English in New York City - An Empirical Case Study

Table of Contents

Acknowledgements

1 Introduction

2 New York City, NY
2.1 The Big Apple – An Ethnical Melting Pot
2.2 A homogenous speech community?

3 The Northern Cities Chain Shift
3.1 Understanding Language Change
3.1.1 The Study of Linguistic Change in Progress
3.1.2 Sociolinguistic Variables
3.1.2.1 Social Status
3.1.2.2 Gender
3.1.2.3 Age
3.1.2.4 Ethnicity
3.1.3 Chain Shifts
3.2 The Northern Cities Shift
3.3 Origin and Occurrence at Present
3.3.1 The Origin and Historical Spread of the NCS
3.3.2 Present Geographical Spread
3.3.3 General Sociolinguistic Aspects of the NCS
3.4 Previous Research
3.4.1 The Quantitative Study of Sound Change in Progress and other work by Labov
3.4.2 Other Studies
3.5 New York City and the NCS

4 Research Design and Techniques
4.1 Interview Methodology
4.2 Data Analysis with Praat
4.3 Organization and Visualization with Microsoft® Office Excel
4.4 Vowel Charts

5. Speaker Analysis
5.1 Annette, 87f
5.2 Merryl, 61f
5.3 Tim, 48m
5.4 Philipp, 48m
5.5 Jake, 24m
5.6 Patrick, 18m
5.7 Samantha, 17f
5.8 Allie, 17f

6 Vowel Analysis
6.1 The TRAP-Vowel
6.1.1 The TRAP-Vowel Preceding a Non-Nasal
6.1.2 The TRAP-Vowel Preceding a Nasal
6.2. The LOT-Vowel
6.3 The CLOTH-Vowel
6.4 The KIT-Vowel
6.5 The DRESS-Vowel
6.6 The STRUT-Vowel
6.7 Vowel Systems

7 Conclusion

8 Bibliography

9 Appendices

Appendix A

Appendix B: F1/F2 Collection

Appendix C: Distribution of Each Vowel’s Phonetic Environment

Appendix D: Diphthongization of /æ/+N

Acknowledgements

First of all, I would like to thank the eight informants whose speech samples have been the basis for the entire thesis and all other participants who patiently and benevolently took part in the interviews and have not been selected for further analysis. Furthermore, thank you, Basia, for having me as your guest in New York City for such a long time and introducing me to so many interesting people.

I would not have been able to create this project without the encouragement and support of my supervisor Prof. Beat Glauser, who has not only suggested the topic to me, but whose love for the study of language and ideas concerning research kindled my interest and influenced my view on linguistics already years ago. I am grateful for his guidance during the development of the present thesis and the meticulous proof-reading he and his Hiwis did in order to constantly improve my work.

Furthermore, I would like to thank my parents, who have always believed in me and supported me in every step, from the idea to stay abroad and conduct my interviews in New York City throughout the entire making to the last page of my thesis with unstinting encouragement.

Finally, I am grateful to Alexander W. for his irreplaceable help with all the technical difficulties that arose during this project.

1 Introduction

“To the delight of linguists and to the chagrin of almost everyone else, language is forever changing” (Gordon 2001: 2). Generally, change may occur in all parts of language, e.g.in its phonology, morphology, syntax or semantics. Traditionally, however, the linguistic focus has been placed on sound change whose progress can be observed by studying language in its social context and the realization by its natives (cf. Labob 1971a).

While staying in New York City for a long period of time earlier this year, my original topic of interest for a thesis was special grammatical features or words typical of New York City English (NYCE). However, my supervisor, Prof. Glauser, informed me that empirical research, in order to obtain useful, representative data, would take too much time with one of the chosen topics. Furthermore, he introduced me to the Northern Cities Chain Shift, i.e. to sound change, which is much easier to analyze.

