prosodic features of speech worksheet pdf

To appear in Xu, Y. \(in press\). Prosody, Tone and Intonation. In The Routledge Handbook of Phonetics. W. F. Katz and P. F. Assmann: Routledge. \ ;

Xu, Y. \(2019\). Prosody, Tone and Intonation. In The Routledge Handbook of Phonetics. W. F. Katz and P. F. Assmann: Routledge, New York.

Xu, Y. \(2019\). Prosody, Tone and Intonation. In The Routledge Handbook of Phonetics. W. F. Katz and P. F. Assmann: Routledge, New York. pp. 314-35\ 6.

• Professional development
• Knowing the subject
• Teaching Knowledge database N-P

Prosodic features

Prosodic features are features that appear when we put sounds together in connected speech.

It is as important to teach learners prosodic features as successful communication depends as much on intonation, stress and rhythm as on the correct pronunciation of sounds.

Example Intonation, stress and rhythm are prosodic features.

In the classroom One way to focus learners on various aspects of prosody is to select a text suitable to be read aloud - for example a famous speech - and ask learners to mark where they think pauses, main stress, linking, and intonation changes occur. They can then practise reading this aloud.

See also: https://www.teachingenglish.org.uk/article/connected-speech-0 https://www.teachingenglish.org.uk/article/intonation https://www.teachingenglish.org.uk/article/stress

Research and insight

Browse fascinating case studies, research papers, publications and books by researchers and ELT experts from around the world.

See our publications, research and insight

Site-wide links

• Rochester Institute of Technology
• Directories
• Search RIT Search

Prosodic Features

Ratings of the prosodic features in speech are restricted to judgments of the speakers ability to blend sounds and words together, to stress syllables and words, and to modify pitch appropriately during oral reading. The slight modifications of pitch which occur over time during normal continuous speech is termed inflection.

PROSODIC FEATURES OF SPEECH

Quiz   by jereza caballes.

Feel free to use or edit a copy

includes Teacher and Student dashboards

Measure skills from any curriculum

Tag the questions with any skills you have. Your dashboard will track each student's mastery of each skill.

• edit the questions
• save a copy for later
• start a class game
• automatically assign follow-up activities based on students’ scores
• assign as homework
• share a link with colleagues
• print as a bubble sheet
• Q 1 / 15 Score 0 It is the variation of speech in a spoken language. 29 Pause Tempo  Stress Intonation

Our brand new solo games combine with your quiz, on the same screen

Correct quiz answers unlock more play!

• Q 1 It is the variation of speech in a spoken language. Pause Tempo  Stress Intonation 60 s
• Q 2 The rising falling intonation pattern is also called __________. 2-3-2 pattern 2-1-3pattern 2-3-1 pattern 2-3-3pattern 60 s
• Q 3 What prosodic feature of speech is easy to use and recognize in spoken language but harder to describe? Intonation Stress Pitch Volume 60 s
• Q 4 What do you call this pattern where the voice is raised on the stressed syllable and goes down on the other remaining syllables in the sentence? Rising Intonation Glide Falling Intonation Shift 60 s
• Q 5 This prosodic feature of speech is a non-fluency feature. Stress Pitch Tempo Pause 60 s
• Q 6 What do you call this pattern where the stressed syllable is the last syllable in the sentence? Rising Intonation Shift Glide Falling Intonation 60 s
• Q 7 7. It is the speaker's voice that varies between low and high that can also affect the meaning. Volume Pause Pitch Shift 60 s
• Q 8 It is an aspect of speech that goes beyond phonemes and deals with the auditory qualities of sound. Prosodic Features Pitch Volume Intonation 60 s
• Q 9 Which of the following is the correct flow of the 2-3-1 pattern? The voice starts with the normal pitch, then goes up on the accented syllable, and remains up until the end of the sentence. Go up to the accented syllable, then start with the normal pitch, and then go down on the last syllable in the sentence. Go down on the last syllable in the sentence, then start with the normal pitch, and remains up until the end of the sentence. Start with the normal pitch, then go up to the accented syllable, and then go down on the last syllable in the sentence. 60 s
• Q 10 It is also called speed, in which fast speech can convey urgency, whereas slower speech can be used for emphasis. Tempo Falling Intonation Volume Stress 60 s

• Q 13 A pause gives the listener time to understand your words. true false True or False 60 s
• Q 14 The 2-3-1 pattern is used for questions that are answerable by yes or no. false true True or False 60 s
• Q 15 Apart from the slight increase in loudness to indicate stress, volume is generally used to show emotions such as fear or anger. true false True or False 60 s

