The European empires around the end of the seventeenth century were hardly cohesive, coherent wholes. Primarily controlled by the developing European states of Portugal, Spain, France, Great Britain, Russia, and the Netherlands, they had been pieced together slowly over the course of several centuries. In describing them we need to keep in mind the difficulties faced by Europeans as they tried to expand their trade, protect their political authority, and export their religious beliefs and ways of life to other parts of the world. The interplay of European political, diplomatic, economic, and cultural concerns with similar aspects of the places being colonized made the development of Europe’s colonial empires not “smooth and linear” but, rather, temporally and spatially “lumpy,” occurring in intense bursts.

France

The French colonies in North America affected by the Great Peace of Montreal were only part of a collection of possessions that stretched from Acadia and Canada to the Caribbean and to South Asia. French colonies in Asia and the Americas developed slowly in the sixteenth and seventeenth centuries: French merchants at that time were more interested in trade on the coast of North Africa. But by the seventeenth century France did control several islands in the Indian Ocean, the Île de France and Île de Bourbon, which produced sugar for export to Europe. In 1674, François Martin, one of the chief proponents of French expansion in India, succeeded in negotiating the concession for what would become Pondichéry on the east coast of the subcontinent for the Compagnie des Indes Orientales, a monopoly company chartered by the French king. While at times taken over by the English, this town would remain a principal French outpost in South Asia. In all, however, French attempts prior to the eighteenth century to develop an empire in Asia remained weak and virtually stillborn (Map 1).

The readings selected for this course are designed to introduce phonological problems and analyses and develop your understanding of phonology. All selections start by presenting a phonological problem and show step by step how this problem is solved and analyzed. In addition, these readings serve as models for you to emulate in constructing your own linguistic analyses of the assigned problems. For these reasons, it is important that you read the assigned chapters carefully and reflect on the targeted aspects of the chapters. Under no circumstances should you rely on class instruction alone to develop your understanding of the course content. You will find that class instruction is not a substitute for trying to make sense of the course content via reading on your own. The central objective of the reading response assignment is to encourage you to read the assigned chapters and to be prepared for the discussions in class.

Write a one-page response for each assigned reading. Two or three questions are provided for each reading. Select one of the questions and write a response addressing the selected question. Organize your response in the form of a mini-essay, meaning that there should be an introduction, a body, and a conclusion. The reading response is due on the day the reading is scheduled to be discussed. Note that I am serious about the one-page limit. Responses that exceed one page will be returned with zero points. So make sure that your response appears on one page. In what follows, I provide the questions on the assigned readings.

Introduction

In Chapters 2 and 3, we introduced two key concepts regarding the distribution of sounds in a language: contrast and complementarity. Two or more sounds are contrastive if they appear in identical environments. They are complementary in their distribution if they appear in non-identical, mutually exclusive contexts. Contrast and complementarity play a key role in determining the status of sounds as either phonemes or allophones. We refer to the sounds whose distributions are contrastive as phonemes. We consider as allophones those sounds that are in complementary distribution and phonetically similar. Clearly, not all sounds are either contrastive or complementary in distribution. We demonstrated through the velar nasal ŋ in English that it appears to be partially contrastive, contrasting with other sounds in the syllable coda, but not in the onset. The exact status of ŋ as either a phoneme or an allophone of some other phoneme requires phonological analyses, a topic we will take up in Chapter 5.

This chapter reinforces your understanding of contrast and complementarity. The data come from standard Thai, the language spoken by the educated, middle-class population in the central region of Thailand, particularly in Bangkok. There are four sets of plosives in standard Thai: (a) voiced plosives [b, d]; (b) voiceless aspirated plosives [ph, th, kh]; (c) voiceless unaspirated plosives [p, t, k]; and (d) voiceless unaspirated and unreleased stops [p¬, t¬, k¬]. Some of these plosives are contrastive, while others are complementary. Part of the challenge of the Thai data is to determine which plosives are contrastive and constitute phonemes and which ones are complementary and are allophones of other phonemes. Moreover, there are potentially more than one set of phonemes from which allophones might be derived. Thus, the Thai data require considering the sources from which the allophones are derived.

This problem requires an essay response. Examine the problem carefully and pay close attention to the instructions given. This problem is related to Chapter 1. Read and review this chapter before attempting this problem. You are strongly encouraged to work with others in solving the problem. But write the response on your own.

