Neural architecture of human language: Hierarchical structure building is independent from working memory

When a sentence loses semantics: Selective involvement of a left anterior temporal subregion in semantic processing

“…To our surprise, we found no main effect of SYNTAX when we compared the OSV with the SOV conditions in the whole‐brain analysis, despite repeated observations of activation in Broca's area for word order changes in Japanese (Iwabuchi et al., 2019, 2020; Kim et al., 2009; Kinno et al., 2008; Koizumi et al., 2012) and in German (Bornkessel et al., 2005; Friederici et al., 2006; Makuuchi et al., 2013; Meyer et al., 2012). This discrepancy may be accounted for by the specific construction of sentences in the current study, in which the subject noun phrase was made “heavier” by adding a modifier.…”

“…We expected a more increased activity in BA 44 in response to OSV sentences with a heavy subject noun phrase (hS) than with a simple subject phrase. However, this maneuver did not yield the expected effect; indeed, a previous study found that the activity in the left BA 44 was not higher for O‐hS‐V sentences than for hS‐O‐V or heavy object (hO)‐S‐V sentences (Iwabuchi et al., 2019). In fact, this area was more activated for hO‐S‐V than for hS‐O‐V, S‐hO‐V, and O‐hS‐V sentences (Iwabuchi et al., 2019).…”

“…The time series data were adjusted for the F‐contrast for the effect of interest (i.e., the contrast for activation in at least one of the four experimental conditions; p < 0.001 uncorrected for multiple comparisons) by removing variance caused by effects of no interest such as head motion parameters. We used the same technique as in our previous studies (Iwabuchi et al., 2019; Makuuchi et al., 2013) to calculate the trial time courses (TTCs) for each participant's preprocessed (realigned, slice time corrected, coregistered, normalized, and upsampled to every 0.5 s) time series data. For each VOI, we defined the time window for statistical testing based on peak latency, and carried out two‐way ANOVAs with within‐subjects factors of SEMANTICS (NAT and SPH) and SYNTAX (SOV and OSV).…”

“…A growing body of evidence supports the view that the ATL represents domain‐general semantic knowledge (Lambon Ralph et al., 2009; Rogers et al., 2006) and integrates semantic information (Jung‐Beeman, 2005; Wilson et al., 2014), thereby serving as a semantic hub (Lambon Ralph et al., 2017; Patterson et al., 2007). However, some studies have implicated the left ATL in syntactic structure building (Bemis & Pylkkänen, 2011; Brennan et al., 2012; Humphries et al., 2005, 2006; Mazoyer et al., 1993; Noppeney & Price, 2004; Rogalsky & Hickok, 2009), in contrast with the view that the left Brodmann area (BA) 44 and the posterior middle temporal gyrus (pMTG) are brain regions responsible for hierarchical syntactic structure building (Ben‐Shachar et al., 2003, 2004; Caplan et al., 2008; Dapretto & Bookheimer, 1999; Friederici et al., 2003; Iwabuchi et al., 2019; Just et al., 1996; Makuuchi et al., 2009, 2012, 2013; Tyler et al., 2013; Zaccarella & Friederici, 2015; Zaccarella et al., 2015, 2017). Neuropsychological studies have also provided mixed results.…”

“…We also manipulated the syntactic processing cost using Japanese sentences with either the basic Subject‐Object‐Verb (SOV) order or a changed Object‐Subject‐Verb (OSV) order to identify the neural loci of syntactic structure building. Changing the word order from SOV to OSV adds one more hierarchy to the linguistic tree structure, thereby increasing the cost of hierarchical structure building (Iwabuchi et al., 2019). A 2 × 2 factorial design crossing SEMANTICS (natural sentences [NAT], sentences with placeholders [SPH]) with SYNTAX (SOV, OSV) allowed us to disentangle the neural correlates of semantic processing from those of syntactic processing.…”