Colour in English: from metonymy to metaphor

Colour words abound with figurative meanings, expressing much more than visual signals. Some of these figurative properties are well known; in English, for example, black is associated with EVIL and blue with DEPRESSION. Colours themselves are also described in metaphorical terms using lexis from other domains of experience, such as when we talk of deep blue, drawing on the domain of spatial position. Both metaphor and colour are of central concern to semantic theory; moreover, colour is recognised as a highly productive metaphoric field. Despite this, comparatively few works have dealt with these topics in unison, and even those few have tended to focus on Basic Colour Terms (BCTs) rather than including non-BCTs. This thesis addresses the need for an integrated study of both BCTs and non-BCTs, and provides an overview of metaphor and metonymy within the semantic area of colour. Conducted as part of the Mapping Metaphor project, this research uses the unique data source of the Historical Thesaurus of English (HT) to identify areas of meaning that share vocabulary with colour and thus point to figurative uses. The lexicographic evidence is then compared to current language use, found in the British National Corpus (BNC) and the Corpus of Contemporary American (COCA), to test for currency and further developments or changes in meaning. First, terms for saturation, tone and brightness are discussed. This lexis often functions as hue modifiers and is found to transfer into COLOUR from areas such as LIFE, EMOTION, TRUTH and MORALITY. The evidence for cross-modal links between COLOUR with SOUND, TOUCH and DIMENSION is then presented. Each BCT is discussed in turn, along with a selection of non-BCTs, where it is revealed how frequently hue terms engage in figurative meanings. This includes the secondary BCTs, with the only exception being orange, and a number of non-BCTs. All of the evidence discussed confirms that figurative uses of colour originate through a process of metonymy, although these are often extended into metaphor.

Machine learning techniques for implicit interaction using mobile sensors

Interactions in mobile devices normally happen in an explicit manner, which means that they are initiated by the users. Yet, users are typically unaware that they also interact implicitly with their devices. For instance, our hand pose changes naturally when we type text messages. Whilst the touchscreen captures finger touches, hand movements during this interaction however are unused. If this implicit hand movement is observed, it can be used as additional information to support or to enhance the users’ text entry experience. This thesis investigates how implicit sensing can be used to improve existing, standard interaction technique qualities. In particular, this thesis looks into enhancing front-of-device interaction through back-of-device and hand movement implicit sensing. We propose the investigation through machine learning techniques. We look into problems on how sensor data via implicit sensing can be used to predict a certain aspect of an interaction. For instance, one of the questions that this thesis attempts to answer is whether hand movement during a touch targeting task correlates with the touch position. This is a complex relationship to understand but can be best explained through machine learning. Using machine learning as a tool, such correlation can be measured, quantified, understood and used to make predictions on future touch position. Furthermore, this thesis also evaluates the predictive power of the sensor data. We show this through a number of studies. In Chapter 5 we show that probabilistic modelling of sensor inputs and recorded touch locations can be used to predict the general area of future touches on touchscreen. In Chapter 7, using SVM classifiers, we show that data from implicit sensing from general mobile interactions is user-specific. This can be used to identify users implicitly. In Chapter 6, we also show that touch interaction errors can be detected from sensor data. In our experiment, we show that there are sufficient distinguishable patterns between normal interaction signals and signals that are strongly correlated with interaction error. In all studies, we show that performance gain can be achieved by combining sensor inputs.