Skip to main content
Fachgebiete
Automobil + Motoren Bauwesen + Immobilien Business IT + Informatik Elektrotechnik + Elektronik Energie + Nachhaltigkeit Finance + Banking Management + Führung Marketing + Vertrieb Maschinenbau + Werkstoffe Versicherung + Risiko
DE
EN
Bücher
Zeitschriften
Themenseiten
Organisationspsychologie Projektmanagement Marketing Smart Manufacturing
Weitere Formate
Podcasts Webinare Technik Webinare Wirtschaft Kongresse Veranstaltungskalender Awards
Jetzt Einzelzugang starten
Zugang für Unternehmen
Referenzkunden
SLX-Digitalkonferenz: Zukunftswerkstatt 2024

Mehr Umsatz mit Top-Kontakten

Am 7. Mai erfahren Sie, wie Vertriebsteams Leadgenerierung, Leadnurturing und Leadstrategien effizient steuern können.
Erweiterte Suche
Anmelden
nach oben
Erschienen in:
Introduction Kapitel lesen Erstes Kapitel lesen

2018 | OriginalPaper | Buchkapitel

5. Syntax and Semantics: Dichotomy Versus Integration

verfasst von : Bernard Scott

Erschienen in: Translation, Brains and the Computer

Verlag: Springer International Publishing

Einloggen

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config
loading …

Abstract

This chapter deals in a general way with the linguistically-oriented mechanisms in the brain that enable it to decode an input stream of individual words into unambiguous, meaningful sentences. Do we know what the mechanism is that tells the brain that a string of words like “He wants an answer” constitutes an intelligible expression and that a gobbledygook string like “Answer he an wants” does not? Is the mechanism behind these judgments initially syntactic or semantic? We examine various views as to which of these mechanisms most accounts for the brain’s handling of language, particularly in the case where an input stream of words must then lead to its translation. And we try to understand how it happens that all of this is done in the brain without complexity effects. The intent here is not academic but practical, namely, to try and glean from among the competing views of cerebral language processing something of value for MT itself. Does the brain’s way of handling syntax and semantics have something to teach MT? In turn, we consider whether MT modeling itself might not have something to offer neuroscience on its open questions regarding language and the brain.

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

  • über 102.000 Bücher
  • über 537 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Maschinenbau + Werkstoffe
  • Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

  • über 67.000 Bücher
  • über 390 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Nachhaltigkeit
  • Maschinenbau + Werkstoffe




 

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

  • über 67.000 Bücher
  • über 340 Zeitschriften

aus folgenden Fachgebieten:

  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Versicherung + Risiko




Jetzt Wissensvorsprung sichern!

Jetzt informieren
Vorheriges Kapitel Language and Complexity: Neurolinguistic Perspectives
Nächstes Kapitel Logos Model: Design and Performance
Anhänge
Nur mit Berechtigung zugänglich
Fußnoten
1
We obviously are only speaking of written language here.
 
2
In Aspects of the Theory of Syntax (1965, p. 162), Chomsky rightly acknowledges that the practice of equating semantics with lexicon “isn’t satisfactory.” The meaning of words in a sentence can be extralexical. For our purposes in this chapter, however, we will keep to this common practice of equating semantics with literal words of the lexicon.
 
3
Ullman (2004, p. 233): “Language depends upon a memorized ‘mental lexicon’ and a computational ‘mental grammar,’” which latter “is not available to conscious access” [i.e., constitutes tacit, implicit knowledge].
 
4
Chomsky (1977, p. 138). He expresses skepticism that syntax is ever based on semantics. See Postscript 5-A for a discussion of this point.
 
5
Schmid (1996) notes the kinship between Langacker’s Cognitive Grammar and the implicit semantico-syntactic grammar of Logos Model.
 
6
Chomsky (1957, p. 15):“The notion of ‘grammatical’ cannot be identified with ‘meaningful’ or ‘significant’ in any semantic sense.”
 
