A central part of knowing a language is the ability to combine basic linguistic units to form com... more A central part of knowing a language is the ability to combine basic linguistic units to form complex representations. While our neurobiological understanding of how words combine into larger structures has significantly advanced in recent years, the combinatory operations that build words themselves remain unknown. Are complex words such as tombstone and starlet built with the same mechanisms that construct phrases from words, such as grey stone or bright star? Here we addressed this with two magnetoencephalography (MEG) experiments, which simultaneously varied demands associated with phrasal composition, and the processing of morphological complexity in compound and suffixed nouns. Replicating previous findings, we show that portions of the left anterior temporal lobe (LATL) are engaged in the combination of modifiers and monomorphemic nouns in phrases (e.g., brown rabbit). As regards compounding, we show that semantically transparent compounds (e.g., tombstone) also engage left anterior temporal cortex, though the spatiotemporal details of this effect differed from phrasal composition. Further, when a phrase was constructed from a modifier and a transparent compound (e.g., granite tombstone), the typical LATL phrasal composition response appeared at a delayed latency, which follows if an initial within-word operation (tomb þ stone) must take place before the combination of the compound with the preceding modifier (granite þ tombstone). In contrast to compounding, suffixation (i.e., star þ let) did not engage the LATL in any consistent way, suggesting a distinct processing route. Finally, our results suggest an intriguing generalization that morpho-orthographic complexity that does not recruit the LATL may block the engagement of the LATL in subsequent phrase building. In sum, our findings offer a detailed spatiotemporal characterization of the lowest level combinatory operations that ultimately feed the composition of full sentences.
This chapter examines the syntactic and semantic properties of heads, e.g. Voice, Appl, and littl... more This chapter examines the syntactic and semantic properties of heads, e.g. Voice, Appl, and little p, that add participants to events. Instead of assuming that such heads exist as distinct primitives in the functional lexicon, it is proposed that there is one such head, which can get different interpretations depending on how it is merged into the structure. The chapter’s approach attributes the relative uniformity of the expression of argument structure to the principles that interpret syntactic structure semantically; thus, syntax is truly autonomous, with the atoms of syntactic representations carrying no inherent semantic values. Once syntactic heads are absolved from the necessity of explicitly carrying certain features relevant to their interpretation, a sparse inventory of functional heads can be developed. The system is applied to a set of constructions that present distinct challenges to theories that demand a kind of transparent reflection of argument structure in underlyi...
Recent attempts to unify linguistic theory and brain science have grown out of recognition that a... more Recent attempts to unify linguistic theory and brain science have grown out of recognition that a proper understanding of language in the brain must reflect the steady advances in linguistic theory of the last forty years. The first Mind Articulation Project Symposium ...
“Cognition in Music” (Serafine, 1983) misrepresents the goals of a cognitive approach to music an... more “Cognition in Music” (Serafine, 1983) misrepresents the goals of a cognitive approach to music and the proper relation between cognitive science and music theory. As a linguist and cognitive scientist, I feel responsible both for the source of Serafine’s major misconception and for the correction of the error. Serafine’s line of reasoning follows a mistaken analogy with arguments from linguistic metatheory. Linguists maintain that a certain structuralist methodology may lead to hypotheses about speakers’ knowledge of language. Serafine appears to interpret this path to a cognitive theory as a restriction on the analysis of language or music to the uncovering of the cognitive processes humans use to deal with linguistic or musical material. She thus asserts a limiting view of what it is interesting or important to do with music. In this paper I hope to liberate music theorists from Serafine’s psychomusical constraints, but the considerations I will bring forward have wider application. Serafine’s main assertion appears at the beginning of her article (p. 120): “every formal analysis of an artwork should be an implicit description of the human cognitive processes that give rise to it, in composing or hearing or both.” Change ‘artwork’ to ‘sentence’ and one derives a constraint on linguistic theory; change ‘hearing’ to ‘reading’ and literary theory is implicated; change ‘hearing’ to ‘looking’ or ‘perceiving’ and one is in the realm of art criticism. If Serafine could support her assertion, it would have wide-ranging implications across a variety of disciplines, but the assertion finds no backing in her article. The linguistic analogy underlying Serafine’s reasoning is clear from the beginning of her article.’ What are the facts of music worthy of investigation?:
