Introduction

In Chapters 10 and 11, we analyzed Ponapean vowel epenthesis – an alternation phenomenon – and showed that epenthesis is caused by the distributional restrictions on syllable structures such as the condition on coda, referred to as Coda-Filter. This filter can prevent a consonant from syllabifying as a coda if it differs in place from the following consonant. This chapter continues this theme by exploring the ramifications of Coda-Filter for three segmental processes in Diola-Fogny: nasal place assimilation, consonant deletion, and vowel epenthesis. These alternations appear to be unrelated to one another or to syllable structure requirements. We demonstrate that this is not the case. We show that there is a relation underlying them, a functional unity that can be traced to the syllable requirements in Diola-Fogny. This chapter highlights the concept of functional unity or “conspiracy” aka Kisseberth (1970). We present and contrast two analyses – one rule-based and one optimal-theoretic – to show how they handle phenomena that are functionally related.

This chapter has three key objectives. First, it develops your understanding of functional unity. We show that functional unity can manifest itself in different processes within one language or across languages. Second, it continues to develop your understanding of Optimality Theory (OT) by highlighting a second reason against the rule approach, its inability to express the functional unity of phonological processes within and across different languages. Related to the concept of functional unity is the idea of typology, a term linguists use to describe the crosslinguistic variations along some parameter. These two terms – functional unity and typology – will be explained later in the chapter. Third, this chapter strengthens your analytical ability. A significant component of critical and analytical thinking is the ability to identify relations between seemingly unrelated things, whether they are ideas, concepts, or processes. By introducing the three seemingly distinct processes in Diola-Fogny and by comparing them with Ponapean, we develop your capability to discern and express the hidden connections beneath what seem to be unrelated processes.

Introduction

In Chapter 21, we presented an optimal-theoretic analysis, which reveals that Swati reduplication, though a morphological process, is governed by principles of phonology, specifically, conditions on foot structures such as Ft-Bin, Par-Syl, and All-Ft-L. This chapter continues the exploration of the interface between phonology and morphology. We focus on two seemingly unrelated phenomena: LuGanda glide epenthesis and Swati infixal reduplication. In LuGanda, the first person singular prefix /N-/ triggers the insertion of a palatal glide y in v(owel)-initial stems, stems with an initial onsetless syllable. This phonological process is juxtaposed with reduplication, a morphological process. In Swati, v-initial stems with three or more syllables display what appears to be infixal reduplication, with the reduplicant placed after the initial onsetless syllable; otherwise, reduplication is prefixal. On the surface, epenthesis and infixal reduplication appear to be as unrelated as two problems can be. But as we show here, there is a connection, which is tied to onsetless syllables. Glide epenthesis and infixal reduplication both happen only in stems with an initial onsetless syllable. We show that syllables without an onset cause misalignment and trigger compensatory strategies like insertion in LuGanda or infixing reduplication in Swati. This chapter highlights another relation between phonology and morphology: how conditions on syllable affect morphology.

This chapter serves three objectives. First, it underscores the interconnectedness of phonology and morphology. We show how Onset, a phonological condition on syllable, affects the outcomes of prefixation and reduplication. Second, we introduce the Theory of Prosodic Alignment developed in Downing (1998a, 1998b, 2000), which seeks a unified treatment of problems like glide epenthesis and infixal reduplication. We show that morpheme concatenation can be subject to alignment constraints, which specify the location of affixation and reduplication. Furthermore, we show that in addition to input–output and base–reduplicant constraints, there is a third type of correspondence condition that controls relations between morphological and prosodic units. Downing’s proposal extends the concepts of alignment and correspondence, first developed in Prince and Smolensky (1993) and McCarthy and Prince (1993, 1995b, 1999). Third, this chapter develops your ability to identify and capture the relations between patterns by showing how two problems that seem unrelated in one analysis receive a unified treatment in another.

Introduction

Chapter 9 analyzed syllable structure requirements in Ponapean, a distributional problem. We shift the lens in this chapter, focusing instead on an alternation problem. Specifically we consider epenthesis, a type of phonological process that inserts segments not present in the input. On the surface, it appears that epenthesis can be analyzed without reference to syllable structure. We show here that an understanding of syllable structure can reveal the cause of epenthesis as well as help us understand and predict the epenthesis site. The phenomenon we choose to illustrate epenthesis here is taken from Ponapean, which has a productive process of vowel epenthesis. We select Ponapean partially because its epenthesis is related to the syllable structure requirements and partially because it creates continuity, allowing us to build on and extend the understanding of Ponapean syllable structures developed in Chapter 9.