7
Hagoort (2015, p. 406). It is hard to imagine MT developers making such a statement.
 
8
Ever since the pioneering work of Reber (1967), many of these neuropsychological studies have involved artificial rather than natural languages. This distinction must surely have some bearing on the relevance of these studies to natural language processes of the brain.
 
9
Bastiaansen and Hagoort (2015) claim that brain frequencies for semantics and syntax judgments were functionally distinguishable during grammaticality testing. But one questions the pertinence of data derived from tests of artificial language to our understanding of natural language processes.
 
10
One of the authors of this 2015 study is Angela Friederici, who a decade earlier had worked within the generativist framework (Optiz and Friederici 2003). Angela Friederici is neuropsychology director of the Max Planck Institute for Human Cognitive and Brain Science in Leipzig. She has authored or co-authored over 400 papers, at least one including co-author Noam Chomsky.
 
11
Our purpose in these remarks is not to denigrate this particular study. It was well done by highly respected neuroscientists, and certainly the exercise is interesting.
 
12
The researchers acknowledge that the participants, who were all native German speakers, may have made correct judgments about BROCANTO simply by virtue of unconscious influence from their native language. Brief German sentences in the present tense are very similar syntactically to the SVO sentences of BROCANTO. If that influence should really be the case, then of the two hypotheses about learning postulated by the authors, a rule-based hypothesis and an associative, similarity-based hypothesis, their study would actually seem to demonstrate the latter.
 
13
Connectionists McClelland and Patterson (2002) contend that rules of grammar do not have neuronal counterparts in the brain, and that the mechanism for stringing words into sentences is probability. See also Kempen and Harbusch (2003).
 
14
Neuropsychologist Pulvermüller (2010) acknowledges that what tests like these prove must remain inconclusive (p. 170): “… it is difficult to decide whether the brain response to an ungrammatical string reflects its syntactic features or rather the conditional probability with which words follow each other. As every ungrammatical string is usually also a very rarely occurring item, it is also possible that the degree of expectedness – or string probability – is reflected.”
 
15
Grodzinsky and Friederici (2006), Optiz and Friederci (2007), Bonhage et al. (2015), which includes Friederici as co-author; Optiz and Hofmann 2015, 77): “… results indicate that rule- and similarity-based mechanisms concur during AGL [artificial grammar learning].” In Mueller et al. (2009, web), co-author Friederici now subscribes to the finding that “present results support learning theories that assume a statistical learning mechanism rather than a rule-based extraction mechanism as an initial acquisition stage.”
 
16
Hauser et al. (2012, web): Researchers found that “Increasing [sentence] complexity was linked to increasing activity in the left inferior pars opercularis [Brodmann Area 44 in the Broca region] whilst ungrammaticality items evoked increased activity in the left operculum [in the broader Broca area.]”
 
17
One wishes the researchers had attempted to correlate the type of knowledge participants preferred with an individual’s inherent language skills, e.g., that the more gifted participants might naturally rely more on similarity, etc., a circumstance we suggested in Chap. 3 and found documentation for in Chap. 4 (Mårtensson et al. 2012).
 
18
Hauser et al. (2012, web): “…. rule and similarity knowledge work in parallel and compete in processing-efficiency, leading to an initial superiority of similarity-based classification, and a subsequent dominance of rule-based processes, once a critical amount of abstract knowledge of adjacent dependencies was acquired.”
 
19
It is interesting that researchers of studies like these generally “propose” rather than find the conclusions they draw, a tacit acknowledgment that the empirical evidence they’ve uncovered still has to be interpreted. Moreover, in all this, one must not lose sight of the fact that BROCANTO is a semantics-free, ambiguity-free, artificial language. Its explanatory value for the brain’s natural language processes may legitimately be questioned.
 
20
The study’s methodology entailed fMRI tracking of anticipatory eye movements as participants read the various kinds of sentences. A study by Rouder and Ratcliff (2006) that compared rule-based theories and similarity-based (exemplar-based) theories found that individuals use rules when new items are confusable or problematic in some way, and use exemplars when they are distinct.
 
21
Optiz and Hofmann 2015, 77): “… results indicate that rule- and similarity-based mechanisms concur during AGL [artificial grammar learning].”
 
