3a r t i c l e i n f o
14 Article history:15 Received 29 January 201516 Revised 1 September 201517 Accepted 3 September 201518 Available online xxxx
19 Keywords:20 EEG21 Language production22 Multi-word utterances23 Picture-word interference24
2 5a b s t r a c t
26This study examined whether the brain operations involved during the processing of successive words in27multi word noun phrase production take place sequentially or simultaneously. German speakers named28pictures while ignoring a written distractor superimposed on the picture (picture-word interference29paradigm) using the definite determiner and corresponding German noun. The gender congruency and30the phonological congruency (i.e., overlap in first phonemes) between target and distractor were manip-31ulated. Naming responses and EEG were recorded. The behavioural performance replicated both the32phonology and the gender congruency effects (i.e., shorter naming latencies for gender congruent than33incongruent and for phonologically congruent than incongruent trials). The phonological and gender34manipulations also influenced the EEG data. Crucially, the two effects occurred in different time windows35and over different sets of electrodes. The phonological effect was observed substantially earlier than the36gender congruency effect. This finding suggests that the processing of determiners and nouns during37determiner noun phrase production occurs at least partly sequentially.38� 2015 Published by Elsevier B.V.39
40
41
42 1. Introduction
43 Psycholinguistic models of language production are typically44 conceived of as a series of processing stages involving different45 types of knowledge: the concept to be expressed, the correspond-46 ing syntactico-semantic representation (grammatical encoding47 process), the phonological representation (phonological encoding48 process) and the corresponding abstract gestures (phonetic encod-49 ing). The broad aspects of such organization are used (quite con-50 sensually) to account for the production of isolated words. In51 most linguistic events, however, speakers do not produce single52 words; rather, they combine words into larger utterances. Impor-53 tantly, it cannot be taken for granted that the production of54 multi-word utterances will amount to the same as performing all55 the single-word steps for each word consecutively.56 Behavioural studies have indicated that speakers can activate57 several words, both at the grammatical and phonological encoding58 levels, before they start articulating an utterance (e.g., Allum &59 Wheeldon, 2007; Costa, Navarette, & Alario, 2006; Damian &
60Dumay, 2009; Konopka, 2012; Lee, Brown-Schmidt, & Watson,612013; Martin, Crowther, Knight, Tamborello, & Yang, 2010;62Meyer, 1996; Oppermann, Jescheniak, & Schriefers, 2010; Schnur,63Costa, & Caramazza, 2006; Smith & Wheeldon, 1999; Wheeldon,64Ohlson, Ashby, & Gator, 2013). Much remains to be described,65however, regarding the temporal alignment of encoding processes66for the different words within such planning units. In particular, a67central and open issue in understanding multi-word production68concerns the amount of sequentiality versus simultaneity in the69encoding of the different words at each encoding step.70Consider for instance the noun phrases the dog, or the black dog.71Are the words planned, encoded and selected one after the other or72simultaneously? If the encoding occurs sequentially, a related issue73would concern the order in which the words are accessed and74encoded. For instance, it has been hypothesized that during the75grammatical encoding process, lexical representations (i.e., lem-76mas) for the different words are processed in parallel (Kempen &77Huijbers, 1983; Schriefers, 1992). At the phonological level of78encoding, several authors have hypothesized that word represen-79tations (known as ‘‘lexemes”) are activated and encoded sequen-80tially within phonological planning units, according to the order81in which they are to be produced (see Dell, 1986; Meyer, 1996).82In early models of sentence processing (e.g., Garrett, 1975, see also83Bock, 1987), by contrast, the phonological content of early occur-84ring determiners was specified only after content words had been
http://dx.doi.org/10.1016/j.cognition.2015.09.0020010-0277/� 2015 Published by Elsevier B.V.
⇑ Corresponding author at: Faculté de Psychologie et des Sciences de l’Education,Université de Genève, Bd du Pont d’Arve 42, 1205 Genève, Switzerland.
E-mail addresses: [email protected] (A. Bürki), [email protected] (J.Sadat), [email protected] (A.-S. Dubarry), [email protected] (F.-Xavier Alario).
Cognition xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Cognition
journal homepage: www.elsevier .com/locate /COGNIT
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
85 phonologically encoded. The empirical evidence available so far86 does not allow providing precise answers to these questions.87 Distinguishing between simultaneous and serial processes on88 the basis of chronometric evidence is a notoriously complex89 endeavour (e.g., Townsend & Wenger, 2004). Behavioural para-90 digms only record the endpoint or final product of the production91 process. Consider for example a study by Alario, Costa, and92 Caramazza (2002a, 2002b, see also Levelt, 2002, for comments)93 in which they observed that latencies of determiner + adjective94 + noun (e.g., the blue kite) phrases were influenced by the lexical95 frequency of both the adjective and the noun, and that the two96 effects were additive. According to the authors, these findings are97 compatible with the view that adjectives and nouns are processed98 in a sequential fashion. However, these findings do not allow us99 rejecting an alternative view in which the two representations
100 are accessed simultaneously, and where the ease of processing101 one representation depends on the ease of processing the other102 one (as suggested for instance by Schriefers, de Ruijter, and103 Steigerwald (1999)).104 Neurophysiological recording techniques (e.g., Electroen-105 cephalography, EEG, and Magnetoencephalography, MEG) provide106 fine-grained temporal information about on-going neural and cog-107 nitive process(es). They can be used to gain further insights into108 the coordination of encoding processes in multi-word production.109 Recent advances in electrophysiological methods have allowed110 psycholinguists to examine the time course and neural dynamics111 of overt word production (for reviews see Ganushchak,112 Christoffels, & Schiller, 2011; Indefrey, 2011; Llorens, Trébuchon,113 Liegeois-Chauvel, & Alario, 2011; Munding, Dubarry, & Alario,114 submitted for publication; Strijkers & Costa, 2011). These studies115 have provided novel insights into the time course of encoding pro-116 cesses during isolated word production. Yet, within this line of117 research, very few neurophysiological studies have investigated118 encoding processes for multi-word utterances. Pylkkänen, Bemis,119 and Blanco Elorrieta (2014) used MEG to compare the production120 of isolated nouns and adjective + noun phrases. They observed dif-121 ferences in neural activity between the two utterance types start-122 ing at about 180 ms after picture onset, which they associated with123 combinatory processes (grammatical encoding). In another study,124 Eulitz, Hauk, and Cohen (2000) compared the EEG activity during125 picture naming with isolated nouns and coloured adjective + noun126 phrases and found no difference in the EEG signal between the two127 types of utterances. By contrast, in a recent study, Bürki and128 Laganaro (2014) reported longer periods of stable electrophysio-129 logical activity for noun phrases with a determiner (e.g., the big130 cat, the cat) than for isolated nouns from around 190 to 300 ms131 after picture onset, and longer periods of stable electrophysiologi-132 cal activity for noun phrases with an adjective than for those with-133 out, from 530 ms after picture onset to 100 ms before the onset of134 articulation.135 Apart from some empirical discrepancies, all three studies com-136 pare different utterance types to explore the encoding processes137 that may differ between them (e.g., presence vs. absence of138 multi-word grammatical encoding). As such, they are not informa-139 tive on our topic of interest, which is the temporal alignment of the140 retrieval and selection of different words contained in an141 utterance.
