Symbol Grounding Tony Belpaeme Stephen J Cowley Karl F Macdorman
Symbol Grounding Tony Belpaeme Stephen J Cowley Karl F Macdorman
Symbol Grounding Tony Belpaeme Stephen J Cowley Karl F Macdorman
Symbol Grounding Tony Belpaeme Stephen J Cowley Karl F Macdorman
1. Symbol Grounding Tony Belpaeme Stephen J Cowley
Karl F Macdorman download
https://ebookbell.com/product/symbol-grounding-tony-belpaeme-
stephen-j-cowley-karl-f-macdorman-4635146
Explore and download more ebooks at ebookbell.com
2. Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Symbol Grounding And Beyond Third International Workshop On The
Emergence And Evolution Of Linguistic Communication Eelc 2006 Rome
Italy September 30 October 1 2006 Proceedings 1st Edition Federico
Divina
https://ebookbell.com/product/symbol-grounding-and-beyond-third-
international-workshop-on-the-emergence-and-evolution-of-linguistic-
communication-eelc-2006-rome-italy-
september-30-october-1-2006-proceedings-1st-edition-federico-
divina-2246980
Symbol Of Hope Amber Paige
https://ebookbell.com/product/symbol-of-hope-amber-paige-50263622
Symbol Of Authority The British District Officer In Africa Anthony
Kirkgreene
https://ebookbell.com/product/symbol-of-authority-the-british-
district-officer-in-africa-anthony-kirkgreene-50674166
Symbol And Politics In Communal Ideology Cases And Questions Sally
Falk Moore Editor Barbara G Myerhoff Editor
https://ebookbell.com/product/symbol-and-politics-in-communal-
ideology-cases-and-questions-sally-falk-moore-editor-barbara-g-
myerhoff-editor-51832790
3. Symbol And Truth In Blakes Myth Course Book Leopold Damrosch
https://ebookbell.com/product/symbol-and-truth-in-blakes-myth-course-
book-leopold-damrosch-51949316
Symbol And Physical Knowledge On The Conceptual Structure Of Physics
1st Edition Massimo Ferrari Auth
https://ebookbell.com/product/symbol-and-physical-knowledge-on-the-
conceptual-structure-of-physics-1st-edition-massimo-ferrari-
auth-4206906
Symbol Spotting In Digital Libraries Focused Retrieval Over
Graphicrich Document Collections 1st Edition Maral Rusiol
https://ebookbell.com/product/symbol-spotting-in-digital-libraries-
focused-retrieval-over-graphicrich-document-collections-1st-edition-
maral-rusiol-4396468
Symbol And Rhetoric In Ecclesiastes The Place Of Hebel In Qohelets
Work Douglas B Miller
https://ebookbell.com/product/symbol-and-rhetoric-in-ecclesiastes-the-
place-of-hebel-in-qohelets-work-douglas-b-miller-4746742
Symbol Pattern And Symmetry The Cultural Significance Of Structure
Michael Hann
https://ebookbell.com/product/symbol-pattern-and-symmetry-the-
cultural-significance-of-structure-michael-hann-5313906
6. Volume 21
Symbol Grounding
Edited by Tony Belpaeme, Stephen J. Cowley and Karl F. MacDorman
These materials were previously published in Interaction Studies 8:1 (2007), under
de guidance of Editor-in-Chief James R. Hurford.
Benjamins Current Topics
Special issues of established journals tend to circulate within the orbit of the
subscribers of those journals. For the Benjamins Current Topics series a number of
special issues have been selected containing salient topics of research with the aim to
widen the readership and to give this interesting material an additional lease of life in
book format.
7. Symbol Grounding
Edited by
Tony Belpaeme
University of Plymouth
Stephen J. Cowley
University of Hertfordshire
Karl F. MacDorman
Indiana University
John Benjamins Publishing Company
Amsterdam/Philadelphia
9. Table of contents
Foreword: Extending symbol grounding 1
Tony Belpaeme and Stephen J. Cowley
Grounding symbols in the physics of speech communication 9
Simon F. Worgan and Robert I. Damper
Social symbol grounding and language evolution 33
Paul Vogt and Federico Divina
How many words can my robot learn? An approach and experiments
with one-class learning 55
Luís Seabra Lopes and Aneesh Chauhan
How human infants deal with symbol grounding 85
Stephen J. Cowley
Semiotic symbols and the missing theory of thinking 107
Robert Clowes
The acquired language of thought hypothesis: A theory of symbol
grounding 127
Christopher Viger
Afterword: Life after the symbol system metaphor 145
Karl F. MacDorman
Index 161
11. foreword
Extending symbol grounding
Tony Belpaeme and Stephen J. Cowley
Faculty of Science and Technology, University of Plymouth /
School of Psychology, University of Hertfordshire
The contributions collected in this volume emerged from an international work-
shop on symbol grounding organised at the University of Plymouth in 2006. The
goal was to extend the classical view of symbol grounding by recognising that
language and cognitive dynamics are mutually constitutive. Specifically, we aimed
to do so by bringing researchers who study human signalling together with others
who focus on simulating intelligence and language. The objectives of the work-
shop were to view language and cognition as linking what goes on in the head
with causal processes that are intersubjective,multimodal,affect-laden,and organ-
ised by historically rooted customs and artefacts. In this, we focus on how symbol
grounding can be reconsidered when language is viewed as a dynamical process
rooted in both culture and biology. This welcomes a cross-disciplinary approach
and this volume contains research related to robotic or computer modelling of
symbol grounding, psychological and linguistic viewpoints on cognitive develop-
ment and semiotic dynamics.
By invoking symbol grounding, Harnad (1990) recognised that any compu-
tational model of mind must explain how an agent’s representations (or symbols)
connect with the external world (cf. Taddeo Floridi, 2005). While Harnad used
‘symbol’ to denote both the agent’s representations and linguistic signals, today
this systematic ambiguity is widely rejected. Most agree that explaining the evolu-
tion and emergence of linguistic signals is quite unlike grounding symbols into
objects beyond the skin. To understand this shift in perspective, one has to re-
visit the historical context of symbol grounding. Harnad (1990) formulated the
symbol grounding problem in response to Searle’s (1980) famous (or notorious)
Minds, Brains, and Programs article. In this, readers are invited to imagine a Chi-
nese Room where a person applies rules to input and generates output. The in-
put consists of written Chinese characters, but to that person these are senseless
squiggles. As is the case for a computer program, this gives rise to output that
12. 2 Tony Belpaeme and Stephen J. Cowley
makes sense outside the room, but not to the person or mechanism in the room.
For many this suggests that computers cannot attain human-like intelligence. Like
a parable, the text is cleverly constructed. Playing the devil’s advocate, Searle first
entertains the view that, because a brain is like a computer executing a program, a
computer could pass the Turing Test.Second,he assumes that syntactic processing
is automatic: symbols manipulated during processing make no use of meaning. By
implication, any device that passes the test is of merely technical interest.Without
intentionality, Searle insinuates, no computer can ever attach meaning to its inner
symbols.1 This is highly controversial. Harnad (1990, 2003) reinterprets it as the
challenge of showing how symbols (linguistic or otherwise) can be connected to
the world beyond the body.Without a person to interpret them,squiggles on paper
or spoken words indeed lack meaning or, for Harnad, are ungrounded. Once in-
wardly represented, however, they connect with referent-indexed symbols and, in
so doing, are grounded. The challenge is to show how agents can causally connect
symbols to the external world. First, they need senses to identify referents which,
eventually, can connect with squiggles or spoken words. To add meaning to the
resulting symbols,they need sensorimotor interaction with the world.To solve the
problem of symbol grounding, a machine must be a robot rather than a computer.
What kinds of symbol do robots need? For Harnad (1990) and computational
theorists (Newell Simon, 1976; Newell, 1980; Pylyshyn, 1984; Fodor, 1987), this
is a matter of definition. Symbols are arbitrary physical tokens or syntactic com-
binations of such tokens that can be manipulated systematically and generatively
and assigned a meaning. An agent’s symbols, therefore, stand in for both physical
properties (e.g., what is common to cups) and repeatable linguistic signals (e.g.,
spoken words). By implication, grounding symbols in cups is broadly similar to
grounding them in the speech stream.
