Temporal Order Processing of Syllables in the Left Parietal Lobe

Discussion

As predicted, the group analysis revealed greater neural activation in the left posterior IPL, specifically the supramarginal gyrus (Fig. 1), during syllable order judgments compared with the control condition (gender identification of the speaker) using the same stimuli. These results support our hypothesis that the left posterior IPL is recruited during TOJs of auditory information such as speech. This is consistent with the role of this region in the Dual Stream model of speech that was proposed by Hickok and Poeppel (2007). They propose that the IPL serves as a translator between auditory speech and articulatory maps in the dorsal stream, including information such as temporal binding. For example, in a multisyllabic word, the translation of the phonological form to the articulatory form would include details regarding not only what syllables were involved, but also in what order they should occur. However, it is important to note that Hickok and Poeppel (2007) propose that integration of information across longer temporal windows (e.g., suprasegmentals) shows greater neural representation in the right hemisphere, whereas integration within shorter temporal windows (e.g., phonemes) shows more bilateral representation. This would suggest that TOJ tasks, such as the one used in the current study, should result in greater right hemisphere recruitment as opposed to left hemisphere activation. However, they also postulate that the left hemisphere may be preferentially recruited for acoustic material. We argue that our left lateralized finding for TOJ of syllables could be reflective of the strong reliance of auditory-motor integration in the left hemisphere during speech acquisition. However, the sequential processing performed by the left hemisphere does not seem to be strictly limited to acoustic material. For example, visual TOJ appears to lead left hemisphere activation even when controlled for stimulus properties (Smith et al., 2003) and duration for salient information (Davis et al., 2009). In addition, patients with damage to the posterior left hemisphere often have difficulty both with sequencing motor movements and detecting sequential errors in observed pantomime action (Weiss et al., 2008). This suggests that the left hemisphere may play a dominant role in serial order processing, regardless of modality.

In addition to temporal order processing, syllable order judgment must also rely on other basic cognitive processes, such as attention, response selection, auditory processing of frequency information, and syllable recognition/discrimination. Much of the activity observed during the syllable order task compared with baseline can be plausibly attributed to these obligatory processes. For example, there are substantial data to support the role of the left superior and middle temporal gyrus in speech perception (Fridriksson and Morrow, 2005; Price et al., 2005). Likewise, a growing body of evidence suggests that posterior inferior frontal regions (e.g., the inferior frontal gyrus and motor cortex) are recruited not only during speech production but also during speech perception (Watkins and Paus, 2004; Fridriksson et al., 2009).

Previous studies have observed posterior IPL activation during phonological processing (for review see Vigneau et al., 2006). We argue that our statistical contrast reveals that this region also plays a special role in auditory temporal order identification. Specifically, if the activity we observed was merely related to general phonological processing, this activity should also have been present in the task of syllable identification, in which no syllable order judgment is required. The results from our second group analysis demonstrate that the posterior IPL is important for TOJ beyond any basic role in phonological processing, at least at the syllable level, as greater activity was observed during the syllable-order condition compared with the syllable identification condition. Based on the conceptualization of the neural language network proposed by Shalom and Poeppel (2008), parietal and frontal areas have similar roles in processing (i.e., analyzing and synthesizing, respectively) compared with the temporal lobe (i.e., memorizing/retrieval). That is, both analyzing and synthesizing require consideration of the internal parts of the representations regardless of whether processing is occurring at a phonological, lexical, or syntactic level. The distinction would be that whereas the parietal regions separate the pieces, the frontal regions combine the pieces together. In support of this hypothesis, our findings implicate the IPL and the IFG in syllable order judgments. Thus, we speculate that the IPL provided temporal analysis of syllable order, whereas the IFG provided subvocal rehearsal in order for this analysis to occur. This is consistent with a number of studies that report neural recruitment of the IFG during phonological working memory tasks.

Alternative explanations are feasible and, therefore, warrant mentioning. It could be argued that, apart from TOJs, the processing of the multisyllabic stimuli requires greater demands on phonological processing (simply because of the number of syllables in the stimuli) and/or greater demands on working memory than the single-syllable stimuli. We attempted to control for the fundamental difference in phonological processing by factoring out activity that was associated with hearing the same stimuli, but responding based on gender (a nonphonological task). Furthermore, the temporal unfolding of the acoustic signal for any given syllable was the same across the two phonologically based tasks (i.e., single-syllable identification and. multisyllabic temporal order). Thus, neural activity related to phonological processing at the syllable and segmental level should be equivalent, particularly when compared with the respective baselines. However, the multisyllabic condition also required processing over a longer temporal window to determine syllable order (i.e., which syllable came first?). Therefore, the primary distinction in the higher-level contrast should be the presence or absence of TOJs. However, at the lower-level contrast, it is important to consider that gender identification would not necessitate the use of verbal working memory. Likewise, working memory demands would be minimal for single-syllable identification. According to Baddeley's theory of working memory (Baddeley, 1992, 2003), the “phonological loop,” which serves as a short-term memory store for auditory information, may be supported by the left IPL and superior temporal lobe. However, more recent works suggest that the temporal lobe is the neural hub for phonological processing rather than the parietal lobe (Rimol et al., 2005; DeLeon et al., 2007; Graves et al., 2007). In the current study, left temporal lobe activity was observed when syllable order was compared with baseline, but this activity did not survive the statistical threshold when compared with the syllable-identification condition. Although the absence of significant activation cannot be interpreted as proof that this particular region is not active, it is worthy to note that left IPL lobe and IFG activity was revealed for this contrast in the absence of left temporal lobe activity. Thus, the difference in the working memory load was likely revealed by the activation in the left IFG. Our findings suggest that phonological processing necessary for identification at the syllable level is supported primarily by the temporal lobe, whereas analysis and construction of syllable order are supported by the left IPL and IFG, respectively.

The fact that the posterior IPL showed significant activation for the “syllable order > syllable identification” contrast supports our hypothesis regarding its involvement in temporal order processing and is consistent with patient data (Wittmann et al., 2004). It seems reasonable to conclude that the posterior IPL plays an important role in processing the temporal order of syllables. Of particular interest, conduction aphasia has been associated with lesions in the IPL (Bartha and Benke, 2003; Quigg et al., 2006; Geldmacher et al., 2007). Most individuals with conduction aphasia present with frequent phonological errors (Ardila, 1992). Typically, these errors do not violate the phonotactic constraints of the persons' native language, suggesting that the phonological form of a word is assembled at the syllable level. Thus, even if a person was able to retrieve the correct lexical item, deficits in the temporal ordering of syllables could contribute to language difficulties by interfering with the translation of auditory speech into articulatory maps. This type of deficit could quite reasonably lead to phonological errors and repetition impairment, such as those seen in conduction aphasia.

In conclusion, we suggest that the left posterior IPL contributes to TOJs of syllables. Whereas the right posterior IPL seems to be associated with TOJs in the visual domain, this same region in the left hemisphere seems to be more important in TOJs in the auditory domain, as suggested by others (Wittmann et al., 2004; Battelli et al., 2007). This may be reflective of left lateralized preference for auditory-motor integration, in general, or for speech processing, more specifically. Although our study only used normal participants, the results broadly support the notion that damage to the left posterior IPL could directly impact speech and language abilities in a manner similar to that which is seen in conduction aphasia, by interfering with the temporal analysis of phonological representations to be sent to frontal regions for the assembly of motor plans. Although our study specifically focused on TOJs, we contend that this is only one of the functions supported by the left IPL for the translation between auditory speech and the articulatory code used for motor speech planning.