Despite developments in phonology over the last few decades, models of verbal working memory make... more Despite developments in phonology over the last few decades, models of verbal working memory make reference to phoneme-sized phonological units, rather than to the features of which they are composed. This study investigates the influence on short-term retention of such features by comparing the serial recall of lists of syllables with varying types and levels of similarity in their onset consonants. Lists are (a) dissimilar (/fa-na-ga/) (b) acoustically similar (/pa-ta-ka/) or (c) articulatorily similar (/da-la-za/). When no overt articulation is required, we find no decrease in performance for articulatorily similar items as compared to the dissimilar list. However, we are able to show that acoustic similarity clearly impairs recall. It is only when participants recall the lists orally, that performance is impaired for both types of similar lists. These results have implications for accounts of the phonological similarity effect in particular and of verbal working memory in general.
This study investigates the relation between vowel identity and emotional state. In Experiment 1,... more This study investigates the relation between vowel identity and emotional state. In Experiment 1, (pseudo)words were invented and articulated in a positive or negative mood condition. Subjects in a positive mood produced more words containing /i:/, a vowel involving the same muscle that is used in smiling--the zygomaticus major muscle (ZMM). Subjects in a negative mood produced more words containing /o:/, involving an antagonist of the ZMM--the orbicularis orbis muscle (OOM). We argue that the link between mood and vowel identity is related to orofacial muscle activity, which provides articulatory feedback to speakers on their emotional state. Experiment 2 tests this hypothesis more specifically. Participants rated the funniness of cartoons while repeatedly articulating either /i:/ (ZMM) or /o:/ (OOM). In line with our hypothesis, the cartoons were rated as funnier by subjects articulating /i:/ than by those articulating /o:/.
Individual Differences in Speech Production and Perception
In this contribution we explore the hypothesis of an interaction between speaker-and listener-spe... more In this contribution we explore the hypothesis of an interaction between speaker-and listener-specific strategies in the encoding and decoding of intonational contrasts� Intonational categories, such as the pitch accents used in the signalling of focus types, can be cued by different phonetic exponents, such as peak alignment or duration of the target words� Through a production task we document speakerspecific strategies: Individual speakers might use more or fewer cues than others (robustness) when encoding intonational contrasts, and each cue can be used to encode one or more contrasts (partitioning)� We show in a subsequent perception task that listeners are sensitive to speaker-specific strategies, since correct identification scores for productions of individual speakers mirror the robustness and partitioning of speakers' productions� Moreover, listeners vary as to how reliably they decode intonational contrasts across speakers� However, in line with the hypothesis of an interaction between speaker-and listener-specific behaviours, some listeners are more reliable at decoding contrasts as encoded by some particular speakers, which in turn are decoded less reliably by other listeners� These findings suggest that phonetic cues to intonational contrasts should not be understood as singly necessary and jointly sufficient features for category membership, but rather as dimensions along which phonological categories cluster, in an individual-specific network of phonological knowledge�
Cette étude examine la variation dans l'alignement des cibles intonatives haute et basse des acce... more Cette étude examine la variation dans l'alignement des cibles intonatives haute et basse des accents de hauteur montants L+H* relativement aux repères acoustiques de la chaîne segmentale au moyen desquels ils sont réalisés. Une explication à cette variation sera présentée en termes d'ajustements dans l'alignement des cibles de F0 en relation avec des gestes articulatoires définis de manière dynamique.
We report on two production experiments which together provide additional support for treating do... more We report on two production experiments which together provide additional support for treating downstep as orthogonal to the tonal structure of utterances, in the sense that certain intonational meanings are expressed by a downstep relation between two H tones, rather than by a particular accent type or accent sequence. The intonational meanings investigated were inferentially accessible information on the one
The incorporation of personality is indispensable for modelling life-like agents, as it functions... more The incorporation of personality is indispensable for modelling life-like agents, as it functions "as a generative engine that contributes to coherence, consistency, and predictability in emotional reactions and responses" of agents .