While each author in this volume accepts the classical view that (inner) sym-
bols must connect with objects and linguistic signals, at the same time they reject
the conflation of language and representation into a symbol. Generally, they pro-
pose two ways of extending the classical view of symbol grounding. First, those
who take a pragmatic view draw on insights from implementing symbol ground-
ing in simulation or on real-world robots. This leads them, first, to reject appeal
to already-combined forms. Instead, symbols are viewed as arising from connec-
tions between a referent, its sensorimotor interpretation and, in some cases, a
communicative signal. This is a semiotic interpretation of symbol (Maturana and
Varela, 1992; Clancey, 1997; Clowes, 2007; Vogt, 2002, 2007). For the pragmatists,
a symbol is ultimately a set of relations which connect a meaning, signal and ref-
erent. As in Peirce’s semiotic model, this is constituted by a triad of interpretant,
representamen and object. Symbols thus unite a system’s sensorimotor capacities
13. Extending symbol grounding 3
with its interpretative ones. To distinguish this from seeing the symbols as mean-
ingless tokens, it has been called the physical symbol grounding problem (see for
example Vogt, 2002). This view is increasingly being supported in several disci-
plines, from psycholinguistics to robotics (for an overview see de Vega, Glenberg
and Graesser, 2008).
Alongside this, a second culturalist group also rejects formal definitions of
symbols. They object to the view that brains use similar forms of symbols to index
both objects and patterns in the speech stream. Underplaying cultural practices,
the classical view exaggerates the importance of notations.By assuming that verbal
patterns are grounded in notational ways,the classical approach forgets that words
are ultimately physical signals that contribute to social activity. For the brain,
indeed, spoken words may entirely lack symbolic representation. In Viger (pp.
127–144), therefore, a capacity to identify referents is the outcome of a process of
acquisition. For Worgan and Damper (pp. 9–32), the classical approach must be
extended by appeal to signal grounding. Finally, for Cowley (pp. 85–106), categori-
sation of spoken words is marginal during the early stages of learning to talk. The
infant-caregiver dyad focuses, not on forms or referents, but on strategic modes of
action that are likely to meet their conflicting needs. To simulate extended sym-
bol grounding, therefore, robots too will need (inner) symbols with richer content
than is posited in the classical model.
Language thus ceases to be a system of formal symbols. Rather, it is viewed as
an external cognitive resource that allows us to acquire categories and concepts. It
functions as agents coordinate behaviour and,in so doing,gain some grasp of how
others perceive the world. During the process of grounding symbols in language,
therefore,agents play social games that always involve (at least) two brains,human
or robotic, or some combination of these systems (cf. Seabra Lopes Chauhan,
2007). In coordinating, of course, they typically interpret each signal in different
ways. Because of individual differences and the dynamic nature of cognition, nei-
ther flesh nor silicon will concur on signal meaning.Although this poses no major
problems for an individual agent, it ensures that cooperation or communication
must take on strategic functions. Since referents are not guaranteed, cultural pro-
cesses must stabilise interaction. We use language to coordinate what we do and
thereby coordinate with what signals represent (Steels and Belpaeme, 2005).
Language thus influences symbol grounding in several ways. It facilitates the
acquisition of meaning: an individual no longer needs to go through a long pro-
cess of exploration to acquire knowledge; it can use language to tap into knowl-
edge present in others (Cangelosi and Harnad, 2001). Language also serves to
coordinate referents between individuals: if knowledge is acquired through senso-
rimotor experience, language serves to align that knowledge between individuals,
14. 4 Tony Belpaeme and Stephen J. Cowley
facilitating communication (Steels and Belpaeme, 2005). Far from being a mono-
lithic system of determinate forms, language is a dynamic process that is shaped
and propagated by users who exploit social and physical constraints. Given that
language also influences symbol grounding, the symbol grounding process too
must be dynamic. Further, while important in human cognition, our cognitive
states (or symbols) depend on much more than verbal patterns. In extending clas-
sical views of symbol grounding, therefore, we emphasise emotion and, equally,
how categories sensitise to culturally based customs and artefacts.Human interac-
tion and,thus,human symbols,depend on more than a syntactic analysis of words.
Beyond this, however, each chapter points in a different direction. First, breaking
with Fodor, Viger argues that the emergence of reference be seen as the culmina-
tion of a learning process. Taking the opposite tack, Cowley (pp. 85–106) focuses
on infancy to stress that the early stages of learning to talk are more dependent on
affective and multimodal processes than spoken words. In an attempt to construct
a general framework, Clowes (pp. 107–126) suggests that language is gradually
internalised. Reporting ambitious work, Vogt and Divina (pp. 33–54) describe a
model that aims at simulating the cultural processes. In building robots that learn
words Seabra Lopes and Chauhan fall in line in finding that teaching — not cat-
egorisation — is a simpler basis for language. Finally, focusing on the evolution
of phonetic forms, Worgan and Damper stress that many properties of human
languages are likely to depend on constraints based in how articulation and audi-
tion favour learning.Such processes,of course,use the physics of speech and,thus,
underpin symbol grounding.
The six chapters extend traditional views of symbol grounding in both ap-
plied and theoretical ways. At the engineering end, Worgan and Damper criticise
the lack of attention paid to the physical aspects of the symbol that is grounded.
Pursuing this, they present a computational model in which a shared speech code
emerges between agents,and show how constraints on production,perception and
acoustics influence speech signals.While Worgan and Damper are concerned with
physical signals, Vogt and Divina focus on simulations of their semantic coun-
terparts. In their paper, they present a large scale simulation in which agents so-
cially transmit and acquire behavioural skills. As the agents communicate, their
signals can have ecological relevance only if they are grounded. Vogt and Divina
present the technical challenges involved and put forward a number of learning
methods, such as cross-situational learning, inspired by psycholinguistic observa-
tions. The contribution of Seabra Lopes and Chauhan takes the study of symbol
grounding into the real world by presenting an experiment in which an adaptive
learning system acquires the meaning of objects through linguistic instruction.
Besides demonstrating the characteristics of their learning method, their prag-
15. Extending symbol grounding 5
matic approach clearly shows that implementing symbol grounding in a real-
world system is a formidable challenge in which many issues – such as perception,
concept representation and concept learning – remain far from solved. Cowley
argues that too much attention is spent on language as the facilitating medium for
symbol grounding, and he observes that infants, from an early age, bootstrap sig-
nal grounding by relying on what agents do together. The infant uses affect-laden
interactions to acquire culturally evolved patterns that enable it to participate in a
grounding process. Clowes starts out by contrasting the view of symbols as formal
tokens with the view that they emerge from a semiotic agreement. In his model,
he seeks to reconcile strengths of these approaches. In so doing, he starts from
Vygotsky’s views on language internalisation and sets out a model of symbol in-
ternalisation. In Clowes’s model this process is dynamic and self-regulatory. In the
most theoretical paper, therefore,Viger takes a philosophical perspective on sym-
bol grounding. He focuses on how abstract symbols, for example, unicorn, can be-
come intrinsic to thinking.While concurring with Fodor that we need a language
of thought,he challenges tradition by suggesting that referents emerge from a long
acquisition process. Finally, we leave the last word to MacDorman who, looking
back over the workshop, proposes that we abandon the symbol system metaphor
to ask how humans – and perhaps robot bodies – can construct themselves into
persons (MacDorman, 2007).
The symbol grounding problem has shown more life in it than Searle’s thought
experiment.Indeed,our papers strongly imply that the Chinese Room parable fails
because of a simplistic assumption that, as in programs, spoken words can be for-
mally represented. In fact, as all contributors stress, semiosis involves more than
form-based computation.Even though symbol grounding has been expanded over
the years, by scholars collected here and elsewhere, a number of open problems
remain. One of these is the role of covariation on the acquisition of meaning: the
meaning of a word or utterance can often be gleaned from that of others occur-
ring in the same context (for example using Landauer Dumais, 1997’s Latent
Semantic Analysis approach). As such, not every word or linguistic construction
needs grounding. But what balance is required between grounded concepts and
indirectly grounded concepts? This leads us to abstract concepts, such as truth
or democracy, which cannot be directly grounded, but which do have meaning.