This chapter has four objectives. First, it introduces a new phonological phenomenon – epenthesis – and develops your understanding of epenthesis. Second, starting with the first chapter on Kikuyu vowel sequencing, we hinted that the distributional restrictions play a significant role in alternation. This chapter continues this theme and develops your understanding of how important a role the distributional properties of Ponapean syllables play in epenthesis. Third, this chapter presents and compares three different analyses of epenthesis. This presentation develops your understanding of the history and development in our thinking about epenthesis from a segment-driven phenomenon to a syllable-driven process. This development has resulted in the rejection of the rule and templatic approaches and the adoption of a constraint-based view of epenthesis introduced in Chapter 11. Finally, this chapter strengthens your ability to develop and evaluate competing analyses of the same phenomenon.

Alternation

This unit introduces the problems of alternation, a type of phonological phenomenon in which a unit of meaning such as a root or prefix appears to vary, that is, alternate, in its pronunciation. As this unit demonstrates, such alternations are often predictable from phonological contexts. This unit develops your ability to identify and analyze alternations. Like Unit 1, this unit focuses on segments, that is, alternations affecting segments. We leave alternations involving tone and stress to later units. This unit opens with Chapter 5, which presents an alternation impacting nasal consonants in English. As we showed in Chapter 2, English nasals are subject to co-occurrence restrictions with immediately following consonants. Chapter 5 shows that such distributional restrictions manifest themselves as conditions on alternation. This chapter highlights, among other things, the relations between distributional restrictions and conditions on alternation. Chapter 6 presents a more complex problem of alternation from Tibetan numerals. Without analyses, the problem represented by Tibetan numerals is not obvious. Only through analyses do we discover that Tibetan numerals alternate and what conditions the alternation. This chapter emphasizes the role of phonological analyses and shows how instrumental the concept of underlying representation is to uncovering the patterns of alternation and the relation between alternating forms of Tibetan numerals. Chapters 7 and 8 analyze even more complex problems of alternation: one involving Tonkawa stems and another affecting roots and suffixes in Yawelmani. You will learn not only that there are alternations but that different alternations can interact, creating complex pattern interactions. We demonstrate how such pattern interactions can be identified and analyzed. These four chapters, together with exercises, present a variety of increasingly more complex problems of alternation. They develop and strengthen your understanding of concepts such as morpheme vs. allomorph, underlying vs. surface representations, natural classes and distinctive features, rules and rule ordering, different types of rule relations and linguistic arguments, and so on.

Introduction

In Chapter 6, we analyzed Tibetan numerals. We showed that identifying the internal morphemic boundary of complex numerals such as jugjig ‘eleven’ was not as straightforward as it first appeared. Two ways of parsing jugjig ‘eleven’ and other complex numerals presented themselves: (a) jug-jig or (b) ju-gjig. In (a), g is analyzed as part of the morpheme meaning ‘ten.’ In (b), it is treated as part of the morpheme ‘one.’ Determining which morpheme this consonant belongs to, that is, identifying the morphemic boundary in complex numerals requires analyses. We see that an otherwise straightforward morphological process of compounding is rendered opaque by a phonological process of initial consonant cluster simplification, obscuring the morphemic boundary in complex numerals. In analyzing alternation as a phonological phenomenon, linguists often start by determining the morphemic structure or composition of complex morphological forms. That is, linguists conduct what is referred to as the Morphemic Analysis. Only by establishing the phonetic makeup of morphemes through a morphemic analysis can we determine whether a morpheme alternates and what triggers the alternation. This chapter introduces a problem that requires a morphemic analysis. This analysis reveals an alternation that involves the stems, with each stem exhibiting four distinct phonetic forms or pronunciations. We compare two approaches to the analysis of this problem and show that only one unearths the patterns that underlie the stem alternation. This problem comes from Tonkawa, an American Indian language spoken in Texas.

