Search

Interpretation is a key aspect to any qualitative research process. Inherent in analysis, researchers must ask what informs interpretations and whose interpretations are accepted as reality. These questions call attention to power and its connection with interpretational practices, while some might argue that it would be impossible to render any interpretations that are data. In this article, we describe our own feminist, collective analytic interpretation as being guided by a set of commitments rather than a set of procedures or a guiding theoretical framework. We highlight how our approach to collective interpretation through a set of feminist commitments is achieved. Then, we outline those commitments, leaving readers with an idea of how to build a feminist collectivist interpretation process into their own work specifically or contemplate the collaborative nature of interpretation in analysis of qualitative data more broadly. [ABSTRACT FROM AUTHOR]

This article looks back on the professional development offered to teachers of English to support them in their teaching of film since the introduction of film into the English syllabus in Ireland in 1998. It focuses on three inter-related elements: the interpretation or reading of film; the pedagogy that supports this interpretation; the potential of exploring issues of social justice and the self-other relation through narrative films. The article outlines four distinct phases in the professional development offered to teachers and shares the emerging findings. The first phase involved the introduction of film and its use in opening up discussion and creating interpretative communities in teacher education workshops for teachers of Leaving Certificate English. The second involved an exploration of film genre in workshops designed for teachers of Transition Year. The third involved the development of a dialogic form of pedagogy in interpreting film in a series of workshops directed at teachers of Junior Cycle English. The fourth involved the use of film in exploring education for justice in a series of workshops for pre-service teachers on Bachelor of Education and Master of Education programmes. The article explores the link among narrative film texts, generative questions, thoughtful interpretation, and the value given to dialog and the movement of question and answer in professional development workshops where narrative films are viewed and interpreted. The article is autoethnographic in character (Holman Jones 2005; Ellis et al. 2010). It involves self-reporting, descriptions of practice and reflection on that practice.

The present paper aims to extract the inference mechanism of neural networks, supposed to be hidden in complicated surface phenomena, by maximizing information in terms of multiple selective constraints. For easy interpretation of the meaning of information content, information is represented in terms of selectivity of components, or selective potentiality. The selective potentiality represents an ability of neurons to respond selectively to inputs, and this selectivity should exclusively increase when going through different neurons. In addition, because the selectivity can be realized by increasing the strength of connection weights, we try to reduce this strength as much as possible, namely, cost minimization. The selectivity and cost are hierarchically applied as multiple constraints, disentangling complicated components to make the functions of neurons and connection weights as clear as possible, leading us to find the inner inference mechanism. The method was applied to the simple qualitative bankruptcy and more complicated bank marketing data sets, where the number of hidden layers increased to 15 to examine how multi-layered networks could be used to disentangle complicated components. Experimental results showed that the selective potentiality could disentangle connection weights and eventually produce linear and individual features for easy interpretation. [ABSTRACT FROM AUTHOR]

This paper aims to extend collective interpretation to networks with complicated components. The collective interpretation is used to generate an internally interpretable model independently of specific inputs and learning conditions. The internally interpretable model is obtained by network compression where multi-layers are sequentially compressed, taking into account all possible routes from inputs to outputs. The network compression is easily applied to fully connected networks, but it cannot be applied to some networks with complicated components. Thus, to make the compression possible, we black-box partially and minimally these components to be replaced by the ordinary components. For demonstrating the effectiveness of this technique, we use here a new model based on the self-organizing map (SOM). Then, we introduce the convolutional neural networks (CNN) for dealing with SOM knowledge, usually represented in two-dimensional lattices. Because our network compression cannot deal with those convolutional components, we temporarily black-box the CNN components. Fixing the other connection weights, we re-train the partially black-boxed network to obtain the simplest prototype model for interpretation. The method was applied to two well-known data sets, and we demonstrated that the present method could compress the networks to get the simplest and interpretable ones. In addition, very stable compressed weights for interpretation could be obtained for easy interpretation. The results suggest that the main mechanism of multi-layered neural networks is based on linear relations between individual inputs and targets, to which peripheral non-linear ones are added. [ABSTRACT FROM AUTHOR]

The present paper aims to propose a new type of information-theoretic method to interpret the inference mechanism of neural networks. We interpret the internal inference mechanism for itself without any external methods such as symbolic or fuzzy rules. In addition, we make interpretation processes as stable as possible. This means that we interpret the inference mechanism, considering all internal representations, created by those different conditions and patterns. To make the internal interpretation possible, we try to compress multi-layered neural networks into the simplest ones without hidden layers. Then, the natural information loss in the process of compression is complemented by the introduction of a mutual information augmentation component. The method was applied to two data sets, namely, the glass data set and the pregnancy data set. In both data sets, information augmentation and compression methods could improve generalization performance. In addition, compressed or collective weights from the multi-layered networks tended to produce weights, ironically, similar to the linear correlation coefficients between inputs and targets, while the conventional methods such as the logistic regression analysis failed to do so.

Sentences with plural expressions can receive at least two interpretations. For example, the sentence The boys hold a balloon could mean that the boys as a group jointly hold one balloon (the collective reading) or that each boy holds one balloon, which would imply that as many balloons were held as there are boys (the distributive reading). Building on Frazier et al. (1999), we show that the human processor favors collective readings. Crucially, the preference for collective readings is only observed when the distributive reading has to be established through the means of phrasal distributivity (e.g., triggered by distributive quantifiers), and the preference disappears in case of lexical distributivity (e.g., the distributive interpretation of win). The findings provide evidence for different mental representations of the two types of distributivity and shed light on why the processor exhibits a default preference for collective interpretations.