142 1.1. The current study
143 The goal of the present study is to investigate the time course of144 retrieval processes that lead to the production of determiner145 + noun phrases (e.g., the cat). More specifically, we ask whether146 the processing of the two words composing these utterances147 occurs simultaneously or is separable in time.
148German native speakers were asked to name pictures (one per149trial) using a determiner and a noun, while ignoring a printed dis-150tractor word superimposed on the picture (picture-word interfer-151ence paradigm). The presence of a distractor allowed us to create152independent markers for the processing of the noun and the153determiner.154To obtain an index of the processing of the noun, we manipu-155lated the phonological overlap between the target noun and dis-156tractor. Previous research has shown that latencies are shorter in157trials with phonological overlap. This effect (hereafter the ‘‘phono-158logical congruency effect”) has been associated with the phonolog-159ical encoding of the noun (Meyer, 1996; Meyer & Schriefers, 1991,160see also Lupker, 1982; Roelofs, 1997; Schriefers, Meyer, & Levelt,1611990; Starreveld, 2000; Starreveld & La Heij, 1995, 1996; for EEG162evidence see for instance Dell’Acqua et al., 2010).163To obtain an index of the processing of the determiner, we164manipulated the congruency between the gender of the target165and that of the distractor. Previous studies have shown that in sev-166eral languages (including Dutch, or German) latencies were slower167with incongruent than with congruent distractors (La Heij, Mak,168Sander, & Willeboordse, 1998; Schiller & Caramazza, 2003;169Schriefers, 1993; Van Berkum, 1997). This gender congruency170effect has been associated with competition between the171determiner forms of the target and distractor in behavioural stud-172ies (see Schiller & Caramazza, 2003) and in neuroimaging studies173(e.g., Heim, Friederici, Schiller, Rüschemeyer, & Amunts, 2009).
1742. Experiment
1752.1. Participants
176Twenty German native speakers (5 men), aged between 18 and17739, took part in the experiment. They were living in France at the178time of the experiment. They were all right-handed with no179reported hearing or language disorders.180Note that the experiment was conducted in France, and that all181speakers were also fluent speakers of French. They had all learned182to speak this language in early adulthood (age of arrival in France183was 24 years on average, number of months since arrival in the184country varied between 11 and 120 months, with a mean of18554 months).186Before the experimental session, the participants were given187details about the experimental procedure and they provided their188informed consent. The study received appropriate ethical approval189(filed under ‘‘ID RCB: 2011-A00562-39” at ‘‘Comite de Protection190des Personnes Sud Méditerranée I”). Participants were paid 30191euros for their participation.192One participant was discarded due to a high number of eye193blinks. Another participant was excluded because her ERPs did194not clearly reveal the usual P100 component associated with visual195tasks. This left 18 participants to be included in the analysis we196report.
1972.2. Materials
198We selected 70 pictures corresponding to German nouns199(i.e., target nouns, see Appendix A). Half of the nouns had feminine200gender and half had masculine gender. We also selected 70 femi-201nine and 70 masculine nouns to be used as distractors. All distrac-202tors had between three and ten letters (mean number of letters:2035.8, S.D. = 1.6). Each picture was associated with four distractors.204The first two distractors overlapped phonologically with the target205noun (one to four of the initial phonemes were identical), the other206two did not. Of the two distractors in each condition (overlap and207no overlap), one had the same gender as the target noun (gender
2 A. Bürki et al. / Cognition xxx (2015) xxx–xxx
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
208 congruent condition), the other had a different gender than that of209 the target noun (gender incongruent condition).210 We formed two lists of distractors. The first list was used in gen-211 der congruent trials and the second in gender incongruent trials.212 Distractors in each of these two lists were used twice, once with213 a phonologically overlapping target and once with a non-214 overlapping target. The mean number of phonemes shared215 between the target and distractor words in the phonological over-216 lap condition did not differ between gender congruent217 (mean = 2.3) and gender incongruent (mean = 2.2, p > 0.4) trials.218 Distractors in the two lists were matched on number of letters219 and syllables, and on lexical frequency as provided by the CELEX220 database (Baayen, Piepenbrock, & Gulikers, 1995). Target nouns221 and the corresponding distractors did not overlap semantically.222 We created four experimental lists, each with the 140 target223 words. In each list, the pictures were paired with a different dis-224 tractor such that each picture appeared once and in a different con-225 dition in each list. Each participant saw the four lists. The order of226 the lists during the experiment was counterbalanced across227 participants.
228 2.3. Procedure
229 The experiment was run with the E-prime 2.0 software (Psy-230 chology Software Tools, Pittsburgh, PA). Participants were tested231 individually, in a quiet booth, at Aix-Marseille University in Mar-232 seille (France). They received instructions in German by a German233 native speaker.234 The experiment started with a familiarization phase. During235 this phase, the pictures of the experiment appeared on the screen236 and the participants had to name them in German together with237 the corresponding definite determiner. A white rectangle showing238 the characters ‘‘XXXX” was superimposed on each picture where239 the distractor nouns were to be placed during the test phase. After240 naming a given picture, participants pressed the space bar to see241 the intended determiner and noun written below the picture. Dur-242 ing the test phase, participants had to name the pictures using the243 definite determiner and corresponding noun while ignoring a244 printed distractor superimposed on the picture. Each trial started245 with a fixation cross, followed, after 500 ms, by a 200 ms blank246 screen interval. The picture with the superimposed distractor247 was then presented on the screen for 2300 ms. The next trial248 started after a 2000 ms blank screen interval.249 Pictures appeared as a black outline on a white square. The250 position and size of the distractor were identical across conditions251 but the position of the distractor differed slightly among the pic-252 tures (see Miozzo & Caramazza, 1999).253 Every three trials participants saw a picture depicting shut eyes254 during 2500 ms. They were told to blink when seeing this picture255 and to avoid blinking at other times. The next trial started after a256 700 ms blank screen interval.257 As noted above, all participants had French as their second lan-258 guage; during the same experimental session, the also performed259 the task in French with different materials. Because the French260 experiment addressed a different issue (i.e., flexibility in encoding261 strategies in L2), it is not presented here (see Bürki, Sadat, & Alario,262 submitted for publication, for more details).