Teachers give this quiz to your class

• Search Menu
• Sign in through your institution
• Browse content in Arts and Humanities
• Browse content in Archaeology
• Anglo-Saxon and Medieval Archaeology
• Archaeological Methodology and Techniques
• Archaeology by Region
• Archaeology of Religion
• Archaeology of Trade and Exchange
• Biblical Archaeology
• Contemporary and Public Archaeology
• Environmental Archaeology
• Historical Archaeology
• History and Theory of Archaeology
• Industrial Archaeology
• Landscape Archaeology
• Mortuary Archaeology
• Prehistoric Archaeology
• Underwater Archaeology
• Zooarchaeology
• Browse content in Architecture
• Architectural Structure and Design
• History of Architecture
• Residential and Domestic Buildings
• Theory of Architecture
• Browse content in Art
• Art Subjects and Themes
• History of Art
• Industrial and Commercial Art
• Theory of Art
• Biographical Studies
• Byzantine Studies
• Browse content in Classical Studies
• Classical History
• Classical Philosophy
• Classical Mythology
• Classical Literature
• Classical Reception
• Classical Art and Architecture
• Classical Oratory and Rhetoric
• Greek and Roman Epigraphy
• Greek and Roman Law
• Greek and Roman Archaeology
• Greek and Roman Papyrology
• Late Antiquity
• Religion in the Ancient World
• Digital Humanities
• Browse content in History
• Colonialism and Imperialism
• Diplomatic History
• Environmental History
• Genealogy, Heraldry, Names, and Honours
• Genocide and Ethnic Cleansing
• Historical Geography
• History by Period
• History of Agriculture
• History of Education
• History of Emotions
• History of Gender and Sexuality
• Industrial History
• Intellectual History
• International History
• Labour History
• Legal and Constitutional History
• Local and Family History
• Maritime History
• Military History
• National Liberation and Post-Colonialism
• Oral History
• Political History
• Public History
• Regional and National History
• Revolutions and Rebellions
• Slavery and Abolition of Slavery
• Social and Cultural History
• Theory, Methods, and Historiography
• Urban History
• World History
• Browse content in Language Teaching and Learning
• Language Learning (Specific Skills)
• Language Teaching Theory and Methods
• Browse content in Linguistics
• Applied Linguistics
• Cognitive Linguistics
• Computational Linguistics
• Forensic Linguistics
• Grammar, Syntax and Morphology
• Historical and Diachronic Linguistics
• History of English
• Language Acquisition
• Language Variation
• Language Families
• Language Evolution
• Language Reference
• Lexicography
• Linguistic Theories
• Linguistic Typology
• Linguistic Anthropology
• Phonetics and Phonology
• Psycholinguistics
• Sociolinguistics
• Translation and Interpretation
• Writing Systems
• Browse content in Literature
• Bibliography
• Children's Literature Studies
• Literary Studies (Asian)
• Literary Studies (European)
• Literary Studies (Eco-criticism)
• Literary Studies (Modernism)
• Literary Studies (Romanticism)
• Literary Studies (American)
• Literary Studies - World
• Literary Studies (1500 to 1800)
• Literary Studies (19th Century)
• Literary Studies (20th Century onwards)
• Literary Studies (African American Literature)
• Literary Studies (British and Irish)
• Literary Studies (Early and Medieval)
• Literary Studies (Fiction, Novelists, and Prose Writers)
• Literary Studies (Gender Studies)
• Literary Studies (Graphic Novels)
• Literary Studies (History of the Book)
• Literary Studies (Plays and Playwrights)
• Literary Studies (Poetry and Poets)
• Literary Studies (Postcolonial Literature)
• Literary Studies (Queer Studies)
• Literary Studies (Science Fiction)
• Literary Studies (Travel Literature)
• Literary Studies (War Literature)
• Literary Studies (Women's Writing)
• Literary Theory and Cultural Studies
• Mythology and Folklore
• Shakespeare Studies and Criticism
• Browse content in Media Studies
• Browse content in Music
• Applied Music
• Dance and Music
• Ethics in Music
• Ethnomusicology
• Gender and Sexuality in Music
• Medicine and Music
• Music Cultures
• Music and Religion
• Music and Culture
• Music and Media
• Music Education and Pedagogy
• Music Theory and Analysis
• Musical Scores, Lyrics, and Libretti
• Musical Structures, Styles, and Techniques
• Musicology and Music History
• Performance Practice and Studies
• Race and Ethnicity in Music
• Sound Studies
• Browse content in Performing Arts
• Browse content in Philosophy
• Aesthetics and Philosophy of Art
• Epistemology
• Feminist Philosophy
• History of Western Philosophy
• Metaphysics
• Moral Philosophy
• Non-Western Philosophy
• Philosophy of Science
• Philosophy of Action
• Philosophy of Law
• Philosophy of Religion
• Philosophy of Language
• Philosophy of Mind
• Philosophy of Perception
• Philosophy of Mathematics and Logic
• Practical Ethics
• Social and Political Philosophy
• Browse content in Religion
• Biblical Studies
• Christianity
• East Asian Religions
• History of Religion
• Judaism and Jewish Studies
• Qumran Studies
• Religion and Education
• Religion and Health
• Religion and Politics
• Religion and Science
• Religion and Law
• Religion and Art, Literature, and Music
• Religious Studies
• Browse content in Society and Culture
• Cookery, Food, and Drink
• Cultural Studies
• Customs and Traditions
• Ethical Issues and Debates
• Hobbies, Games, Arts and Crafts
• Natural world, Country Life, and Pets
• Popular Beliefs and Controversial Knowledge
• Sports and Outdoor Recreation
• Technology and Society
• Travel and Holiday
• Visual Culture
• Browse content in Law
• Arbitration
• Browse content in Company and Commercial Law
• Commercial Law
• Company Law
• Browse content in Comparative Law
• Systems of Law
• Competition Law
• Browse content in Constitutional and Administrative Law
• Government Powers
• Judicial Review
• Local Government Law
• Military and Defence Law
• Parliamentary and Legislative Practice
• Construction Law
• Contract Law
• Browse content in Criminal Law
• Criminal Procedure
• Criminal Evidence Law
• Sentencing and Punishment
• Employment and Labour Law
• Environment and Energy Law
• Browse content in Financial Law
• Banking Law
• Insolvency Law
• History of Law
• Human Rights and Immigration
• Intellectual Property Law
• Browse content in International Law
• Private International Law and Conflict of Laws
• Public International Law
• IT and Communications Law
• Jurisprudence and Philosophy of Law
• Law and Politics
• Law and Society
• Browse content in Legal System and Practice
• Courts and Procedure
• Legal Skills and Practice
• Primary Sources of Law
• Regulation of Legal Profession
• Medical and Healthcare Law
• Browse content in Policing
• Criminal Investigation and Detection
• Police and Security Services
• Police Procedure and Law
• Police Regional Planning
• Browse content in Property Law
• Personal Property Law
• Study and Revision
• Terrorism and National Security Law
• Browse content in Trusts Law
• Wills and Probate or Succession
• Browse content in Medicine and Health
• Browse content in Allied Health Professions
• Arts Therapies
• Clinical Science
• Dietetics and Nutrition
• Occupational Therapy
• Operating Department Practice
• Physiotherapy
• Radiography
• Speech and Language Therapy
• Browse content in Anaesthetics
• General Anaesthesia
• Neuroanaesthesia
• Browse content in Clinical Medicine
• Acute Medicine
• Cardiovascular Medicine
• Clinical Genetics
• Clinical Pharmacology and Therapeutics
• Dermatology
• Endocrinology and Diabetes
• Gastroenterology
• Genito-urinary Medicine
• Geriatric Medicine
• Infectious Diseases
• Medical Oncology
• Medical Toxicology
• Pain Medicine
• Palliative Medicine
• Rehabilitation Medicine
• Respiratory Medicine and Pulmonology
• Rheumatology
• Sleep Medicine
• Sports and Exercise Medicine
• Clinical Neuroscience
• Community Medical Services
• Critical Care
• Emergency Medicine
• Forensic Medicine
• Haematology
• History of Medicine
• Browse content in Medical Dentistry
• Oral and Maxillofacial Surgery
• Paediatric Dentistry
• Restorative Dentistry and Orthodontics
• Surgical Dentistry
• Medical Ethics
• Browse content in Medical Skills
• Clinical Skills
• Communication Skills
• Nursing Skills
• Surgical Skills
• Medical Statistics and Methodology
• Browse content in Neurology
• Clinical Neurophysiology
• Neuropathology
• Nursing Studies
• Browse content in Obstetrics and Gynaecology
• Gynaecology
• Occupational Medicine
• Ophthalmology
• Otolaryngology (ENT)
• Browse content in Paediatrics
• Neonatology
• Browse content in Pathology
• Chemical Pathology
• Clinical Cytogenetics and Molecular Genetics
• Histopathology
• Medical Microbiology and Virology
• Patient Education and Information
• Browse content in Pharmacology
• Psychopharmacology
• Browse content in Popular Health
• Caring for Others
• Complementary and Alternative Medicine
• Self-help and Personal Development
• Browse content in Preclinical Medicine
• Cell Biology
• Molecular Biology and Genetics
• Reproduction, Growth and Development
• Primary Care
• Professional Development in Medicine
• Browse content in Psychiatry
• Addiction Medicine
• Child and Adolescent Psychiatry
• Forensic Psychiatry
• Learning Disabilities
• Old Age Psychiatry
• Psychotherapy
• Browse content in Public Health and Epidemiology
• Epidemiology
• Public Health
• Browse content in Radiology
• Clinical Radiology
• Interventional Radiology
• Nuclear Medicine
• Radiation Oncology
• Reproductive Medicine
• Browse content in Surgery
• Cardiothoracic Surgery
• Gastro-intestinal and Colorectal Surgery
• General Surgery
• Neurosurgery
• Paediatric Surgery
• Peri-operative Care
• Plastic and Reconstructive Surgery
• Surgical Oncology
• Transplant Surgery
• Trauma and Orthopaedic Surgery
• Vascular Surgery
• Browse content in Science and Mathematics
• Browse content in Biological Sciences
• Aquatic Biology
• Biochemistry
• Bioinformatics and Computational Biology
• Developmental Biology
• Ecology and Conservation
• Evolutionary Biology
• Genetics and Genomics
• Microbiology
• Molecular and Cell Biology
• Natural History
• Plant Sciences and Forestry
• Research Methods in Life Sciences
• Structural Biology
• Systems Biology
• Zoology and Animal Sciences
• Browse content in Chemistry
• Analytical Chemistry
• Computational Chemistry
• Crystallography
• Environmental Chemistry
• Industrial Chemistry
• Inorganic Chemistry
• Materials Chemistry
• Medicinal Chemistry
• Mineralogy and Gems
• Organic Chemistry
• Physical Chemistry
• Polymer Chemistry
• Study and Communication Skills in Chemistry
• Theoretical Chemistry
• Browse content in Computer Science
• Artificial Intelligence
• Computer Architecture and Logic Design
• Game Studies
• Human-Computer Interaction
• Mathematical Theory of Computation
• Programming Languages
• Software Engineering
• Systems Analysis and Design
• Virtual Reality
• Browse content in Computing
• Business Applications
• Computer Security
• Computer Games
• Computer Networking and Communications
• Digital Lifestyle
• Graphical and Digital Media Applications
• Operating Systems
• Browse content in Earth Sciences and Geography
• Atmospheric Sciences
• Environmental Geography
• Geology and the Lithosphere
• Maps and Map-making
• Meteorology and Climatology
• Oceanography and Hydrology
• Palaeontology
• Physical Geography and Topography
• Regional Geography
• Soil Science
• Urban Geography
• Browse content in Engineering and Technology
• Agriculture and Farming
• Biological Engineering
• Civil Engineering, Surveying, and Building
• Electronics and Communications Engineering
• Energy Technology
• Engineering (General)
• Environmental Science, Engineering, and Technology
• History of Engineering and Technology
• Mechanical Engineering and Materials
• Technology of Industrial Chemistry
• Transport Technology and Trades
• Browse content in Environmental Science
• Applied Ecology (Environmental Science)
• Conservation of the Environment (Environmental Science)
• Environmental Sustainability
• Environmentalist Thought and Ideology (Environmental Science)
• Management of Land and Natural Resources (Environmental Science)
• Natural Disasters (Environmental Science)
• Nuclear Issues (Environmental Science)
• Pollution and Threats to the Environment (Environmental Science)
• Social Impact of Environmental Issues (Environmental Science)
• History of Science and Technology
• Browse content in Materials Science
• Ceramics and Glasses
• Composite Materials
• Metals, Alloying, and Corrosion
• Nanotechnology
• Browse content in Mathematics
• Applied Mathematics
• Biomathematics and Statistics
• History of Mathematics
• Mathematical Education
• Mathematical Finance
• Mathematical Analysis
• Numerical and Computational Mathematics
• Probability and Statistics
• Pure Mathematics
• Browse content in Neuroscience
• Cognition and Behavioural Neuroscience
• Development of the Nervous System
• Disorders of the Nervous System
• History of Neuroscience
• Invertebrate Neurobiology
• Molecular and Cellular Systems
• Neuroendocrinology and Autonomic Nervous System
• Neuroscientific Techniques
• Sensory and Motor Systems
• Browse content in Physics
• Astronomy and Astrophysics
• Atomic, Molecular, and Optical Physics
• Biological and Medical Physics
• Classical Mechanics
• Computational Physics
• Condensed Matter Physics
• Electromagnetism, Optics, and Acoustics
• History of Physics
• Mathematical and Statistical Physics
• Measurement Science
• Nuclear Physics
• Particles and Fields
• Plasma Physics
• Quantum Physics
• Relativity and Gravitation
• Semiconductor and Mesoscopic Physics
• Browse content in Psychology
• Affective Sciences
• Clinical Psychology
• Cognitive Neuroscience
• Cognitive Psychology
• Criminal and Forensic Psychology
• Developmental Psychology
• Educational Psychology
• Evolutionary Psychology
• Health Psychology
• History and Systems in Psychology
• Music Psychology
• Neuropsychology
• Organizational Psychology
• Psychological Assessment and Testing
• Psychology of Human-Technology Interaction
• Psychology Professional Development and Training
• Research Methods in Psychology
• Social Psychology
• Browse content in Social Sciences
• Browse content in Anthropology
• Anthropology of Religion
• Human Evolution
• Medical Anthropology
• Physical Anthropology
• Regional Anthropology
• Social and Cultural Anthropology
• Theory and Practice of Anthropology
• Browse content in Business and Management
• Business Strategy
• Business History
• Business Ethics
• Business and Government
• Business and Technology
• Business and the Environment
• Comparative Management
• Corporate Governance
• Corporate Social Responsibility
• Entrepreneurship
• Health Management
• Human Resource Management
• Industrial and Employment Relations
• Industry Studies
• Information and Communication Technologies
• International Business
• Knowledge Management
• Management and Management Techniques
• Operations Management
• Organizational Theory and Behaviour
• Pensions and Pension Management
• Public and Nonprofit Management
• Strategic Management
• Supply Chain Management
• Browse content in Criminology and Criminal Justice
• Criminal Justice
• Criminology
• Forms of Crime
• International and Comparative Criminology
• Youth Violence and Juvenile Justice
• Development Studies
• Browse content in Economics
• Agricultural, Environmental, and Natural Resource Economics
• Asian Economics
• Behavioural Finance
• Behavioural Economics and Neuroeconomics
• Econometrics and Mathematical Economics
• Economic Systems
• Economic Methodology
• Economic History
• Economic Development and Growth
• Financial Markets
• Financial Institutions and Services
• General Economics and Teaching
• Health, Education, and Welfare
• History of Economic Thought
• International Economics
• Labour and Demographic Economics
• Law and Economics
• Macroeconomics and Monetary Economics
• Microeconomics
• Public Economics
• Urban, Rural, and Regional Economics
• Welfare Economics
• Browse content in Education
• Adult Education and Continuous Learning
• Care and Counselling of Students
• Early Childhood and Elementary Education
• Educational Equipment and Technology
• Educational Strategies and Policy
• Higher and Further Education
• Organization and Management of Education
• Philosophy and Theory of Education
• Schools Studies
• Secondary Education
• Teaching of a Specific Subject
• Teaching of Specific Groups and Special Educational Needs
• Teaching Skills and Techniques
• Browse content in Environment
• Applied Ecology (Social Science)
• Climate Change
• Conservation of the Environment (Social Science)
• Environmentalist Thought and Ideology (Social Science)
• Natural Disasters (Environment)
• Social Impact of Environmental Issues (Social Science)
• Browse content in Human Geography
• Cultural Geography
• Economic Geography
• Political Geography
• Browse content in Interdisciplinary Studies
• Communication Studies
• Museums, Libraries, and Information Sciences
• Browse content in Politics
• African Politics
• Asian Politics
• Chinese Politics
• Comparative Politics
• Conflict Politics
• Elections and Electoral Studies
• Environmental Politics
• European Union
• Foreign Policy
• Gender and Politics
• Human Rights and Politics
• Indian Politics
• International Relations
• International Organization (Politics)
• International Political Economy
• Irish Politics
• Latin American Politics
• Middle Eastern Politics
• Political Methodology
• Political Communication
• Political Philosophy
• Political Sociology
• Political Theory
• Political Behaviour
• Political Economy
• Political Institutions
• Politics and Law
• Politics of Development
• Public Administration
• Public Policy
• Quantitative Political Methodology
• Regional Political Studies
• Russian Politics
• Security Studies
• State and Local Government
• UK Politics
• US Politics
• Browse content in Regional and Area Studies
• African Studies
• Asian Studies
• East Asian Studies
• Japanese Studies
• Latin American Studies
• Middle Eastern Studies
• Native American Studies
• Scottish Studies
• Browse content in Research and Information
• Research Methods
• Browse content in Social Work
• Addictions and Substance Misuse
• Adoption and Fostering
• Care of the Elderly
• Child and Adolescent Social Work
• Couple and Family Social Work
• Direct Practice and Clinical Social Work
• Emergency Services
• Human Behaviour and the Social Environment
• International and Global Issues in Social Work
• Mental and Behavioural Health
• Social Justice and Human Rights
• Social Policy and Advocacy
• Social Work and Crime and Justice
• Social Work Macro Practice
• Social Work Practice Settings
• Social Work Research and Evidence-based Practice
• Welfare and Benefit Systems
• Browse content in Sociology
• Childhood Studies
• Community Development
• Comparative and Historical Sociology
• Economic Sociology
• Gender and Sexuality
• Gerontology and Ageing
• Health, Illness, and Medicine
• Marriage and the Family
• Migration Studies
• Occupations, Professions, and Work
• Organizations
• Population and Demography
• Race and Ethnicity
• Social Theory
• Social Movements and Social Change
• Social Research and Statistics
• Social Stratification, Inequality, and Mobility
• Sociology of Religion
• Sociology of Education
• Sport and Leisure
• Urban and Rural Studies
• Browse content in Warfare and Defence
• Defence Strategy, Planning, and Research
• Land Forces and Warfare
• Military Administration
• Military Life and Institutions
• Naval Forces and Warfare
• Other Warfare and Defence Issues
• Peace Studies and Conflict Resolution
• Weapons and Equipment

• < Previous chapter
• Next chapter >

11 Prosodic structure

• Published: October 2019
• Cite Icon Cite
• Permissions Icon Permissions

This chapter investigates the relationship between the phonological or prosodic structure of a spoken utterance and its syntactic, semantic, and information structural analysis. A full theory of the form-meaning correspondence must account for the effect of prosodic features such as intonation patterns on interpretation. In line with other work in LFG that is concerned with the contribution made by phonology or prosody to grammatical structure and interpretation, the existence of a separate level of prosodic structure or p-structure within the projection architecture is assumed. The chapter reviews previous LFG approaches to prosody and the place of prosodic structure within the grammar (Section 11.3), before presenting the approach that is adopted which relies on analyzing a string as having two distinct aspects: one syntactic, the s-string, the other phonological/ prosodic, the p-string (Section 11.4). This approach is exemplified with an account of declarative questions and prosodic focus marking.