This chapter has four key objectives. First, it introduces yet another case of phonological alternation. Unlike the alternations we discussed in the two preceding chapters, this alternation illustrates in a more dramatic way the extent to which a morpheme might alternate. It is more complex in that it involves not just one phonological process. Consequently, it requires more untangling. Second, this chapter introduces the Morphemic Analysis and develops your ability to conduct the Morphemic Analysis. Third, this chapter strengthens your ability to analyze phonological alternation as a phenomenon. In particular, it highlights the types of arguments phonologists often advance in support of a particular analysis. Lastly, this chapter reinforces your understanding of distinctive features and the role they play in the expression of natural classes and phonologically significant generalizations.

Syllable

Beginning with this unit, we shift the lens from segmental phenomena to syllable and syllable-related phonological processes. Syllables are phonological units; they organize sounds into larger constituents, similar to phrasal units such as verb phrases in syntax. There are four chapters in this unit. Chapter 9 considers the distribution of syllable in Ponapean and shows that syllables, when examined in relation to where they can appear in a word (i.e. initial, medial, and final) exhibit restrictions. Not all attested syllable types can freely appear in these positions. Ponapean syllables, like segments, are subject to distributional restrictions. This chapter demonstrates how to identify such restrictions and to analyze them via the rule and templatic theories of syllables. Chapter 10 changes the focus from examining syllable itself to analyzing syllable-related problems. Building on what we learned about Ponapean syllables, this chapter explains how requirements on Ponapean syllables help us understand vowel epenthesis: why it occurs and where epenthesis takes places. This chapter evaluates the rule and templatic theories of syllable construction against Ponapean epenthesis and discusses how and where they differ. Starting in Chapter 11, we introduce Optimality Theory (OT), a theory that uses constraints and constraint ranking as opposed to rules and rule ordering in Derivational Theory (DT). Using Ponapean as an example, we demonstrate how OT can explain its syllable distribution and epenthesis. This OT analysis shows how different requirements on syllables can be expressed as constraints and how constraint ranking can account for both the distribution of syllables and vowel epenthesis, an alternation problem. Finally, in Chapter 12, we examine three segmental processes in Diola-Fogny and show how they can be traced to one requirement on syllable. This chapter highlights one key argument against the rule-based DT in favor of the constraint-based OT. Together, the four chapters of this unit showcase not only different phonological problems but also distinct and competing analyses of these problems. In addition, they cover a range of concepts: that is, syllable, onset, rime, nucleus, coda, mora, constraint, constraint ranking, markedness vs. faithfulness constraints, and so on.

Introduction

This chapter examines the restrictions on the distribution of syllable and syllable-internal structures. We show that the distribution of syllable, just like the distribution of segments, may exhibit co-occurrence restrictions and that an understanding of the restrictions is important not only in defining what is a well-formed syllable but also in understanding alternation processes such as epenthesis. The data we use to illustrate the syllable distribution come from Ponapean, a Micronesian language spoken by about 18,000 people on the island of Ponape, an island located roughly halfway between Hawaii and Indonesia. We select Ponapean because Ponapean syllables reveal a number of common restrictions. These restrictions make it far from straightforward to define the well-formed syllable and determine what is or is not allowed. By analyzing the syllable and syllable-internal restrictions in Ponapean, we highlight what an analysis of syllable distribution entails and how linguists go about analyzing these restrictions.

This chapter has three key objectives. First, it introduces the distributional restrictions on syllable and syllable-internal constituents. It develops your understanding of these restrictions. Second, we compare two theories of syllabification. In one theory, syllable structures are defined by rules; syllabification is viewed as applying a set of ordered rules (Steriade 1982; Levin 1985). The second theory characterizes syllable as a template and conceives syllabification as mapping segments to the template (Itô 1989; Archangeli 1991). This comparison shows how syllable is represented and how syllabification is viewed in these two theories that attempt to capture the predictable nature of syllable structures. This presentation provides the background to analyses of phenomena such as epenthesis, tone, and stress because an understanding of syllable plays an important role in accounts of these phenomena. Finally, through the analysis of Ponapean syllables, this chapter develops your ability to analyze the syllable-related distributional restrictions.