(1) re-prefixation creates the restitutive of the verb, not the repetitive. a. The walls in the h... more (1) re-prefixation creates the restitutive of the verb, not the repetitive. a. The walls in the house were green long before they purchased it; they plan to re-paint them (white) as soon as they can. b. The door of the cabinet was built open, and John closed it for the first time when he brought it home. Mary then reopened the door. (2) As has been observed by many, re-Verb doesn't refer to the repetition of the activity described by the whole VP but rather describes the re-occurrence of the state within the VP. That's why re-, unlike "again," is incompatible with simple activity predicates: a. *John re-smiled b. John smiled again. Wrong view, Lieber (2004, 147), "what we mean when we say that re-means 'to do again' is that re-induces an iteration of the action denoted by the verb." c. This door was built open and hasn't been touched since. d. I just closed the door and reopened it. e. I.e., I opened it for the first time, but restored it to the open state it was built in. Lieber: "[re-] also does not attach to verbs which imply a result which cannot be reversed. For example, it is impossible to *reeat the apple…"
Standard practice in linguistics often obscures the connection between theory and data, leading s... more Standard practice in linguistics often obscures the connection between theory and data, leading some to the conclusion that generative linguistics could not serve as the basis for a cognitive neuroscience of language. Here the founda-tions and methodology of generative grammar are clarified with the goal of explaining how generative theory already functions as a reasonable source of hypotheses about the representation and computation of language in the mind and brain. The claims of generative theory, as exemplified, e.g., within Chom-sky’s (2000) Minimalist Program, are contrasted with those of theories en-dorsing parallel architectures with independent systems of generative phonol-ogy, syntax and semantics. The single generative engine within Minimalist ap-proaches rejects dual routes to linguistic representations, including possible extra-syntactic strategies for semantic structure-building. Clarification of the implications of this property of generative theory undermines the fou...
The striking parallels across languages in the syntactic expression of “argument structure” broad... more The striking parallels across languages in the syntactic expression of “argument structure” broadly construed has led within generative grammar, at least since Generative Semantics, to generative theories that demand a kind of transparent reflection of argument structure in
Abstract—We have developed a method suitable for recon-structing spatio-temporal activities of ne... more Abstract—We have developed a method suitable for recon-structing spatio-temporal activities of neural sources by using magnetoencephalogram (MEG) data. The method extends the adaptive beamformer technique originally proposed by Borgiotti and Kaplan to incorporate the vector beamformer formulation in which a set of three weight vectors are used to detect the source activity in three orthogonal directions. The weight vectors of the vector-extended version of the Borgiotti–Kaplan beamformer are then projected onto the signal subspace of the measurement covariance matrix to obtain the final form of the proposed beamformer’s weight vectors. Our numerical experiments show that both spatial resolution and output signal-to-noise ratio of the proposed beamformer are significantly higher than those of the minimum-variance-based vector beamformer used in previous investigations. We also applied the proposed beamformer to two sets of auditory-evoked MEG data, and the results clearly demon-strat...
We discuss theoretical approaches to blocking effects, with particular emphasis on cases in which... more We discuss theoretical approaches to blocking effects, with particular emphasis on cases in which words appear to block phrases (and perhaps vice versa). These approaches share at least one intuition: that syntactic and semantic features create possible ‘‘cells’ ’ or slots in which particu-lar items can appear, and that blocking occurs when one such cell is occupied by one form as opposed to another. Accounts of blocking differ along two primary dimensions: the size of the objects that com-pete with one another (morphemes, words, phrases, sentences); and whether or not ungrammatical forms are taken into consideration in determining the correct output (relatedly, whether otherwise well-formed objects are marked ungrammatical by competition). We argue that blocking in the sense of competition for the expression of syntactic or semantic features is limited to insertion of the phonological expo-nents of such features (the Vocabulary items of Distributed Morphol-ogy) at terminal nodes fr...