Opinions on how abstract concepts acquire meaning abound, some positing that
abstract concepts are grounded as good as directly in sensorimotor experience,
such as democracy being grounded in the experience of going voting (Glenberg
Kaschak, 2002), some believing abstract concepts to be grounded through a long
series of links to grounded, perceptual concepts. These at the moment are just
theories and as such are in need empirical support or, perhaps more feasibly, in
16. 6 Tony Belpaeme and Stephen J. Cowley
need of implementation and exploration on a robotic platform. Finally, the new
directions in symbol grounding thinking hinted at in introduction and further in
this volume – distributed symbol grounding,signal grounding,emotion – warrant
further exploration. In the book we show how new methods, such as computer
simulations and robotics, and the increased collaboration between fields as dis-
parate as philosophy and computer science sparks debate. Who knows, perhaps
all this results in a new theory of symbol grounding: extended symbol grounding.
Note
1. While computational syntax makes no use of meaning, human language may differ. Accord-
ingly, we face philosophical sticking points. Whereas Searle (1980) views intentionality (and
meaning) as a causal feature of the brain, Harnad (1990) operationalises‘aboutness’ in terms of
how meaning can be made intrinsic to the system itself. Others trace meaning to, for example,
adaptivity of living systems in an environment (di Paolo, 2005) or biosemiosis (Barbieri, 2007).
References
Barbieri, M. (2007). Is the cell a semiotic system? In: M. Barbieri (Ed.) Introduction to biosemi-
otics, 179-207. Springer: Berlin.
Blackmore, S. (1999). The meme machine. Oxford: Oxford University Press.
Cangelosi, A. Harnad, S. (2001). The adaptive advantage of symbolic theft over sensorimo-
tor toil: Grounding language in perceptual categories. Evolution of Communication, 4(1),
117–142.
Clowes,R.(2007).Semiotic symbols and the missing theory of thinking.Interaction Studies,8(1),
105–124.
Cowley, S. J. (2007). How human infants deal with symbol grounding. Interaction Studies, 8(1),
83–104.
De Vega, M., Glenberg,A. M., Graesser,A. C. (Eds.). (2008). Symbols and embodiment: Debates
on meaning and cognition. Oxford, UK: Oxford University Press.
Di Paolo, E. (2006). Autopoiesis, adaptivity teleology, agency. Phenomenology and the Cognitive
Sciences, 4 (4) 429-452.
Fodor, J. A. (1987). Modules, frames, fridgeons, sleeping dogs and the music of the spheres. In
Pylyshyn,Z.W.(Ed.) The robot’s dilemma: The frame problem in artificial intelligence (Chap-
ter 8). Norwood, NJ: Ablex..
Glenberg, A. M., Kaschak, M. P. (2002). Grounding language in action. Psychonomic Bulletin
Review, 9(3), 558-565.
Harnad, S. (1990). The Symbol grounding problem. Physica D, 42, 335–346.
Harnad, S. (2003). The Symbol grounding problem. Encyclopedia of Cognitive Sciences.
London:Macmillan/Nature.
17. Extending symbol grounding 7
Landauer, T. K., Dumais, S. T. (1997). A Solution to Plato’s Problem: The Latent Semantic
Analysis Theory of Acquisition,Induction,and Representation of Knowledge.Psychological
Review, 104(2), 221-240.
MacDorman, K. F. (2007). Life after the symbol system metaphor. Interaction Studies, 8(1), 143–
158.
Newell, A. Simon, H.A. (1976). Computer science as empirical enquiry: Symbols and search.
Communications of the ACM, 19(3), 113–126.
Newell,A. (1980). Physical symbol systems. Cognitive Science, 4(4), 135–183.
Pylyshyn, Z.W. (1984). Computation and cognition. Cambridge, MA: The MIT Press.
Seabra Lopes, L. Chauhan,A. (2007). How many words can my robot learn? An approach and
experiments with one-class learning. Interaction Studies, 8(1), 53–81.
Searle, J. (1980). Minds, brains and programs. Behavioral and Brain Sciences, 3(3), 417–457.
Steels, L. and Belpaeme, T. (2005). Coordinating perceptually grounded categories through lan-
guage.A case study for colour. Behavioral and Brain Sciences, 24(8), 469–529.
Taddeo,M. Floridi,L.(2005).The symbol grounding problem:A critical review of fifteen years
of research. Journal of Experimental and Theoretical Artificial Intelligence, 17(4), 419–445.
Viger, C. (2007). The acquired language of thought hypothesis: A theory of symbol grounding.
Interaction Studies, 8(1), 125–142.
Vogt, P. (2002). The physical symbol grounding problem. Cognitive Systems Research, 3(3), 429–
457.
Vogt, P. Divina, F. (2007). Social symbol grounding and language evolution. Interaction Stud-
ies,8(1), 31–52.
Worgan, S. F. Damper, R. I. (2007). Grounding symbols in the physics of speech communica-
tion. Interaction Studies, (8)1, 7–30.
19. Grounding symbols in the physics of
speech communication
Simon F.Worgan and Robert I. Damper
School of Electronics and Computer Science, University of Southampton
The traditional view of symbol grounding seeks to connect an a priori internal
representation or‘form’ to its external referent. But such a‘form’ is usually itself
systematically composed out of more primitive parts (i.e., it is‘symbolic’), so
this view ignores its grounding in the physics of the world. Some previous work
simulating multiple talking/listening agents has effectively taken this stance, and
shown how a shared discrete speech code (i.e., vowel system) can emerge. Taking
the earlier work of Oudeyer, we have extended his model to include a dispersive
force intended to account broadly for a speaker’s motivation to increase auditory
distinctiveness. New simulations show that vowel systems result that are more
representative of the range seen in human languages. These simulations make
many profound abstractions and assumptions. Relaxing these by including more
physically and physiologically realistic mechanisms for talking and listening is
seen as the key to replicating more complex and dynamic aspects of speech, such
as consonant-vowel patterning.
Keywords: origins of speech sounds, symbol grounding, signal grounding, multi-
agent simulation, self-organisation, emergent phenomena
Introduction
The computational metaphor that underpins cognitive science, much of artifi-
cial intelligence and functionalist philosophy of mind sees intelligent behaviour
as the product of the workings of a formal symbol manipulation system (e.g.,
Newell, 1973; Minsky, 1974; Fodor, 1975; Newell and Simon, 1976; Newell, 1980,
1990; Pylyshyn, 1984; Dietrich, 1990). But this view faces a formidable problem,
famously articulated by Harnad (1990) as: “How can the semantic interpreta-
tion of a formal symbol system be made intrinsic to the system, rather than just
parasitic on the meanings in our heads?” (p.335). Harnad calls this the symbol
20. 10 S. F.Worgan and R. I. Damper
grounding problem (SGP) and comments: “The handicap has been noticed in
various forms since the advent of computing” (p.338). The earliest reference that
we know is that of Mays (1951), who writes “if we grant that these machines [i.e.,
digital computers] are complex pieces of symbolism,… to acquire a significance
the symbols need to be linked with a set of referents” (p.249). So if the com-
putational metaphor is to offer any purchase in modelling and understanding
cognition, the SGP poses a challenge that cannot be neglected (Cangelosi, Greco
and Harnad, 2002). We take this challenge seriously, because the long-term goal
of our research is to understand, via computer modelling and simulation, how
speech sound categories (broadly, ‘phonemes’) could have emerged during lan-
guage evolution, and then how these could be combined systematically to lead to
utterances with semantic content.
To some the SGP is symptomatic of an incorrect view of AI and cognitive
science, famously parodied as “good old-fashioned AI,” or GOFAI, by Haugeland
(1985). For instance, as Pfeifer and Scheirer (1999, p.71) write, “… the symbol
grounding problem is really an artifact of symbolic systems and ‘disappears’ if a
different approach is used.” The different approach they have in mind is, of course,
embodied or nouvelle AI as spearheaded by Brooks (1990,1991,1999),which seeks
to replace the central role played by symbolic representation with nonsymbolic
interfacing to the physical world through cycles of perception and action, usually
conceived as based on some connectionist or statistical machine learning prin-
ciples. However, the complete banishment of symbolism from the scene is rather
too radical for most AI scientists and cognitive psychologists, who continue to see
a role for formal symbol systems, albeit in combination with some sort of con-
nectionist component (e.g., Minsky, 1990; Harnad, 1990, 1993) in modelling and
explaining the higher cognitive functions involved in,for example,using language,
doing mathematics, and decision making under uncertainty, where nouvelle AI
has arguably promised more than it has delivered.