Overview

This is a textbook intended for an introductory course in phonology. The central goal of such a course, we believe, should be the development of students’ abilities to analyze sound patterns in natural languages, hence the title, Analyzing Sound Patterns. By “analytical abilities,” we mean roughly the abilities to: (a) observe, identify, and describe patterns in the data; (b) form hypotheses and construct analyses of the patterns; (c) compare and evaluate competing hypotheses and analyses; (d) develop and evaluate linguistic arguments; and (e) draw conclusions from the analyses and explore their implications. This book is organized around phonological problems and analyses. It is designed to guide students step by step through the analysis, from observation to pattern discovery and from hypothesis formation to analyses and explanations. The book is problem- and analysis-driven rather than concept-driven. It is not organized according to key linguistic and phonological concepts. Key concepts are introduced, but they are embedded in the analyses.

This book consists of six units. They cover six broad areas of phonological study: (a) distribution; (b) alternation; (c) syllable; (d) tone; (e) stress; and (f) prosodic morphology. Unit 1 and Unit 2 focus on two main types of phonological patterns known as distribution (including complementary distribution) and alternation. These patterns are exemplified mostly with data from segmental phonology. These two units introduce a wide range of segmental phenomena from Kikuyu vowel co-occurrence to the distribution of English nasals, and from segment deletion in Tibetan and Tonkawa to assimilatory processes in English and Yawelmani. Starting with Unit 3, the book shifts to suprasegmental phenomena such as syllable, tone, or stress. These units build on earlier themes of distribution and alternation and extend the investigation to processes that affect domains larger than segments. Unit 3 and Unit 4, for instance, open with a chapter on syllable and tonal distribution followed by a chapter that investigates syllable-based and tonal alternations. These chapters not only reinforce the understanding of distribution and alternation but also expand it to new phenomena. They also make transparent the relations between segmental and suprasegmental problems.

Introduction

Up to this unit, we have focused on phonological problems such as distributional restrictions and morphophonemic puzzles such as alternation. Although alternation impacts the phonological forms of words and their internal constituents, the studies of alternation concern the phonological changes caused by morphological processes such as affixation. These phonological changes, while affecting the surface forms of morphological units, generally do not determine how words are formed. Starting from this chapter, we shift our focus away from phonological issues to consider problems of word formation, the central concern of morphology. There is an area of study within morphology that investigates how morphological units take the phonological or prosodic forms they do. This area of morphology is known as prosodic morphology. Prosodic morphology seeks to identify and explain the patterns in how the prosodic shapes of morphological units come about from the ways they are constructed. To illustrate some of the prosodic morphology concerns, we introduce a puzzle from Arabic, in which the plural forms of nouns are built not from concatenating phonological materials to the left or right of the base such as suffixation in English. They appear to be constructed through modifying the base internally, giving rise to the description of this area of study as nonconcatenative morphology. In addition, these broken plurals exhibit a fixed shape, regardless of the base forms from which they are formed. We present two analyses: the template-and-circumscription account of McCarthy and Prince (1990) and the parafixation-and-transfer account in Hammond (1988). Even though both proposals use templates to explain the fixed shape of broken plurals (which leads linguists to describe it as templatic morphology as well), they differ radically in their views of how broken plurals are constructed. Thus, these analyses highlight many of the issues posed by problems of templatic morphology.

Distribution

This unit is concerned with the problems of distribution. Distributional problems arise from restrictions imposed on the co-occurrence of sounds or on the co-occurrence of sounds with structural positions. This unit focuses on the distribution of sound segments. Distributional problems concerning suprasegmental phenomena such as syllable, tone, and stress will be discussed in Units 3–5. This unit includes four chapters. Chapter 1 is concerned with vowel co-occurrence in Kikuyu, which shows that Kikuyu vowels cannot freely combine in verb roots. This chapter introduces the concept of pattern and related concepts. Chapter 2 continues the exploration of distribution by showing that consonants are subject to co-occurrence restrictions. We show that English nasal consonants exhibit co-occurrence restrictions with following consonants and with certain structural positions. Chapter 3 shifts the focus and examines a sub-type of distributional phenomena, problems referred to as complementary distribution. Complementary distribution is a type of distribution which concerns the relations of sounds. In problems of complementary distribution, not only are sounds restricted, but restrictions on some sounds are opposite to the restrictions on other sounds. By showcasing a relatively simple example of complementary distribution involving Luganda liquids, this chapter demonstrates not only how to identify such restrictions, but also how to analyze them. Chapter 4 presents a more complex type of complementary distribution involving Thai plosives. This chapter strengthens your understanding of both complementarity and contrast as concepts of sound relations. These four chapters, together with accompanying exercises, present a variety of segment-related distributional problems and introduce a range of phonological concepts, concepts such as phoneme versus allophone, underlying versus surface representation, natural classes and distinctive features.