The Northern Cities Chain Shift (NCS) strongly influences the order of six short vowels in the vowel system of General American (GA), which have not been changed much by other important chain shifts, e.g. the Great Vowel Shift (GVS), in the historical development of English. The NCS has been in progress for several centuries, but still has not been completed. In addition, although the vowel changes triggered by the NCS are severe, only a few linguists have conducted empirical research on that topic. The present thesis will contribute its share to the empirical data collected with respect to the NCS, by analyzing speech samples taken of eight different speakers during a stay in New York City, N.Y. with respect to the occurrence of the NCS in the dialect of New York City.

Since the term dialect, according to Francis (1958: 43), denominates “the variety of language spoken by the members of a single homogeneous speech community”, the fact thatNew York City is such a “highly complicated speech area” (Bronstein 1962: 12) brings up the question of how homogenous it is as a speech community and whether one can speak of a New York dialect at all. Hence, chapter two of the present thesis deals with the special role of New York City English as an American accent and describes why it can be treated as belonging to a distinct speech community.

Before analyzing to what extent NYCE already contains features of the NCS, chapter three of the present thesis gives reasons for linguistic change and chain shifts. Moreover, it provides a definition of NCS and describes its origins, as far as they are known, their geographic occurrence and general facts about social factors. This is followed by a short listing of previous research by linguists with empirical studies. A chapter on lacunae in research was neglected, since the speech of New York City has been analyzed thoroughly by several researchers, e.g. Labov (1966) or Hubbell (1950). Its peculiar vowel system will be presented in chapter 3, as well.

Furthermore, since Weinrich, Labov, and Herzog (1968) already argued that a great deal of progress in the understanding of phonological change can come from investigation of the minute progress of individual sound changes, a small empirical case study was conducted with eight native speakers of New York City English. Chapter four describes the interview methodology as well as the data collection and its subsequent arrangement.

The next two chapters will attempt to find an “order in the apparent chaos of language variation” (Gordon 2001: 3). Therefore, chapter five provides the analyses of each individual speaker and chapter six describes the characteristics of the individual vowels possibly affected by the NCS. These chapters will try to provide answers to the questions how far the NCS is developed in the speech of New York City inhabitants and how it combines with the local accent.It is important to note that “sound change theoretically can proceed from any location within the social spectrum” (Labov 1980: 253) and that the analyses are strictly from a viewpoint of theoretical phonology and thus avoid any discussion on the social nature and status of this dialect or vowel shift.

Finally, a conclusion will be drawn of whether and how the NCS is present in New York City English, how it combines with specific pronunciation patterns found in the speech of New York City and what advantages and difficulties occur when studying language change in progress.

2 New York City, NY

2.1 The Big Apple –AnEthnical Melting Pot

“New York has always been, and continues to be, a city mostly populated by newly arrived immigrants and migrants, eager to benefit from the greater economic incentives of a huge metropolis” (Garcia 1997: 18).

Shortly after Peter Minuit had bought Manhattan Island for the Dutch West India Companyin 1626 from the Indians in a barter deal with knives, beads and trinkets worth 60 guilders, or about $24 (cf. Federal Writer’s Project 1938: 39), the island’s early economic activity as a trading portattracted settlers of numerous nationalities (cf. Berrol 1997: 7). By 1664, when New Amsterdam, as the Dutch had called it, became an English colony, it was already inhabited by the Dutch, by Jews, Walloons, French, English, Portuguese, Swedish, Finnish, Germans, English, Irish, Scots and Negroes from Brazil (cf. Federal Writer’s Project 1938: 88).By that time, “eighteen different languages had been identified in the island of Manhattan” (Garcia 1997: 20). Flourishing trade and the later founding of the New York stock exchange in 1792 rapidly increased the city’s value and made it America’s most important financial and insurance center (cf. Godfrey 1995: 441). It has not lost its economic importance ever since, but has become the largest city in the United States(cf. Encyclopædia Britannica 2009) and today “considers itself [...] the money center of the world” (Strom&Mollenkopf 2006: 140)^[1]. Thisprosperityhasalways been a reason for massive immigration. New settlers were driven by the hope for a better, wealthier life in the New World, which they mostly entered through New York Harbor, certainly one of the most important gateways for immigration to the United States of all times^[2].