In this chapter, we investigate the relationship between the phonological or prosodic structure of a spoken utterance and its syntactic, semantic, and information structural analysis. A full theory of the form-meaning correspondence must account for the effect of prosodic features such as intonation patterns on interpretation. Work in LFG that is concerned with the contribution made by phonology or prosody to grammatical structure and interpretation usually assumes the existence of a separate prosodic structure (sometimes called phonological structure ) within the overall grammatical architecture. We will review previous LFG approaches to prosody and the place of prosodic structure within the grammar, before presenting and exemplifying the approach that we adopt.

11.1 Prosody and grammar

Prosody refers to the patterns of rhythm and intonation in spoken language. Prosodic (also referred to as suprasegmental) features of speech are acoustic characteristics associated with a particular phonological domain. The relevant features are generally assumed to be pitch, length, and loudness; their physical correlates are fundamental frequency (F0, measured in Hz), duration, and intensity (measured in dB), respectively. Our concern here is with prosody at the sentence level; we are interested in word-level prosody (for example, the lexically determined location of accent or tonal pattern) insofar as it interacts with sentence-level prosody, such as intonational contours that distinguish declaratives from questions. Reflecting the vast majority of research in LFG, the data we consider in this chapter will overwhelmingly focus on patterns of intonation and the analysis of F0; for LFG work that considers duration and pitch as well, see Bögel ( 2015 : Chapter 3).

The approach that we adopt in this book, in line with the majority of work in LFG on the interface between intonation and the other aspects of linguistic structure discussed in Part II , is a phonological model of intonation designed to capture generalizations about this system and its meaningful contrastive elements. 1 Under such a model, contours are analyzed as melodies, abstracting away from their precise shape and the specific Hz values with which they are produced. This kind of model is inherently flexible with respect to the relationship between prosody and other aspects of linguistic structure. This is a necessary condition if the model is to account for the attested variability in intonation that can relate to factors such as speech rate, speaker, style, and dialect, inter alia (see, for example, Arvaniti and Ladd 2009 ; Arvaniti 2016 ). For a general introduction to phrasing and intonation, which also includes data referred to in this chapter, see Hayes and Lahiri ( 1991 ).

The central concerns of a theory of the relation between prosody and other parts of the grammar are the understanding and analysis of the relation between the phonological features of a word or utterance and its abstract grammatical properties. It is clear enough, for example, that some connection must exist between the phonological realization of a word and the c-structural, f-structural, and s-structural features and structures that correspond to that phonological realization. However, the precise nature of that connection and how it is constrained are less clear. The problem is particularly acute above the word level: how is a particular phonological or prosodic domain related to a particular syntactic or semantic unit?

This is not merely a matter of incorporating an additional, ancillary aspect of linguistic analysis into an otherwise complete model of grammar. Prosodic features such as pitch and duration can, and very often do, make crucial contributions to the interpretation of an utterance. For example, in Japanese a question is distinguished from a declarative sentence in writing by the presence of a sentence-final particle such as ka . In spoken Japanese, the inclusion of this particle is optional (Hinds 1986 ); what is critical is the intonation pattern that is used. If the version of ( 1 ) without ka is spoken with a final fall in intonation, the utterance is interpreted as a declarative; if there is a final rise in intonation, it is interpreted as a yes-no question. In the latter case, the contribution to meaning that prosody makes effectively “overrides” the apparently unambiguous declarative syntax of the sentence. This means that the pitch movement associated with the end of the utterance is crucial to the interpretation of ( 1 ) when it does not include the question particle. Such data illustrate how prosody can make an independent contribution to meaning:

We represent intonational contours, which may also be referred to as melodies or tunes, using two abstract tone specifications: High (H) and Low (L). For example, our representation of the final falling declarative intonational contour in Japanese is shown in ( 2a ) as a High tone preceding a Low tone, while the final rising intonational contour associated with the interrogative reading of ( 1 ) is represented in ( 2b ) as the sequence Low–High:

Prosody may also be used to resolve ambiguity. For example, it has been claimed that differences in prosodic phrasing can be used to distinguish between candidate syntactic structures, though speakers in general do not do this consistently (see, for instance, Allbritton et al. 1996 ). Consider the following examples, taken from Price et al. ( 1991 ), in which square brackets delimit the subordinate clause, and parentheses indicate the prosodic constituency that can be associated with each reading. Note how the boundaries of the two types of constituents coincide:

Which of the two syntactic analyses is preferred may depend upon the location of cues such as pauses which signify boundaries between prosodic constituents: when a pause occurs after gradually , the interpretation is the one given in ( 3a ), i.e. gradually modifies the verb learn in the subordinate clause; when a pause occurs after learn , the interpretation is the one given in ( 3b ), i.e. gradually is part of the main clause. Such data indicate that a close relationship can exist between prosodic phrasing and syntactic structure (though one should be cautious about claiming that disambiguation is the result of a one-to-one relationship between the two). Furthermore, they demonstrate the importance of integrating prosody into any account of the form-meaning relation.

Debate in the literature on prosody has centered on one fundamental issue in particular, which we too must address before presenting the approach that we adopt, namely whether the phonological/prosodic component is an independent module with its own internal structure.

11.2 Prosody: an independent level of structure?

At the heart of the debate on the place of prosody in the grammar is the issue of its relation to syntactic structure. Specifically, the question is whether phonological processes make direct reference to syntactic information or indirect reference via an interface relation, the nature of which must be identified and defined.

11.2.1 Direct versus indirect reference

Support for the direct reference approach comes from data which indicate that phonological output is defined on the basis of syntactic structure, as one might argue is the case in ( 3 ). This information is generally assumed to concern syntactic constituency (for example Kaisse 1985 ; Odden 1995 ) but could in principle refer to other types of syntactic relation, as Pak ( 2008 ) points out. Such an approach to the syntax-phonology interface is problematic on a view which regards syntax and phonology as fundamentally distinct. For instance, Scheer ( 2011 : 347) holds that syntax and phonology are separate aspects of linguistic structure and as such are each unable to interpret units, features, and structures specific to the other.

Even setting aside this incompatibility, the direct reference approach has been criticized in the light of data which show that syntactic and prosodic structure are not, as one would predict under such a model, necessarily isomorphic; for discussion of relevant data, see Chen ( 1987 ), Selkirk and Shen ( 1990 ), and Bošković ( 2001 ). Lahiri and Plank ( 2010 ) seek to emphasize, based on observations about Germanic languages that date back at least to Steele ( 1775/1779 ), that the prosody-syntax relation is in fact characterized by extensive misalignment; a lack of isomorphism between the two is more the rule than the exception (see example 12).

The alternative to direct reference is an indirect reference approach, according to which an interface relation serves to connect the syntax and phonology macromodules. An indirect reference approach copes well with data which demonstrate a lack of isomorphism between syntactic and prosodic structure. While the two are systematically related to one another via a mapping algorithm, under such an approach there is no expectation that isomorphism is the result. With respect to modularity and domain specificity, the distinction between syntax and prosody can be maintained by assuming indirect reference: the objects and structures native to one module cannot be interpreted or manipulated by the other module.

An indirect reference approach to the syntax-prosody interface is not by definition restricted to one particular theoretical framework, but it clearly fits well with the parallel grammatical architecture and general commitment to modular specificity of LFG (§ 7.2 ). Of course, such an approach requires an independent level of representation with its own primitives and organizing principles, which provides the domains of application relevant to postlexical phonological processes. The issue of precisely how to define and represent prosodic structure continues to be the subject of research and debate in the wider literature. In the following section, we review the ideas which have been most influential in the modeling of prosody and its interfaces within the LFG framework.

11.2.2 Units, constraints, and internal structure

Early work on the representation of prosodic features in generative phonology, for example by Chomsky and Halle ( 1968 ), assumed binary features such as [± stress ]. Such features could be assigned to particular syllables by rules applied in particular phonological contexts, but no cohesive theory existed of the prosodic structure of whole phrases or utterances. Building on earlier work by Liberman ( 1975 ) and Liberman and Prince ( 1977 ), which explored hierarchical structure in phonology at the level of the word and below, Selkirk ( 1978 , 1980a , b , 1981 ) first proposed that the phonological or prosodic features of phrases and utterances could be described by reference to a structured prosodic representation. Under this “Prosodic Phonology” approach, developed further by Nespor and Vogel ( 1986 ), an utterance can be analyzed not only in terms of hierarchical syntactic structure, i.e. constituent structure, but also in terms of hierarchical prosodic structure .

According to Prosodic Phonology, the categories of prosodic structure are, like the categories of constituent structure, finite in number and available crosslinguistically. In contrast to constituent structure, prosodic phrases are not projected by a head, but are related to one another according to the Prosodic Hierarchy. The precise inventory of categories which form the Prosodic Hierarchy varies in the literature (as does the inventory of constituent structure categories, as discussed in Chapter 3 ); we assume the inventory and set of hierarchical relations shown in ( 4 ), taken (with modified labels) from Selkirk ( 1995 ), with these units being defined independently of units at any other level such as c-structure. Much research has investigated the units of the Prosodic Hierarchy, focusing on prosodic/phonological phenomena that have these units as their domain, and the associated prosodic features which can be taken as cues for their detection: 2

The syllable is the smallest unit of the Prosodic Hierarchy that is of concern to us. 3 In many languages, syllables are the smallest units with which certain phonological features, such as stress, are associated. 4 A foot is a prosodic constituent consisting of one or more syllables. Patterns in the alternation of stressed and unstressed syllables in languages such as English are often stated by reference to rules of foot formation. One or more feet make up a prosodic word . Feet are crucial in the formation of prosodic words, but beyond this the foot is not relevant for the purposes of modeling prosody's interfaces with other modules of the grammar. (For this reason, and in the interests of clear presentation, the foot level is omitted in the analyses of prosodic structure which we present in this book.) Prosodic words are usually assumed to be the domain within which lexical phonological processes, and often other processes such as cliticization, apply. One or more prosodic words make up a phonological phrase , the unit within which a single prosodic contour may apply. The intonational phrase consists of one or more prosodic phrases. This is the unit within which the smaller prosodic contours of prosodic phrases are subject to larger overarching processes, such as the gradual declination of median pitch; the boundaries of intonational phrases are therefore generally characterized by pitch reset. Finally, the utterance is the largest prosodic constituent, comprising any single unit of continuous speech consisting of one or more intonational phrases. 5

As with constituent structure, it is possible to state rules or constraints on the formation of valid prosodic structures. These constraints (see Selkirk 1984 for details) are usually subsumed under the “Strict Layer Hypothesis” (Nespor and Vogel 1986 : 7; Selkirk 2011 : 347), which constitutes a general wellformedness condition on prosodic structure. The Strict Layer Hypothesis requires that an utterance be parsed exhaustively into non-recursive prosodic constituents, which form “layers” that correspond to categories of the Prosodic Hierarchy. Thus, an Intonational Phrase (IntP) can immediately dominate only Phonological Phrases (PhPs), and a PhP can immediately dominate only Prosodic Words (PWs). 6 For example:

The basic approach to prosodic structure illustrated by ( 5 ) is widely assumed in LFG approaches to postlexical phonological processes, though the hierarchical structuring of prosodic constituents has not always been represented as a tree structure, as we will see.