263 2.4. EEG recording and pre-processing
264 The EEG was recorded with an Active-Two Biosemi system265 (BIOSEMI, Amsterdam), using 128 channels. The signal was sam-266 pled at 512 Hz. ERP extraction and pre-processing were conducted267 with the Cartool software (Brunet, Murray, & Michel, 2011) and268 EEGLAB Matlab toolbox (Delorme & Makeig, 2004). Offline, the269 EEG signals were averaged referenced and band-pass filtered
270between 0.2 and 30 Hz (2nd order Butterworth filter). We also271applied a 50 Hz notch filter. We did not apply any baseline correc-272tion. Several arguments against the use of baseline corrections can273be found for instance in Michel, Koenig, and Brandeis (2009). In274addition, in the present study, the use of a baseline correction275would potentially have generated arbitrary differences across tri-276als. The time interval between stimulus presentation and the end277of the analyzed period was indeed quite large and such long inter-278vals may result in large inter-trial variation (e.g., Luck, 2014). This279would have been especially true in the last part of the time win-280dow of interest.281Epochs corresponding to incorrect naming responses were282excluded from further processing. The remaining epochs were283visually inspected for artefacts (eye blinks, movements). Epochs284with artefacts or amplitudes exceeding ±100 lV were rejected. A2853-D spline interpolation was applied to all electrodes with a bad286signal (between 3% and 17% of electrodes per participant,287mean = 12%). The same interpolation was applied to all the individ-288ual epochs of a given participant. Stimulus-locked (from picture289onset to 250 time frames [TF] – or 488 ms) and response-locked290epochs (from 480 ms to 100 ms before the onset of the verbal291response) were extracted and exported in the EEGLAB format in292order to be processed with LIMO (Pernet, Chauveau, Gaspar, &293Rousselet, 2011). Several studies have indeed shown that response294related processes (e.g., phonological encoding processes) can be295examined by aligning the EEG averages on the onset of the vocal296responses (see for instance Laganaro, 2014; Riès, Janssen, Burle,297& Alario, 2013).298Response-locked analyses capture activities and effects whose299timing is locked to the response and are especially appropriate300when the late effects are targeted. Response times in naming tasks301vary greatly among items and participants. The further away we302move from the point of alignment, the greater the variability303among trials and the smaller the chance to capture an effect. More-304over, given that the window of analysis is set to equal the duration305of the shortest trial a stimulus locked analysis alone contains no306information on the late encoding processes for most of the trials,307similarly a response-locked analysis alone contains no information308on the early processes for these same trials.
3092.5. Analyses and results
3102.5.1. Naming responses311Responses and naming latencies (i.e., time interval between pic-312ture onset and articulation onset) were manually checked offline313for accuracy with the software Praat (Boersma & Weenink, 2014).314Within the 5040 trials recorded, there were 237 errors (5%), most315of them due to dysfluencies (73% of the errors) or to the use of a316noun other than the intended one (15%). Among the remaining317errors, 10% involved the wrong determiner, and 2% were missing318responses or mispronunciations. Five items (0.001%) had unclear319voice onsets and were removed. Visual inspection of the distribu-320tion of naming latencies led to the removal of 36 data points321(0.7%) below 580 ms. The item ‘‘Schaufel” (shovel) was removed322from further analyses because of extremely high mean naming323latencies (1018 ms) compared to the other items. The dataset324was further restricted to the 3762 trials (75% of trials) with EEG325uncontaminated epochs.326We analyzed naming latencies with mixed-effects regression327models (e.g., Baayen, Davidson, & Bates, 2008; Goldstein, 1987)328using the statistical software R (R Development Core Team, 2014)329and library LmerTest (Kuznetsova, 2014). Order of presentation,330gender congruency and phonological overlap were entered in the331models as fixed effects. The model had random intercepts for par-332ticipants and items, and random slopes allowing for the effects of333gender congruency and phonological overlap to differ across par-
A. Bürki et al. / Cognition xxx (2015) xxx–xxx 3
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
334 ticipants and items (see for instance Baayen & Milin, 2010; Barr,335 Levy, Scheepers, & Tily, 2013). To ensure that the results were336 not driven by few atypical data points, residuals larger than 2.5337 times the standard deviation were considered outliers and338 removed (Baayen, 2008). Following the results of the Box–Cox test339 (Box & Cox, 1964), we used the inverse of the latencies as the340 dependent variable. Denominator degrees of freedom and341 p-values were computed based on Satterthwaite’s approximations.342 Responses to gender congruent trials were initiated with343 shorter naming latencies than responses to gender incongruent tri-344 als (777 ms and 819 ms respectively, SE of the difference between345 means = 5.7 ms). Responses in the phonological overlap condition346 were initiated with shorter naming latencies than responses in347 the no overlap condition (789 ms and 806 ms respectively, SE of348 the difference between means = 5.7 ms). The statistical model349 revealed significant fixed effects of gender congruency350 (F(1,329.1) = 14.1, p < 0.001; b = 1.6 � 10�05, SE = 4.3 � 10�06,351 t = 3.8, p < 0.001) and of phonological overlap (F(1,26.3) = 57.1,352 p < 0.0001; b = 5.1 � 10�05, SE = 6.7 � 10�06, t = 7.6, p < 0.0001).353 We note that naming latencies also decreased with order of pre-354 sentation (F(1,5852.2) = 859.3; b = �7.5 � 10�07, t = �29.3,355 SE = 2.6 � 10�07, p < 0.0001).
356 2.5.2. ERP357 Stimulus-locked (from picture onset to 380 ms) and response-358 locked (from 480 ms to 100 ms before the onset of articulation)359 single trial ERPs were analyzed over all space and time dimensions360 using a hierarchical general linear model, following the procedure361 described in Pernet et al. (2011). In LIMO, the data of each partic-362 ipant were first analyzed separately to estimate the parameters363 of the General Linear Model based on information at each time364 point and electrode. These estimated parameters were then used365 to test for statistical significance across participants with robust366 parametric tests. We highlight that this analysis is performed with-367 out any a priori restriction over time or spatial localisation of the368 effects (e.g., a priori time-windows or sets of electrodes).369 We examined the effects of two categorical predictors, gender370 congruency and phonological overlap, and of one covariate, the371 position of the trial within the experimental session (i.e., order of372 presentation). The effects of the two categorical predictors were373 assessed with robust paired t-tests, the effect of the covariate374 was assessed with a one sample t-test (see Pernet et al., 2011, for375 a detailed account of the exact statistical procedures we followed).376 Each of these analyses was performed using 1000 bootstraps. Cor-377 rections for multiple comparisons were applied using spatio-378 temporal clustering (Maris & Oostenveld, 2007, see also Pernet379 et al., 2011) with alpha set to 0.05. In addition, an effect at a given380 electrode was only considered significant when it lasted for at least381 5 time frames (i.e., 10 ms).382 Fig. 1 shows the stimulus-locked grand-averaged ERPs at383 selected electrodes, and corresponding topographies. The N1/P2384 components were clearly visible and were followed by a broader385 component, as is typically observed in stimulus-locked activity386 for the picture naming task (e.g., Laganaro, 2014; Llorens,387 Trébuchon, Riès, Liégeois-Chauvel, & Alario, 2014).388 The analysis of the stimulus-locked ERPs revealed no significant389 differences across the experimental conditions for either the390 phonological or the gender manipulations (see Fig. 2). In contrast,391 both manipulations showed effects in the response-locked ERPs.392 Figs. 2 and 3 show the results of the robust paired t-tests on ampli-393 tudes for the stimulus-locked and response-locked ERPs, respec-394 tively, for the phonological and gender manipulations.395 As can be seen on the left-hand side of Fig. 3, amplitudes differ396 as a function of phonological overlap at several time windows. The397 earliest difference (1) starts 480 ms before the onset of articulation398 and lasts for about 50–60 ms. The second difference (2) lasts from
399about 390 to 350 ms before the onset of articulation. A third time400period (3) affected by the phonological overlap between target and401distractor starts about 280 ms before the onset of articulation and402lasts for about 20 ms. Finally, there is a late effect of phonological403overlap (4), from about 165 to 130 before the onset of articulation.404The difference in amplitudes between gender-congruent and405gender-incongruent trials is much less widespread in time. It starts406about 210 ms before the onset of articulation and lasts for about40760 ms. Notably, the two effects do not overlap in time.