Historically, the city at the mouth of the Hudson River and the shore of the Atlantic Ocean has always been “center stage for European immigrant groups’ struggle for material success” (Keogan 2002: 227).As a consequence, New York City was continuously transformed and renewed by the influx of a constantly changing population (cf. Lankevich 1998: preface). It has long been ethnically diverse “in terms of national origin” (Keogan 2002: 227) and is still considered an ever-bubbling ‘melting pot’ of ethnicities (cf. Becker&Coggshall 2009:751).

Located in southeastern New York State, in the north east of the U.S., the Big Apple, as New York City is frequently called, today is home to over 8 million people. Currently, 36.7% of them are foreign-born (cf. Online Sources US Census 2008^[3] ), the numbers of foreign-bornsranging from 18% to 40% over the last decade^[4]. The population is only 44.6% white^[5] (cf. US Census 2008), who may be born in the U.S. mainly, but whose ancestors, considering the figures of foreign-born people over the last centuries, are also likely to have had a nationality other than American. The non-white population of New York City holds a considerable number of ethnic minorities, such as Hispanic or Latino (27.5%), Black or African American (25.1%), Asian (11.8%), a considerable percentage summarized as ‘others’ (16.9%), and inhabitants identifying themselves as two or more races (2.1%). These figures stand in direct contrast to the U.S.A. as a whole, which counts 74.3% whites and only 12.6% foreign-borns (cf. US Census 2008).

However, the term ‘melting pot’ must not convey the impression that ethnic distinctions and identities are being erased or assimilated to a general American one. It rather refers to the fact that many different nationalities and ethnic groups live together, but in a certain conscious distinction in the New York City area of 309 square miles (cf. Encyclopædia Britannica 2009).

The relatively strong and constant flow of first generation European immigrants to the New York region has facilitated the reproduction of an inclusive immigrant identity, as other immigrant groups – mainly from Latin America/Caribbean and Asia – have been gradually added to the foreign born mix over the past few decades. These demographic conditions have helped to perpetuate a dominant “immigrant as us” identity in the New York City area. (Keogan 2002: 228)

Ethnic communities are proud of their origins; they are strongly represented and act self-confidently in the Big Apple. They concentrate in cultural and native language areas of their own, which can be found throughout the city^[6] and spread over all five boroughs, i.e. Manhattan, Brooklyn, the Bronx, Queens and Staten Island^[7] (cf. Bronstein 1962: 22). Thus, people of a specific nationality or ethnic group are able to live in a culturally and linguistically homogenous community. As a consequence, in New York City 47.8%, i.e. nearly half of the population older than 5 years speaks a language other than English (LOTE) at home (cf. US Census 2008). Spanish, for instance, is spoken by almost a quarter of the city’s population (24.3%), the largest groups of speakers having arrived from Puerto Rico, the Dominican Republic, Mexico, Ecuador, Colombia, and Cuba^[8] (cf. Becker&Coggshall: 752).

Businesses, organizations and state departments recognize and promote^[9] New York’s world-famous culturally vibrant spirit and its linguistic diversity because they know they can sell more “in the language of the ‘heart’” (Garcia 1997: 42). The common wisdomstill holds that New York City is full of ‘ethnies’, is incredibly diverse and deserves the title of being one of the most multilingual cities in the world: “Where more than in New York can one truly say, ‘we have met the ethnics [sic] and they are us’?” (Fishman 1997: 342)