The data we examined in Chapter 1 come exclusively from the bi-syllabic verb roots in Kikuyu. On the basis of these data, we concluded that Kikuyu imposes restrictions on the co-occurrence of vowels such that two types of mid vowels do not mix in bi-syllabic roots, with one exception. The purpose of this problem is to encourage you to compare the bi-syllabic verb roots with the data in (I) and (II) and to identify whether there is any relation between them. The data in (I) and (II) differ from bi-syllabic roots in that they consist of a mono-syllabic verb root – CVC or VC – followed by an extension suffix -VC. So while V1 belongs to the root, V2 is part of the suffix. In Kikuyu, the extension suffix -ek/-εk expresses susceptibility or potentiality. For instance, the stem haat-a means ‘sweep.’ The stem with -ek/-εk, haat-ek-a, means ‘sweepable’ or ‘be swept.’ The suffix -o r/- ɔ r marks reversiveness. Its meaning is similar to that of the English prefix un- in un-tie or un-do.

Introduction

We start this unit by investigating a different type of phonological phenomenon called stress. Stress is different from emphasis or emphatic stress. Although both can have phonetic properties such as higher pitch, longer duration, and/or increased loudness, stress and emphasis perform different linguistic functions. When a form or a part of it is emphasized, the intention of a speaker is to draw attention to this form and/or contrast it with other entities. Stress is used not for attention-drawing or contrastive purposes. It is used to demarcate, among other things, the borders of morphological and phonological units such as word boundaries. Unlike emphasis, which shifts depending what is emphasized, stress is fixed and exhibits predictable patterns. Moreover, stress is similar to segments in that both are an integral part of the phonetic makeup of a word, unlike emphasis which is superimposed. This chapter develops your understanding of the patterns of stress. We showcase stress from three languages: Pintupi, Wargamay, and Choctow. These languages highlight three types of stress patterns. We demonstrate that stress is the manifestation of rhythm in natural languages. Linguists express this rhythm by hypothesizing a level of phonological structure known as foot. Like syllables, feet are phonological constituents; both gather strings of sounds into larger units. But as units, feet are larger and can include more than one syllable. This chapter highlights the crucial role feet play in predicting stress placement in natural languages. As meters are defined in feet, this theory of stress is known as Metrical Theory (henceforth, MT).

This chapter has three objectives. First, it introduces stress and highlights its characteristic patterns and crosslinguistic properties. Second, this chapter presents MT, a proposal initiated in Liberman (1975) and Liberman and Prince (1977) and many subsequent works. We focus on the version of MT in Hayes (1995). According to Hayes (1995), foot structures are restricted, with only three types: syllabic trochee, moraic trochee, and iamb. The crosslinguistic variations in stress result mainly from these foot types. Finally, this chapter expands your ability to analyze stress. This analytic ability, we believe, can be developed only by learning how stress is analyzed and analyzing it yourself.

Introduction

We continue the exploration of phonological alternation as a phenomenon in this chapter. The data we consider here come from a set of number words in Tibetan, a Sino-Tibetan language spoken in present-day Tibet. Even though Tibetan numerals, as we show here, are no different from other phonological alternation phenomena in which a single morpheme has two or more phonetic variants, Tibetan numerals are much more deceptive on the surface. At first glance, it is not obvious that Tibetan numerals present a case of phonological alternation. As a matter of fact, it is not obvious whether they represent a phonological puzzle at all. Consequently, Tibetan numerals are more interesting and challenging to analyze than the phonological alternation phenomena introduced in earlier chapters, and the pattern to Tibetan numerals is harder to unearth.

This chapter has three central objectives. First, it introduces an additional and more challenging phonological alternation problem so that you can become familiar with the range of phenomena we call phonological alternation. Second, it develops and strengthens your ability to tackle more complex phonological alternation puzzles. Third, this chapter reinforces the significance of understanding the concept of underlying representation, a level of representation that may be distinct from the surface representation of a given form. In fact, it is imperative that you start viewing every sound, every word or every sentence as having an underlying form or representation and a surface form or representation. For now, the surface form or representation of a sound, word, or sentence may be equated with what is spoken; the underlying form or representation may be equated with the base form, which may or may not be identical to the surface representation. As we show through this analysis of Tibetan numerals, unless we look at and compare these numerals in terms of these two levels of representation, it is difficult to determine what type of phenomenon they represent and what pattern they exhibit.