4082.6. General discussion
409Using EEG, we tested the extent to which empirical effects asso-410ciated with noun and determiner processing overlap in time and411space during the preparation of noun phrase utterances. Beha-412vioural results replicated both the phonology1 and the gender con-413gruency effects (e.g., Meyer & Schriefers, 1991; Schriefers, 1993). In414addition, the phonological and gender manipulations affected the415EEG data. Crucially, the two effects occurred in different time win-416dows over different sets of electrodes (i.e., over non-overlapping417topographies). This finding is as predicted if the processing of two418words in determiner noun phrases occurs at least partly419sequentially.420In the remainder of the discussion we interpret the present421findings in the light of prior data and hypotheses regarding the422source of the phonological and gender congruency effects. Doing423so allows explicit proposals regarding the temporal alignment of424underlying encoding processes in multi-word noun phrase425production.426Empirical evidence from previous studies suggests that both the427phonological congruency and gender congruency effects occur dur-428ing the phonological processing of the utterance. The phonological429congruency effect has been associated with various cognitive sub-430processes involved in phonological encoding, i.e., lexeme access,431the activation of the individual phonemes of the word, and the432insertion of the phonemes in the metrical frame (metrical spell-433out, Levelt, Roelofs, & Meyer, 1999). The gender congruency effect434has been associated with competition between the determiner435forms of the target and distractor (Heim et al., 2009; Schiller &436Caramazza, 2003). Under the assumption that the ERP and beha-437vioural effects we observe in the present study have a common ori-438gin (more on this below), the time course of the gender and439phonological congruency effects reveals that the phonological440encoding of the determiner and that of the noun occur sequen-441tially. In addition, this time course suggests that accessing the442noun word form (as indexed by the onset of the phonological facil-443itation effect) occurs prior to accessing the determiner word form444(as indexed by the gender congruency effect).445The idea that the determiner word form is accessed after the446noun word form is not new. Previous research led to the formula-447tion of the ‘‘early” and ‘‘late-selection” hypotheses for determiner448selection (Caramazza, Miozzo, Costa, Schiller, & Alario, 2001;449Costa, Alario, & Sebastián-Gallés, 2007). These hypotheses state450that determiner selection will occur ‘‘relatively early” in the451response preparation process when selection is mostly guided by452grammatical gender (e.g., in several Germanic languages), and that453it will occur ‘‘relatively late” when grammatical gender and454phonology fully determine the determiner form (as in several
1 The effect of phonological congruency may appear relatively small (17 ms) in thepresent study when compared to similar effects reported in previous studies (e.g.,>30 ms in Meyer and Schriefers (1991), with auditory distractors, or in Dell’Acquaet al. (2010) with written distractors). This may be due to a smaller number ofoverlapping phonemes in the present study, with some target-distractor pairsoverlapping by only one phoneme. Smaller phonological overlaps indeed lead tosmaller effect sizes (e.g., Schiller, 2008).
4 A. Bürki et al. / Cognition xxx (2015) xxx–xxx
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
Fig. 1. Stimulus locked ERPs averaged across all participants and conditions with: (a) Waveform at selected electrodes and (b) topographies at different time windows.
Fig. 2. (a) Results of robust paired t-tests (representation of significant t-values) for phonological overlap (left) and gender congruency (right) after correction for multiplecomparisons; each line is an electrode. (b) Mean waveforms for the different conditions at selected electrodes.
A. Bürki et al. / Cognition xxx (2015) xxx–xxx 5
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
455 Romance languages2). This proposal was solely based on response456 time data. The present study shows that even in languages with457 reputedly ‘‘early determiner selection” (e.g., German), where the458 form of the determiner does not strictly depend on the noun’s
459phonology, determiner form processing is detected significantly later460than noun form processing. This finding raises the issue of whether461the late selection of determiners (relatively to nouns) is a general462feature of languages with determiners or whether this late selection463is present only in languages where the phonological form of the464noun is related, even partially, to the noun’s gender. In German,465the phonology of a subset of nouns provides partial information466about the gender of the noun. Previous research has shown that467speakers use this phonological information to guide their responses468in gender decision tasks (Schiller, Münte, Horemans, & Jansma,4692003). Possibly, the phonological form of determiners is processed470later in naming tasks in this language because participants partly
Fig. 3. (a) Results of robust paired t-tests (representation of significant t-values) for phonological congruency (left) and gender congruency (right) after correction for multiplecomparisons; with the colours representing electrodes with higher (red) and lower (orange) amplitudes for trials with phonological or gender congruent trials; each line is anelectrode. (b) Schematic of the electrodes with significant differences at selected time frames within the time window of the effects. (c) Mean waveforms for the differentconditions at selected electrodes; note that the shape of the ERPs is typical of response-locked analyses in speech production tasks (e.g., Laganaro, 2014). (For interpretation ofthe references to colour in this figure legend, the reader is referred to the web version of this article.)
2 In these languages (as well as in English) the form of the determiner may bedetermined by the phonology of the noun (e.g., le chien ‘the dog’ vs. l’âne ‘the donkey’in French). Studies have reported that in these languages, speakers are not sensitive tothe congruency in gender between target and distractor (e.g., Alario & Caramazza,2002, see also Costa et al., 2007, for a review). Caramazza and colleagues (e.g.,Caramazza et al., 2001) suggested that the gender congruency effect is absentprecisely because the selection of the form of the determiner is delayed until theproperties of the noun are made available.