The present paper aims to extend the potential learning method to overcome the problem of collective interpretation, which aims to interpret multi-layered neural networks by compressing them into the simplest ones. In the process of compression, positive, negative, and complicated weights have had unfavourable effects for interpretation. To deal with the problems of collective interpretation, the potential learning is extended only to use positive weights. In addition, to obtain more appropriate weights for interpretation, the number of candidate weights for higher potentialities is first increased as much as possible. Then, from among many candidates, more appropriate weights are selected as more important ones. This extended potentiality learning is expected to produce more stable and more simple representations for easy interpretation. The extended method was applied to three datasets, namely, an artificial dataset, a real eye-tracking dataset, and a student evaluation dataset. In all cases, it was observed that the selectivity of connection weights could be increased. Correspondingly, the majority of connection weights became positive, and the collective weights were quite similar to the regression coefficients of the logistic regression analysis. Finally, for the third dataset (student evaluations), the extended method could extract more explicit input-output relations, compared with the logistic regression analysis, while improving generalisation performance. [ABSTRACT FROM AUTHOR]

Abstract The present paper proposes a new method called "neural self-compressors" to compress multi-layered neural networks into the simplest possible ones (i.e., without hidden layers) to aid in the interpretation of relations between inputs and outputs. Though neural networks have shown great success in improving generalization, the interpretation of internal representations becomes a serious problem as the number of hidden layers and their corresponding connection weights becomes larger and larger. To overcome this interpretation problem, we introduce a method that compresses multi-layered neural networks into ones without hidden layers. In addition, this method simplifies entangled weights as much as possible by maximizing mutual information between inputs and outputs. In this way, final connection weights can be interpreted as easily as by the logistic regression analysis. The method was applied to four data sets: a symmetric data set, ovarian cancer data set, restaurant data set, and credit card holders' default data set. In the first set, the symmetric data set, we tried to explain how the present method could produce interpretable outputs intuitively. In all the other cases, we succeeded in compressing multi-layered neural networks into their simplest forms with the help of mutual information maximization. In addition, by de-correlating outputs, we were able to transform connection weights from those close to the regression coefficients to ones with more explicit features. [ABSTRACT FROM AUTHOR]

We propose a cognitively plausible formal model of collective interpretation. The model represents how members of a collective interact to interpret their environment. Current theories of collective interpretation focus on how heedful communication among members of a collective (i.e., how much individuals pay attention to others' interpretations) improves interpretive performance; their general assumption is that heed tends to be uniformly beneficial. By unpacking the micromechanisms that underlie such performance, our model reveals a more complex story. Heedfulness can benefit interpretive performance. It can help collectives properly interpret situations that are especially ambiguous, unknown, or novel. Conversely, heedfulness also generates conformity pressures that induce agents to give too much weight to others' interpretations, even if erroneous, thereby potentially degrading interpretive performance. These two effects join into a nonmonotonic trajectory that represents how heed relates to interpretive performance: due to its beneficial properties, performance increases with heed until it peaks before degrading due to conformity pressures. The form of this nonmonotonic relationship is contingent on the nature of the task: ambiguous situations make collectives vulnerable to too much heed: ambiguity ignites conformism; novel situations make collectives dependent on heed: novelty requires multiple eyes to be seen. In addition to these results, our model offers a flexible platform that future work can use to explore collective interpretation in a variety of organizational and supraorganizational contexts. [ABSTRACT FROM AUTHOR]

Centre (center) and milieu (middle) are Nouns of Internal Localization (NIL). They combine with the preposition à to form complex locative preposition phrases such as NP0 au milieu de NP1 (at the middle of) and NP0 au centre de NP1 (at the center of). Referentialist semantics characterizes these constructions with reference to a geometric interpretation of centre and milieu. This paper challenges the referentialist approach and shows that the distinction between au milieu de and au centre de hinges on two semantic criteria. The first criterion has to do with whether NP1 is bounded or not. The second criterion concerns the functional meaning of NP0 with respect to NP1. These two criteria account for linguistic phenomena that range from the concrete domain (concrete nouns in the singular and in the plural, mass nouns, collective nouns) to the abstract domain (intensifying names and action names).The study of these two preposition phrases is also a good opportunity to propose a renewed approach to localization based on philosophy, and more precisely phenomenology. We develop the following idea : au centre de/au milieu de impose distinct perspectivizing effects and localize radically different elements, even when these elements are apparently synonyms. To substantiate these claims, we use the Gestalt-based opposition between figure and ground. We show that au centre de makes a functional link with a bounded whole (a figure), whereas au milieu de implies a coincidence with a substance whose boundaries are not specified.

Two experiments are reported that show that introducing event participants in a conjoined noun phrase (NP) favours a single event (collective) interpretation, while introducing them in separate clauses favours a separate events (distributive) interpretation. In Experiment 1, acceptability judgements were speeded when the bias of a predicate toward separate events versus a single event matched the presumed bias of how the subjects' referents were introduced (as conjoined noun phrases or in conjoined clauses). In Experiment 2, reading of a phrase containing an anaphor following conjoined noun phrases was facilitated when the anaphor was they, relative to when it was neither/each of them; the opposite pattern was found when the anaphor followed conjoined clauses. We argue that comprehension was facilitated when the form of an anaphor was appropriate for how its antecedents were introduced. These results address the very general problem of how we individuate entities and events when presented with a complex situation and show that different linguistic forms can guide how we construe a situation. The results also indicate that there is no general penalty for introducing the entities or events separately—in distinct clauses as “split” antecedents. [ABSTRACT FROM AUTHOR]