Introduction

This chapter continues the exploration of stress from metrical and optimal-theoretic perspectives. In Chapter 18, we identified three areas that distinguish the Metrical Theory (MT) of stress from Optimality Theory (OT). They are: MT’s derivational vs. OT’s non-derivational design, the inviolability of rules vs. the violability of constraints, and the simplicity vs. complexity of analyses. This chapter examines these differences against a pattern interaction phenomenon. In Yimas, syllables with epenthetic nuclei are both stressed and not stressed. This pattern interaction between stress and epenthesis puts both MT and OT to the test, highlights the implications of the three differences, and brings to light the advantages of OT as well as the challenge it faces.

This chapter has three objectives. First, it shows that stress is not unrelated to other phenomena. Stress impacts and is impacted by other phenomena. This chapter introduces a phenomenon of this kind and shows how stress interacts with epenthesis. Second, Yimas stress–epenthesis interactions represent an example of pattern interactions. This chapter showcases how MT and OT treat pattern interactions and develops your ability to analyze them. Lastly, this chapter strengthens your capability to develop, compare, and evaluate competing analyses.

Introduction

In this chapter, we continue to explore problems of prosodic morphology. One morphological process used widely by many languages is reduplication, a word-formation process that relies on repeating or copying a form or part of it and concatenating it to this form. Reduplication varies from language to language. But one unifying characteristic is that the copied portion of a reduplicated form tends to take on a specific shape, even though the base – the form upon which the copy is based – varies in its shape. A central challenge posed by reduplication, like that posed by Arabic broken plurals, is how to explain this invariant shape of reduplicated forms. The data we use to illustrate reduplication come from Swati verbs. Swati verb reduplication is selected for three reasons. First, it involves both base reduction and expansion. Second, Swati reduplication shows up with segments not supplied by the base. Finally, Swati reduplication treats syllables with onsets differently from those without onsets. These characteristics are seen in other languages with reduplication. For these reasons, Swati reduplication illustrates some of the key issues in analyses of reduplication. We present two analyses here. One relies on copying and template mapping, similar in key respects to the template-and-circumscription proposal in McCarthy and Prince (1990). The second proposal looks at reduplication from Optimality Theory (OT). We demonstrate how issues not raised by earlier analyses are asked and addressed in OT-based analyses of reduplication.

Introduction

Up to this chapter, we have approached the sound patterns of human languages from the perspective of rules. One shared characteristic of the rule approaches including the templatic view of syllabification and epenthesis is that they generate only the attested patterns of a language. Unattested patterns cannot be derived. This is accomplished by the rules as well as the conditions, which restrict what rules can generate. An example of a condition on rules is Coda-Filter introduced in Chapter 10, which blocks a rule from syllabifying a consonant as a coda if this consonant has an independent place of articulation. We shift the focus from this chapter and introduce a drastically different approach to the analysis of sound patterns. Unlike the earlier rule approaches, this theory does not block the unattested patterns. The attested and unattested patterns, referred to as candidates, are all submitted for evaluation. Rather than generating the attested patterns through rules, this approach relies on a set of ranked constraints to select the best or optimal candidate out of competing candidates. Under this approach, the optimal form selected by the constraints is the attested form. This approach is known as Optimality Theory (OT). We show that OT does away with rules. Instead, it elevates conditions such as Coda-Filter to a central role in the analysis.

This chapter has three objectives. First, it introduces OT, a theory of language and linguistic analysis that has become very popular since the earlier 1990s. Second, this chapter develops your ability to conduct linguistic analyses, in particular, to analyze sound patterns using OT. Lastly, by presenting an optimal-theoretic analysis of syllabification and epenthesis in Ponapean, this chapter develops your ability to analyze the same phonological phenomena from multiple perspectives and to compare and evaluate competing analyses.

Introduction

We continue to explore distribution in this chapter. As a type of phonological phenomenon, distribution displays co-occurrence restrictions. By co-occurrence restrictions, we refer to constraints imposed on the ability of two or more entities to appear together. With respect to distribution in phonology, these entities can refer to sounds (vowels and consonants), stress, tone, and structural positions such as word-initial, word-final, or syllable positions such as onset, nuclei, and coda, etc. Languages limit the ability of certain sounds to co-occur or appear in such positions as those in a word or syllable. In the units on tone and stress to be presented later, we show that stress and tone are subject to distributional restrictions as well. For instance, a high tone might not co-occur with another high tone in some languages. In others, all syllables containing a long vowel or ending in a vowel–consonant sequence must be stressed. These phenomena have in common restrictions on the co-occurrence of two or more entities, resulting in gaps in the distribution. Gaps in distributions stem from the fact that not all logical possibilities are attested.