6 A. Bürki et al. / Cognition xxx (2015) xxx–xxx
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
471 rely on the phonology of the noun to determine its gender. Up to this472 point we have interpreted the ERP effects in the context of earlier473 interpretations of behavioural findings. Yet, ERPs can potentially474 reveal richer and more subtle differences in processing than475 response times do and could reflect other/additional processes. We476 note for instance that the phonological congruency effect is mostly477 observed before the gender congruency effect, but that it also shows478 a later component. We speculate that the later ‘‘rebound” of the479 effect might reflect metrical spell-out or a pre-response monitoring480 process in relation to the insertion of the determiner form in the481 metrical phrase. The suggestion that the different temporal occur-482 rences of a phonological effect in the EEG data may reflect different483 underlying processes echoes back to a proposal by Schuhmann,484 Schiller, Goebel, and Sack (2011). The authors used TMS to deter-485 mine the functional relevance and time course of the left posterior486 inferior frontal gyrus (IFG), left middle temporal gyrus (MTG) and487 left posterior superior temporal gyrus (STG) during a picture naming488 task. They observed that the MTG is involved 225 ms after picture489 onset, the IFG at around 300 ms, and the MTG again around490 400 ms together with the posterior STG. The authors assume that491 the early involvement of the MTG reflects lexical retrieval, that the492 involvement of the IFG at 300 ms reflects phonological encoding pro-493 cesses, and that the late effects reflect monitoring processes.494 Similarly the question may arise as to whether the ERP manifes-495 tation of the gender congruency effect could reflect a late monitor-496 ing process, in which the appropriateness of the determiner form is497 checked. Distinguishing between cognitive processes and the498 monitoring of these processes on empirical grounds is not straight-499 forward without an ad-hoc manipulation, as any ERP effect can500 probably be argued to reflect one, the other or both. Crucially for501 our purposes, an account of the late occurring gender and phonol-502 ogy effects in terms of monitoring would not alter the key conclu-503 sion of the present study, i.e., that the determiner and noun are504 processed with some amount of sequentiality. In addition, we note505 that the monitoring account would imply that ERP and response506 time effects have a different source. As such, it is arguably less par-507 simonious than a single locus account. The locus of the gender con-508 gruency effect in response times has been examined in several509 studies, and these studies all converge on the conclusion that the510 effect arises because the determiner form from the distractor com-511 petes with the determiner to be produced.512 Another relevant issue concerns the extent to which the time513 course of the ERP effects also reflects the time course of encoding514 of the distractor word. In the picture word interference paradigm,515 effects emerge as a consequence of the relationship between the516 target and distractor words. They thus provide information on517 the processing of both words. It could be argued for instance that518 the phonological information of the distractor is accessed before519 the information about its gender. The retrieval of the target word520 would be facilitated by the amount of overlapping information in521 a first step and by the overlap of gender in a second step. First note522 that in this account, the effects observed on the ERPs and on the523 response time would have different loci (see remark above about524 parsimony). In addition, this account assumes that the phonologi-525 cal encoding of the target word precedes the grammatical encoding526 of this word, in contradiction with the widely held assumption that527 speakers access the grammatical features of the words prior to528 their phonological content (e.g., van Turennout, Hagoort, &529 Brown, 1998 for admittedly debated empirical evidence). Most530 importantly, in the present study, the effects are observed when531 the ERP is time-locked to the production of the target sequence,532 not on the presentation of the distractor word. Compared to a533 stimulus-locked analysis, the response locked analysis blurs the534 potentials evoked by stimulus processing and highlights the activ-535 ity synchronized to response preparation (Riès et al., 2013). For this536 reason, the influence of the distractor word on the ERP signal
537should primarily reveal the encoding of the target word at every538particular time point. Irrespective of when the phonological539information from the distractor word is processed, this phonolog-540ical information should not affect the processing of the target word541unless the target word is being phonologically encoded at this par-542ticular time point. The finding that the gender congruency effect543occurs after the phonological congruency effect in the EEG can thus544not readily be interpreted as reflecting solely the time course of545encoding of the distractor word.546So far we have discussed the relative time course of the two547effects. Their actual timing does not inform us on the onset or off-548set of the phonological encoding of the target word, but points to549time windows in which the phonological information of the target550word is being processed. This timing appears congruent with pre-551vious evaluations of the time course of phonological encoding esti-552mated in bare noun production studies. In the present study, the553phonological congruency effect becomes apparent in a time win-554dow around 480 ms before the onset of articulation (or counting555forward from the stimulus, about 300 ms after picture onset). This556timing is consistent with the findings of a previous EEG study of557the same paradigm. Using a jackknife procedure, the data in558Dell’Acqua et al. (2010) revealed a phonological congruency effect559in the picture word interference paradigm at about 320 ms after560picture onset (for mean naming latencies of about 870 ms). Unlike561in the present study, this effect was apparent in the stimulus-562locked ERPs (but see Blackford, Holcomb, Grainger, & Kuperberg,5632012 who found no difference between phonologically related564and phonologically unrelated masked primes in the stimulus-565locked ERPs, despite having found an advantage for phonologically566related trials in the response times). More generally, according to567Indefrey’s (2011) meta-analysis (see Indefrey & Levelt, 2004;568Strijkers & Costa, 2011 for a critical standpoint), word form access569starts 275 ms after picture onset for a mean response time of570600 ms (i.e., 325 ms before the onset of articulation) and the inser-571tion of phonemes in the metrical frame (segmental spell-out) lasts572from 355 to 465 ms after picture onset (i.e., 245–135 ms before the573onset of articulation). The phonological congruency effect we574observed for the noun (once again, about 300 ms after picture575onset) broadly occurs in a similar time window. Interestingly,576ERP differences were observed in the response-locked but not in577the stimulus-locked analysis. Response-locked analyses capture578activities and effects whose timing is locked to the response. The579fact that we observe robust effects in the response-locked analysis580and no effect in the stimulus-locked analysis suggests that the581phonological and gender congruency effects are better locked to582the initiation of the vocal response than to the visual processing583of the stimulus.584To conclude, neurolinguistic studies of word production have585started to provide increasingly detailed information on the time586course of cognitive mechanisms underlying the production of iso-587lated words. A major challenge for neurolinguistic research today588is to extend this study to multi-word utterances, such as deter-589miner noun phrases, or natural speech. The present work provides590crucial information on this issue by showing sequential processes591during the production of determiner noun phrases.
592Acknowledgements
593We acknowledge funding from the Swiss National Science594Foundation (grant no. IZK0Z1-146008), the University of Geneva595(Subside tremplin), the European Research Council under the Euro-596pean Community’s Seventh Framework Program (FP7/2007-2013597Grant agreement no. 263575), and the Brain and Language598Research Institute (Aix-Marseille Université: A⁄MIDEX grant599ANR-11-IDEX-0001-02 and Labex grant ANR-11-LABX-0036). We
A. Bürki et al. / Cognition xxx (2015) xxx–xxx 7
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
600 thank the ‘‘Féderation de Recherche 3C” (Aix-Marseille Université)601 for institutional support.