Against this background, a new view of the SGP has recently arisen in which
the physics of the external world plays an important and simplifying role (Sun,
2000; Vogt, 2002). Vogt (2002) coins the term physical symbol grounding problem
and writes: “It is based on the idea that symbols should be grounded (cf. Harnad,
1990) and… they should be grounded by physical agents that interact with the
world (cf. Brooks, 1990)” (p.435). Our work is broadly consonant with this view,
treating the SGP (as does Vogt, 2002) as a technical problem by way of computer
simulation, although we have also been influenced in our thinking by the work of
Barsalou (1999).
Quite apart from the intrinsic scientific interest in studying the emergence
of human speech and language for its own sake (Damper, 2000), it makes an
21. Grounding symbols in the physics of speech communication 11
excellent context in which to consider the SGP. First and foremost, we believe
human communication to be the clearest, certainly best-developed, example of
externally-grounded cognition. As Vogt (2002, p.431) writes, “language through
its conventions offers a basis for invariant labeling of the real world.” Since human
communication is a social phenomenon, we pursue an approach of multi-agent
simulation, not unlike much previous work in ‘language games’ but with one im-
portant difference (see below).
Inparticularinthispaper,wearguethattheemergenceof speechsoundcatego-
ries can and should be grounded in the physics of speech communication between
agents,recognising that the human’s contact with the external world of sound is via
their articulatory and auditory systems.Important previous work along these lines
is that of Steels (1997, 1998, 1999, 2003), de Boer (2000, 2001, 2005), and Oudeyer
(2005a,2005b,2005c),who have explored grounded speech-category formation by
computer simulation of multi-agent systems, with agents equipped with rudimen-
tary articulatory and auditory systems and associated ‘neural’ processing. Broadly
speaking, this line of work had its beginnings in the early and influential efforts
of Lindblom (1986) and his colleagues to explain the origins of vowel systems in
the world’s languages (Liljencrantz and Lindblom, 1972; Lindblom, MacNeilage
and Studdert-Kennedy,1984; Lindblom,1986,2000) based on“adaptive dispersion
theory.” In their numerical simulations, the clustering of vowels in some metric
space was predicted by minimising an energy function designed to reflect percep-
tual distinctiveness.An important question is exactly how realistic the simulations
have to be (e.g., in terms of faithfully modelling the articulatory/auditory systems
and brain mechanisms). Hence, our longer-term goal is to answer this question,
although at this stage we will restrict ourselves to relatively simple simulations
such as have been used in previous work.
Although Steels (1997) argues for a “limited rationality constraint” in multi-
agent simulations (i.e., agents should not have access to each other’s internal
states), this constraint is typically violated in language games where nonlinguis-
tic feedback figures importantly. For instance, de Boer (2001) writes, “the initia-
tor then communicates the success or failure to the imitator using nonlinguistic
communication” (p.52). In our view, this amounts to a form of ‘mind-reading,’ se-
riously undermining the credibility of the simulations. Hence, we wish to avoid
this aspect of language games, and favour Oudeyer’s alternative approach where
he dispenses with nonlinguistic feedback. As he writes, “it is crucial to note that
agents do not imitate each other… The only consequence of hearing a vocalization
is that it increases the probability, for the agent that hears it, of vocalizations…
similar to those of the heard vocalization”(Oudeyer,p.443).In spite of the absence
of structured, coordinated interactions between agents, he achieves two results in
22. 12 S. F.Worgan and R. I. Damper
his simulations which mirror important aspects of real language:“on the one hand
discreteness and compositionality arise thanks to the coupling between perception
and production within agents, on the other hand shared systems of phonemic cat-
egories arise thanks to the coupling across agents” (Oudeyer, p.445).
A related line of investigation is that of Kirby (2001) and Kirby and Hurford
(2002) who describe the iterated learning model (ILM). This, however, operates
at the syntactic level, that is, learning agents receive from adult agents “meaning-
signal pairs”(p.103) that act as training data.Thus,the ILM already tacitly assumes
the emergence of phonetic distinctiveness. Whereas the language-game style of
simulations are concerned with language change once the basic mechanisms are
in place, by contrast, Oudeyer is concerned with the earliest origins of a phonemic
sound system, as are we. Further, Oudeyer’s model is based on horizontal cultural
interaction between agents of the same generation, following the works of Steels
and colleagues,whereas the ILM is based on iterated learning among agents of one
generation and agents of the previous generation (so this is more vertical learning).
However, Oudeyer’s work has its own drawback in that he ignores the tenets
of dispersion theory.“There are no internal forces which act as a pressure to have a
repertoire of different discrete sounds,”he writes (p.443).But to cite de Boer (2001,
p.61),a successful vowel system has“its vowel clusters…dispersed (for low energy)
and compact (for high imitative success).” These ideas are broadly consistent with
notions of HH theory (Lindblom, 1990) and the dispersion-focalisation theory
(DFT) of Schwartz et al. (1997). Although Oudeyer tries to argue that the lack of
a dispersion force is a virtue of his simulations (it is one less assumption), he also
seems to recognise that it causes problems for the emergence of sound systems
with realistically large numbers of vowels,writing,“Functional pressure to develop
efficient communication systems might be necessary here” (p.447).
Accordingly,the principal purpose of the present paper is to introduce ideas of
HH theory and DFT into Oudeyer-style simulations in the belief that more re-
alistic vowel systems (i.e., more representative of those seen in a variety of human
languages) will result. We will do this by extending the topological spaces in the
neural maps used to couple auditory and articulatory processing as a vastly-sim-
plified form of brain.We call these extensions contour spaces.The work is intended
to form a baseline for future work in which we will study the impact of increased
realism of the agents’ articulatory and auditory capabilities, as well as extending
our simulations beyond prediction of static vowel systems to the emergence of
connected speech sounds with appropriate consonant-vowel patterning.
The remainder of this paper is structured as follows.In the next section,we set
out our conception of physical symbol grounding, which we call signal grounding,
and relate this to more traditional views of symbol grounding. Then, as a baseline
24. And while to thoughts refined they make pretence,
Hate all that's common, ev'n to common sense.
[196] In the first edition the reading was dull believers, which Pope in
the second edition altered to plain. The change was occasioned by the
outcry against the couplet. An ordinary man, he wrote to Caryll, would
imagine the author plainly declared against these schismatics for quitting
the true faith out of contempt of the understanding of some few of its
believers. But these believers are called 'dull,' and because I say that
these schismatics think some believers dull, therefore these charitable
well-disposed interpreters of my meaning say that I think all believers
dull. There is a culpable levity in the language of Pope's lines, but he
could not intend to espouse the cause of the sceptics when he selects
them as an instance of people who purposely go wrong because the
crowd go right.
[197] If this couplet is interpreted by the grammatical construction, the
unfortified towns daily changed their sides in consequence of vacillating
betwixt sense and nonsense. Of course Pope only meant that in war
weak towns frequently changed sides, but not for the same reason that
weak heads changed their opinions.
[198] The Book of Sentences was a work of Peter Lombard, which
consisted of subtle disquisitions on theology. Thomas Aquinas wrote a
commentary upon it.
[199] St. Thomas Aquinas died in 1274. Scotus, who died in 1308,
disputed the doctrines of his predecessor, and their respective disciples
divided for a century the theological world.—Croker.
[200] Cowley speaks of the cobwebs of the schoolmen's trade, and says
in a note, the distinctions of the schoolmen may be likened to cobwebs
either because of the too much fineness of the work, or because they
take not the materials from nature, but spin it out of themselves.
[201] A place where old and second-hand books were sold formerly, near
Smithfield.—Pope.
[202] Between this and verse 448:
25. The rhyming clowns that gladded Shakespear's age,
No more with crambo entertain the stage.
Who now in anagrams their patron praise,
Or sing their mistress in acrostic lays?
Ev'n pulpits pleased with merry puns of yore;
Now all are banished to th' Hibernian shore!
[And thither soon soft op'ra shall repair,
Conveyed by Sw——y to his native air.
There, languishing awhile, prolong its breath,
Till like a swan it sings itself to death.]
Thus leaving what was natural and fit,
The current folly proved their ready wit:
And authors thought their reputation safe,
Which lived as long as fools were pleased to laugh.—Pope.