22
E.g., Nakazawa et al. (2002, web): “The ability to retrieve complete memories on the basis of incomplete sets of cues, is a crucial function of biological memory systems. The extensive recurrent connectivity of the CA3 area of hippocampus has led to suggestions that it might provide this function.” Kumaran et al. (2016, web): “Pattern completion: recurrent connections from the active CA3 neurons onto other active CA3 neurons are strengthened during the experience, such that if a subset of the same neurons later becomes active, the rest of the pattern will be reactivated.”
 
23
N400 is a negative-going event-related potential (ERP) that peaks at around 400 ms after the onset of a content word. It was discovered by Friederici in 1980 when negative N400 amplitudes were found to be elicited by semantically incongruous words, e.g., He spread the warm bread with socks. P600 is a positive ERP response elicited by syntactic ambiguities or by violations in agreement or word order (Kuperberg 2007, 24f).
 
24
Yaxu Zhang et al. (2010, 765). Researchers conclude: “… we found that semantic interpretation proceeded despite the impossibility of a well-formed syntactic analysis.”
 
25
Kuperberg (2007) argues that syntax-related P600 responses can also be trigged by semantic factors such as verb-argument violations and violations in associative semantic relationships, therefore raising doubt to the attribution of P600 to syntactic violations alone. According to Kuperberg, P600 responses to both syntactic and semantic factors suggest that normal language comprehension proceeds dynamically along two complementary streams, one semantic and memory-based, the other syntactic and rule-based, neither of which are purely one or the other. Moreover, the two streams compete, which he says helps explain individual differences in mode of comprehension.
 
26
Chomsky (2000, 69) himself envisions generative grammar as a means for mapping input preparatory to its interpretation. (See Postscript 5-D).
 
27
Pothos (2007) argues that semantic patterns rather than syntactic rules form the deeper basis of linguistic learning, and he connects these patterns with hippocampal exemplars. We contend that these exemplar-patterns are semantico-syntactic, allowing them to function at an abstract level semantically as well as syntactically. We argue that semantico-syntactic abstraction accounts for the brain’s effectiveness and efficiency in the use and processing of language, an effectiveness Logos Model feebly attempts to simulate.
 
28
See Part Two for a full presentation of SAL on the open parts of speech.
 
29
The term “function” in construction grammar denotes meaning or semantic intent.
 
30
Note how this combination of exposure together with cognitive processes seems to comport with the empirical, analytical and analogical operations ascribed to language acquisition in Chap. 3. Relative to this, see Postscript 5-F for a construction grammarian’s discussion of a child’s acquisition of language through an initial exposure to patterns of language and a subsequent mechanism of generalization upon these patterns.
 
31
Note how this comports with Logos Model, where the store of pattern-rules constitutes Logos Model’s working knowledge of a given source language. See Postscript 5-G for further discussion.
 
32
There is little similarity between SAL and Fillmore’s subsequently developed Frame Semantics that has now become a part of construction grammar. Just the opposite was the case with Fillmore’s earlier work with Case Grammar, where similarities clearly exist between it and SAL. SAL however evolved inductively from issues relating to sentence analysis in MT, independently of Case Grammar. Moreover, Case Grammar’s syntactic subdivisions never came to be developed into a full-blown, second-order language anywhere near resembling the semantico-syntactic elaborations of SAL.
 
33
The neurolinguist Pulvermüller (2010, 2013)) holds that the tendency of most grammatical frameworks to list syntactic rules and lexicon as separate components is not neurologically founded. In Pulvermüller 2010, 28) he attempts to “define mechanisms … that may help to neurologically underpin the tight link between syntax and semantics postulated in the context of Cognitive and Construction Grammar.”
 
34
Tomasello is a co-director of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany. His usage-based, psycholinguistic studies fall within the general constructionist framework of cognitive linguists.
 
35
7th International Conference on the Evolution of Language.
 
36
The adjective modifier red constrains the semantics of the noun ball much the way the adjectival noun kitchen constrains the semantics of the noun table in Fig. 4.​1 of Chap. 4.
 