11.3 Representing prosodic structure in LFG: early approaches

Early work in LFG made no attempt to integrate a theory of prosody or its relation to the syntactic component of grammar. The first significant steps in this direction were made by Butt and King ( 1998a ), who made an explicit proposal regarding the LFG representation of prosodic features relevant to syntax. Butt and King ( 1998a ) propose a p(honological)-structure, projected from c-structure and parallel to but separate from f-structure. Butt and King's phonological structure encodes only phonological information that is relevant to the syntax and can contribute to the full interpretation of an utterance. Butt and King ( 1998a ) illustrate their proposals with reference to ambiguous verb phrases in Bengali such as this sentence from Hayes and Lahiri ( 1991 ):

When spoken, the two readings of the sentence in ( 6 ) are associated with different phrasings: the idiomatic reading ‘was startled' is licensed when the VP corresponds to a single PhP ( 7a ), while the literal reading ‘saw a ghost' emerges when the verb form and b h ut ‘ghost' constitute two distinct PhPs ( 7b ), as indicated by the labeled brackets:

Butt and King ( 1998a ) represent p-structure as a feature structure, similar to f-structure. The p-structure they propose for the idiomatic reading composed of two PhPs is given in ( 8 ), and the p-structure for the literal reading composed of three PhPs is given in ( 9 ). The p-structure feature (prosodic dom ain) defines the level of constituent within the Prosodic Hierarchy that a particular feature structure represents, while the feature p-form represents a word's phonological form: 7

Butt and King ( 1998a ) propose that p-structure is projected from c-structure just as f-structure is, and that prosodic constituents such as IntPs and PhPs are derived from syntactic constituents by means of mapping processes. The relevant mapping processes permit not only isomorphism, but also some degree of difference between the two structures (for example, by increasing or decreasing the levels of embedding in p-structure relative to the c-structure input). With respect to the ambiguity of ( 6 ), Butt and King argue that the prosodic distinction between the two readings corresponds to a difference in c-structure. In the case of the idiomatic reading, Butt and King analyze the N b h ut ‘ghost' and the V dek h lam ‘saw' as sisters at c-structure; together they constitute a V′, as shown in ( 10 ). This single bar-level syntactic constituent corresponds to a single PhP in the relevant prosodic representation ( 8 ). By contrast, in the literal reading the NP b h ut ‘ghost' occupies specifier position in the VP, as shown in ( 11 ). The syntactic relationship between the NP and the verb is not as close as in the case of the idiomatic reading, and this is also the case in the prosodic representation ( 9 ): the two relevant syntactic constituents correspond to two separate PhPs rather than forming one PhP.

Butt and King ( 1998a ) state that their feature-structure representation of p-structure is equivalent to a tree structure representation of the type shown in ( 5 ), but they acknowledge a deficiency in the representation: the feature structure's contents are unordered. They therefore find it necessary to rely on a notion of projection precedence, assuming that information about ordering is indirectly available from the c-structure, as for the f-precedence relation on f-structures (§ 6.11.4 ). Another issue with feature structures as a means of representing prosodic structure, highlighted by O’Connor ( 2006 : 151–2), is that an enriched theory of phonological structure is required in order to associate text (i.e. the words in an utterance) with a tune (i.e. an intonational contour) because tonal information at the phrasal level cannot be directly associated with the relevant text in the feature-structure approach. In subsequent work, the original feature-structure approach of Butt and King ( 1998a ) has been rejected in favor of other means of representation.

On Butt and King's view, p-structure is fundamentally derivative of the c-structure, even though it is a distinct level of linguistic structure. While the overall approach is one of indirect reference, by conceiving of p-structure as a projection from c-structure Butt and King ( 1998a ) ultimately base their definitions of p-structure units on syntactic constituency. On this view, isomorphism is the default; when p-structure differs from c-structure, it is as a result of additional processes applying. As noted in § 11.2.1 , this assumption about the nature of the relationship between syntax and prosody has been challenged.

The next significant work on prosodic structure within LFG was on English and Bosnian/Croatian/Serbian by O’Connor ( 2005a , b , 2006 ) and by Mycock ( 2006 ) in the context of the typology of constituent question formation. They assume, with Butt and King ( 1998a ), that p-structure is a separate level within the parallel architecture, but both abandon the feature-structure representation in favor of tree structures. In this respect, O’Connor ( 2006 ) and Mycock ( 2006 ) share a good deal of common ground. They propose “tune structure” and “contour” rules respectively to capture facts about intonational contours that apply to specific prosodic domains. O’Connor adopts the Autosegmental-Metrical/Tones and Break Indices (AM/ToBI) framework for the expression of tune structure rules. 8 Mycock (in her PhD thesis and in later work) stops short of doing the same, but the p-structure rules with which she works are broadly compatible with an AM/ToBI approach. In terms of empirical focus, O’Connor ( 2006 ) and Mycock ( 2006 ) are both concerned with how prosody interfaces with other parts of the grammar. In particular, they seek to capture the important role that prosody can play in signaling information structure status. Mycock ( 2006 ) also analyzes scope marking in constituent questions with reference to p-structure, offering a way to model the prosody-semantics interface in LFG. Where these two approaches differ significantly is in the location of p-structure in the architecture. Mycock ( 2006 ) follows Butt and King ( 1998a ) in assuming that p-structure is projected from c-structure (and thus inherits the issues with such an approach discussed above), but augments this with the proposal that p-structure maps to s-structure (and i-structure) independently of syntax in order to account for the type of phenomena exemplified in ( 2 ), which show that prosody can make a crucial contribution to meaning. O’Connor ( 2006 ) takes a different view, according to which p-structure has a direct relationship only with a level which he refers to as discourse structure. The architecture which O’Connor proposes is non-standard and has not been adopted beyond his work; we do not consider it further here.

Perhaps surprisingly, given LFG's co-description architecture and the general commitment to modularity, the next major proposal concerning the analysis of prosody, put forward by Bögel et al. ( 2009 ), did not include a separate level of prosodic representation. As a result, this work has less in common with the indirect reference approaches outlined in § 11.2.1 than the LFG analyses which preceded it. Bögel et al. ( 2009 ) proposed a “pipeline architecture” as a way of capturing facts about the (mis)alignment of prosodic and syntactic constituents. Rather than treating prosodic structure as projected from c-structure, Bögel et al. argue that prosodic information feeds into c-structure analysis. In their model, the input to c-structure is a prosodically bracketed string. Phrase structure rules then apply as normal to produce a valid c-structure for the string, but with one significant augmentation: phrase structure rules are formulated so as to make reference not only to syntactic categories but also to prosodic brackets. These rules are used in Bögel et al. ( 2009 ) to capture the misalignment which characterizes the syntax-prosody relation in Germanic languages that, as Lahiri and Plank ( 2010 ) discuss, is often due to the differing behavior of function words in the formation of syntactic and prosodic constituents. Lahiri and Plank exemplify using an old advertising slogan:

The bracketing in ( 12 ) shows that the English function words a and of group with a constituent to their right in the syntax, but to their left in the prosody. (In fact, the slogan appeared in writing as Drinka pinta milka day , reflecting the prosodic bracketing shown.) In their analysis of this example, Bögel et al. ( 2009 ) propose that the phonological string undergoes prosodic parsing, which introduces the relevant prosodic boundaries into the string. This “prosodically annotated string” is then the input to the syntactic component:

A key feature of this pipeline architecture is that syntax can interpret the prosodic boundaries that are inserted into the string. Though this is not relevant for the example in ( 13 ), it is crucial to Bögel et al.'s ( 2009 ) account of the role of prosody in resolving syntactic ambiguity. For example, the phrase in ( 14 ) is ambiguous, depending on what exactly the AdjP old modifies, and is compatible with either of the syntactic analyses indicated:

The ambiguity is resolved when prosodic phrasing is available. ( 15a ) is compatible with the syntactic analysis in ( 14a ), while ( 15b ) pairs with ( 14b ). That is, the preference is for alignment of syntactic and prosodic boundaries:

Bögel et al. capture this fact about alignment with their Principle of Prosodic Preference: “syntactic structures with constituent boundaries that do not coincide with prosodic boundaries are dispreferred.” While this approach successfully accounts for the role that prosody can play in resolving syntactic ambiguity and captures cases of mismatch between syntax and prosody, it does so at the cost of introducing prosodic information into the syntax macromodule. In an important respect, this violates the grammatical principle of modularity (§ 7.2 ): the syntactic module should not be able to interpret nonsyntactic objects. This is circumvented by proposing that a prosodic boundary has an additional role as one of two syntactic categories (left boundary, LB, or right boundary, RB), with the result that prosodic objects are effectively “disguised” as being syntactic: under this approach, LB and RB are terminal nodes in the c-structure tree. This does not represent a strictly modular approach to the issue. Moreover, the rather simplified prosodic representation involving LB and RB that Bögel et al. ( 2009 ) adopt does not allow easy reference to the different types of units that belong to the Prosodic Hierarchy, something which is essential in the analysis of many phenomena. Presumably, reference to distinct prosodic units could be made via a separate, additional prosodic component introducing a large number of differently labeled prosodic brackets into the string, but this adds further complexities. Bögel et al. ( 2009 ) do not explore the details or the wider implications of including such a component. Subsequent work on prosody and its interfaces within LFG has moved away from the “pipeline” architecture proposed by Bögel et al. ( 2009 ) and returned to an approach in which prosodic structure represents a distinct level in the parallel architecture, a position more compatible with strict modularity. In later work such as Bögel ( 2015 ), Bögel rejects the prosodic hierarchy as the basis for p-structure.

11.4 Modeling the prosody-syntax interface

We now present the model of the prosody-syntax interface that we adopt in this book. Espousing a strong commitment to modular specificity and the indirect reference approach, Dalrymple and Mycock ( 2011 ) propose a new model for integrating prosody with the wider LFG grammar. This model conforms to three important criteria, addressing issues raised in the two previous sections. Firstly, it enables the contribution of prosody to different components of the grammar to be modeled with relative ease. Secondly, it does not assume any sort of close correspondence between the syntactic and prosodic phrase structures of an utterance, though it does permit any such correspondence to be accounted for. Thirdly, it respects the principle of modularity of the grammar, maintaining a strict separation between syntactic and prosodic information. In Dalrymple and Mycock's model, although prosody may contribute to syntax, semantics, and information structure, no structure is able to interpret prosodic information except the prosodic component, and the prosodic component itself is unable to interpret anything but prosodic information.

Dalrymple and Mycock's ( 2011 ) model of the prosody-syntax interface was further developed and streamlined by Mycock and Lowe ( 2013 ). In the following sections, the model that we present is essentially the Mycock and Lowe version.

11.4.1 The p-string and the s-string

In § 3.5 , we adopted the proposal of Kaplan ( 1987 , 1995 ) that the string of words usually treated as constituting the terminal nodes of the c-structure tree should be thought of as a separate structure from the c-structure. Under this proposal, the string is related to the c-structure by a projection function, labeled π , from string elements to terminal c-structure nodes; terminal c-structure nodes are therefore not words, but X 0 or X ^ level categories. This is illustrated in ( 16 ) for the sentence David arrived :

The string can be thought of as a grammatical signal parsed into minimal units. The majority of analyses assume this view of the string: it is regarded as comprising minimal syntactic units which are then subject to phrase-structural analysis in the c-structure, via the π projection. However, this ignores the fact that a string can also be parsed into minimal phonological/prosodic units. The string, as Asudeh ( 2009 : 107) observes, thus represents part of the syntax-phonology interface.