602 Appendix A
603 Target and distractor words used in Experiment.
Target word(Gender)
Distractor type
Same gender Different gender
Overlap Nooverlap
Overlap No overlap
Besen (M) Berg Markt Beere HupeAdler (M) Adel Kredit Ader HuldigungArm (M) Argwohn Koffer Armut ButterAxt (F) Aktion Witterung Akzent WindBanane (F) Bank Trägheit Bann KittelBaum (M) Bau Zucker Baustelle KachelBirne (F) Binde Hantel Biber TüftlerBlume (F) Bluse Erwartung Blutegel KarpfenBombe (F) Borste Last Bock GartenBrunnen (M) Bruch Schall Bruchbude SchalotteBrust (F) Brutalität Tapete Bruder StaatBürste (F) Bürde Nahrung Bürger HandelBus (M) Busch Filter Butter KammerErdbeere (F) Erde Staffelei Erwerb KalkErdnuß (F) Erwartung Strafe Erfolg BannFeder (F) Fee Gießkanne Fehler HanfFinger (M) Filz Busch Firma BruchbudeFisch (M) Filter Adel Finsternis BaustelleFliege (F) Fliese Kasse Flieder StrahlGabel (F) Gabe Erde Garten ErwerbGeist (M) Geier Pickel Geige HaftGitarre (F) Gießkanne Bank Gipfel KesselHamburger
(M)Hammel Palast Haltung Knolle
Hammer (M) Hals Bruch Haft GeigeHand (F) Hantel Brutalität Hanf BiberHarfe (F) Harke Tüte Handel SchaumHubschrauber
(M)Hunger Kaffee Huldigung Nase
Hut (M) Huf Bau Hupe MannschaftKaktus (M) Kaffee Geier Kachel BeereKamm (M) Kampf Wunsch Kammer VorlageKartoffel (F) Karre Fee Karpfen LeimKassette (F) Kasse Bürde Kalk GipfelKette (F) Kelle Waffe Kessel LachsKirsche (F) Kiste Stange Kittel NapfKnoten (M) Knoblauch Lehm Knolle FirmaKorb (M) Koffer Hals Konditorei TrachtKrebs (M) Kredit Nabel Creme SchwarteLampe (F) Last Urkunde Lachs TrägerLeiter (F) Leitung Gabe Leim TanzLeopard (M) Lehm Schwager Lehre WurstMantel (M) Markt Knoblauch Mannschaft CremeMatrose (M) Mangel Filz Matratze LehreMixer (M) Mist Kampf Milch HaltungNadel (F) Nahrung Kiste Napf BockNagel (M) Nabel Termin Nase PiniePanzer (M) Palast Mist Pappe MilchPilz (M) Pilger Mangel Pinie ArmutPinsel (M) Pickel Argwohn Pille MatratzeRakete (F) Rache Schau Radar Urlaub
Appendix A (continued)
Target word(Gender)
Distractor type
Same gender Different gender
Overlap Nooverlap
Overlap No overlap
Schalter (M) Schall Berg Schalotte ZungeSchaufel (F) Schau Karre Schaum RadarSchwan (M) Schwager Pilger Schwarte PilleSpiegel (M) Spieß Vorstand Spinne TendenzStadt (F) Stange Fliese Stall BlutegelStatue (F) Staffelei Kelle Staat FliederStraße (F) Strafe Bluse Strahl ErfolgTasse (F) Tapete Aktion Tanz FehlerTeppich (M) Termin Hunger Tendenz WahlTraktor (M) Trafo Huf Tracht SpinneTräne (F) Trägheit Binde Träger WaldTrompete (F) Trophäe Wand Troll AkzentTür (F) Tüte Harke Tüftler BruderUhr (F) Urkunde Rache Urlaub WandelVogel (M) Vorstand Wahn Vorlage AderVulkan (M) Wunsch Trafo Wurst PappeWaffel (F) Waffe Leitung Wandel TrollWagen (M) Wahn Spieß Wahl KonditoreiWalnuss (F) Wand Borste Wald BürgerWindel (F) Witterung Trophäe Wind StallZug (M) Zucker Hammel Zunge Finsternis
988
989References
990Alario, F.-X., & Caramazza, A. (2002). The production of determiners: Evidence from991French. Cognition, 82(3), 179–223. http://dx.doi.org/10.1016/S0010-0277(01)99200158-5.993Alario, F.-X., Costa, A., & Caramazza, A. (2002a). Frequency effects in noun phrase994production: Implications for models of lexical access. Language and Cognitive995Processes, 17(3), 299–319. http://dx.doi.org/10.1080/01690960143000236.996Alario, F.-X., Costa, A., & Caramazza, A. (2002b). Hedging one’s bets too much? A997reply to Levelt (2002). Language and Cognitive Processes, 17(6), 673–682. http://998dx.doi.org/10.1080/01690960143000461.999Allum, P. H., & Wheeldon, L. R. (2007). Planning scope in spoken sentence1000production: The role of grammatical units. Journal of Experimental Psychology:1001Learning, Memory, and Cognition, 33(4), 791–810. http://dx.doi.org/10.1037/10020278-7393.33.4.791.1003Baayen, R. H. (2008). Analyzing linguistic data: A practical introduction to statistics1004using R. New York: Cambridge University Press.1005Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with1006crossed random effects for subjects and items. Journal of Memory and Language,100759(4), 390–412. http://dx.doi.org/10.1016/j.jml.2007.12.005.1008Baayen, R. H., & Milin, P. (2010). Analyzing reaction times. International Journal of1009Psychological Research, 3(2), 12–28.1010Baayen, R. H., Piepenbrock, R., & Gulikers, L. (1995). CELEX2 LDC96L14. Philadelphia:1011Linguistic Data Consortium.1012Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for1013confirmatory hypothesis testing: Keep it maximal. Journal of Memory and1014Language, 68(3), 255–278. http://dx.doi.org/10.1016/j.jml.2012.11.001.1015Blackford, T., Holcomb, P. J., Grainger, J., & Kuperberg, G. R. (2012). A funny thing1016happened on the way to articulation: N400 attenuation despite behavioral1017interference in picture naming. Cognition, 123(1), 84–99. http://dx.doi.org/101810.1016/j.cognition.2011.12.007.1019Bock, K. (1987). Exploring levels of processing in sentence production. In G. Kempen1020(Ed.), Natural language generation (pp. 351–363). Boston: Martinus Nijhoff1021Publishers.1022Boersma, P., & Weenink, D. (2014). Praat: Doing phonetics by computer (Version10235.4.01).1024Box, G. E. P., & Cox, D. R. (1964). An analysis of transformations. Journal of the Royal1025Statistical Society. Series B (Methodological), 26(2), 211–252.1026Brunet, D., Murray, M. M., & Michel, C. M. (2011). Spatiotemporal analysis of1027multichannel EEG: CARTOOL. Computational Intelligence and Neuroscience, 2011,1028e813870. http://dx.doi.org/10.1155/2011/813870.1029Bürki, A., Sadat, J., & Alario, F.-X. (submitted for publication). Time course of1030phonological encoding in proficient bilinguals.