The lines between brackets are from the manuscript, and were not
printed by Pope. The whole passage was probably written after the poem
was first published, since the topics seem to have been suggested by
Addison's papers upon false wit in the Spectator of May, 1711, where the
anagrams, acrostics, and punning sermons of the reign of James I. are all
enumerated. Swiney was the director of the Italian opera, which, at the
commencement of 1712, failed to meet with adequate support, and he
withdrew, not to Ireland, but to the continent. He remained there, says
Cibber, twenty years, an exile from his friends and country.
[203] An additional couplet follows in the manuscript:
To be spoke ill of, may good works befall,
But those are bad of which none speak at all.
[204] The parson alluded to was Jeremy Collier; the critic was the Duke of
Buckingham; the first of whom very powerfully attacked the profligacy,
and the latter the irregularity and bombast of some of Dryden's plays.
These attacks were much more than merry jests.—Warton.
[205] Dryden himself, Virg. Geor. iv. 729:
But she returned no more to bless his longing eyes.—Wakefield.
[206] Blackmore's attack upon Dryden occurs in a poem which appeared
in 1700, called a Satire against Wit. The author treats wit as money, and
proposes that the whole should be recoined for the purpose of separating
the base metal from the pure.
26. Into the melting pot when Dryden comes
What horrid stench will rise, what noisome fumes!
How will he shrink when all his lewd allay
And wicked mixture shall be purged away!
When once his boasted heaps are melted down,
A chestful scarce will yield one sterling crown.
This is exaggerated, but the censure is directed against the indecency
which was really infamous. The invectives of Milbourne in his Notes on
Dryden's Virgil, 1698, had not the same excuse. The strictures are
confined to the translation of the Eclogues and Georgics, and are
throughout rabid, insolent, coarse, and contemptible. To demonstrate his
own superiority, Milbourne inserted specimens of a rival translation, which
is on a par with his criticisms. He was in orders, and acknowledges that
one of his reasons for not sparing Dryden was that Dryden never spared a
clergyman. I am only, replied the poet, with exquisite sarcasm, to ask
pardon of good priests, and am afraid his part of the reparation will come
to little. Dryden retaliated upon both antagonists together in the couplet,
Wouldst thou be soon dispatched, and perish whole?
Trust Maurus with thy life, and Milbourne with thy soul.
Pope's line in the first edition was
New Bl——s and new M——s must arise.
In the second edition he substituted S——s, which meant Shadwells, for
Bl——s, but in the quarto of 1717 he again coupled Blackmore with
Milbourne, and printed both names at full length. Blackmore was living,
and the changes indicate Pope's varying feelings towards him.
[207] In the fifth book of Vitruvius is an account of Zoilus's coming to the
court of Ptolemy at Alexandria, and presenting to him his virulent and
brutal censures of Homer, and begging to be rewarded for his work;
instead of which, it is said, the king ordered him to be crucified, or, as
some said, stoned. His person is minutely described in the eleventh book
of Ælian's various History.—Warton.
Boileau's Art of Poetry, translated by Soame and Dryden:
Let mighty Spenser raise his reverend head,
Cowley and Denham start up from the dead.
[208] A beautiful and poetical illustration. Pope has the art of enlivening
his subject continually by images and illustrations drawn from nature,
which by contrast have a particularly pleasing effect, and which are
indeed absolutely necessary in a didactic poem.—Bowles.
27. The passage originally stood thus in the manuscript:
Wit, as the sun, such pow'rful beams displays,
It draws up vapours that obscures its rays,
But, like the sun eclipsed, makes only known
The shadowing body's grossness, not its own;
And all those clouds that did at first invade
The rising light, and interposed a shade,
When once transpierced with its prevailing ray
Reflect its glories, and augment the day.
[209] His instance refuted his position that bare threescore was the
duration of modern fame. It is now a hundred years, said Dennis in
1712, since Shakespeare began to write, more since Spenser flourished,
and above three hundred years since Chaucer died. And yet the fame of
none of these is extinguished. Another century and a half has elapsed,
and the reputation of Chaucer, Spenser, and Shakespeare is greater than
ever. The notion of the failing language is not more sound. Though it is
a hundred and fifty years since the Essay on Criticism was published,
there is not a line which has an antiquated air.
[210]
The treach'rous colours in few years decay.—Pope.
The next line is from Addison:
And all the pleasing landscape fades away.
[211] That is, of which those, who do not possess it, form an erroneous
estimate, as productive of more happiness and enjoyment to the owner,
than he really receives from it.—Wakefield.
[212] In the previous paragraph Pope admitted that the fame of a modern
might last three-score years. Here, contradicting himself and the facts, he
limits its duration to the youth of the author. He applies to poets in
general what was only true of inferior writers. The ephemeral versifiers
were examples of deficient wit, and not of the unhappy consequences
of genuine poetic power.
[213]
Like some fair flow'r that in the spring does rise.—Pope.
This line was an example both of the feeble expletive and of the ten
low words. Supplies in the amended version is, as Wakefield observes,
a poor expression.
28. [214] The Duke of Buckingham's Vision:
The dearest care that all my thought employs.
[215] Wakefield objects to the slovenly superfluity of words, and asks
to whom can a wife possibly belong but the owner? He misunderstood
Pope, who, by the wife of the owner, meant the wife of the owner of the
wit. The metaphor is coarse, and out of keeping with the theme.
[216] Thus in the first edition:
The more his trouble as the more admired,
Where wanted scorned, and envied where acquired.
Against this Pope wrote, To be altered. See Dennis, p. 20. How, said
Dennis, can wit be scorned where it is not? The person who wants this
wit may indeed be scorned, but such a contempt declares the honour that
the contemner has for wit. Pope, in a letter to Caryll, admitted that he
had been guilty of a bull, and the reading in the second edition was,
'Tis most our trouble when 'tis most admired,
The more we give, the more is still required.
[217] In the first edition,
Maintained with pains, but forfeited with ease;
and in the second edition,
The fame with pains we gain, but lose with ease.
The original version appears better than the readings which successively
replaced it.
[218] Another couplet follows in the manuscript:
Learning and wit were friends designed by heav'n;
Those arms to guard it, not to wound, were giv'n.
[219] Dryden's Prologue to the University of Oxford:
Be kind to wit, which but endeavours well,
And, where you judge, presumes not to excel.
The feelings of antiquity were doubtless represented truly by Horace
when he said that indifferent poets were not tolerated by anybody. There
is not the least foundation for Pope's statement that it was the habit of
old to praise bad authors for endeavouring well, and if it had been, the
authors would not have cared for commendations on their abortive
industry to the disparagement of their intellect.
29. [220] Wakefield remarks upon the unhappy effect of crowns and
crown in consecutive lines, and thinks the phrase some others in the
next verse too mean and elliptical. Soame and Dryden, in their translation
of Boileau's Art of Poetry, speak of the base rivals who
aspire to gain renown
By standing up and pulling others down.
[221] Mr. Harte related to me, that being with Mr. Pope when he received
the news of Swift's death, Harte said to him, he thought it a fortunate
circumstance for their friendship, that they had lived so distant from each
other. Pope resented the reflection, but yet, said Harte, I am convinced it
was true.—Warton.
[222] That is, all the unsuccessful authors maligned the successful. The
unsuccessful writers never said anything more slanderous.
[223] Boileau's Art of Poetry, translated by Soame and Dryden:
Never debase yourself by treach'rous ways
Nor by such abject methods seek for praise.
Pope's own life is the strongest example upon record of the degradation
he deplores.
[224] In the margin of the manuscript Pope has written the passages of
Virgil from which he took his expressions. Æn. iii. 56:
quid non mortalia pectora cogis
Auri sacra fames?
Geor. i. 37:
Nec tibi regnandi veniat tam dira cupido,
which Dryden translates,
Nor let so dire a thirst of empire move.
[225] Such a manly and ingenuous censure from a culprit in this way, as
in the case of Pope, is entitled to great praise.—Wakefield.
If his indecorums had been the failing of youth and thoughtlessness, and
he had publicly recanted his errors, his self-condemnation would be
meritorious. The larger portion of his offences were, on the contrary,
committed after he had declared indecency to be unpardonable. Any man,
however persistently reprobate, might earn great praise on terms like
these.