The incorporation of personality is indispensable for modelling life-like agents, as it functions... more The incorporation of personality is indispensable for modelling life-like agents, as it functions "as a generative engine that contributes to coherence, consistency, and predictability in emotional reactions and responses" of agents .
In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA syste... more In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA system. The NECA system generates interactions between two or more animated characters. The RRL is a formal framework for representing the information that is exchanged at the interfaces between the various NECA system modules.
A perception experiment investigated the appropriateness of a number of accent patterns in contex... more A perception experiment investigated the appropriateness of a number of accent patterns in contexts in which a referring expression can be regarded as neither completely given (already active in the listener's consciousness at the time of utterance) nor completely new (inactive), but rather in between the two, i.e. accessible (semi-active). Results clearly show that, for the purposes of intonation, accessible information cannot be treated as a uniform category. In a number of cases, one particular type of pitch accent, H+L* (early peak accent), is significantly preferred over another accent type, H* (medial peak accent), as well as over deaccentuation. These cases comprise whole-part relations, where the referent constituted a part of an already mentioned whole, and the scenario condition, where the referent was predictable from the contextually given schema, or frame. The remaining types of accessible information were shown to be preferentially deaccented, with a second choice for H+L* rather than H*. These cases comprised relations such as converseness, part-whole (in that order only), synonymy, and hypernymhyponym (in either order). These findings point to the intermediate status of H+L* for the signaling of information states. #
German Intonation in Autosegmental-Metrical Phonology Martine Grice, Stefan Baumann, and Ralf Ben... more German Intonation in Autosegmental-Metrical Phonology Martine Grice, Stefan Baumann, and Ralf Benzmuller 3.1. INTRODUCTION English, Dutch, and German are often claimed to have very similar prosody and intonation, an observation that might be related to the fact that ...
A first attempt was made to align gestures to F0 targets by D'Imperio et ... more A first attempt was made to align gestures to F0 targets by D'Imperio et al (2003). Based on optoelectronic data they investigated H pitch targets in Neapolitan Italian statements/questions and their relationship to orofacial movements. Their study in fact showed a close relationship between F0 maxima and inter-lip distances for statements (peak lip aperture between two labials) and questions (minimum
In this paper we describe a commercial application of a net environment, and present the user dat... more In this paper we describe a commercial application of a net environment, and present the user data we have collected so far from three launches of this application. A net environment in our definition is a virtual space inhabited by avatars which have been created and are subsequently visited and instructed by users via the internet. Net environments are a useful means for studying user behaviour in general, and they are particularly well suited for presentation of multimedia content and systematic gathering of user responses on the appropriateness or effectiveness of the different presentations.
In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA syste... more In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA system. The NECA system generates interactions between two or more animated characters. The RRL is a formal framework for representing the information that is exchanged at the interfaces between the various NECA system modules.
In this paper we investigate acoustic and articulatory anchors for F0 targets corresponding to pr... more In this paper we investigate acoustic and articulatory anchors for F0 targets corresponding to prenuclear and nuclear accent peaks in German, both across two different articulation rates (normal and fast) and across two different syllable structures (CV: and CVC). For the articulatory measurements we used Electromagnetic Midsagittal Articulography (EMMA).
This paper reports on a production experiment in German eliciting focus domains of various sizes,... more This paper reports on a production experiment in German eliciting focus domains of various sizes, ranging from broad to narrow focus, as well as contrastive focus. Results show that speakers use categorical as well as gradient prosodic means to indicate different focus structures, with an increase of prominence-lending cues as the focus domain narrows. Contrast is shown to enhance certain differences between narrow and broad focus. There is a clear indication that speakers differ considerably as to the combination of strategies they employ for marking focus structure.