Dalrymple and Mycock ( 2011 ) were the first to make a detailed proposal concerning this dual nature and function of the string. They argue that the string must be analyzed as having two distinct aspects: one syntactic, the s-string , the other phonological/prosodic, the p-string . Any linguistic signal can be parsed in two ways: the s-string represents the parsing of a signal into minimal syntactic units, and so corresponds to the conception of the string in most earlier works, while the p-string represents the parsing of a signal into minimal phonological/prosodic units. Just as the s-string is the basis of the syntactic phrase-structural analysis of an utterance, so the p-string is the basis of the prosodic phrase-structural analysis. To give an example, in Germanic languages units are formed according to the types of rhythmic principles highlighted by Lahiri and Plank ( 2010 ), along with the structure of the Prosodic Hierarchy; see § 11.2 . The units of p(rosodic)-structure are native to the phonology macromodule, and the projection of p-structure is based on phonological and prosodic features; there is no sense in which p-structure is derived from syntactic phrase structure, and no assumption of a necessarily close correlation between syntactic and prosodic phrase structure. The two are, of course, related to one another, but it is the string which represents the point of interface; thus, modularity is respected. Dalrymple and Mycock ( 2011 ) illustrate their proposals using the sentence Anna was studying at the university , providing a “double-tree” analysis of the two aspects of the string. The s-string, p-string, c-structure, and p-structure which they propose—with some emendations to bring it in line with proposals by Mycock and Lowe ( 2013 )—are shown in ( 17 ). The units of the s-string are each related to a terminal node of the c-structure via the π projection. The prosodic units related to these syntactic units form the p-string. Parallel to the syntactic analysis of this utterance, the units of the p-string are each related to a terminal node of the p-structure via the β projection, with “S” standing for syllable. Thus, the two aspects of the string receive equivalent but distinct analyses: 9

The relation between s-string and p-string units or, to put it another way, between the syntactic and prosodic parses of an utterance, is represented in ( 17 ) using dotted lines. Dalrymple and Mycock ( 2011 ) propose that the s-string–p-string relation is defined by information stored in lexical entries. Underpinning their proposal is a more fine-grained understanding of the nature and content of lexical entries, according to which a lexical entry specifies both a s(yntactic)-form and a p(honological)-form , as well as the c-structure category and f-description. The co-occurrence of s-forms and p-forms in lexical entries constrains the analysis of a string by requiring that any lexical item identified as a syntactic unit necessarily corresponds to a phonological/prosodic unit or units, in the same relative position in the string. The role of the lexical entry in mediating between the s-string and the p-string is illustrated in ( 18 ). The diagram in ( 19 ) exemplifies the portion of the grammatical architecture for the noun university which appears in ( 17 ):

Dalrymple and Mycock ( 2011 ) therefore use the p-string, the s-string, and the lexicon to model the interface between the syntax and phonology macromodules.

Under this analysis, information of various kinds can be associated with s-string and p-string units. Mycock and Lowe ( 2013 ) develop Dalrymple and Mycock's ( 2011 ) approach to lexical entries by proposing that s-string and p-string units are not atomic, but in fact are bundles of information which they represent as feature structures.

A key component of any feature structure representing a unit of the s-string or p-string is the feature fm ( form ), whose value is related to the s-form or p-form of a lexical entry. For example, the lexical entry for university includes:

The symbol • is used to refer to s-string units (Mycock and Lowe 2013 ). In the case of an s-string fm feature such as that shown in ( 20 ), the relation with the s-form in the relevant lexical entry (i.e. university) is one of identity. That is, the s-form of a word as it appears in a lexical entry is identical to the value of the fm feature in any s-string instantiation of that word. In the case of the p-string, on the other hand, this is not the case. For instance, the p-form of university includes five syllables (shown separated by periods), of which the second and fourth syllables contain the vowel /ɪ/: 10

Notice that in the p-string in ( 17 ) the vowels in those same syllables are realized as /ə/. Phonological processes related to speech tempo and other contextual factors may apply, rendering the relation between p-forms and p-string fm features opaque. (It is important that the underlying vowels are retrievable though, as speakers can clearly access and produce them in instances of slow, careful production.) In informal terms, it is not difficult to see how the application of such phonological processes affects the derivation of p-string fm features from p-forms, and likewise how the corresponding undoing of those same rules affects the reconstruction of p-forms from p-string fm features. For the present purposes, however, we make no proposals regarding the precise formalization of this aspect of phonology. 11

The other respect in which fm features differ from the s-forms and p-forms of lexical entries is the possibility of many-to-one correspondences. Again, this is clearly seen by comparison of ( 21 ) with ( 17 ). The single p-form of the lexical entry in ( 21 ) corresponds to five distinct units in the p-string representation. Under a feature-structure approach, each of these p-string units is represented as a distinct p-string feature structure with its own p-string fm feature; that is, there is one p-string unit, represented as a separate feature structure, for each syllable. We assume that the specification of p-string units is part of the lexical information stored in lexical entries. 12

Of course, fm features and their values are not the only components of a lexical entry. Each lexical entry also includes a c-structure category and f-description. At this point, we must note an inconsistency in the formal details of the presentation so far. Following long-established tradition in LFG, we have been using the variable ↑ in f-descriptions in lexical entries. Thus, for example, the f-structure number feature for university is specified by the f-description (↑ index   num ) = sg . However, as discussed in § 5.3.1 , the variable ↑ is an abbreviation for ϕ ( ∗ ^ ) ⁠ , that is, the f-structure projected, via the ϕ function, from the mother of the current c-structure node. This variable was appropriate when a syntactic element such as university was conceived of as a terminal node of the c-structure tree, with the preterminal node N as its mother. In that case, ↑ refers to the f-structure corresponding to the N node, as required. However, as discussed in § 3.5 and at the start of the current section, most major approaches to the LFG architecture follow Kaplan ( 1987 , 1995 ) in understanding the (s-)string to be distinct from the c-structure, related to it by the projection function π . Since ↑ is defined in terms of the mother function ℳ on c-structure nodes, our use of ↑ in a lexical entry is therefore technically incorrect. What should rather be used is the distinct function ϕ ( π (•)), which represents the f-structure projected, via the ϕ function, from the c-structure terminal node projected, via the π function, from the s-string unit with which the lexical entry is associated, as shown here:

The lexical entry for university (with its parts labeled) is therefore: 13

While a lexical entry of this type is technically more accurate, the use of ↑ is a well-established tradition in LFG, and is certainly more readable. We therefore retain this traditional notation, on the understanding that ↑ appearing in a lexical entry refers to a different function from ↑ when it appears in phrase structure annotations. In a lexical entry, ↑ is defined as ϕ ( π (•)), where the π function maps from units of the s-string to terminal nodes of the c-structure. In a phrase structure rule, ↑ is defined as ϕ ( ∗ ^ ) (or, to be completely explicit, as ϕ ( ℳ ( * ) ) ⁠ , in terms of the mother function ℳ on c-structure nodes).

11.4.2 Edge features

Besides the fm feature, whose value is specified in each lexical entry, a range of other features can appear in string feature structures. Some of these are lexically specified; for example, a lexically stressed syllable is specified in the lexicon as having the feature syllstress p . Other features represent properties that string elements may possess in a particular context. The most important of these contextually specified properties for our purposes is the association of string elements with information about the location of the beginning or end of syntactic or prosodic constituents. Reliance on such edge information is an acknowledged characteristic of prosody and its interaction with other aspects of linguistic structure, and as such has been integral to models of prosody and its interfaces over the years, for instance the end-based (also known as the edge-based) approach of Selkirk ( 1986 ) and the Align family of constraints in Optimality-Theoretic approaches including Selkirk ( 1995 ). Following Dalrymple and Mycock ( 2011 ), Mycock and Lowe ( 2013 ) assume that string units appearing in particular contexts are associated with sets of labels. These labels represent the left and right edges of the constituents in the associated structure that belongs to the relevant module of the grammar (i.e. c-structure or p-structure). So, in ( 17 ), the s-string element ‘Anna’ is associated with the left edge of the IP in the c-structure, and also with both the left and right edges of an NP and N. This is represented in the relevant string unit by means of features l (left edge) and r (right edge) whose values are sets comprising information concerning which constituents this particular s-string unit represents the left or right edge of. Thus, because the N Anna is the left edge of the IP, NP, and N constituents and the right edge of the NP and N constituents, the representation of the s-string unit for Anna in ( 17 ) is:

There are two p-string elements corresponding to the s-string unit for Anna : ‘æ’ and ‘nə’. The first of these is associated with the left edge of a PW, a PhP, the IntP, and the Utt in the p-structure. As is the case for edge information related to s-string units, this edge information about p-string units appears as values of the features l and r within p-string attribute structures, as shown in ( 25 ). 14

The features fm , l , and r are not the only features that have been proposed for s-string and p-string attribute structures. Mycock and Lowe ( 2013 ) also include prosodic information associated with particular p-string units in p-string feature structures. For example, they assume the feature syllstress with value p in order to represent the location of primary stress. Thus, the full representation of the two p-string units for Anna in ( 17 ) is:

The mechanism by which c-structure and p-structure category edge information is specified as appearing in s-string and p-string features is defined by Mycock and Lowe ( 2013 ) in terms of the relation between a c-structure or p-structure node and its rightmost or leftmost daughter. They define a relation D which, when applied to a c-structure or p-structure node, finds the set of immediate daughter nodes of the node to which it applies; it is therefore the inverse of the mother relation ℳ discussed in § 5.3.1 . The leftmost immediate daughter of the node in question can then be defined as the member of the set of daughter nodes which is not preceded, in linear terms, by any other member of the same set. Likewise, the rightmost immediate daughter of the node can be defined as the member of the set of daughter nodes which is not followed, in linear terms, by any other member of the same set. These functions, from nodes to leftmost and rightmost immediate daughters, are labeled D l and D r respectively, and are defined, in relation to c-structure nodes, as follows:

This is shown graphically in ( 27 ), where the mother node is represented as c 1 , the leftmost daughter is D l ( c 1 ), and the rightmost daughter is D r ( c 1 ):

The leftmost and rightmost terminal nodes dominated by a node can be defined by the recursive application of these functions until a node is reached that has no daughters. Mycock and Lowe ( 2013 ) label the functions that find these terminal nodes as T l and T r respectively. The definitions they provide are given in ( 28 ). These rules can be informally read as follows: the leftmost/rightmost terminal node from the current node is the current node if the current node has no daughters; otherwise, it is the leftmost/rightmost terminal node of the leftmost/rightmost daughter (respectively) of the current node. The rule applies recursively to find the appropriate terminal descendant of any node:

In ( 29 ), we display a sample tree in which the current node (* in the definitions above) is represented as c 1 , the leftmost terminal daughter is T l ( c 1 ), and the rightmost terminal daughter is T r ( c 1 ):

Finally, the s-string elements corresponding to these terminal nodes are straightforwardly obtained by applying the inverse function π −1 from terminal nodes of the c-structure to elements of the s-string. In ( 30 ), the leftmost s-string element of the node labeled c 1 is π −1 ( T l ( c 1 )), and the rightmost s-string element is π −1 ( T r ( c 1 )):

Mycock and Lowe ( 2013 ) propose to use arrows as abbreviations for the paths from c-structure nodes to leftmost and rightmost string elements. The arrow ⇙ abbreviates the function from a c-structure node to the leftmost s-string element dominated by it, and the arrow ⇘ abbreviates the function from a c-structure node to the rightmost s-string element dominated by it:

The tree in ( 32 ) provides an example of the use of ⇙ and ⇘:

The tree in ( 35 ) explicitly shows these specifications on all c-structure and p-structure nodes. Notice that the labels appearing in the sets which are the values of each string unit's l and r features correspond exactly to the specifications made by the arrows in the c- and p-structures.

As mentioned previously, in addition to the features fm , l , and r , Mycock and Lowe ( 2013 ) include prosodic information associated with particular p-string units in p-string feature structures. The feature syllstress with value p , which appears in ( 35 ), represents the location of primary stress: 15

The approach to prosody and its interfaces developed by Dalrymple and Mycock ( 2011 ) and illustrated in ( 35 ) makes crucial reference to edge information associated with string elements. Dalrymple and Mycock ( 2011 ) themselves note that “passing information about the edges of all major constituents into the p-string and s-string … may seem excessive,” but also point out that in at least some cases, for instance when speech is slower and more careful, a greater degree of alignment exists between syntactic and prosodic constituents, meaning that “information about constituent boundaries must be available at the interface in order that the relevant alignment principles can apply to give phrasing which reflects the appropriate degree of increased isomorphism.” However, Dalrymple and Mycock also state that “this does not mean that we necessarily expect all of this information to emerge as being relevant at the interface. In fact, it is an important feature of this architecture that it enables us to explore the question of precisely which aspects of prosodic and syntactic structure are important at the interface and to investigate why this should be, both crosslinguistically and on a language-by-language basis.”