30. [226] No one has expressed himself upon this subject so pithily as
Cowley:
'tis just
The author blush, there where the reader must.
[227] Hamlet:
And duller shoulds't thou be than the fat weed.—Bowles.
[228] Wits, says he, in Charles the Second's reign had pensions, when all
the world knows that it was one of the faults of that reign that none of
the politer arts were then encouraged. Butler was starved at the same
time that the king had his book in his pocket. Another great wit
[Wycherley] lay seven years in prison for an inconsiderable debt, and
Otway dared not to show his head for fear of the same fate.—Dennis.
[229] The young lords who had wit in the court of Charles II. were, says
Dennis, Villiers Duke of Buckingham, the Earl of Mulgrave, afterwards
Sheffield Duke of Buckingham, Lord Buckhurst afterwards Earl of Dorset,
the Marquess of Halifax, the Earl of Rochester, Lord Vaughan, and several
others. The jilts who ruled the state were the mistresses of the king. The
Duke of Buckingham and his Rehearsal are chiefly aimed at in the
expression statesmen farces writ.—Croker.
[230] Pepys, under the date of June 12, 1663, notices that wearing masks
at the theatre had of late become a great fashion among the ladies.
Cibber states that the immorality of the plays was the cause of the usage.
When he wrote in 1739 the custom had been abolished for many years in
consequence of the ill effects which attended it.
[231] He must mean in everyday life. There was no use for the modest
fan at the theatre after the ladies had adopted the more effectual plan of
wearing masks. Pope, ver. 535, ascribes the introduction of obscenity to
the Restoration. In the theatre the grossness was a legacy from the older
drama, and particularly from the plays of Beaumont and Fletcher, which
were the most popular pieces on the stage.
[232] The author has omitted two lines which stood here as containing a
national reflection, which in his stricter judgment he could not but
disapprove, on any people whatever.—Pope.
The cancelled couplet was as follows:
Then first the Belgian morals were extolled,
We their religion had, and they our gold.
31. This sneer was dictated by the poet's dislike to William III. and the Dutch,
for displacing the popish king James II.—Croker.
This ingenious and religious author seems to have had two particular
antipathies—one to grammatical and verbal criticism, the other to false
doctrine and heresy. To the first we may ascribe his treating Bentley,
Burman, Kuster, and Wasse with a contempt which recoiled upon himself.
To the second we will impute his pious zeal against those divines of king
William, whom he supposed to be infected with the infidel, or the
socinian, or the latitudinarian spirit, and not so orthodox as himself, and
his friends Swift, Bolingbroke, etc. Thus he laid about him, and censured
men of whose literary, or of whose theological merits or defects, he was
no more a judge than his footman John Searle.—Dr. Jortin.
[233] Jortin asserted that by the unbelieving priests Pope alluded to
Burnet. If Jortin is right, the passage was not satire but falsehood. That
there was, however, much infidelity and socinianism during the reign of
William III. is proved by an address of the House of Commons to the
king, quoted by Bowles, beseeching his majesty to give effectual orders
for suppressing all pernicious books and pamphlets which contained
impious doctrines against the Holy Trinity, and other fundamental articles
of the protestant faith, tending to the subversion of the christian religion.
This address was presented in February 1698.
[234] In this line he had Kennet in view, who was accused of having said,
in a funeral sermon on some nobleman, that converted sinners, if they
were men of parts, repented more speedily and effectually than dull
rascals.—Jortin.
[235] The published sermons of the reign of William III. do not answer to
this description, which is certainly a calumny.
[236] So Lucretius, iv. 333:
Lurida præterea fiunt quæcunque tuentur Arquati.
Besides, whatever jaundice-eyes do view,
Looks pale as well as those, and yellow too.—Creech.
This notion of the transfusion of the colour to the object from a jaundiced
eye, though current in all our authors, is, I believe, a mere vulgar error.—
Wakefield.
It is still a disputed point whether jaundice ever affects the eye in a
degree to permit only the passage of the yellow rays. The instances are at
least very rare; but popular belief is a sufficient ground for a poetical
32. comparison. Pope had just exemplified his simile; for everything looked
yellow to him in the reign of William III.
[237] In the first edition,
Speak when you're sure, yet speak with diffidence.
Dennis objected that a man when sure should speak with a modest
assurance, and Pope wrote on the margin of the manuscript, Dennis, p.
21. Alter the inconsistency.
Pope's maxim was commended by Franklin. He found that his
overbearing, dictatorial manner roused needless opposition, and he
resolved never to use a word that imported a fixed opinion, but he
employed instead the qualifying phrases, I conceive, I imagine, or it
so appears to me at present. To this, he says, after my character of
integrity, I think it principally owing that I had early so much weight with
my fellow citizens when I proposed new institutions or alterations in the
old; for I was but a bad speaker, subject to much hesitation, and yet I
generally carried my point. He admits that his humility was feigned. Had
it been real there would have been no need for I conceive, I imagine,
which are implied without a tiresome, superfluous repetition. Unless the
dogmatism is in the mind opinions have not the tone of decrees.
[238] Warton praises Pope for practising the precept in correcting the
poems of Wycherley, and condemned Wycherley for ungenerously
resenting the candour of his critic. Bowles extenuates the conduct of
Wycherley, and says that the superannuated bard bore the corrections
with great temper till Pope seriously advised him to turn the whole into
prose. Warton and Bowles were deceived by the printed correspondence
of Pope and Wycherley, and were not aware that the letters were garbled
in the very particulars which relate to the cause of the quarrel,—a quarrel
so discreditable to Pope that he had recourse to forgery to shield himself
and throw the blame upon Wycherley. It is certain that the
superannuated bard did not take offence at the advice to turn his works
into prose, and there is no reason to doubt the contemporaneous report
that his anger arose from discovering that Pope, while professing
unlimited friendship, had made him the subject of some satirical verses.
[239] This picture was taken to himself by John Dennis, a furious old critic
by profession, who, upon no other provocation, wrote against this Essay
and its author in a manner perfectly lunatic: for, as to the mention made
of him in ver. 270, he took it as a compliment, and said it was
treacherously meant to cause him to overlook this abuse of his person.—
Pope.
33. Pope's acrimonious note on his early antagonist first appeared in the
edition of 1743, when Dennis had been some years dead. His book
against me, the poet wrote to Caryll, Nov. 19, 1712, made me very
heartily merry in two minutes' time, and here we find him still smarting
with resentment after thirty years and upwards had gone by, and his
enemy was in the grave. The original reading in the manuscript of ver.
585 was But D—— reddens. The substituted name is taken from
Dennis's tragedy of Appius and Virginia, which appeared in 1709. The
stare was one of his characteristics. He starts, stares, and looks round
him at every jerk of his person forward, says Sir Richard Steele, when
describing his walk. The tremendous was not only a sarcasm on his
appearance, but on his partiality for the epithet, which was an old topic of
ridicule. If, said Gildon, in 1702, there is anything of tragedy in the
piece, it lies in the word 'tremendous,' for he is so fond of it he had rather
use it in every page than slay his beloved Iphigenia. Gay, in 1712,
jeeringly dedicated his Mohocks to Mr. D[ennis], and assigned, among the
reasons for the selection, that his theme was horrid and tremendous.
[240] This thought occurs also in Donne's fourth Satire, which our poet
has modernised:
And though his face be as ill
As theirs, which in old hangings whip Christ, still
He strives to look worse.—Wakefield.
[241] It may not be known to every reader that noblemen and the sons of
noblemen are admitted of course at our universities to the degree of M.A.,
after keeping the terms of two years.—Wakefield.
The privilege is now abolished.
[242] If Cibber was the dull fellow Pope would have had him thought, no
conduct could have been more proper towards him than that which Pope
here recommends. Pope seems to have anticipated Colley's subsequent
resolution to write as long as Pope could rail.—Bowles.
[243] Duke of Buckingham's Essay on Satire,
But who can rail so long as he can sleep?
[244] Pope may have derived this comparison from the Epilogue, written
by a person of honour, to Dryden's Secret Love:
But t'other day I heard this rhyming fop
Say critics were the whips, and he the top:
For as a top spins best the more you baste her,
So ev'ry lash you give, he writes the faster,
34. The author of the Epilogue was more exact than Pope, whose application
of the simile is inaccurate, for the top is in full spin when it is popularly
said to be asleep.