■ Many previous studies have shown that predictable words are read faster and lead to reduced neu... more ■ Many previous studies have shown that predictable words are read faster and lead to reduced neural activation, consistent with a model of reading in which words are activated in advance of being encountered. The nature of such preactivation, however, has typ-ically been studied indirectly through its subsequent effect on word recognition. Here, we use magnetoencephalography to study the dynamics of prediction within serially presented adjective–noun phrases, beginning at the point at which the pre-dictive information is first available to the reader. Using corpus transitional probability to estimate the predictability of a noun, we found an increase in activity in the left middle temporal gyrus in response to the presentation of highly predictive adjectives (i.e., adjectives that license a strong noun prediction). Moreover, we found that adjective predictivity and expected noun frequency interacted, such that the response to the highly predictive adjec-tives (e.g., stainless) was ...
Previous magnetoencephalography (MEG) studies reported that morphologically complex words are dec... more Previous magnetoencephalography (MEG) studies reported that morphologically complex words are decomposed into morphemes around 170 ms after the onset of visual stimuli (M170) in the left fusiform gyrus and inferior temporal gyrus (FG/ITG) (Solomyak and Marantz 2010). Moreover, another MEG study found that transition probability (TP) between morphemes was correlated with the amplitude of the M170 (Lewis et al. 2011). As these studies targeted English, in which morphological boundaries are always letter boundaries, it is difficult to examine whether the M170 is modulated by morphological TP or TP between letters. In the present study, we targeted the Japanese language since it uses kanji that basically represent verbal roots, as well as kana, each of which represents a mora. For example, in the Japanese verb mawa-s-u (turn), the verbal root mawa-and remaining parts -u are represented by a single kanji and kana, respectively. Here, a voice morphemes and a tense morpheme-u are written as a single kana. In the present MEG experiment, we compared the effects of morphTP (e.g., TP between mawa-and-s) and letterTP (e.g., TP between mawa-and-s-u) on the left FG/ITG activation. We recruited 22 right-handed native speakers of Japanese (nine males, 35.5 ± 7.3 yrs.). We used 112 Japanese verbs for each of intransitives, transitives, intransitive-causatives, and transitive-causatives, as well as the same number of nonwords (total 896 stimuli). The participants performed a visual lexical decision task. A 157-channel whole-head MEG system (Kanazawa Institute of Technology, Japan) was used. For the MEG data analyses, we used spatiotemporal cluster permutation tests as implemented in MEG-Python (Gramfort et al. 2013) and Eelbrain (https://pythonhosted.org/eelbrain/) (see Gwilliams et al. 2016 for the details). As our primary target was the M170, the region of interest was anatomically defined as the left FG/ITG and the analysis time window was restricted to 50-250 ms after word onset. Behavioral data showed that the mean accuracies of the lexical decision task of intransitives, transitives, intransitive-causatives, and transitive-causatives were higher than 90%, indicating that the participants correctly performed this task. A two-way repeatedmeasures analysis of variance (rANOVA) with the transitivity × causativeness for the accuracy showed significant main effects of transitivity and causativeness, as well as a significant interaction (transitivity: p = 0.046; causativeness: p = 0.0001; interaction: p = 0.042). The RTs also showed a significant main effect of causativeness (p < 0.0001), but a main effect of transitivity and an interaction were not significant (transitivity: p = 0.39; interaction: p = 0.34). These results suggest that the causative conditions had higher processing loads. For the MEG data, we first examined whether the causative conditions elicited larger activation in the left FG/ITG. We found that the causative verbs showed a significantly larger activation in this region (corrected p = 0.044) (Fig. 1A). We further examined whether the morphTP, as well as the length of verbs, modulated the left FG/ITG activation, by using spatiotemporal cluster regression analyses. We discovered significantly negative correlation of the morphTP in the posterior part of the left FG/ITG (corrected p = 0.033) (Figure 1B). We also found significantly positive correlation of the length of verbs in the left FG/ITG (Figure 1C). Finally, we examined whether the letterTP modulated the left FG/ITG activation. In contrast to the morphTP, we did not find any significant correlation. These results demonstrated that morphologically complex verbs in Japanese are indeed decomposed into morphemes, but not into letters, similar to morphologically complex words in English examined in the previous MEG studies.
Dans une grammaire de l'anglais appliquant correctement la convention X-barre, la categorie S... more Dans une grammaire de l'anglais appliquant correctement la convention X-barre, la categorie S doit etre remplacee par V.