8 A. Bürki et al. / Cognition xxx (2015) xxx–xxx
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
1031 Bürki, A., & Laganaro, M. (2014). Tracking the time course of multi-word noun1032 phrase production with ERPs or on when (and why) cat is faster than the big cat.1033 Language Sciences, 5, 586. http://dx.doi.org/10.3389/fpsyg.2014.00586.1034 Caramazza, A., Miozzo, M., Costa, A., Schiller, N. O., & Alario, F.-X. (2001). A1035 crosslinguistic investigation of determiner production. In E. Dupoux (Ed.),1036 Language, brain and cognitive development: Essays in honor of Jacques Mehler1037 (pp. 209–226). Cambridge, MA: MIT Press.1038 Costa, A., Alario, F.-X., & Sebastián-Gallés, N. (2007). Cross-linguistic research on1039 language production. In M. G. Gaskell (Ed.), The oxford handbook of1040 psycholinguistics (pp. 531–546). Oxford: Oxford University Press.1041 Costa, A., Navarette, E., & Alario, F.-X. (2006). Accessing object names when1042 producing complex noun phrases: Implications for models of lexical access.1043 Cognitiva, 18(1), 3–23. http://dx.doi.org/10.1174/021435506775462454.1044 Damian, M. F., & Dumay, N. (2009). Exploring phonological encoding through1045 repeated segments. Language and Cognitive Processes, 24(5), 685–712. http://dx.1046 doi.org/10.1080/01690960802351260.1047 Dell, G. S. (1986). A spreading-activation theory of retrieval in sentence production.1048 Psychological Review, 93(3), 283–321. http://dx.doi.org/10.1037/0033-1049 295X.93.3.283.1050 Dell’Acqua, R., Sessa, P., Peressotti, F., Mulatti, C., Navarrete, E., & Grainger, J. (2010).1051 ERP Evidence for ultra-fast semantic processing in the picture-word1052 interference paradigm. Frontiers in Psychology, 1, 177. http://dx.doi.org/1053 10.3389/fpsyg.2010.00177.1054 Delorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of1055 single-trial EEG dynamics including independent component analysis. Journal of1056 Neuroscience Methods, 134(1), 9–21. http://dx.doi.org/10.1016/j.1057 jneumeth.2003.10.009.1058 Eulitz, C., Hauk, O., & Cohen, R. (2000). Electroencephalographic activity over1059 temporal brain areas during phonological encoding in picture naming. Clinical1060 Neurophysiology, 111(11), 2088–2097. http://dx.doi.org/10.1016/S1388-24571061 (00)00441-7.1062 Ganushchak, L. Y., Christoffels, I. K., & Schiller, N. O. (2011). The use of1063 electroencephalography in language production research: A review. Frontiers1064 in Psychology, 2, 208. http://dx.doi.org/10.3389/fpsyg.2011.00208.1065 Garrett, M.-F. (1975). The analysis of sentence production. In G. Bower (Ed.),1066 Psychology of learning and motivation (pp. 133–177). New York: Academic Press.1067 Goldstein, H. (1987). Multilevel models in educational and social research. London:1068 Griffin.1069 Heij, W. L., Mak, P., Sander, J., & Willeboordse, E. (1998). The gender-congruency1070 effect in picture-word tasks. Psychological Research, 61(3), 209–219. http://dx.1071 doi.org/10.1007/s004260050026.1072 Heim, S., Friederici, A. D., Schiller, N. O., Rüschemeyer, S.-A., & Amunts, K. (2009).1073 The determiner congruency effect in language production investigated with1074 functional MRI. Human Brain Mapping, 30(3), 928–940. http://dx.doi.org/1075 10.1002/hbm.20556.1076 Indefrey, P. (2011). The spatial and temporal signatures of word production1077 components: A critical update. Language Sciences, 2, 255. http://dx.doi.org/1078 10.3389/fpsyg.2011.00255.1079 Indefrey, P., & Levelt, W. J. M. (2004). The spatial and temporal signatures of word1080 production components. Cognition, 92(1–2), 101–144. http://dx.doi.org/1081 10.1016/j.cognition.2002.06.001.1082 Kempen, G., & Huijbers, P. (1983). The lexicalization process in sentence production1083 and naming: Indirect election of words. Cognition, 14(2), 185–209. http://dx.doi.1084 org/10.1016/0010-0277(83)90029-X.1085 Konopka, A. E. (2012). Planning ahead: How recent experience with structures and1086 words changes the scope of linguistic planning. Journal of Memory and Language,1087 66(1), 143–162. http://dx.doi.org/10.1016/j.jml.2011.08.003.1088 Kuznetsova, A. (2014). Package « lmerTest »: Tests for random and fixed effects for1089 linear mixed effect models (lmer objects of lme4 package).1090 Laganaro, M. (2014). ERP topographic analyses from concept to articulation in word1091 production studies. Frontiers in Psychology, 5, 493. http://dx.doi.org/10.3389/1092 fpsyg.2014.00493.1093 Lee, E.-K., Brown-Schmidt, S., & Watson, D. G. (2013). Ways of looking ahead:1094 Hierarchical planning in language production. Cognition, 129(3), 544–562.1095 http://dx.doi.org/10.1016/j.cognition.2013.08.007.1096 Levelt, W. J. M. (2002). Picture naming and word frequency: Comments on Alario,1097 Costa and Caramazza. Language and Cognitive Processes, 17(3), 299–319. http://1098 dx.doi.org/10.1080/01690960143000443.1099 Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech1100 production. The Behavioral and Brain Sciences, 22(1), 1–38. http://dx.doi.org/1101 10.1017/S0140525X99001776.1102 Llorens, A., Trébuchon, A., Liegeois-Chauvel, C., & Alario, F.-X. (2011). Intra-cranial1103 recordings of brain activity during language production. Language Sciences, 2,1104 375. http://dx.doi.org/10.3389/fpsyg.2011.00375.1105 Llorens, A., Trébuchon, A., Riès, S., Liégeois-Chauvel, C., & Alario, F.-X. (2014). How1106 familiarization and repetition modulate the picture naming network. Brain and1107 Language, 133, 47–58. http://dx.doi.org/10.1016/j.bandl.2014.03.010.1108 Luck, S. J. (2014). An introduction to the event-related potential technique. MIT Press.1109 Lupker, S. J. (1982). The role of phonetic and orthographic similarity in picture-word1110 interference. Canadian Journal of Psychology, 36(3), 349–367. http://dx.doi.org/1111 10.1037/h0080652.1112 Maris, E., & Oostenveld, R. (2007). Nonparametric statistical testing of EEG- and1113 MEG-data. Journal of Neuroscience Methods, 164(1), 177–190. http://dx.doi.org/1114 10.1016/j.jneumeth.2007.03.024.1115 Martin, R. C., Crowther, J. E., Knight, M., Tamborello, F. P., & Yang, C.-L. (2010).1116 Planning in sentence production: Evidence for the phrase as a default planning