[245] Dryden's Aurengzebe:
The dregs and droppings of enervate love.—Steevens.
It has been suggested that he alludes to Wycherley.—Warton.
Whom else could the lines suit at that period, when Pope says, Such
bards we have? If Wycherley was intended, what must we think of Pope,
who could wound, in this manner, his old friend, for whom he professed
so much kindness, and who first introduced him to notice and patronage.
—Bowles.
The application was too obvious for Pope to have ventured on the lines
unless he had designed to expose his former ally. The original reading of
ver. 610 in the manuscript was,
But if incorrigible bards we view,
Know there are mad, c.
And the alteration turned an unappropriated description into a particular
censure on living men. Wycherley would be the last person to detect the
likeness, and relaxing, some months after the Essay appeared, in his
indignation against Pope, he praised the poem, according to a letter of
Cromwell, dated Oct. 26, 1711, but which rests on the authority of Pope
alone.
[246] In allusion to this class of pedants Gray said, Learning never
should be encouraged; it only draws out fools from their obscurity.
[247] A common slander at that time in prejudice of that deserving
author. Our poet did him this justice, when that slander most prevailed;
and it is now (perhaps the sooner for this very verse) dead and forgotten.
—Pope.
The accusation from which he defended Garth was brought against Pope
himself. This poem, says Johnson, of Cooper's Hill, had such reputation
as to excite the common artifice by which envy degrades excellence. A
report was spread that the performance was not Denham's own, but that
he had bought it of a vicar for forty pounds. The same attempt was made
to rob Addison of his Cato and Pope of his Essay on Criticism. The story
told was that Wycherley sent the Essay to Pope for his revision, and that
Pope published it as his own. Authors are not the only persons who are
exposed to such calumnies. The victories of a great general are almost
35. invariably imputed to some subordinate officer, and it was long a favourite
theory of the malignant that Napoleon owed his successes to Berthier, and
the Duke of Wellington to Sir George Murray.
[248] There is an ellipse of that after sacred, and of it after fops,
or the line is not English, and when the omitted words are supplied the
inversion is intolerable.
[249] The propriety of the specification in this proverbial remark is
founded on a circumstance no longer existing in our poet's time, and
derived, therefore, by him from older writers. In the reigns of James I.
and Charles I., says Pennant, the body of St. Paul's cathedral was the
common resort of the politicians, the newsmongers, and the idle in
general. It was called Paul's Walk, and the frequenters known by the
name of Paul's walkers.—Wakefield.
[250] Between this and ver. 624—
In vain you shrug and sweat and strive to fly:
These know no manners but in poetry.
They'll stop a hungry chaplain in his grace,
To treat of unities of time and place.—Pope.
[251] This stroke of satire is literally taken from Boileau:
Gardez-vous d'imiter ce rimeur furieux,
Qui, de ses vains écrits, lecteur harmonieux,
Aborde en récitant quiconque le salue,
Et poursuit de ses vers les passants dans la rue.
Il n'est temple si saint, des anges respecté,
Qui soit contre sa muse un lieu de sûreté.
Which lines allude to the impertinence of a French poet called Du Perrier,
who finding Boileau one day at church, insisted upon repeating to him an
ode during the elevation of the host.—Warton.
Boileau tells the incident of an individual poetaster. Pope generalises the
exceptional trait, and represents it to have been the usual practice of
foppish critics to talk criticism at the altar. The probability is that he had
never known an instance. The line For fools rush in, is certainly
fashioned, says Bishop Hurd, on Shakespeare, Richard iii. Act 1, Sc. 3:
Wrens make prey where eagles dare not perch.
[252] Virgil, Geo. iv. 194:
36. Excursusque breves tentant.
Nor forage far, but short excursions make.
Dryden.—Wakefield.
[253] Humanly is improperly put for humanely. The only authorised
sense of the former is belonging to man; of the latter, kindly,
compassionately.—Dr. George Campbell.
[254] Love to praise means a love of bestowing praise, but, as
Wakefield says, it is an obscure expression, and repugnant to usage.
37. [255] This is followed by two additional lines in the manuscript:
Such did of old poetic laws impart,
And what till then was fury turned to art.
[256] Between ver. 646 and 647, I have found the following lines, since
suppressed by the author:
That bold Columbus of the realms of wit,
Whose first discovery's not exceeded yet.
Led by the light of the Mæonian star,
He steered securely, and discovered far.
He, when all nature was subdued before,
Like his great pupil, sighed and longed for more;
Fancy's wild regions yet unvanquished lay,
A boundless empire, and that owned no sway.
Poets, c.—Warburton.
[257] Dr. Knightly Chetwood to Lord Roscommon:
Hoist sail, bold writers! search, discover far;
You have a compass for a polar star.—Wakefield.
[258] After ver. 648, in the first edition, came this couplet:
Not only nature did his laws obey,
But fancy's boundless empire owned his sway.
Dennis denied that nature obeyed the laws of Aristotle. The laws of
nature, he said, are unalterable but by God himself. Pope's language is
inaccurate.
[259] The obvious interpretation of this passage would be that poets,
Homer excepted, indulged in savage liberty till they were restrained by
the laws of Aristotle, which is inconsistent with ver. 92-99, where Pope
says that the Greek critics framed their laws from the practice of the
poets.
[260] He presided over wit by his Rhetoric and Poetics, and gave proofs
by his Physics that he had conquered nature. Pope's panegyric on the
dominion exercised by Aristotle is far inferior to Dryden's celebration of
the deliverance from it.
38. The longest tyranny that ever swayed
Was that wherein our ancestors betrayed
Their free-born reason to the Stagyrite,
And made his torch their universal light.
Had we still paid that homage to a name,
Which only God and nature justly claim,
The western seas had been our utmost bound,
Where poets still might dream the sun was drowned,
And all the stars that shine in southern skies
Had been admired by none but savage eyes.
[261] Oldham—
Each strain a graceful negligence does wear.—Wakefield.
[262] Before he goes ten lines further, said Dennis, he forgets himself,
and commends Longinus for the very contrary quality for which he
commended Horace. He commends Horace for judging coolly in verse,
and extols Longinus for criticising with fire in prose. With very little faith
in the traits he had ascribed to Horace, Pope was tempted in the
manuscript to reverse the characteristic, and write
He judged with spirit as he sung with fire.
He subsequently affixed to the original reading the note, Not to be
altered. Horace judged with coolness as Longinus with fire.
[263] Dennis notices that this couplet is borrowed from Roscommon's
Essay on Translated Verse:
Thus make the proper use of each extreme,
And write with fury, but correct with phlegm.
[264] In all Pope's works there cannot be found a couplet so paltry and
impertinent as this.—Wakefield.
The construction of the last line is deplorably faulty. Horace does not
suffer more by wits than he suffers by critics is Pope's meaning, but
interpreted by his language, he would be read as asserting that Horace
did not suffer more by wrong translations than critics suffered by wrong
quotations.
[265] Dionysius of Halicarnassus.—Pope.
These prosaic lines, this spiritless eulogy, are much below the merit of the
critic whom they are intended to celebrate.—Warton.
39. A most meagre account of a very excellent and judicious critic. But what
can we expect when men overstep the limit of their enquiries, and rush in
where learning has not authorised them to tread.—Wakefield.
The lines first appeared in the second edition. In a pretended letter to
Addison, dated October 10, 1714, Pope speaks of having found a
particular remark in one of the treatises of the Greek critic, but he had
probably never looked into the original when this couplet was written, and
seems to have falsely inferred from chance quotations that the comments
upon Homer were the special characteristic of the works of Dionysius.
Pope was indebted for the leading phrase in his couplet to a passage of
Rochester, quoted by Wakefield:
Compare each phrase, examine ev'ry line,
Weigh ev'ry word, and ev'ry thought refine.
[266] This dissolute and effeminate writer little deserved a place among
good critics for only two or three pages on the subject of criticism.—
Warton.
It is to be suspected that Pope had never read Petronius, and mentioned
him on the credit of two or three sentences which he had often seen
quoted, imagining that where there was so much, there must necessarily
be more. Young men, in haste to be renowned, too frequently talk of
books which they have scarcely seen.—Johnson.
If Pope had been acquainted with the general tenor of the fragments
which remain of Petronius, he would not have celebrated the most corrupt
and disgusting writer of antiquity for an unalloyed combination of
charming qualities.