A central part of knowing a language is the ability to combine basic linguistic units to form com... more A central part of knowing a language is the ability to combine basic linguistic units to form complex representations. While our neurobiological understanding of how words combine into larger structures has significantly advanced in recent years, the combinatory operations that build words themselves remain unknown. Are complex words such as tombstone and starlet built with the same mechanisms that construct phrases from words, such as grey stone or bright star? Here we addressed this with two magnetoencephalography (MEG) experiments, which simultaneously varied demands associated with phrasal composition, and the processing of morphological complexity in compound and suffixed nouns. Replicating previous findings, we show that portions of the left anterior temporal lobe (LATL) are engaged in the combination of modifiers and monomorphemic nouns in phrases (e.g., brown rabbit). As regards compounding, we show that semantically transparent compounds (e.g., tombstone) also engage left anterior temporal cortex, though the spatiotemporal details of this effect differed from phrasal composition. Further, when a phrase was constructed from a modifier and a transparent compound (e.g., granite tombstone), the typical LATL phrasal composition response appeared at a delayed latency, which follows if an initial within-word operation (tomb þ stone) must take place before the combination of the compound with the preceding modifier (granite þ tombstone). In contrast to compounding, suffixation (i.e., star þ let) did not engage the LATL in any consistent way, suggesting a distinct processing route. Finally, our results suggest an intriguing generalization that morpho-orthographic complexity that does not recruit the LATL may block the engagement of the LATL in subsequent phrase building. In sum, our findings offer a detailed spatiotemporal characterization of the lowest level combinatory operations that ultimately feed the composition of full sentences.
This chapter examines the syntactic and semantic properties of heads, e.g. Voice, Appl, and littl... more This chapter examines the syntactic and semantic properties of heads, e.g. Voice, Appl, and little p, that add participants to events. Instead of assuming that such heads exist as distinct primitives in the functional lexicon, it is proposed that there is one such head, which can get different interpretations depending on how it is merged into the structure. The chapter’s approach attributes the relative uniformity of the expression of argument structure to the principles that interpret syntactic structure semantically; thus, syntax is truly autonomous, with the atoms of syntactic representations carrying no inherent semantic values. Once syntactic heads are absolved from the necessity of explicitly carrying certain features relevant to their interpretation, a sparse inventory of functional heads can be developed. The system is applied to a set of constructions that present distinct challenges to theories that demand a kind of transparent reflection of argument structure in underlyi...
Recent attempts to unify linguistic theory and brain science have grown out of recognition that a... more Recent attempts to unify linguistic theory and brain science have grown out of recognition that a proper understanding of language in the brain must reflect the steady advances in linguistic theory of the last forty years. The first Mind Articulation Project Symposium ...
“Cognition in Music” (Serafine, 1983) misrepresents the goals of a cognitive approach to music an... more “Cognition in Music” (Serafine, 1983) misrepresents the goals of a cognitive approach to music and the proper relation between cognitive science and music theory. As a linguist and cognitive scientist, I feel responsible both for the source of Serafine’s major misconception and for the correction of the error. Serafine’s line of reasoning follows a mistaken analogy with arguments from linguistic metatheory. Linguists maintain that a certain structuralist methodology may lead to hypotheses about speakers’ knowledge of language. Serafine appears to interpret this path to a cognitive theory as a restriction on the analysis of language or music to the uncovering of the cognitive processes humans use to deal with linguistic or musical material. She thus asserts a limiting view of what it is interesting or important to do with music. In this paper I hope to liberate music theorists from Serafine’s psychomusical constraints, but the considerations I will bring forward have wider application. Serafine’s main assertion appears at the beginning of her article (p. 120): “every formal analysis of an artwork should be an implicit description of the human cognitive processes that give rise to it, in composing or hearing or both.” Change ‘artwork’ to ‘sentence’ and one derives a constraint on linguistic theory; change ‘hearing’ to ‘reading’ and literary theory is implicated; change ‘hearing’ to ‘looking’ or ‘perceiving’ and one is in the realm of art criticism. If Serafine could support her assertion, it would have wide-ranging implications across a variety of disciplines, but the assertion finds no backing in her article. The linguistic analogy underlying Serafine’s reasoning is clear from the beginning of her article.’ What are the facts of music worthy of investigation?:
(1) re-prefixation creates the restitutive of the verb, not the repetitive. a. The walls in the h... more (1) re-prefixation creates the restitutive of the verb, not the repetitive. a. The walls in the house were green long before they purchased it; they plan to re-paint them (white) as soon as they can. b. The door of the cabinet was built open, and John closed it for the first time when he brought it home. Mary then reopened the door. (2) As has been observed by many, re-Verb doesn't refer to the repetition of the activity described by the whole VP but rather describes the re-occurrence of the state within the VP. That's why re-, unlike "again," is incompatible with simple activity predicates: a. *John re-smiled b. John smiled again. Wrong view, Lieber (2004, 147), "what we mean when we say that re-means 'to do again' is that re-induces an iteration of the action denoted by the verb." c. This door was built open and hasn't been touched since. d. I just closed the door and reopened it. e. I.e., I opened it for the first time, but restored it to the open state it was built in. Lieber: "[re-] also does not attach to verbs which imply a result which cannot be reversed. For example, it is impossible to *reeat the apple…"
Standard practice in linguistics often obscures the connection between theory and data, leading s... more Standard practice in linguistics often obscures the connection between theory and data, leading some to the conclusion that generative linguistics could not serve as the basis for a cognitive neuroscience of language. Here the founda-tions and methodology of generative grammar are clarified with the goal of explaining how generative theory already functions as a reasonable source of hypotheses about the representation and computation of language in the mind and brain. The claims of generative theory, as exemplified, e.g., within Chom-sky’s (2000) Minimalist Program, are contrasted with those of theories en-dorsing parallel architectures with independent systems of generative phonol-ogy, syntax and semantics. The single generative engine within Minimalist ap-proaches rejects dual routes to linguistic representations, including possible extra-syntactic strategies for semantic structure-building. Clarification of the implications of this property of generative theory undermines the fou...
The striking parallels across languages in the syntactic expression of “argument structure” broad... more The striking parallels across languages in the syntactic expression of “argument structure” broadly construed has led within generative grammar, at least since Generative Semantics, to generative theories that demand a kind of transparent reflection of argument structure in
Abstract—We have developed a method suitable for recon-structing spatio-temporal activities of ne... more Abstract—We have developed a method suitable for recon-structing spatio-temporal activities of neural sources by using magnetoencephalogram (MEG) data. The method extends the adaptive beamformer technique originally proposed by Borgiotti and Kaplan to incorporate the vector beamformer formulation in which a set of three weight vectors are used to detect the source activity in three orthogonal directions. The weight vectors of the vector-extended version of the Borgiotti–Kaplan beamformer are then projected onto the signal subspace of the measurement covariance matrix to obtain the final form of the proposed beamformer’s weight vectors. Our numerical experiments show that both spatial resolution and output signal-to-noise ratio of the proposed beamformer are significantly higher than those of the minimum-variance-based vector beamformer used in previous investigations. We also applied the proposed beamformer to two sets of auditory-evoked MEG data, and the results clearly demon-strat...
We discuss theoretical approaches to blocking effects, with particular emphasis on cases in which... more We discuss theoretical approaches to blocking effects, with particular emphasis on cases in which words appear to block phrases (and perhaps vice versa). These approaches share at least one intuition: that syntactic and semantic features create possible ‘‘cells’ ’ or slots in which particu-lar items can appear, and that blocking occurs when one such cell is occupied by one form as opposed to another. Accounts of blocking differ along two primary dimensions: the size of the objects that com-pete with one another (morphemes, words, phrases, sentences); and whether or not ungrammatical forms are taken into consideration in determining the correct output (relatedly, whether otherwise well-formed objects are marked ungrammatical by competition). We argue that blocking in the sense of competition for the expression of syntactic or semantic features is limited to insertion of the phonological expo-nents of such features (the Vocabulary items of Distributed Morphol-ogy) at terminal nodes fr...