One example of a case where alignment between syntactic and prosodic constituents is key is the type of disambiguation analyzed by Butt and King ( 1998a ) and presented in § 11.3 . To recap briefly, two readings are associated with the Bengali sentence in ( 36 ), with the ambiguity being resolved by prosodic phrasing:

Crucial to disambiguation of the sentence in ( 36 ) is the alignment of phrase-level p-structure and c-structure boundaries: the difference between Reading 1 and Reading 2 is whether b h ut (‘ghost') is a phrase in its own right (i.e. a separate NP in the c-structure tree, and a separate PhP at prosodic structure). If b h ut is simultaneously the left and the right edge of a PhP in the p-structure, then it is also the left and right edge of a NP in the c-structure and the appropriate reading is Reading 2; otherwise, it is Reading 1.

In Dalrymple and Mycock's model (see also Mycock and Lowe 2013 ), a principle of “Interface Harmony” is important to the analysis of this and other instances of prosodic disambiguation, as well as to the increasing alignment which occurs as speech rate decreases. Interface Harmony requires that certain information introduced at c-structure (such as the location of a phrase's left or right edge, for example), which has its presence recorded via a label in the s-string (by virtue of being a member of the l or r set in such a unit's feature structure), must be matched with a label in the corresponding p-string unit; both labels should be values of the same feature (i.e. l or r ) in the relevant string units. That is, at the interface between s-string and p-string, a principle of Interface Harmony applies which can require that two labels co-occur—one in an s-string unit and one in a p-string unit. Note that these are labels : they are associated with but do not in themselves constitute specific syntactic or prosodic information. Prosodic disambiguation relies on the alignment of c-structure and p-structure boundaries and therefore can be understood in terms of Interface Harmony. The Bengali example provided by Butt and King ( 1998a ) illustrates this well. In ( 37 ), we see how the boundaries of phrase-level prosodic constituents (parentheses) and syntactic constituents (square brackets) align in the respective readings:

Each phrasal boundary in the syntax (NP or VP) has a corresponding PhP boundary in the prosodic analysis. 16 (There is no “double marking” or “double edge effect” if an edge of the VP also represents an edge of an NP.) The two readings are therefore associated with distinct c-structure and p-structure pairs. As proposed by Dalrymple and Mycock ( 2011 ), information about the edges of all major constituents is recorded as labels in l or r feature-value sets, by means of requirements like those given in ( 34 ). The string representations for the Bengali utterances, based on the analysis presented by Butt and King ( 1998a ) (which itself is based on that of Hayes and Lahiri 1991 ), are as shown in ( 38 ) and ( 39 ). Under neutral focus, a High tone, represented in the p-string feature structure as tone h , is associated with the leftmost PW of the rightmost PhP in Bengali (Hayes and Lahiri 1991 ). This serves as an indicator of prosodic constituency, as examples ( 38 ) and ( 39 ) show. The labels that align and upon which the disambiguation relies are given in bold in both examples; the first word ami ‘I' is omitted.

In examples ( 38 ) and ( 39 ), the principle of Interface Harmony, coupled with the rules of alignment in Bengali, mean that XP labels in an s-string feature structure are matched with phrase labels in the values of the l or r attributes of the corresponding p-string feature structure. This means that, ordinarily at least, the s-string in ( 38 ) is not paired with the p-string in ( 39 ), for example. Such constraints on wellformedness at the interface between the syntax and phonology macro- modules, situated in the string, account for the relationship between c-structure and p-structure in cases of prosodic disambiguation. They also account for the ambiguity that remains when only the written form of the sentence is available, because without p-structure constituent edges to match with, in principle either of the two c-string/s-string analyses is possible.

When it comes to Interface Harmony, determining exactly what prosodic information (for example, boundary type) is important to the syntax, and vice versa, is a key area for future research. For instance, in the Bengali examples we see that each NP forms a separate PhP comprising the relevant PWs, with the attendant result that in the literal reading (Reading 2), the V dek h lam forms a PhP on its own. However, as we will see in the following sections, information about the location of syntactic and prosodic constituent boundaries is not the only sort of information that may be associated with string units and may therefore appear as values of the l and r features; importantly, any grammatical or prosodic feature may contribute a label to the feature structure of the s-string or p-string unit at its left or right edge. In the next two sections, we show how this approach to p-structure can be used to analyze phenomena in which prosody plays an important role.

11.5 Declarative questions

The first phenomenon which we will explore with respect to prosody and its interfaces is that of English declarative questions. Similar to the Japanese example at the start of this chapter, in declarative questions prosody makes a crucial and unique contribution to the semantic and pragmatic interpretation of the utterance. Specifically, the intonational contour (or tune) is the only means by which a declarative question is marked as a type of question with interrogative semantics, rather than simply an ordinary declarative statement. 17 The relevant tune is characterized by a L(ow) tone preceding a final H(igh) tone; each tone is associated with a particular syllable, as shown:

Following a considerable body of work on intonational phonology (for an overview, see Ladd 2008 ), we assume that a tune such as the one associated with interrogative semantics in ( 40 ) minimally consists of a nuclear tone and a boundary tone. The nuclear tone is the pitch target which bears the main stress in an IntP. Boundary tones appear at one edge or both edges of an IntP. A left boundary tone is associated with the first syllable in an IntP, while a right boundary tone is associated with the final syllable in an IntP.

The English interrogative tune in ( 40 ) can be characterized as consisting of a l nuclear tone and a h right boundary tone. In the English interrogative tune, the nuclear tone is associated with the stressed syllable of the first PW in the last PhP in the IntP; see, for example, Hayes and Lahiri ( 1991 ). A tone is rendered simultaneously with the associated syllable. In order to model this, we must refer to the rightmost and leftmost p-string units in a prosodic projection that are marked for primary stress ; that is, the leftmost and rightmost syllables within a projection that are specified as being the location of primary stress (represented as the p-string feature syllstress   p ). Recall that T l (⋄) is defined as β −1 ( T l (⋄)), the leftmost p-string element of the prosodic structure node ⋄; similarly, T r (⋄) is defined as β −1 ( T r (⋄)), the rightmost p-string element. β is the function from p-string elements to terminal nodes of the prosodic structure, and β −1 is the inverse of β . An example configuration is:

Thus, we define the relations T ls and T rs , which pick out the stressed syllable closest to the left and right edge of a constituent ⋄, as follows: 18

According to the first line of this definition, T ls (⋄) is the leftmost syllable T l (⋄) in the prosodic phrase if the p-string element corresponding to T l (⋄) contains the feature syllstress with value p . A sample configuration meeting these requirements is given in ( 43 ):

The second line of the definition of T ls applies in case the leftmost syllable does not correspond to a stressed element of the p-string. In that case, we check the next p-string element to the right, and continue rightwards until we encounter a p-string element containing the feature syllstress with value p .

As usual, we can introduce a convenient abbreviation for these concepts, using arrows superscripted with s :

For the prosodic part of the analysis of a declarative question, the interrogative tune is a property associated with an IntP. The specification of this tune is introduced by a prosodic phrase structure rule, parallel to the syntactic phrase structure rules with which we are familiar:

Line 1 of the annotation below the final PhP in ( 45 ) means that the value of n_tone is l for the leftmost stressed syllable (i.e. the leftmost p-string unit including the feature-value pair syllstress   p ) in this PhP. Line 2 requires the value of the rightmost p-string unit in this PhP to have an rb_tone (right boundary tone) feature whose value is h . ( 45 ) captures the intonational contour we are concerned with, but what it does not do is link it to the semantic contribution associated with this interrogative tune. To capture this, we must turn now to the syntactic part of the analysis.

As discussed in Chapter 8 , s(emantic)-structure is projected from f-structure, that is, from the syntax. As a consequence, it is necessary to associate the meaning constructor for interrogative semantics with a syntactic unit representing the clause as a whole. We assume a constructional meaning, introduced in the IP phrase structure rule, which contributes the relevant meaning constructor (§ 8.6 ). For present purposes, we define the semantics of polar interrogativity as follows (which represents a simplification, but suffices for the purposes of illustration in this case):

We abbreviate this meaning constructor as [ PolarInt ]. This interrogative meaning contribution is associated with the right edge of the root node, IP. The phrase structure rule for English declarative questions is therefore:

This rule ensures that a label, “PolarIntSem,” appears as a member of the set value of r in the rightmost s-string unit corresponding to the root IP. This specification always appears together with the polar interrogative meaning constructor [ PolarInt ], as illustrated in ( 52 ).

Returning now to the prosodic part of our analysis, just as on the syntactic side, the interrogative tune specification is always accompanied by a further specification which defines a label PolarInt . This label appears as a member of the set value of r in the rightmost p-string unit in the IntP: 19

Putting together ( 48 ) and ( 45 ), we have a specification of the interrogative tune which combines its tonal features with a label relating to polar interrogativity: 20

Following the analysis of Lahiri and Plank ( 2010 ), adopted here, the last three syllables of university form a prosodic word (PW). A partial representation of the p-structure and p-string for these three syllables illustrates the way this rule determines certain feature values:

We are now in a position to put together the syntactic and prosodic aspects of our analysis of declarative questions. Crucial to this analysis are the feature values in the rightmost string feature structures, specifically “PolarIntSem” in the final s-string unit (see 47) and PolarInt in the final p-string unit (see ( 49 )), which are shown in this partial representation:

But what is the connection between the labels “PolarIntSem” and PolarInt ? Both are clearly intended to link with polar interrogativity, which is characteristic of this construction, but the labels and the syntactic/prosodic structures with which they are associated are not directly dependent on one another: neither is derived from the other and their sphere of application is entirely separate, in line with strict modularity. And yet it is clear that the two must be related if the analysis is to achieve its aim of pairing the interrogative tune with the appropriate meaning constructor. Once again, the key assumption is the principle of Interface Harmony.

Interface Harmony requires that information about a meaning constructor introduced at c-structure (for example [ PolarInt ]), which has its presence recorded via a label in the s-string (i.e. “PolarIntSem”), must be matched with an equivalent label in the corresponding p-string unit (i.e. PolarInt ). As in other cases, both labels should be associated with the same feature (i.e. l or r ) in the relevant string units. Because the labels “PolarIntSem” and PolarInt are intrinsically related to one of the two separate phrase structure rules which introduce either the relevant meaning constructor or the relevant tune, the appropriate semantics and intonational contour must also co-occur. In the case of a declarative question, then, the principle of Interface Harmony applies so that, if “PolarIntSem” appears as a member of the set value of r in any s-string unit, then the equivalent label PolarInt must appear in the corresponding p-string unit as a member of the set value of r (along with the feature values associated with the rest of the interrogative tune phrase structure rule, as shown in ( 49 ) and ( 50 )). In the case of ( 52 ), the relevant labels co-occur and correspond to one another; if they did not, Interface Harmony would be violated and the resulting structures would be ungrammatical. The phrases to which these labels are related (IP on the syntactic side, IntP on the prosodic side) are also wellformed according to ( 47 ) and ( 49 ). This means that the structure shown in ( 52 ) is grammatical. In this way, the contribution of prosody to syntax and semantics can be modeled in a framework that does not require a direct relation between syntactic and prosodic structure, and does not blur the distinction between these two modules of grammar:

In this section, we have seen how to model p-structure and account for the ways in which it interacts with syntax and semantics. In the next section, we turn to another aspect of interpretation that can have a close relationship with prosody: information structure.

11.6 Prosodic focus marking

In this section, we explore how prosody can contribute to the interpretation of an utterance as a result of its relationship to information structure. In the following, we analyze data first discussed by Mycock and Lowe ( 2013 ); we broadly follow the proposals of Lowe and Mycock ( 2014 ), but alter their account slightly in order to maintain consistency with the model of information structure described in Chapter 10 .