This chapter examines a case of distribution involving English nasals. Those of you who are familiar with English believe that English has three nasal sounds. One is the bilabial m, which appears as the first and last sound in ‘mat’ or ‘damn.’ The second is the alveolar n, as in ‘nap’ or ‘tan’. The third nasal is the ng sound, which is the last sound in ‘long’ or ‘strong.’ Linguists use the symbol ŋ to represent the ng sound, a sound linguists call engma. Unlike m and n, engma is a velar sound, articulated by raising the back of the tongue toward the velar region. We see shortly that these three nasal sounds are not equal in their distribution. One of them is restricted to a specific position in a word. Moreover, all three nasals are subject to restrictions when they appear as part of a consonant cluster.

Introduction

We have introduced three cases of phonological alternation in three languages in this unit. These three cases of alternation, though varying in complexity, are limited to variations in a specific part of a morphologically complex form. Chapter 5 is concerned with affix alternation in English. The alternations involve the stems in Tibetan and Tonkawa. In this chapter, we introduce yet another case of phonological alternation. This case is somewhat different and more complex in that the components of a morphologically complex form all alternate and these alternations seem to interact. The data for this case of phonological alternation come from Yawelmani, a native American language spoken in California. The puzzling and interesting aspect of Yawelmani alternations is centered round a set of forms that appear to be exceptions to an otherwise straightforward process of vowel harmony. We show that we can come to an understanding of these exceptional forms through phonological analysis.

This chapter has four objectives. First, it continues to strengthen your ability to analyze increasingly more complex phonological alternations, phenomena that involve pattern interactions. Second, it extends your understanding of the two types of linguistic arguments – predictability and simplicity – that linguists use to evaluate competing analyses. In addition, this chapter introduces a third type of linguistic argument, naturalness. We show that all three types of linguistic arguments play a role in determining the final analysis of Yawelmani. Third, this chapter introduces some of the distinctive features linguists use to classify vowels. It reinforces your understanding of distinctive features and highlights the role they play in capturing the similarity in seemingly unrelated alternations. Finally, this chapter introduces the concept of rule ordering. We show that the different processes in Yawelmani interact in specific ways and that such pattern interactions can be formally expressed through rule ordering, that is, by arranging the rules to apply in a chronological sequence.

Introduction

We analyzed Mende tone distribution and alternation in Chapters 13 and 14. Mende contrasts two tones: H(igh) and L(ow). They exhibit co-occurrence restrictions such that only some of the logically possible tone patterns are attested in mono-morphemic nouns. Examination of poly-morphemic nouns reveals that suffixes may alternate in tone, with their tones dependent on the tone of the noun roots. We concluded that the tone distribution and the suffix tone alternation support the view that Mende tones are not always properties of the units that bear these tones on the surface. In terms of representation, this means that tones can be floating in Mende and its surface tones result from the rules that map tones to tone-bearing units (TBUs). This chapter continues the examination of tone, focusing on a series of tonal alternations that result from vowel contact in Yoruba. Yoruba contrasts three tones: H, M(id), and L. These tones appear similar. But as we demonstrate, one tone behaves very differently, resulting in tone asymmetry.

Tonal phenomena such as those in Yoruba raise the question of the proper treatment of phonological asymmetry, of which tone asymmetry is a sub-type. We consider two theories of asymmetry here: Contrastive Specification and Radical Underspecification. Contrastive Specification (henceforth, CS), which was developed in works such as Mester and Itô (1989) and Steriade (1987), among others, holds that all contrastive properties must be specified in underlying representation. In Yoruba, this means that H, M, and L are all specified. According to this view, tone asymmetry follows from phonological rules, not representations, as contrastive tones have similar representations. In contrast with CS, Radical Underspecification (RU), which was developed in Archangeli (1984, 1988), Archangeli and Pulleyblank (1989, 1994), and Pulleyblank (1986, 1988), etc., claims that not all contrastive properties are specified. When applied to Yoruba, this means that the tone that behaves differently may be underspecified or unspecified. Tone asymmetry, under this view, is the result of phonological representations. This chapter evaluates these two theories against the tonal alternations in Yoruba. We show that while some evidence is consistent with both theories, neither is sufficient to explain the full range of tone behaviors in Yoruba.

