Digital intelligence: models, processes and technologies

1995

1987 articulates this view further, and it is popular among computationalists. Each of the elds has its more empirical and more theoretical aspects|the distinction at hand is orthogonal. Linguistics has its descriptive and theoretical perspectives, and so does computational linguistics. Computationalists describe concrete algorithms and architectures (and report on their relative successes), but they also analyze these theoretically|in terms of their decidability, time/space complexity, the data structures they require, and, in the case of parallel algorithms, the communication protocols needed. The division of theoretical labor suggested here is sometimes obscured by the many other purposes which computers serve in linguistic research, e.g., as vehicles for projects in applied linguistics (natural language interfaces, information retrieval, computer-assissted language learning, etc.); as visualization tools; as laboratories for linguistic experiments; as channels to immense data reserves in the form of corpora; as repositories for data organization, storage, and retrieval; etc. But clearly the use of computers cuts across the usual divisions of theory/application, theory/experiment and theory/data.

2003

Musicology Today, 2020

In the last few years, the field of machine learning has experienced unprecedented development. The availability of big data offers automation of almost all human domains, with algorithmic composition being no exception. The primary motivation for using machine learning to generate musical scores is the generality of these models. Unlike rule-based architectures, the system can adapt to any style or genre through the training process. However, by directly applying machine learning algorithms to generate musical content, we encounter certain limits. This paper analyzes the methods of music notation and compares the processing of a symbolic knowledge representation with the processing of a raw audio signal.

2013

A model of computation on multi-dimensional words, based on the overlapping operation, induces a categorical structure, to which a new type of logic corresponds, whose formulae express properties of computations in a language containing all first order formulae. While the resulting deductive system is strictly less powerful than (intuitionistic) first order logic, it is more powerful than coherent logic. The approach is illustrated through an example of an online game of map-colouring.

Proceedings of the 20th annual meeting on Association for Computational Linguistics -, 1982

We argue that because the very concept of computation rests on notions of interpretation, the semantics of natural languages and the semantics of computational formalisms are in the deepest sense the same subject. The attempt to use computational formalisms in aid of an explanation of natural language semantics, therefore, is an enterprise that must be undertaken with particular care. We describe a framework for semantical analysis that we have used in the computational realm, and suggest that it may serve to underwrite computadonally-oriented linguistic ser.antics as well. The major feature of this framework is the explicit recognition of both the declarative and the procedural import of meaningful expressions; we argue that whereas these two viewpoints have traditionally been taken as alternative, any comprehensive semantical theory must account for how both aspects of an expression contribute to its overall significance.

In the paper it is argued that the relation between the socalled »Gutenberg galaxis« of printculture and the »Turing galaxis« of digital media is not one of op- position and substitution, but rather one of co-evolution and integration. Or more precisely: that the Gutenberg–galaxis on the one hand can be inscribed into the Turing–galaxis, which on the other hand is textual in character since it is based on linear and serially processed representations manifested in a binary alphabet. In continuation of this text and hypertext is described as notions of related but different sorts of complex systems and it is argued that both texts and hypertexts are best thought of as sequentially organised. The first part of the paper contains a description of the basic properties of computers and digital representation answering the question: what are the properties common to all kinds of use of computers – whether used for one purpose or another? The second part deals with the relationship of printed and electronic text and the third part adresses the idea of hypertext as a system which exploits modal shifts between node-mode and link-mode as a significant part of the semantics of the system.

Leonardo electronic almanac, 2013

As language (both writing/speech) rapidly changes due to on-going developments in speech recognition systems, text-to-speech and chat bots, this paper focuses on the various attempts to synthesize, and mechanize language over time: to submit it to the logical, rational, and mechanical: to atomize and/or render it as pure code. This involves looking afresh at the kinds of philosophical questions these developments raise with respect to language – as a ”˜human’ phenomenon – which is increasingly being mediated by technology. The question – what is language when it is made by a machine? – touches upon ethical concerns; the notion of linguistic agency, and the shifting relationship between language and thought. While attempts at synthesising speech can be traced back as far as Roger Bacon (1200s) and Christian Kratzenstein, (1770s) more recent attempts to mechanize speech include early 20th c. mechanisms for encoding speech, such as the 1939 World’s Fair ”˜Parallel Bandpass Vocoder’ and...

1975

1989

of the formal system. Part IImay beread independently of Part I, though Part II provides the formal de nitions underlying the syntactic analysis outlined in Part I. A brief conclusion is followed by three appendices describing di erent aspects of ECAT. ECAT is an LA-grammar of a sizeable fragment of English, which is implemented in Lisp as a parser. The parsers for English ECAT and German DCAT were originally developed at CSLI-Stanford in INTERLISP-D on a Xerox 1109 machine. After translation into Common Lisp, ECAT was substantially expanded. The Common Lisp version of semantically interpreted ECAT presented in this book was developed on an HP-9000 work station, but has been successfully brought up on Micro-VAXes, TI-explorers, and SUN work stations. The parsers for the natural and formal languages described in this book are all running at the Laboratory for Computational Linguistics of Carnegie Mellon University. The sample derivations contained in the following pages are undoctored verbatim outputs of the parsers.