In the following English examples, the focused element can be identified as the element which bears the main or nuclear stress in the utterance (see, for instance, Ladd 2008 ). We identify this element as bearing the Nuclear Tone. The Nuclear Tone is the final pitch accent within the relevant IntP domain. This pitch accent is perceived as being the most prominent within the intonational contour under consideration. In a declarative statement such as the answer in ( 53 ), the Nuclear Tone is a h (igh) tone:

In this example, Norman in the answer corresponds to the queried constituent in the question, and thus functions as the focus in the reply sentence. Contrast this with the following example, where Anna corresponds to the questioned constituent, and hence is the focus. In this case, Anna bears the Nuclear Tone:

The answer sentences in ( 53 ) and ( 54 ) exemplify narrow focus : the word that bears prosodic focus marking is also the only element that bears focus status at information structure. However, it is also possible for a single word to bear prosodic focus marking when more than one word has focus status at information structure. This is known as broad focus , and is exemplified in ( 55 ). In ( 55a ), the whole VP in the answer sentence has focus status, but the Nuclear Tone is associated with Norman (specifically, with the initial stressed syllable of Norman ). 21 In ( 55b ), the subject NP some old woman is the focus of the answer sentence, but the Nuclear Tone is associated with the initial stressed syllable of woman ; cf. ( 54 ):

Although the distinction between broad and narrow focus is generally accepted in the literature on focus marking, in an important sense the analysis that we adopt here transcends the distinction between broad and narrow focus. This distinction depends on different correlations between focus in the syntax and information structure on the one hand, and the prosodic marking of focus status on the other. In the modular grammatical architecture which we assume, these different aspects of focus and its encoding are specified separately. Important to understanding these inextricably linked dimensions of the notion “focus” are two concepts which we refer to as Extent of Focus (Foc-Extent) and Exponent of Focus (Foc-Exponent). Foc-Extent (also known as the Focus Domain) refers to the portion of a sentence which can be said to have focus status in information structure terms, while Foc-Exponent is the indication at some level of representation, for example prosody (p-structure), of the focus status of part or all of a sentence. 22 In the examples in ( 55 ), square brackets enclose the syntactic elements that constitute the Foc-Extent, while the h Nuclear Tone annotation indicates the word (given in bold) which bears the focus marking (the main stress in the sentence).

Of course, the precise definition of Foc-Extent depends on the general approach taken to information structure and its relation to other aspects of linguistic structure within the grammar. We define the Foc-Extent as the set of meaning constructors corresponding to syntactic elements that are associated with focus. In the approach introduced in Chapter 10 , elements of a sentence's meaning—i.e. meaning constructors, which appear in bold in ( 56 )—are categorized according to their discourse function ( df ) at s(emantic)-structure, and consequently belong to the relevant set (for example topic , focus ) at the level of i(nformation)- structure.

In ( 56 ), Anna is the subject of a clause, and is also the topic at i-structure by virtue of appearing in the specifier of IP. Key to this analysis is the attribute-value pair df topic included in the s-structure for ‘Anna', a σ . This information, combined with the annotations on the terminal node and the specifier of IP node in the c-structure, serves to categorize the relevant meaning constructor as belonging to the topic set in the clause's i-structure, f σι :

Given the approach to i-structure exemplified in ( 56 ), Foc-Extent is equivalent to the set of meaning constructors which are the value of the attribute focus at i-structure. These meaning constructors are semantic units and thus correspond—sometimes imperfectly—to units at other levels of representation, such as syntactic constituents.

It is possible to provide a relatively straightforward informal generalization concerning the relationship between Foc-Extent and Foc-Exponent in English: the Foc-Exponent is associated with the Prosodic Word which corresponds to the rightmost syntactic word that is the syntactic realization of the Foc-Extent. However, a full formal analysis represents a significant challenge, in large part because of the extensive misalignment that is a feature of the correspondences between units belonging to different levels of representation.

When we consider Foc-Exponent and Foc-Extent separately, the difference between narrow and broad focus collapses to an extent. Based purely on their syntax, these two types of focus are fundamentally the same: in all of the examples in ( 53 ), ( 54 ) and ( 55 ), the Foc-Extent is a single syntactic constituent. Similarly, the basic facts about the Foc-Exponent are the same in all cases: the rightmost Prosodic Word of the focus constituent bears the Nuclear Tone.

In order to provide an analysis of prosodic focus marking, we employ the mechanisms introduced previously, including the arrows defined in § 11.4.1 .

As with declarative questions, we represent the prosodic contour of a phrase with focus marking by means of prosodic phrase structure rules. Focus marking of the kind under discussion involves sentences that are declarative statements. We assume this prosodic phrase structure rule for declarative statements: 23

The Foc-Exponent, captured by the rule in ( 58 ), serves to delimit the Foc-Extent. Foc-Extent is expressed via separate c-structure rules which, together with the p-structure rule in ( 58 ) and the principle of Interface Harmony, play an equally important role in the analysis of prosodic focus marking as an interface phenomenon.

In the case of the sentence in ( 53 ), the focus constituent in the answer sentence ( Norman ) could be identified with either the NP, the N′, or the N which dominates the word Norman in the c-structure. Following Mycock and Lowe ( 2013 ), we assume here that the phrase in question is the NP, since focused constituents often correspond to XP categories, but it makes no difference to the analysis of the present example.

We propose the following node annotation principle for Extent of Focus in English:

These constraints specify that the label “DFFoc” appears as a member of the set value of the features l in the leftmost string element (⇙ l ) and r in the rightmost string element (⇘ r ) corresponding to the c-structure node on which the annotations appear. Thus this label serves to effectively delimit the focused constituent. In the present example, the annotated node is an NP consisting of only a single word ( Norman ), so the leftmost and rightmost string elements dominated by the NP are the same. This is not the case when the phrase in focus consists of more than one word, as, for example, in instances of “broad” focus (see ( 55 )), and it is for this reason that separate specifications are required for the left edge and right edge of the phrase.

The principle in ( 59 ) specifies only the presence of a label, “DFFoc,” as a member of the l and r sets in particular string elements. On its own, it does not say anything about the information structure status of the words in the focused phrase. In order to specify this, we must introduce a wellformedness constraint on strings, or a string constraint (see Lowe and Mycock 2014 and § 6.11.1 ). String constraints define the valid s-strings or p-strings for a language. A string constraint consists of a regular expression encoding the constraint, and a label which specifies the aspect of the string referred to by the constraint (here s-string , the other possibility being p-string ). We make use of the symbol Σ, which represents a string element of any form.

The labels specified in ( 59 ) serve to delimit a span in the s-string: the leftmost s-string element in the span is marked by having the label “DFFoc” as a member of the l value set, while the rightmost one is marked by having the label “DFFoc” in its r value set. Our aim is to define a string constraint which requires every word in the relevant span to be marked in such a way that its meaning constructor appears in the focus set at i-structure. For clarity, we define the following abbreviatory templates:

The template inclLeft ensures that its argument _ α is a member of the l set of the current string element, and similarly for the template inclRight . The template sdf specifies that its argument _ α is the value of the df attribute of the s-structure ↑ σ , and nsdf specifies that the df value of ↑ σ is different from _ α . These basic templates are used in the definition of the templates lfocus and rfocus :

The f-description lfocus holds of a string element if “DFFoc” is a member of its l set, and the value of the df attribute in its mother's semantic structure is focus . Recall that we use the up arrow ↑ in lexical entries to refer to the c-structure node accessible from an s-string element via the π −1 relation (§ 11.4.1 ):

Similarly, rfocus holds of a string element if “DFFoc” is a member of its r set, and the df value of its mother's semantic structure is focus :

We then define the metacategory in ( 64 ), representing the Foc-Extent in terms of s-string items:

The definition in ( 64 ) is a disjunction over two possible sequences: in both sequences, the leftmost element must be associated with the feature df   focus at s-structure and must have the label “DFFoc” as a member of its l value set (in both cases, as a result of application of the template @ lfocus ); likewise in both sequences, the rightmost element must be associated with df   focus at s-structure and must have the label “DFFoc” as a member of the r value set (by the application of the template @ rfocus ). The only difference between the two possibilities is whether the Foc-Extent is more than one s-string unit (that is, whether the s-string element with l {DFFoc} is different from the element with r {DFFoc}), or only one s-string unit with “DFFoc” as a member of both its l and r value sets. If there are more than two elements in the string, the elements that are neither leftmost nor rightmost must be associated with df   focus at s-structure. This is achieved by application of the template @ sdf ( focus ).

We can use these definitions in formulating the s-string constraint in ( 65 ), which we call focus . It allows for the appearance of several focused constituents in a string, each preceded and followed by string elements that are not in focus (marked with @ nsdf ( focus )). The regular expression focus string constraint constitutes one component of the s-string constraints that are relevant for English. It is intersected with all of the other s-string constraints for English to obtain the fully specified s-string wellformedness condition, as discussed in § 6.11.1 .

The rule in ( 65 ) permits one or more sequences consisting of any number of string elements that are not associated with focus at s-structure (by the template @ nsdf ( focus )), optionally followed by a sequence fitting the definition of Σ Foc , followed by any number of string elements that are not associated with focus at s-structure. That is, for any string in English the rule in ( 65 ) licenses one or more substrings of which every element is associated with df   focus at s-structure, and of which the leftmost and rightmost elements are marked by the appearance of “DFFoc” in their l or r value sets respectively. Following the approach to information structure presented in Chapter 10 , the association with df   focus at s-structure has the effect that the meaning constructors corresponding to the s-string unit(s) in this sub-string appear in the focus set at i-structure.

The p-structure rules in ( 57 ) and ( 58 ), the c-structure constraint in ( 59 ), and the s-string constraint in ( 65 ) are the basis for the analysis shown in ( 66 ) for the example given in ( 53 ).

In this utterance, the stressed syllable of the word ‘Norman' /nɔː/ is associated with the Nuclear Tone ( n_tone h ). The prosodic phrase structure rule in ( 58 ) may thus apply to the Prosodic Word dominating that syllable, in this case the final Prosodic Word in the p-structure pronounced /nɔː.mən/. As a result of ( 58 ) applying, the label DFFoc appears as a member of the set value of r in the syllable which has primary stress within this Prosodic Word, i.e. the first syllable of ‘Norman'. On the syntactic side, the phrase-structure rule in ( 59 ) applies to the NP that corresponds to the s-string element ‘Norman'. By this rule, the label “DFFoc” must appear in the l and r sets in the corresponding leftmost and rightmost s-string elements respectively, which in this case are the same element. By the string constraint in ( 65 ), the meaning constructor corresponding to this s-string element, that is the meaning constructor introduced in the lexical entry for the name Norman , appears in the focus set at information structure. The principle of Interface Harmony requires that an s-string element which has a feature “DFFoc” in its r value set must be associated with a p-string element that has the feature DFFoc in its r value set. In ( 66 ), this requirement is satisfied, and the analysis succeeds. If, say, the c-structure rule in ( 59 ) were to apply not to the final NP in the c-structure but to the first one (meaning that Anna was in focus at information structure), the analysis would fail because “DFFoc” would appear in the r set in the s-string element Anna but would not be associated with a DFFoc feature in a corresponding p-string element. That is, prosodic and syntactic specifications are effectively required to match up, such that whichever prosodic element bears the Nuclear Tone necessarily corresponds to the syntactic element associated with the meaning constructor(s) that represent the focus at information structure.

Precisely the same set of rules can equally well account for instances of broad focus, where the Foc-Extent in the c-structure and s-string spans more than one word. Example ( 67 ) shows the analysis for ( 55a ), an example of broad focus in which the Foc-Extent is hit Norman . The only difference from the structure in ( 66 ) is that the c-structure rule in ( 59 ) applies at the VP level, meaning that the label “DFFoc” appears in the l set in the s-string element corresponding to the word hit . This in turn means that the meaning constructor for hit appears in the focus set at i-structure along with that for Norman , whose r set in the s-string contains the label “DFFoc” due to the annotation DFFoc ∈ (⇘ r ) on the VP node.

In this section, we have provided an analysis of prosodic focus marking in English. The approach which we have presented can also be used to analyze prosodic marking of this and other information structure categories crosslinguistically.

11.7 Further reading and related issues

Bögel ( 2012 , 2013 , 2014 , 2015 ) has developed a distinct approach to prosody within the LFG framework whose formulation makes crucial reference to actual values of the acoustic characteristics of the speech signal. Bögel's proposals are in part influenced by the proposals of Dalrymple and Mycock ( 2011 ), but in certain respects depart significantly from other approaches to prosody within LFG. Bögel ( 2015 ) argues against a dependence on the prosodic hierarchy as a means of modeling p-structure, and proposes a model based on the “p-diagram.” The “p-diagram” represents speech as a series of feature vectors , each of which encodes the phonetic and phonological properties of a single syllable. Bögel's prosodic analysis is therefore fundamentally syllable-based, although in principle it could be made more fine-grained. The phonetic and phonological properties of individual syllables are accessible by other components of the grammar via a projection function. Bögel uses this approach in her analysis of prosodic resolution of syntactic ambiguities (see Bögel 2015 : 70–7), as well as other phenomena.