2.2 A homogenous speech community?

The sociological evidence given above shows that New York City’s population is one of the most complex ones in the U.S. (cf. Bronstein 1962: 17). Its extreme diversity of nationalities and ethnicities makes it a unique area not only for ethnic investigations, but also for dialect research (cf. Becker&Coggshall 2009: 752). However, when it comes to dialectology, it is important to bear in mind that the city is not only inhabited by different ethnicities with LOTEs, but also that the remaining 52.2% who, according to the U.S. Census, speak English as a mother tongue are not limited to General American, but include both African American varieties and varieties of Caribbean English (cf. Garcia 1997: 7). Hence, the question of the existence of a specific New York City dialect arises; it “has been studied quite extensively by linguists” (Gordon 2007: 67). The first to report on New York City English (NYCE) was Babbitt in 1896, then there was Thomas (1942), followed by a very thorough analysis by Hubbell (1950) who investigated the relation between speech differences and varying degrees of cultivation and social status, and later Labov (1966), whose Stratification of English in New York City and later works are the most detailed and representative ones when it comes to the English spoken in the Big Apple.

All these main works mentioned above demonstrate by many kinds of evidence that New York is “not a collection of speakers living side by side, borrowing occasionally from each other’s dialects” (Labov 1966: 6), but must be considered one single speech community.

Already Babbitt (1896) stated that, despite its diverse origins, the complex population of New York City is an entity connected by “economic, cultural and social ties” (Bronstein 1962: 17) when it comes to the population below the distinct upper classes, and must be treated as such. In addition, all the authors named above argue for the separateness of New York City speech because of its clearly distinctive linguistic features, which were characterized as remarkably uniform by subjective evaluations of native New Yorkers (cf. Labov 1966: 6). Babbitt (1896) was the first to point out the distinct New York variety of English pronunciation found in the lower classes, which do not travel and are not influenced by outside patterns. Later works about special vocabulary or so-called New York Cityisms are not lacking (e.g. Allen 1993, Gold 1981). This existing “stability of norms helps to define a speech community” (Labov 1972: 117) and even fits McDavid’s definition of a dialect as a “variety of language […] set off (more or less sharply) from other varieties by (more or less clear) features of pronunciation, grammar, or vocabulary“(1958: 480). Hence, by definition New York City is a clearly separate dialect area.

Despite the argument brought up by some linguists that “in a city with a population with over 8 million people it is clearly a fiction to talk of a New York accent” (Gordon 2007: 69), New Yorkers do not speak Standard American English as their daily vernacular, nor have they ever done so (cf. Garcia 1997: ix). Not only do they have a dialect by definition, as shown above, but surprisingly, are considered one of the most distinctive and recognizable accents in the United States (cf. Gordon 2007: 69), which was already recognized by Babbitt in his early report (1896).

The English of New York City holds a special place in American dialects and the public consciousness, which has long been debated (cf. McDavid 1958: 160). Although it is spoken in only a small area of the U.S.^[10], “Americans in general are more aware of this accent than of any other local accent except possibly American southern accents” (Wells 1982b: 501f) and can clearly classify it as New York City English (regardless of stereotypes and prejudice). In interviews, New Yorkers have reported to have been “spotted instantly, innumerable times” (Labov 1966: 327) by conversational partners from other American regions as speakers coming from the Big Apple. However, at the same time it is apparently not distinctive enough to be recognized by the British as in any way different from other American accents (cf. Wells 1982b: 502). This is supported by the fact that it clearly belongs to the English spoken in the northern cities of the United States since it is affected by the Northern Cities Chain Shift (NCS).

Another remarkable aspect of New York City English is that despite a noticeable “linguistic insecurity of its speakers” (Becker&Coggshall 2009: 752), it has influence on New Yorkers’ ethnic identities, which is confirmed by the fact that “many show adoption of some NYCE features” (Becker&Coggshall 2009: 761). Contrariwise, the speech of the considerable number of foreign-bornspopulation evidently has “no effect on the dialect spoken by the natives of New York, except to enlarge the vocabulary of the latter with a certain number of expressions native to many languages” (Bronstein 1962: 23f).

Although ethnically as well as linguistically diverse and highly complex, New York City has been shown to be a single speech community and can be treated as a dialect area when it comes to linguistic research, as it was done, for instance, by Labov (1966: 19). Thus, an empirical case study of the NCS is possible, efficient and promising, if the speakers analyzed are selected carefully according to criteria, to bedescribed in chapter 4. However, “readers are reminded of the earlier caveat about the diversity of accents in a city of this size” (Gordon 2008: 69).