1117scope. Cognition, 116(2), 177–192. http://dx.doi.org/10.1016/j.1118cognition.2010.04.010.1119Meyer, A. S. (1996). Lexical access in phrase and sentence production: Results from1120picture–word interference experiments. Journal of Memory and Language, 35(4),1121477–496. http://dx.doi.org/10.1006/jmla.1996.0026.1122Meyer, A. S., & Schriefers, H. (1991). Phonological facilitation in picture-word1123interference experiments: Effects of stimulus onset asynchrony and types of1124interfering stimuli. Journal of Experimental Psychology: Learning, Memory, and1125Cognition, 17(6), 1146–1160. http://dx.doi.org/10.1037/0278-7393.17.6.1146.1126Michel, C. M., Koenig, T., & Brandeis, D. (2009). Electrical neuroimaging in the time1127domain. In C. M. Michel, T. Koenig, D. Brandeis, L. R. R. Gianotti, & J.1128Wackermann (Eds.), Electrical neuroimaging (pp. 111–144). Cambridge:1129Cambridge University Press.1130Miozzo, M., & Caramazza, A. (1999). The selection of lexical-syntactic features in1131noun phrase production: Evidence from the picture-word interference1132paradigm. Journal of Experimental Psychology: Learning Memory and Cognition,113325, 907–922.1134Munding, D., Dubarry, A.-S., & Alario, F.-X. (submitted for publication). On the1135cortical dynamics of word production: A review of the MEG evidence.1136Oppermann, F., Jescheniak, J. D., & Schriefers, H. (2010). Phonological advance1137planning in sentence production. Journal of Memory and Language, 63(4),1138526–540. http://dx.doi.org/10.1016/j.jml.2010.07.004.1139Pernet, C. R., Chauveau, N., Gaspar, C., & Rousselet, G. A. (2011). LIMO EEG: A toolbox1140for hierarchical LInear MOdeling of ElectroEncephaloGraphic data.1141Computational Intelligence and Neuroscience, 2011, e831409. http://dx.doi.org/114210.1155/2011/831409.1143Pylkkänen, L., Bemis, D. K., & Blanco Elorrieta, E. (2014). Building phrases in1144language production: An MEG study of simple composition. Cognition, 133(2),1145371–384. http://dx.doi.org/10.1016/j.cognition.2014.07.001.1146R core team (2014). R: A language and environment for statistical computing. Vienna,1147Austria: R Foundation for Statistical Computing. <http://www.R-project.org>.1148Riès, S., Janssen, N., Burle, B., & Alario, F.-X. (2013). Response-locked brain dynamics1149of word production. PLoS ONE, 8(3), e58197. http://dx.doi.org/10.1371/journal.1150pone.0058197.1151Roelofs, A. (1997). The WEAVER model of word-form encoding in speech1152production. Cognition, 64(3), 249–284. http://dx.doi.org/10.1016/S0010-02771153(97)00027-9.1154Schiller, N. O. (2008). The masked onset priming effect in picture naming. Cognition,1155106, 952–962. http://dx.doi.org/10.1016/j.cognition.2007.03.007.1156Schiller, N. O., & Caramazza, A. (2003). Grammatical feature selection in noun1157phrase production: Evidence from German and Dutch. Journal of Memory and1158Language, 48(1), 169–194. http://dx.doi.org/10.1016/S0749-596X(02)115900508-9.1160Schiller, N. O., Münte, T., Horemans, I., & Jansma, B. M. (2003). The influence of1161semantic and phonological factors on syntactic decisions: An event-related1162brain potential study. Psychophysiology, 40, 869–877. http://dx.doi.org/10.1111/11631469-8986.00105.1164Schnur, T. T., Costa, A., & Caramazza, A. (2006). Planning at the phonological level1165during sentence production. Journal of Psycholinguistic Research, 35(2), 189–213.1166http://dx.doi.org/10.1007/s10936-005-9011-6.1167Schriefers, H. (1992). Lexical access in the production of noun phrases. Cognition, 451168(1), 33–54. http://dx.doi.org/10.1016/0010-0277(92)90022-A.1169Schriefers, H. (1993). Syntactic processes in the production of noun phrases. Journal1170of Experimental Psychology: Learning, Memory, and Cognition, 19(4),1171841–850.1172Schriefers, H., de Ruijter, J., & Steigerwald, M. (1999). Parallelism in the production1173of noun phrases: Experiments and reaction time models. Journal of Experimental1174Psychology: Learning, Memory, and Cognition, 25(3), 702–720. http://dx.doi.org/117510.1037/0278-7393.25.3.702.1176Schriefers, H., Meyer, A. S., & Levelt, W. J. M. (1990). Exploring the time course of1177lexical access in language production: Picture-word interference studies.1178Journal of Memory and Language, 29(1), 86–102. http://dx.doi.org/10.1016/11790749-596X(90)90011-N.1180Schuhmann, T., Schiller, N. O., Goebel, R., & Sack, A. T. (2011). Speaking of which:1181Dissecting the neurocognitive network of language production in picture1182naming. Cerebral Cortex, bhr155. http://dx.doi.org/10.1093/cercor/bhr155.1183Smith, M., & Wheeldon, L. (1999). High level processing scope in spoken sentence1184production. Cognition, 73(3), 205–246. http://dx.doi.org/10.1016/S0010-02771185(99)00053-0.1186Starreveld, P. A. (2000). On the interpretation of onsets of auditory context effects in1187word production. Journal of Memory and Language, 42(4), 497–525. http://dx.1188doi.org/10.1006/jmla.1999.2693.1189Starreveld, P. A., & La Heij, W. (1995). Semantic interference, orthographic1190facilitation, and their interaction in naming tasks. Journal of Experimental1191Psychology: Learning, Memory, and Cognition, 21(3), 686–698. http://dx.doi.org/119210.1037/0278-7393.21.3.686.1193Starreveld, P. A., & La Heij, W. (1996). The locus of orthographic-phonological1194facilitation: Reply to Roelofs, Meyer, and Levelt (1996). Journal of Experimental1195Psychology: Learning, Memory, and Cognition, 22(1), 252–255. http://dx.doi.org/119610.1037/0278-7393.22.1.252.1197Strijkers, K., & Costa, A. (2011). Riding the lexical speedway: A critical review on the1198time course of lexical selection in speech production. Frontiers in Psychology, 2,1199356. http://dx.doi.org/10.3389/fpsyg.2011.00356.1200Townsend, J. T., & Wenger, M. J. (2004). The serial-parallel dilemma: A case study in1201a linkage of theory and method. Psychonomic Bulletin & Review, 11(3), 391–418.1202http://dx.doi.org/10.3758/BF03196588.
A. Bürki et al. / Cognition xxx (2015) xxx–xxx 9
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
1203 Van Berkum, J. J. A. (1997). Syntactic processes in speech production: The retrieval1204 of grammatical gender. Cognition, 64(2), 115–152. http://dx.doi.org/10.1016/1205 S0010-0277(97)00026-7.1206 van Turennout, M., Hagoort, P., & Brown, C. M. (1998). Brain activity during1207 speaking: From syntax to phonology in 40 milliseconds. Science, 280(5363),1208 572–574. http://dx.doi.org/10.1126/science.280.5363.572.
1209Wheeldon, L., Ohlson, N., Ashby, A., & Gator, S. (2013). Lexical availability and1210grammatical encoding scope during spoken sentence production. Quarterly1211Journal of Experimental Psychology, 66(8), 1653–1673. http://dx.doi.org/10.1080/121217470218.2012.754913.
1213
10 A. Bürki et al. / Cognition xxx (2015) xxx–xxx
COGNIT 3073 No. of Pages 10, Model 5G
5 September 2015
Please cite this article in press as: Bürki, A., et al. Sequential processing during noun phrase production. Cognition (2015), http://dx.doi.org/10.1016/j.cognition.2015.09.002