Clitic positioning and the treatment of syntax-prosody mismatches are discussed by O’Connor ( 2002a , b , 2005a ), Lowe ( 2011 , 2016a ), and Bögel ( 2014 , 2015 ). Mycock ( 2006 , 2007 ) presents an LFG approach to constituent question formation crosslinguistically which includes description and analysis of question intonation, and Mycock ( 2010 ) discusses the modeling of prosody and its interfaces, exemplifying the analysis with data from Hungarian. It must be noted that in these works the authors do not necessarily assume the version of LFG's parallel architecture that is adopted here.

Bögel's ( 2012 , 2013 , 2014 , 2015 ) p-diagram approach, grounded in details of the speech signal, represents an alternative approach within the LFG framework; this is briefly discussed in § 11.7 .

For an interesting critical review of the Prosodic Hierarchy and Prosodic Phonology in general, see Scheer ( 2011 ).

Syllables consist of one or more morae, the smallest prosodic units, but we will not make reference to morae here. Morae themselves are analyzed as consisting of smaller phonological units, segments and features , units which are generally considered to be sub-prosodic.

As it is not our intention to provide a comprehensive introduction to prosody and phonology, we set aside the issue of precisely how stress should be defined. For an overview, see Fox ( 2000 ).

There are many alternative definitions of these prosodic constituents. See, for example, Selkirk ( 1978 ), Nespor and Vogel ( 1986 ), Levelt ( 1989 ), Wheeldon ( 2000 ), and Frota ( 2012 ). As stated previously, we seek to define these constituents in purely prosodic terms, whereas many of these authors define them in a combination of syntactic and prosodic terms.

This strict view of the constraints on prosodic domination has been called into question by a number of authors, including Inkelas ( 1989 ) and Itô and Mester ( 2003 ). In particular, it has been noted that exhaustivity and the bar on recursivity may be best viewed as tendencies, given that they can be violated in a number of languages. For a recasting of the Strict Layer Hypothesis in Optimality-Theoretic terms that takes this into account, see Selkirk ( 1995 ). Bögel ( 2015 ) provides an alternative approach within LFG which explicitly avoids the prosodic hierarchy as a basis for prosodic structure.

Note that the p-structure feature p-form is not the same as the f-structure feature pform encoding the form of a preposition (§ 2.5.4 ).

For an introduction to AM/ToBI see Beckman et al. ( 2005 ) and Ladd ( 2008 ); for ToBI analyses of data from a range of languages, see the papers in Jun ( 2005 , 2014 ).

In ( 17 ), each PW is also a separate PhP, but it could be the case that, for example, the PWs are grouped into two PhPs, giving (æ nə wəz stʌ di ɪŋ ət ðə ju nə) PhP (vɜ: sə ti) PhP . Our approach is flexible in order to account for the attested variability in intonation mentioned in § 11.1 . While for our purposes, nothing in ( 17 ) hinges on the placement of PhP boundaries, their inclusion is important because they can be associated with other aspects of prosody. For instance, voicing assimilation in Bengali is bounded within PhPs (Hayes and Lahiri 1991 ). Thus, even if a PhP comprises a single PW, this does not mean that the two can or should be conflated.

This is a simplification. As Bögel ( 2015 ) observes, following Levelt et al. ( 1999 ), segments and the metrical frame are most likely stored separately in the lexicon. See Bögel ( 2015 ) for more on the precise contents of the p-form portion of a lexical entry.

For a formalization of some postlexical phonological processes in an LFG setting, see Bögel ( 2015 ).

This issue affects not only the p-form–p-string relation, but also the s-form–s-string relation, since single lexical entries can specify multiple s-string elements; see Lowe ( 2016a ).

On the function λ , which relates a node of the tree to its label, see § 5.1.2 .

For a different approach to capturing prosodic phrasing and its encoding in LFG, see Bögel's work on the p-diagram. Bögel ( 2015 ) provides full details; see also § 11.7 .

Lowe ( 2016a ) also proposes a feature clitic , which can appear in both s-string and p-string units, to distinguish syntactic or prosodic clitics from non-clitics at the level of the string, and an s-string or p-string feature field , which contains labels referring to all syntactic categories dominating a particular element in the c-structure or to all the prosodic categories dominating a particular element in the p-structure.

These boundaries are crucial in accounting for different voicing assimilation possibilities in these two examples; see Hayes and Lahiri ( 1991 ).

In what follows, we assume that a declarative question has the same interpretation as an equivalent polar interrogative that is syntactically marked as such by subject-auxiliary inversion, for example Was Anna studying at the university? This is a simplification, as declarative questions exhibit contextual restrictions which polar interrogatives do not; on this, see, for example, Gunlogson ( 2003 ). Our aim here is to illustrate the mechanism whereby prosody can have an effect on meaning within the LFG framework. The full details of an LFG analysis of declarative questions await further research.

The function N , which appears in ( 42 ), finds the next element in linear order when applied to string elements, as discussed in § 6.11.2 ; N −1 finds the preceding element.

We follow Mycock and Lowe ( 2013 ) in distinguishing syntactic labels like “PolarIntSem” from prosodic labels like PolarInt by putting the latter in italics. Note that these are simply labels, and do not imply the presence of semantic properties in either the c-structure or p-structure: PolarInt could equally well be labeled XYZ , or “PolarIntSem” could be “XYZ.”

Note that we do not assume a one-to-one relationship between a tune and meaning.

Here, we set aside the issue of an event, but not the type of that event, being presupposed in the context of What happened? or What did X do (to Y)? type questions. When such a question is asked, it is in fact only the type of the event that is in focus: the occurrence of some event is presupposed. On this, see Mycock ( 2006 ). For the purpose of illustrating the analysis of prosodic focus marking in LFG, we make the simplifying assumption that a single meaning constructor is associated with a verb, and thus that the verb is associated with a single discourse function at information structure. If one were to adopt a neo-Davidsonian approach to verb meaning instead, the distinction between event and event type could be captured using the s-structure attributes rel and event (see § 9.10.5.2 ), and these two aspects of verb meaning could be analyzed as belonging to different information structure categories.

Of course, information structure may also be indicated at other levels of structure, for example c-structure, but our concern here is with prosodic marking.

For the use of the ampersand here, see § 6.1.1 .

As with PolarInt above, note that DFFoc is simply a label; it does not situate discourse features in p-structure.

We specify and require harmony for only the r features in the s-string and p-string. This reflects the fact that focus marking in English is fundamentally right-edge based. We make no claims as to whether or not the same is true for other languages.

• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Have an account?

Prosodic Features of Speech

8th - 9th grade.

15 questions

Introducing new   Paper mode

No student devices needed.   Know more

1. This is a prosodic feature of speech that deals with variety of pitch.

A. Intonation

B. Juncture

2. __________ is the highness or lowness of voice.

A. Prosodic feature

D. Intonation

3. The level of pitch that conveys strong emotions

D. very high

4. “I won the lottery.” If read in a normal and low pitches, it shows _______.

B. Surprised

5. My heart is torn into two! If read with normal and high pitches, it shows _____.

A. Distress

B. Sarcastic

D. Happiness

6. A meaning of sentence changes with a slight change of intonation.

B. Partly True

D. Not Applicable

7. There is no intonation without pitch.

B. Partly true

D. None of the above

8. Pitch: Voice:: _____: Loudness

D. Suprasegmentals

9. Stress, Pitch, Intonation, Projection, Speed and Voice are all called:

A. Prosodic features of speech

B. Parts of Speech

C. Figures of Speech

D. All of the above

10. Juncture: Pauses:: Stress:Accent: Prosodics: ________

C. Suprasegmentals

11. Which of the following statements is FALSE?

A. Prosodic features of speech is all about the rhythm that is involved in speaking.

B. Misunderstandings happen because the suprasegmental are neglected.

C. Changing the volume of the voice has no effect on the speakers ability to convey idea.

D. Prosodic features are important in public speaking

12. The ability to adjust the volume of your voice in relation to the size of your audience.

D. Projection

13. Covid-19 is contagious

If the intonation goes high, what is implication of the sentence?

14. Suprasegmental sounds deal with ____________.

The resultant pattern of rising and falling of sounds

A. Inflection

B. Intonation

Explore all questions with a free account

Continue with email

Continue with phone

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

Enter the email address you signed up with and we'll email you a reset link.

• We're Hiring!
• Help Center

Prosody and Suprasegmental Features

Related Papers

Language, Cognition and Neuroscience

Jennifer Cole

Hassan ( E A S Z Y T O U C H ) A . Gbelekale

This paper underscores that prosodic features are the crucial elements in spoken languages that propel and enhance the meaning-making potential of language in use in a communication process. It is dominantly realised in interlocutors’ speech in their effort to produce utterances and communicate intentions. It also helps the audience to infer contextual and situational meaning, entailments, presuppositions and implied meaning in utterances that would otherwise lack meaning if treated only in content or denotatively.

Francis Picavet

Ryan Sandell

ANU Press eBooks

Henrike Frye

Brain and Cognition

Henri Cohen , Josée Douaire

jacqueline vaissiere

Geert Booij

Journal of Pragmatics

Anne Wichmann

UPenn Ph.D. dissertation

Yong-cheol Lee

The fact that “purely” prosodic marking of focus may be weaker in some languages than in others, and that it varies in certain circumstances even within a single language, has not been commonly recognized. Therefore, this dissertation investigated whether and how purely prosodic marking of focus varies within and across languages. We conducted production and perception experiments using a paradigm of 10-digit phone-number strings in which the same material and discourse contexts were used in different languages. The results demonstrated that prosodic marking of focus varied across languages. Speakers of American English, Mandarin Chinese, and Standard French clearly modulated duration, pitch, and intensity to indicate the position of corrective focus. Listeners of these languages recognized the focus position with high accuracy. Conversely, speakers of Seoul Korean, South Kyungsang Korean, Tokyo Japanese, and Suzhou Wu produced a weak and ambiguous modulation by focus, resulting in a poor identification performance. This dissertation also revealed that prosodic marking of focus varied even within a single language. In Mandarin Chinese, a focused low/dipping tone (tone 3) received a relatively poor identification rate compared to other focused tones (about 77% vs. 91%). This lower identification performance was due to the smaller capacity of tone 3 for pitch range expansion and local dissimilatory effects around tone 3 focus. In Seoul Korean, prosodic marking of focus differed based on the tonal contrast (post-lexical low vs. high tones). The identification rate of high tones was twice as high than that of low tones (about 24% vs. 51%), the reason being that low tones had a smaller capacity for pitch range expansion than high tones. All things considered, this dissertation demonstrates that prosodic focus is not always expressed by concomitant increased duration, pitch, and intensity. Accordingly, “purely” prosodic marking of focus is neither completely universal nor automatic, but rather is expressed through the prosodic structure of each language. Since the striking difference in focus-marking success does not seem to be determined by any previously-described typological feature, this must be regarded as an indicator of a new typological dimension, or as a function of a new typological space.

RELATED PAPERS

Wojciech Wciórka

Pilar Diez del Corral Corredoira

Annals of Neurology

Belinda Perez Zuniga

Frontiers in Education

Josefina Amanda Suyo-Vega

European Psychiatry

Silvia Alboni

Sains Malaysiana

StephenL White

Frontiers in Pediatrics

Andromachi Katsonouri

Health Psychology

Jeannette Ickovics

Elio Dávila Hilarión

Lecture Notes in Computer Science

Dieter Landes

Research in Higher Education

Alberto Cabrera

IFIP Advances in Information and Communication Technology

Todor Ganchev

Demetris Athanasoulis

Anales de Psicología

Yolanda Salat Cuscó

Asmiyenti Djaliasrin

Jordan George

Physical Review E

Yan-Chr Tsai

Carl Stahle

DÉBORA SANTOS DE OLIVEIRA , Araceli García-Rodríguez

DergiPark (Istanbul University)

Arash Omidi

Tidsskrift for Den norske legeforening

Hanna Oommen

RELATED TOPICS

•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024