3 The Northern Cities Chain Shift

3.1 Understanding Language Change

3.1.1 The Study of Linguistic Change in Progress

It was long felt to be an empirical impossibility to study the mechanisms of sound change in progress since it was held to proceed “by imperceptible degrees” (Gordon 2001: 3) and, thus, it was treated as a historical phenomenon and only examined in retrospect. Only with the seminal paper by Weinreich, Labov and Herzog (1968) and the later work and pioneering methodologies of Labov and his colleagues did the interest in the analysis of sound change in progress flourish. From that moment on, “finding order in the apparent chaos of language variation has been a guiding principle in sociolinguistics” (Gordon 2001: 3), at least since Labov’s empirical case studies on Martha’s Vineyard (1963) and in New York City (1966).

Studies on sound change make use of empirical data to establish a status quo of the rearrangement of language and then look for factors which have an impact on this alteration in order to find reasons for the change in progress. Concerning the latter, Bloomfield points out:

No permanent factor [...] can account for specific changes which occur at one time and place and not at another [...]. Although many sound changes shorten linguistic forms, simplify the phonetic system, or in some other way lessen the labor of utterance, yet no student has succeeded in establishing a correlation between sound change and any antecedent phenomenon: the causes of sound change are unknown. (1933:386)

Despite this warning, “linguists have continued to put forward simplistic theories that would attempt to explain sound change by a single formal principle, such as the simplification of rules, the maximization of transparency and so on” (Labov 1980: 252).

This statement primarily holds for major linguistic theories on sound change, of which three main approaches shall briefly be presented to give an overview, i.e. the Neogrammarian, the Structuralist and the Generativist approach.

The Neogrammarians around Paul, Brugmann and Osthoff were looking for mechanical and physiological motivations in sound change, such as ease of articulation, since the Neogrammarian approach indicates that there must be a rule for every irregularity in language and nothing happens coincidentally (cf. McMahon 1994: 21). Hence, they were looking for reasons and rules of sound change.

The Structuralists, whose models were established in the early 20th century and whose most famous member is the Swiss linguist Ferdinand de Saussure, claim that every modification is part of the superordinated language system and must have a function in it (McMahon 1994: 24). Nevertheless they were aware of the influence of external factors on the direction of linguistic changes, as it was indicated by Martinet’s theory on the influence of other languages mentioned below (cf. McMahon 1994: 32). Hence, they are not looking for rules like the Neogrammarians, but rather for purposes and functions of sound changes. The last important linguistic group are the Generativists, who came to prominence from the late 1950s with Noam Chomsky as the leading figure. They represent the theory that reasons for sound change lie in the grammar of a particular language. Hence, what changes over time are not the sounds themselves, but rather the grammar by which they are altered (cf. McMahon 1994: 32). The generative belief is that all change has the purpose of simplification, which is achieved by the development of simpler and more economical grammars (cf. McMahon 1994: 36).

However, sound change cannot be ascribed to only one single reason; It is rather influenced by a number of factors whose joint impact causes language to alter.Martinet stated that changes originate from the developments of society, changes in settlement or customs. They may be influenced by the contact with other languages or dialects which have impact on them. Gaps in a language system may even remain empty for prolonged periods of time until they are filled by an external factor such as la nguage contact. This might result in new variation in the language in question. Furthermore, language or, in this case, sound changes cannot evolve in one person, but in the collective, i.e. a speech community (Martinet 1955: 229f). Following this theory, Labov adds that change originates from irregular linguistic fluctuation in the speech of a social subgroup that, for one reason or another, unconsciously sees a weakening of their separate identity within the community (cf. 1972: 178ff). Additionally, he points out that language change can be analyzed and described in its past or present state, but is not predictable (cf. 2001: 227).