Prosodic morphology

In this last unit, we move the investigation from phonological problems to problems of word formation, the domain of morphology. One area of morphological studies, known as prosodic morphology, is concerned with the prosodic shapes of morphological units and how they come about from the ways words themselves are formed. This unit introduces three distinct problems of prosodic morphology and shows how they are analyzed. There are three chapters in this unit. We open with Chapter 20, which introduces a problem from Arabic. In Arabic, the plural forms of nouns referred to as broken plurals surface with a consistent initial shape. In addition, part of broken plurals appears to be correlated to singular stems. This chapter evaluates two analyses of these properties of Arabic broken plurals: the template-and-circumscription account of McCarthy and Prince (1990) and the parafixation-and-transfer account of Hammond (1988). As we show, they present two drastically different views of how Arabic broken plurals are constructed. Then in Chapter 21, we analyze a common word-formation process, that is, reduplication. Reduplication refers to a word-formation process, which constructs new forms of words by copying or duplicating an existing word or part of it and then concatenating it to this word. A crosslinguistic property of reduplication is that the reduplicant, that is, the copy, tends to have a fixed phonological form similar to that of Arabic broken plurals. Using Swati reduplication as an example, this chapter presents the templatic and optimal-theoretic explanations of this invariant property of reduplication. Finally, Chapter 22 juxtaposes two problems that appear on the surface to be totally unrelated: Swati infixing reduplication versus Luganda glide epenthesis. This chapter demonstrates that these two problems – one morphological and one phonological – are driven partially by one requirement of syllables, that is, syllables must have onsets. As a matter of fact, a key claim of prosodic morphology is that word-formation processes such as Arabic broken plurals and Swati reduplication are governed at least partially by principles of phonology. One main investigative focus of prosodic morphology is to uncover the relations between phonology and morphology. These three chapters, together with the exercises that follow, are designed to familiarize you with the issues and concerns of prosodic morphology and develop your abilities to identify and analyze problems associated with Prosodic Morphology.

Introduction

In Chapter 17, we presented a metrical analysis of Pintupi, Wargamay, and Choctaw stress. The key claim of the metrical analysis is that stress reflects the rhythmic structures of natural languages. To express the rhythm, Metrical Theory (MT) proposes that syllables are organized into feet and feet are grouped into a word unit. This view of stress is derivational in nature. It claims that the underlying form is stressless and devoid of metrical structures. The surface form, with metrical structures and stress, stems from ordered rules. In this respect, the metrical view isnot unlike derivational analyses of segmental, syllable, and tonal phenomena. We introduce the Optimality Theory (OT) view of stress here. A significant departure of OT from MT is that metrical structures are not constructed by ordered rules. In OT, the input is fed into Gen (Generator), which generates candidates or outputs with a variety of metrical structures including the one with the attested metrical parse. Eval (Evaluator) evaluates these outputs against a set of ranked constraints. The attested output, OT claims, corresponds to the candidate with the best metrical parse. This view shifts the burden of analyses from deriving the attested forms to evaluating competing outputs. This chapter presents this view and compares it with the metrical analysis of stress.

This chapter has three main objectives. First, it introduces the constraints on metrical structures and shows how they account for the stress in the three languages. Even though OT differs from MT with its derivational design, many insights of MT form the basis for the metrical structure constraints. Second, we evaluate the OT analysis of stress against the metrical view and show how competing analyses can be distinguished. Lastly, this chapter develops your ability to conduct OT-based analyses of stress and to evaluate competing analyses.