37
None of the particular computational approaches the author suggests for keeping this explosion “in check” seem relevant to Logos Model, or to what we have postulated for the brain. One indication of this is that neither the brain nor Logos Model seems troubled in the least by threats of combinatorial explosion. Some other mechanism is obviously at work in the brain (and Logos Model), quite different from those the author speaks of. But in other respects Van den Broeck makes interesting observations that bear on our discussion.
 
38
To remind the reader, the methods discussed in earlier chapters are, in the order Van den Broeck places them: (i) the analytical, procedure-driven methods of rule-based MT; (ii) the empirical, probabilistic, data-driven methods of statistical MT; (iii) the analogical, symbolic net-like approach of Logos Model. The methods of course are not mutually exclusive. (The statistically-rooted neural-net approach of NMT would appear to fall between (ii) and (iii).)
 
39
If we think of the concept chair as the set of all particular chairs, then the SAL concept support surface may be seen as the set of sets that share this common property (sets such as shelf, floor, seat, etc.).
 
40
The NLP advantage manifested by Logos Model stems precisely from its ability to process semantics at an abstract level. Note, for example, in the depiction of Logos Model in Fig. 4.​1 in Chap. 4, the role that second-order semantics plays in interpreting the ambiguous terms from and take, allowing these words to be semantically “grounded,” preparatory to their translation.
 
41
Note in this connection the observation by Pothos (2007, 228): “An influential tradition in categorization assumes that category exemplars are stored, and the potential membership of new instances to a category is determined through a process of (whole) exemplar similarity.” Of course, what exactly is meant by similarity is another matter. Most commonly, similarity pertains to the overlapping of features (Daltrozzo and Conway 2014).
 
42
In the first instance, check has the meaning of consult, in the second instance the meaning of examine. The sense of the preposition for differs in each instance also. See discussion in the early part of Chap. 3.
 
43
The intricate, cognition-related interactions of the hippocampus and the prefrontal cortex (including the Broca area) now seems fully established in the literature, as reflected, e.g., in Jadhav et al. (2016): “Interactions between the hippocampus and the prefrontal cortex (PFC) are critical for learning and memory.”
 