Labov attempted to resolve the question of the reasons for sound change by taking samples of changes in progress and assessing which model characterized those best (cf. 1981: 269). However, “his data did not fall unambiguously into one category or the other” (McMahon 1994: 57). Since the causes of language change include a multiplicity of factors (cf. Aitchison 1981: 197), when it comes to the study and analysis of linguistic changes in progress, certain research strategies are required in order to cover as many sources and causes of change as possible.

First, Fasold points out, the structure of the language in question has to be acquired before it can be analyzed (cf. 1990: 227). According to Labov, three issues need to be solved, i.e. the transition problem, the embedding problem and the evaluation problem (cf. 1972: 161f). The transition problem refers to the necessity to trace the development of a particular linguistic phenomenon back. One has to “find the route by which one stage of a linguistic change has evolved from an earlier stage” (Labov 1991: 161). Second, the embedding problem requires the analysis of sociolinguistic variables. One has to find social and linguistic circumstances and speakers’ behavior “in which the linguistic change is carried” (Labov 1991: 162), since it cannot be understood without the social life of a speech community. Third, “the evaluation problem is to find the subjective (or latent) correlates of the objective (or manifest) changes which have been observed” (Labov 1991: 162), i.e. comparing each informant’s general attitudes with their linguistic behavior and measuring their subjective, unconscious reactions “to values of the linguistic variable itself” (Labov 1991: 162).

“It is essential to realize that language is both a social and a mental phenomenon in which sociolinguistic factors are likely to be inextricably entwined. ‘Nothing is simple’ might be a useful motto for historical linguists to hang in their studies” (Aitchison 1981: 197).

3.1.2 Sociolinguistic Variables

“One cannot understand the development of language change apart from the social life of the community in which it occurs. Or, to put it another way, social pressures are continually operating upon language [...] as an immanent social force” (Labov 1972: 3). Change can be triggered by internal linguistic causes such as the structure of the linguistic system (cf. Trudgill 1986: 182) and social factors, such as prestige, fashion, attitude, foreign influence and social need (cf. Trudgill 1986: 182; Aitchison 1981: 197). These factors are covered by Labov’s term sociolinguistic variable (cf. 1972: 7ff.). It means that alternative utterances can refer to the same topic or item, but at the same time give evidence of a social factor and, hence, have social significance (cf. Fasold 1990: 233f). When studying language change in a speech community, salient sociolinguistic variables which serve as a focus on the given community need to be selected (cf. Labov 1972: 7f).

First, a sociolinguistic variable is an item that is frequent and is used so often in the spontaneous, unmonitored and undirected speech of a community that its behavior and usage can easily be traced, captured and understood in unstructured conversations and recordings. Second, it should be integrated into the structure of the language and the common speech of a community as much as possible. Third, it should show a high stratification, i.e. the phenomenon in question should have an asymmetric distribution among different social subgroups which may differ in sociolinguistic criteria such as age, gender, ethnicity or social class (cf. Labov 1972: 8).

However, survey categories like the ones mentioned above are not categories people live by, but rather categories that group and value the things which provide social orientation and influence human beings subconsciously. Nevertheless, they have certain reality in people’s everyday lives, existing as cultural terminology and discourses that people stick and orient to when using language (cf. Eckert 1996: 67). Each individual has the opportunity to decide which categories and discourses are important for his or her life. Any factor “cannot take effect unless the language is ‘ready’ for a particular change. Factors influencing change simply make use of inherent tendencies which reside in the physical and mental make-up of human beings. Causality needs therefore to be explored on a number of different levels” (Aitchison 1981: 197).

3.1.2.1 Social Status

Language change is strongly connected with different social classes in the sense that it does not take place simultaneously at all social levels (cf. Trudgill 1986: 188). “Instead, it originates in a particular social group and then spreads from that point to other strata, just as regional dialect change typically starts in a focal area and spreads outward from that point” (Trudgill 1986: 188).