[267] To commend Quintilian barely for his method, and to insist merely
on this excellence, is below the merit of one of the most rational and
elegant of Roman writers. Considering the nature of Quintilian's subject,
he afforded copious matter for a more appropriate and poetical character.
No author ever adorned a scientifical treatise with so many beautiful
metaphors.—Warton.
[268] In the early editions,
Nor thus alone the curious eye to please,
But to be found, when need requires, with ease.
[269]
The Muses sure Longinus did inspire.—Pope.
40. The taste and sensibility of Longinus were exquisite; but his observations
are too general, and his method too loose. The precision of the true
philosophical critic is lost in the declamation of the florid rhetorician.
Instead of showing for what reason a sentiment or image is sublime, and
discovering the secret power by which they affect a reader with pleasure,
he is ever intent on producing something sublime himself, and strokes of
his own eloquence.—Warton.
[270] This verse is ungrammatical. With respect to the thought, Boileau,
whose translation of Longinus our poet had most probably read, has said,
in his preface to that work, exactly the same thing: Souvent il fait la
figure qu'il enseigne; et, en parlant du sublime, il est lui-même très-
sublime. Pope's couplet seems indebted also to the Prologue of Dryden's
Tempest, speaking of Shakespeare:
He, monarch-like, gave those his subjects law;
And is that nature, which they paint and draw.—Wakefield.
Wakefield calls ver. 680 ungrammatical, because, literally construed, it
reads, And whose own example is himself, etc.
[271] Felt is a flat, insipid word in this place.—Wakefield.
[272] Rome, as invariably pronounced by Pope's contemporaries had
the same sound with doom, and the pronunciation is not quite obsolete
in our own time among persons who were born in the last century. In the
previous part of the poem he had made Rome rhyme to dome, which
itself was often pronounced like doom.
[273] The superstition of some ages after the subversion of the Roman
Empire, wrote Pope to Caryll, July 19, 1711, is too manifest a truth to
be denied, and does in no sort reflect upon the present catholics, who are
free from it. Our silence on these points may, with some reason, make our
adversaries think we allow and persist in those bigotries, which in reality
all good and sensible men despise, though they are persuaded not to
speak against them. Most of Pope's associates were men of letters or
men of the world, and he could know little of the spirit of the church to
which he nominally belonged, when he was simple enough to believe that
the Romanists of his day would sanction a sweeping denunciation of the
ignorance and superstition of the monks.
[274]
All was believed, but nothing understood.—Pope.
[275] Between ver. 690 and 691, the author omitted these two:
41. Vain wits and critics were no more allowed,
When none but saints had licence to be proud.—Pope.
[276] Here he forms the tenses wrong.—Wakefield.
Pope told Caryll that he did not speak in this couplet of learning in
general, but of polite learning,—criticism, poetry, etc.—which was the only
learning concerned in the subject of the Essay. He at the same time
confessed his belief that the learning which the monks possessed was
barely kept alive by them. The explanation would not contribute to
conciliate the offended catholics.
[277] The glory from his own greatness, the shame from the rancour
with which some of his brother priests assailed him.—Croker.
Oldham in his Satire:
On Butler, who can think without just rage,
The glory and the scandal of the age.—Wakefield.
Pope avowed his conviction to Caryll that the priests had openly accused
him of heterodoxy in other passages of his poem, because they were
secretly exasperated at his eulogy upon Erasmus. What in their own
opinion, he said, they are really angry at is that a man whom their tribe
oppressed and persecuted should be vindicated after a whole age of
obloquy by one of their own people, who is free and bold enough to utter
a generous truth in behalf of the dead, whom no man sure will flatter, and
few do justice to.
[278] If the restoration of learning consisted in recovering the works and
reviving the spirit of the ancients, it had been in a great degree
accomplished before the time of Erasmus.—Roscoe.
[279] Genius is here personified, and this person is said by Pope to have
been spread over the ruins of Rome. The poet has evidently mixed up
genius considered as a quality of the remains of antiquity with genius
considered as a presiding being.
[280] For the expression in the last half of this verse, Wakefield quotes
Addison on sculpture in the letter from Italy,
Or teach their animated rocks to live.
And for the expression in the first half, he quotes Dryden's Religio Laici:
Or various atoms, interfering dance,
Leaped into form.
42. Wakefield ascribes the origin of the phrase to the fable that the stones of
Thebes moved into their places at the music of Amphion, and it is thus
used by Waller in his poem upon his Majesty's Repairing of St. Paul's:
He like Amphion makes those quarries leap
Into fair figures from a confused heap.
[281] Leo was not only an admirer of music, but a skilful performer, and
we are informed by Pietro Aaron that though he had acquired a
consummate knowledge in most arts and sciences, he seemed to love,
encourage, and exalt music more than any other. To sacred music he
paid a more particular attention, and sought throughout Europe for the
most celebrated performers, both vocal and instrumental.—Roscoe.
[282] M. Hieronymus Vida, an excellent Latin poet, who writ an Art of
Poetry in verse. He flourished in the time of Leo X.—Pope.
But Vida was by no means the most celebrated poet that adorned the age
of Leo X. His merits seem not to have been particularly attended to in
England till Pope had bestowed this commendation upon him, although
the Poetics had been correctly published at Oxford by Basil Kennet some
time before. They are perhaps the most perfect of his compositions; they
are excellently translated by Pitt.—Warton.
[283] The ancients, says the writer of the Supplement to the Profound,
always gave ivy to the poets, as may appear from numberless places in
the classics, nor was it ever applied to patrons or critics, in
contradistinction to poets, by any but this ingenious author.
[284] Alluding to
Mantua, væ miseræ, nimium vicina Cremonæ. Virg.—Warburton.
This application is made in Kennet's edition of Vida.—Warton.
To say that the birth-place of Vida would be next in fame to the birth-
place of Virgil was to rank him before all the other poets that Italy had
produced—before Dante, Petrarch, Tasso, and Ariosto. The antithesis is
marred by its want of truth.
[285] This, Warburton says, refers to the sack of Rome by the Duke of
Bourbon, which Pope assumed had driven poetry out of Italy. The
assigned cause is inadequate to account for the effect.
[286] The born to serve is a sarcasm on the readiness with which the
French submitted to the despotism of Louis XIV.
43. [287] May I be pardoned for declaring it as my opinion, that Boileau's is
the best Art of Poetry extant? The brevity of his precepts, the justness of
his metaphors, the harmony of his numbers, as far as Alexandrine lines
will admit, the exactness of his method, the perspicacity of his remarks,
and the energy of his style, all duly considered, may render this opinion
not unreasonable. It is scarcely to be conceived, how much is
comprehended in four short cantos. He that has well digested these,
cannot be said to be ignorant of any important rule of poetry.—Warton.
Boileau is said by Pope to sway in right of Horace because the Frenchman
avowed that he based his Art of Poetry on that of the Roman. The English
poet has been indebted to both.
[288] The comparison fails. The Romans of old subdued the Britons, and
ruled over them for centuries.
[289] Essay on Poetry, by the Duke of Buckingham. Our poet is not the
only one of his time who complimented this Essay and its noble author.
Mr. Dryden had done it very largely in the Dedication to his Translation of
the Æneid; and Dr. Garth, in the first edition of his Dispensary, says:
The Tyber now no courtly Gallus sees,
But smiling Thames enjoys his Normanbys;
though afterwards omitted, when parties were carried so high in the reign
of Queen Anne, as to allow no commendation to an opposite in politics.
The duke was all his life a steady adherent to the church of England party,
yet an enemy to the extravagant measures of the court in the reign of
Charles II. On which account, after having strongly patronized Mr. Dryden,
a coolness succeeded between them on that poet's absolute attachment
to the court, which carried him some length beyond what the duke could
approve of. This nobleman's true character had been very well marked by
Mr. Dryden before:
The muse's friend,
Himself a muse. In Sanadrin's debate
True to his prince, but not a slave of state.
Abs. and Achit.
Our author was more happy; he was honoured very young with his
friendship, and it continued till his death in all the circumstances of a
familiar esteem.—Pope.
The Duke of Buckingham, in his Essay, has followed the method of
Boileau, in discoursing on the various species of poetry in their different
gradations, to no other purpose than to manifest his own inferiority. His
44. Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com