A diverse set of speech data was labelled in three sites by 13 transcribers with differing levels... more A diverse set of speech data was labelled in three sites by 13 transcribers with differing levels of expertise, using GToBI, a consensus transcription system for German intonation. Overall inter-transcriber-consistency suggests that, with training, labellers can acquire sufficient skill with GToBI for large-scale database labelling
Despite developments in phonology over the last few decades, models of verbal working memory make... more Despite developments in phonology over the last few decades, models of verbal working memory make reference to phoneme-sized phonological units, rather than to the features of which they are composed. This study investigates the influence on short-term retention of such features by comparing the serial recall of lists of syllables with varying types and levels of similarity in their onset consonants. Lists are (a) dissimilar (/fa-na-ga/) (b) acoustically similar (/pa-ta-ka/) or (c) articulatorily similar (/da-la-za/). When no overt articulation is required, we find no decrease in performance for articulatorily similar items as compared to the dissimilar list. However, we are able to show that acoustic similarity clearly impairs recall. It is only when participants recall the lists orally, that performance is impaired for both types of similar lists. These results have implications for accounts of the phonological similarity effect in particular and of verbal working memory in general.
This study investigates the relation between vowel identity and emotional state. In Experiment 1,... more This study investigates the relation between vowel identity and emotional state. In Experiment 1, (pseudo)words were invented and articulated in a positive or negative mood condition. Subjects in a positive mood produced more words containing /i:/, a vowel involving the same muscle that is used in smiling--the zygomaticus major muscle (ZMM). Subjects in a negative mood produced more words containing /o:/, involving an antagonist of the ZMM--the orbicularis orbis muscle (OOM). We argue that the link between mood and vowel identity is related to orofacial muscle activity, which provides articulatory feedback to speakers on their emotional state. Experiment 2 tests this hypothesis more specifically. Participants rated the funniness of cartoons while repeatedly articulating either /i:/ (ZMM) or /o:/ (OOM). In line with our hypothesis, the cartoons were rated as funnier by subjects articulating /i:/ than by those articulating /o:/.
Individual Differences in Speech Production and Perception
In this contribution we explore the hypothesis of an interaction between speaker-and listener-spe... more In this contribution we explore the hypothesis of an interaction between speaker-and listener-specific strategies in the encoding and decoding of intonational contrasts� Intonational categories, such as the pitch accents used in the signalling of focus types, can be cued by different phonetic exponents, such as peak alignment or duration of the target words� Through a production task we document speakerspecific strategies: Individual speakers might use more or fewer cues than others (robustness) when encoding intonational contrasts, and each cue can be used to encode one or more contrasts (partitioning)� We show in a subsequent perception task that listeners are sensitive to speaker-specific strategies, since correct identification scores for productions of individual speakers mirror the robustness and partitioning of speakers' productions� Moreover, listeners vary as to how reliably they decode intonational contrasts across speakers� However, in line with the hypothesis of an interaction between speaker-and listener-specific behaviours, some listeners are more reliable at decoding contrasts as encoded by some particular speakers, which in turn are decoded less reliably by other listeners� These findings suggest that phonetic cues to intonational contrasts should not be understood as singly necessary and jointly sufficient features for category membership, but rather as dimensions along which phonological categories cluster, in an individual-specific network of phonological knowledge�
Cette étude examine la variation dans l'alignement des cibles intonatives haute et basse des acce... more Cette étude examine la variation dans l'alignement des cibles intonatives haute et basse des accents de hauteur montants L+H* relativement aux repères acoustiques de la chaîne segmentale au moyen desquels ils sont réalisés. Une explication à cette variation sera présentée en termes d'ajustements dans l'alignement des cibles de F0 en relation avec des gestes articulatoires définis de manière dynamique.
We report on two production experiments which together provide additional support for treating do... more We report on two production experiments which together provide additional support for treating downstep as orthogonal to the tonal structure of utterances, in the sense that certain intonational meanings are expressed by a downstep relation between two H tones, rather than by a particular accent type or accent sequence. The intonational meanings investigated were inferentially accessible information on the one
The incorporation of personality is indispensable for modelling life-like agents, as it functions... more The incorporation of personality is indispensable for modelling life-like agents, as it functions "as a generative engine that contributes to coherence, consistency, and predictability in emotional reactions and responses" of agents .