Literatur
Bastiaansen M, Hagoort P (2015) Frequency-based segregation of syntactic and semantic unification during online sentence level language comprehension. J Cogn Neurosci 27(11):2095–2107CrossRef
Bendor D, Spiers HJ (2016) Does the hippocampus map out the future? Trends Cogn Sci 20(3):167–169CrossRef
Bleys J (2008) Expressing second order semantics and the emergence of recursion. In: Smith ADM, Smith K, Ferrer-i-Cancho R (eds) Proceedings of the 7th international conference on the evolution of language. World Scientific Publishing Co, Ltd., Singapore, pp 34–41
Chadwick MJ, Bonnici HM, Maguire EA (2012) Decoding information in the human hippocampus: a user’s guide. Neuropsychologia 50(13):3107–3121. Paris: ElsevierCrossRef
Chomsky N (1957) Syntactic structures. Mouton, The Hague/ParisMATH
Chomsky N (1965) Aspects of the theory of syntax. MIT Press, Cambridge
Chomsky N (1977) Essays on form and interpretation. Elsevier North-Holland, New York/Amsterdam
Chomsky N (2000) New horizons in the study of language and mind. Cambridge University Press, CambridgeCrossRef
Fillmore C, Kay P, O’Connor C (1988) Regularity and idiomaticity in grammatical constructions: the case of let alone. Language 64:501–538CrossRef
Glaser YG, Martin RC, Van Dyke JA, Chris Hamilton A, Tan Y (2013) Neural basis of semantic and syntactic interference in sentence comprehension. Brain Lang 126:314–326CrossRef
Goldberg AE (2003) Constructions: a new theoretical approach to language. Trends Cogn Sci 7(5):219–224CrossRef
Grodzinsky Y, Friederici AD (2006) Neuroimaging of syntax and syntactic processing. Curr Opin Neurobiol 16(2):240–246CrossRef
Hagoort P (2015) The binding problem for language, and its consequences for the neurocognition of comprehension. In: Gibson EA, Pearlmutter NJ (eds) The processing and acquisition of reference. A Bedford Book/MIT Press, London, pp 403–436
Hoffman T, Trousdale G (eds) (2013) The Oxford handbook of construction grammar. Oxford University Press, Oxford
Jadhav S, Rothschild G, Roumis D, Frank L (2016) Coordinataed excitation and inhibition of prefrontal ensembles during awake hippocampal sharp-wave ripple events. Neuron 90(1):113–127CrossRef
Kempen G, Harbusch K (2003) An artificial opposition between grammaticality and frequency: comment on Bornkessel, Schlesewsky, and Friederici (2002). Cognition 90(2):205–210CrossRef
Kurczek J, Brown-Schmidt S, Duff M (2013) Hippocampal contributions to language: evidence of referential processing deficits in amnesia. J Exp Psychol Gen 142(4):1346–1354CrossRef
Lakoff G (1987) Women, fire, and dangerous things: what categories reveal about the mind. CSLI, ChicagoCrossRef
Langacker R (1991) Foundations of cognitive grammar, 2 vols. Stanford University Press, Stanford
Langacker R (2008) Cognitive grammar: a basic introduction. Oxford University Press, New YorkCrossRef
Mårtensson J, Eriksson J, Bodammer NC, Lindgren M, Johansson M, Nyberg L, Lövdén M (2012) Growth of language-related brain areas after foreign language learning. Neuro Image 63(1):240–244
McClelland JL, Patterson K (2002) Rules or connections in past-tense inflections: what does the evidence rule out? Trends Cogn Sci 6(11):465–472CrossRef
Musso M, Moro A, Glauche V, Rijntjes M, Reichenbach J, Büchel C, Weiller C (2003) Broca’s area and the language instinct. Nat Neurosci 6:774–781CrossRef
Optiz B, Friederci AD (2007) Neural basis of processing sequential and hierarchical syntactic structures. Hum Brain Mapp 28(7):585–592CrossRef
Optiz B, Hofmann J (2015) Concurrence of rule- and similarity-based mechanisms in artificial grammar learning. Cogn Psychol 77:77–99CrossRef
Osterhout L, Holcomb PJ (1992) Event-related brain potentials elicited by syntactic anomaly. J Mem Lang 31(6):785–896CrossRef
Pothos EM (2007) Theories of artificial grammar learning. Psychol Bull 133:227–244CrossRef
Pulvermüller F (2010) Brain embodiment of syntax and grammar: discrete combinatorial mechanisms spelt out in neuronal circuits. Brain Lang 112(3):167–179CrossRef
Reber AS (1967) Implicit learning of artificial grammars. J Verbal Learn Verbal Behav 6(6):855–863CrossRef
Rouder JN, Ratcliff R (2006) Comparing exemplar- and rule-based theories of categorization. Curr Dir Psychol Sci 15(1):9–13. Hoboken: Wiley-BlackwellCrossRef
Schmid P (1996) Clausal constituent shifts: a study in cognitive grammar and machine translation. Ph.D. dissertation, Georgetown University. UMI Dissertation Services
Shanks DR (1995) The psychology of associative learning. Cambridge University Press, CambridgeCrossRef
Ullman MT (2004) Contributions of memory circuits to language: the declarative/procedural model. Cognition 92:231–270CrossRef
Van den Broeck W (2008) Constraint based compositional semantics. In: Smith ADM, Smith K, Ferrer-i-Cancho R (eds) Proceedings of the 7th international conference on the evolution of language. World Scientific Publishing Co, Ltd., Singapore, pp 338–345
Metadaten
Titel
Syntax and Semantics: Dichotomy Versus Integration
verfasst von
Bernard Scott
Copyright-Jahr
2018
Buch
Translation, Brains and the Computer

Print ISBN: 978-3-319-76628-7

Electronic ISBN: 978-3-319-76629-4

Copyright-Jahr: 2018

DOI
https://doi.org/10.1007/978-3-319-76629-4_5