Introduction

Up to now we have focused only on problems of distribution. The shared feature of distributional problems is the co-occurrence restrictions, restrictions imposed on, say, the types of segment sequences allowed, the types of segments appearing in certain contexts such as word boundaries or syllable positions. For example, we find that English can have mp and nt sequences such as in bʌ mp ‘bump’ and te nt ‘tent’ within a morpheme, but no English morpheme allows np or mt as a sequence. In analyzing distributional problems, linguists consider the forms that may or may not be semantically related. But in analyses of alternation, linguists examine forms that are semantically related – pairs such as ‘damn’ and ‘damnation,’ ‘rewrite’ and ‘rethink,’ or ‘undoable’ and ‘unthinkable.’ These pairs are related semantically in that they share the same root damn, the same prefixes re- and un-, or the same suffix -able. Linguists compare these semantically related forms to see whether the root damn, the prefixes re- and un-, or the suffix -able have identical or different pronunciations when they appear alone or with different affixes or roots. In other words, they compare, say, damn when it appears alone and damn when it appears with the suffix -ation to see if the two occurrences of damn are identical or different in their phonetic forms. This chapter introduces a relatively simple case of phonological alternation from English. The central goal of this chapter is to develop your understanding of alternation as a type of phonological phenomenon and develop your ability to analyze alternation problems.

Introduction

We introduced the phenomena of distribution in Unit 1, the patterns that emerged from co-occurrence restrictions. The phenomena we analyzed involve the distribution of segments such as Kikuyu vowel co-occurrence. We examine a different type of phenomenon, the phenomenon of tone, starting from this chapter. Tone is related to pitch, a term that may be familiar to you. Pitch is correlated with the rate of vocal cord vibration in the production of voiced sounds. This rate is called the fundamental frequency or F0. The higher the F0 of a voiced sound, the higher its pitch is perceived to be. Pitch is a perceptual term, referring to the perceptual sensations in hearing voiced sounds. Tone, on the other hand, is a linguistic term, referring to pitch when it is used to signal meaning differences in languages. Just as English uses voicing to signal meaning differences such as bɪt ‘bit’ versus pɪt ‘pit,’ languages may use pitch variations to differentiate meanings. For instance, though pù ‘European’ and pû ‘cave’ may consist of identical sounds in Mende, they differ in meaning because their tones (marked by ` and ˆ) are different. When pitch variations signal contrasts in meaning, we refer to them as tones.

Just as segments may be subject to restrictions when they co-occur, tones are subject to restrictions as well, resulting in predictable tonal patterns. Tonal restrictions may result from the co-occurrence of different tones, the co-occurrence of tones with segments, or the constraints on tones imposed by phonological or morphological units. In this chapter, we introduce a case of tonal distribution involving the co-occurrence of different tones. The data come from Mende, a Mande language spoken in Sierra Leone. The co-occurrence of tones is highly restricted in Mende nouns, with only a small number of attested tone patterns. To gain an understanding of the Mende tonal restrictions, we conduct two analyses. In one analysis, we treat tone as a property of segments that bear the tones. In a second analysis, tone is analyzed as a property of morphemes. The goal of this presentation is to develop your understanding of the view that tone, while realized on segments on the surface, may behave as if it were independent of them. This view has led to the development of the theory known as Autosegmental Phonology.

Introduction

In Unit 2, we introduced the phenomena of alternation, which are illustrated mainly by alternations that involve segmental changes. Recall that alternations are the patterns that result from morpheme concatenation such as affixation, compounding, or reduplication. We demonstrated that the phonetic form of a morpheme, whether it is a root, stem, or an affix, can change or alternate, depending on the morpheme(s) it comes in contact with. Such changes generate the alternations where a single morpheme has two or more distinct phonetic forms or pronunciations. We continue the study of tone in this chapter, showing that tones of a morpheme can alternate as a result of morphological processes as well. Just like segmental alternations, tonal changes can manifest in the root, stem, or affix. Moreover, just as segmental alternations may be related to the distributional restrictions on segments, tonal alternation and distribution may be caused by identical constraints. For this reason, we continue to focus on the data from Mende to see whether there is any relation between its tonal distribution and alternation.

The tonal alternation data come from poly-morphemic nouns in Mende. They exhibit a number of tonal changes affecting the roots and the suffixes. We show that some of these changes can be traced back to the distributional restrictions on tone discussed in Chapter 13. The alternation data shed light on the debate regarding the segmental versus autosegmental view of tone. We demonstrate that neither view at the extreme is correct. Though some relations between tones and TBUs (tone-bearing units) are unpredictable and consistent with the segmental view, there exist predictable relations between tones and TBUs. Consequently, tones may be unlinked or floating in underlying representation and the mapping of tones to TBUs can be accomplished by rules, as advocated by the autosegmental analysis.