The incorporation of personality is indispensable for modelling life-like agents, as it functions... more The incorporation of personality is indispensable for modelling life-like agents, as it functions "as a generative engine that contributes to coherence, consistency, and predictability in emotional reactions and responses" of agents .
In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA syste... more In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA system. The NECA system generates interactions between two or more animated characters. The RRL is a formal framework for representing the information that is exchanged at the interfaces between the various NECA system modules.
A perception experiment investigated the appropriateness of a number of accent patterns in contex... more A perception experiment investigated the appropriateness of a number of accent patterns in contexts in which a referring expression can be regarded as neither completely given (already active in the listener's consciousness at the time of utterance) nor completely new (inactive), but rather in between the two, i.e. accessible (semi-active). Results clearly show that, for the purposes of intonation, accessible information cannot be treated as a uniform category. In a number of cases, one particular type of pitch accent, H+L* (early peak accent), is significantly preferred over another accent type, H* (medial peak accent), as well as over deaccentuation. These cases comprise whole-part relations, where the referent constituted a part of an already mentioned whole, and the scenario condition, where the referent was predictable from the contextually given schema, or frame. The remaining types of accessible information were shown to be preferentially deaccented, with a second choice for H+L* rather than H*. These cases comprised relations such as converseness, part-whole (in that order only), synonymy, and hypernymhyponym (in either order). These findings point to the intermediate status of H+L* for the signaling of information states. #
German Intonation in Autosegmental-Metrical Phonology Martine Grice, Stefan Baumann, and Ralf Ben... more German Intonation in Autosegmental-Metrical Phonology Martine Grice, Stefan Baumann, and Ralf Benzmuller 3.1. INTRODUCTION English, Dutch, and German are often claimed to have very similar prosody and intonation, an observation that might be related to the fact that ...
A first attempt was made to align gestures to F0 targets by D'Imperio et ... more A first attempt was made to align gestures to F0 targets by D'Imperio et al (2003). Based on optoelectronic data they investigated H pitch targets in Neapolitan Italian statements/questions and their relationship to orofacial movements. Their study in fact showed a close relationship between F0 maxima and inter-lip distances for statements (peak lip aperture between two labials) and questions (minimum
In this paper we describe a commercial application of a net environment, and present the user dat... more In this paper we describe a commercial application of a net environment, and present the user data we have collected so far from three launches of this application. A net environment in our definition is a virtual space inhabited by avatars which have been created and are subsequently visited and instructed by users via the internet. Net environments are a useful means for studying user behaviour in general, and they are particularly well suited for presentation of multimedia content and systematic gathering of user responses on the appropriateness or effectiveness of the different presentations.
In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA syste... more In this paper, we describe the Rich Representation Language (RRL) which is used in the NECA system. The NECA system generates interactions between two or more animated characters. The RRL is a formal framework for representing the information that is exchanged at the interfaces between the various NECA system modules.
In this paper we investigate acoustic and articulatory anchors for F0 targets corresponding to pr... more In this paper we investigate acoustic and articulatory anchors for F0 targets corresponding to prenuclear and nuclear accent peaks in German, both across two different articulation rates (normal and fast) and across two different syllable structures (CV: and CVC). For the articulatory measurements we used Electromagnetic Midsagittal Articulography (EMMA).
This paper reports on a production experiment in German eliciting focus domains of various sizes,... more This paper reports on a production experiment in German eliciting focus domains of various sizes, ranging from broad to narrow focus, as well as contrastive focus. Results show that speakers use categorical as well as gradient prosodic means to indicate different focus structures, with an increase of prominence-lending cues as the focus domain narrows. Contrast is shown to enhance certain differences between narrow and broad focus. There is a clear indication that speakers differ considerably as to the combination of strategies they employ for marking focus structure.
A diverse set of speech data was labelled in three sites by 13 transcribers with differing levels... more A diverse set of speech data was labelled in three sites by 13 transcribers with differing levels of expertise, using GToBI, a consensus transcription system for German intonation. Overall inter-transcriber-consistency suggests that, with training, labellers can acquire sufficient skill with GToBI for large-scale database labelling
Uploads
Papers by Martine Grice