■ Many previous studies have shown that predictable words are read faster and lead to reduced neu... more ■ Many previous studies have shown that predictable words are read faster and lead to reduced neural activation, consistent with a model of reading in which words are activated in advance of being encountered. The nature of such preactivation, however, has typ-ically been studied indirectly through its subsequent effect on word recognition. Here, we use magnetoencephalography to study the dynamics of prediction within serially presented adjective–noun phrases, beginning at the point at which the pre-dictive information is first available to the reader. Using corpus transitional probability to estimate the predictability of a noun, we found an increase in activity in the left middle temporal gyrus in response to the presentation of highly predictive adjectives (i.e., adjectives that license a strong noun prediction). Moreover, we found that adjective predictivity and expected noun frequency interacted, such that the response to the highly predictive adjec-tives (e.g., stainless) was ...
Previous magnetoencephalography (MEG) studies reported that morphologically complex words are dec... more Previous magnetoencephalography (MEG) studies reported that morphologically complex words are decomposed into morphemes around 170 ms after the onset of visual stimuli (M170) in the left fusiform gyrus and inferior temporal gyrus (FG/ITG) (Solomyak and Marantz 2010). Moreover, another MEG study found that transition probability (TP) between morphemes was correlated with the amplitude of the M170 (Lewis et al. 2011). As these studies targeted English, in which morphological boundaries are always letter boundaries, it is difficult to examine whether the M170 is modulated by morphological TP or TP between letters. In the present study, we targeted the Japanese language since it uses kanji that basically represent verbal roots, as well as kana, each of which represents a mora. For example, in the Japanese verb mawa-s-u (turn), the verbal root mawa-and remaining parts -u are represented by a single kanji and kana, respectively. Here, a voice morphemes and a tense morpheme-u are written as a single kana. In the present MEG experiment, we compared the effects of morphTP (e.g., TP between mawa-and-s) and letterTP (e.g., TP between mawa-and-s-u) on the left FG/ITG activation. We recruited 22 right-handed native speakers of Japanese (nine males, 35.5 ± 7.3 yrs.). We used 112 Japanese verbs for each of intransitives, transitives, intransitive-causatives, and transitive-causatives, as well as the same number of nonwords (total 896 stimuli). The participants performed a visual lexical decision task. A 157-channel whole-head MEG system (Kanazawa Institute of Technology, Japan) was used. For the MEG data analyses, we used spatiotemporal cluster permutation tests as implemented in MEG-Python (Gramfort et al. 2013) and Eelbrain (https://pythonhosted.org/eelbrain/) (see Gwilliams et al. 2016 for the details). As our primary target was the M170, the region of interest was anatomically defined as the left FG/ITG and the analysis time window was restricted to 50-250 ms after word onset. Behavioral data showed that the mean accuracies of the lexical decision task of intransitives, transitives, intransitive-causatives, and transitive-causatives were higher than 90%, indicating that the participants correctly performed this task. A two-way repeatedmeasures analysis of variance (rANOVA) with the transitivity × causativeness for the accuracy showed significant main effects of transitivity and causativeness, as well as a significant interaction (transitivity: p = 0.046; causativeness: p = 0.0001; interaction: p = 0.042). The RTs also showed a significant main effect of causativeness (p < 0.0001), but a main effect of transitivity and an interaction were not significant (transitivity: p = 0.39; interaction: p = 0.34). These results suggest that the causative conditions had higher processing loads. For the MEG data, we first examined whether the causative conditions elicited larger activation in the left FG/ITG. We found that the causative verbs showed a significantly larger activation in this region (corrected p = 0.044) (Fig. 1A). We further examined whether the morphTP, as well as the length of verbs, modulated the left FG/ITG activation, by using spatiotemporal cluster regression analyses. We discovered significantly negative correlation of the morphTP in the posterior part of the left FG/ITG (corrected p = 0.033) (Figure 1B). We also found significantly positive correlation of the length of verbs in the left FG/ITG (Figure 1C). Finally, we examined whether the letterTP modulated the left FG/ITG activation. In contrast to the morphTP, we did not find any significant correlation. These results demonstrated that morphologically complex verbs in Japanese are indeed decomposed into morphemes, but not into letters, similar to morphologically complex words in English examined in the previous MEG studies.
Dans une grammaire de l'anglais appliquant correctement la convention X-barre, la categorie S... more Dans une grammaire de l'anglais appliquant correctement la convention X-barre, la categorie S doit etre remplacee par V.
Uploads
Papers by Alec Marantz