Tone

This unit investigates tone, a term that refers to pitch when it is used contrastively to distinguish one word from another. In tonal languages, words can have different tones and tones can display co-occurrence restrictions. Moreover, just as morpheme concatenation can trigger segmental alternations, tones may alternate as well. This unit examines both tonal distribution and tonal alternation. It has four chapters. Chapter 13 considers tonal distribution data from Mende nouns and shows that only sixteen out of the eighty-four tone sequences are attested. This chapter presents two analyses of Mende tone distribution. One analysis views tone as a property of segments. This Segmental Theory of tone is juxtaposed with Autosegmental Theory, which sees tone as “independent” of segments. In Chapter 14, Segmental Theory and Autosegmental Theory are tested against related tonal alternation in Mende. These comparisons not only introduce these theories but also highlight their respective strengths and limitations. Chapters 15 and 16 examine tonal alternations that result from vowel contact in Yoruba. This examination reveals that out of the three tones – high, mid, and low – in Yoruba, the mid tone is most likely to lose out when they come in contact. This tone asymmetry provides the testing ground for three approaches to asymmetry: Contrastive Specification and Radical Underspecification in Chapter 15 and Optimality Theory in Chapter 16. Together, the four chapters of this unit introduce tone as a phenomenon, familiarize you with some of its key patterns and properties, illustrate how tonal patterns are identified, and demonstrate how they are handled by both pre-OT and OT approaches.

Introduction

We start this first chapter by introducing the phenomena of distribution in the sound systems of human languages. As a label, distribution refers to a wide range of phonological phenomena whose shared feature is the co-occurrence restrictions imposed on sounds. Languages may impose restrictions on the co-occurrence of vowels, consonants or vowel-plus-consonant sequences, on the co-occurrence of stress and tone, or on the placement of vowels, consonants, stress, or tone in certain positions in a syllable or word, etc. These co-occurrence restrictions on sounds result in patterns or regularities, which are “distortions” in the distribution of sounds. These “distortions” give rise to asymmetries in the distribution such that some patterns are attested or attested in abundant quantities while others are attested rarely or not at all. We introduce an example of distribution from bi-syllabic verbal roots in Kikuyu, a Bantu language spoken by roughly 6 million people in Kenya. Kikuyu bi-syllabic roots are generally of the shape CV1CV2C (C=Consonant; V=Vowel). You will see from this chapter that Kikuyu limits the co-occurrence of V1 and V2 in bi-syllabic verbal roots, such that some V1. . .V2 sequences are attested while others are not. Kikuyu verbal roots present a common example of co-occurrence restrictions imposed on vowel sequencing.

This chapter has three key objectives. First, it explains what a pattern – specifically, a sound pattern – is. To develop your understanding of patterns, we introduce three key concepts: logical possibilities, even distribution, and uneven distribution. Second, this chapter introduces distribution as a type of phonological phenomenon using the restrictions on vowel sequencing in Kikuyu verb roots as an example. Finally, this chapter guides you step by step through the process of identifying the vocalic patterns in Kikuyu verb roots. The process of determining the distributional patterns involves three steps: (a) determining the logical possibilities; (b) identifying which logical possibilities are attested and which ones are not; and (c) verifying the patterns. Using Kikuyu as an example, we explain what these steps entail and show how one might go about identifying and verifying the patterns, all in an attempt to develop your ability to identify co-occurrence restrictions in the sound systems of human languages.

Stress

This unit examines stress. Stress manifests itself in pitch, duration, and loudness. In a word, the stressed units, that is, syllables or moras, tend to have higher pitch, longer duration, and increased amplitude, though their importance as markers of stress varies from language to language. As this unit shows, stress is predictable. Languages can differ in how and where stress is assigned, but they exhibit shared characteristics. This unit, which comprises three chapters, is devoted to this phenomenon. Chapter 17 presents stress data from Pintupi, Wargamay, and Choctaw. These three languages are selected to highlight the typical patterns of stress and their crosslinguistic variations. This chapter also introduces the Metrical Theory (MT) of stress and shows how it handles stress in the three languages. Two key claims of MT are that syllables are organized into larger constituents known as feet and that feet are responsible for the rhythmic property of stress. This metrical account of stress is juxtaposed with the view of stress based on Optimality Theory (OT) in Chapter 18. We show that though OT abandons the rule approach to foot construction, it draws significant insights from MT. These two chapters present a comparison of MT and OT and highlight their strengths and drawbacks. Chapter 19 analyzes the interactions between stress and epenthesis in Yimas, whose normal stress assignment seems to be both impacted and not impacted by epenthesis. This chapter highlights the problem posed by this pattern interaction problem for Derivational Theory and some of the advantages for OT. Through these three chapters, this unit develops your understanding of the phonetic properties and phonological patterns of stress and showcases how patterns of stress are identified and analyzed.