The feature of social class is a way of subdividing the members of a society on a continuum scale by means of factors such as their power, education, career, profession or housing. However, ascribing an individual to a particular social stratum is difficult and only possible approximately, since the unique features of each social class are not easy to distinguish. It is the members of a specific community that are best able to identify and distinguish the social groups and their indicators of the system they are living in (cf. Wolfram&Schilling-Estes 1998: 152f).

Language change often has a social stratification, i.e. a certain speech variable is frequently used by, for instance, “the highest-status class, least frequently by the lowest-status class, and at intermediate frequency by the classes in between” (Fasold 1990: 224). Explanations why a specific group of speakers adopts a certain feature can be found by considering why a social group does or does not adopt that feature (cf. Gordon 2001: 6).

This brings up the question of which classes are most likely to start language change and which ones follow their lead for which reasons. The common belief assumes that change originates in the uppermost classes whose speech is seen as prestigious or dominant and is then adopted by other social strata (cf. Trudgill 1998: 1986). However, more than the upper classes, the second highest or middle-status groups tend to lead in language change, since they “have the strongest loyalty to their local communities, while the lowest social groups have neither strong affinity to their local community nor broader community allegiance” (Trudgill 1986: 188). The latter fact involves the theory of Milroy and Milroy, i.e. the concept of social networks, which was established in their famous Belfast studies (Milroy&Milroy 1978). Milroy and Milroy stated that the density of ties between its members, i.e. its cohesion and integration affects speakers’ receptiveness in the sense that networks which show a high density take over new linguistic features much faster (cf. Milroy 2002: 551).

The “role of social class is strongly connected with the level of conscious awareness” (Trudgill 1986: 188), i.e. whether a certain language variable is a change from below, i.e. below the level of conscious awareness, or a change from above, i.e. at the level of conscious awareness. In general, changes from below mostly take place in lower social classes, while the upper classes are more likely to be active in changes from above (cf. Trudgill 1986: 188), since the latter often turn out to be “a movement away from socially stigmatized features or toward external prestige forms” (Trudgill 1986: 188). Socially prestigious variants share a positive association with high-status groups, serve as a linguistic marker^[11] of status and are, therefore, desirable for speakers from lower social strata, whereas socially stigmatized variants carry negative connotations and are associated with lower-status groups. Prestigious variants are relatively rare in phonology, but one example is the use of an unflapped t in words like better or latter. In contrast, stigmatized forms are more frequent and include grammatical features like multiple negation (e.g. I didn’t do nothing) or phonological features like the use of – in ’ for – ing (e.g. doin ’, swimmin ’, workin ’). Note that prestigious and stigmatized forms do not exist in complementary distribution, i.e. if a variant is not stigmatized, it does not mean that it is already prestigious and vice versa (cf. Trudgill 1986: 182f).

“The connection between linguistic form and social meaning is tenuous and frequently subject to reinterpretation” (Gordon 2001: 6), as, for instance, in the case of the changing prestige of /r/ in New York City (cf. Labov 1966: 40ff). Generally speaking, sound changes usually start out simply as social indicators and as changes from below. As they proceed, new pronunciations may become prestigious or stigmatized, social markers or stereotypes and even rise to the level of speakers’ conscious awareness (cf. Trudgill 1986: 187).Moreover, the general tendency for linguistic changes is that their origin lies in the upper working class or lower middle class, while both the very lowest and the highest social groups lag behind (cf. Labov 1980: 254). However, theoretically “sound change could proceed from any location within the social spectrum” (Labov 1980: 253).

3.1.2.2 Gender

Gender differences play an important role in language use and language change.Since both genders perceive, rate and treat prestigious variants differently, their speech behavior reflects these attitudes. Labov (1990) has stated some basic principles concerning the relation between language and gender:“Principle 1: For stable sociolinguistic variables, men use a higher frequency of nonstandard forms than women” (Labov 1990: 210). Thus, women use forms reflecting a higher prestige associated with them. This is also expressed by principle 1a: “In change from above, women favor the incoming prestige form more than men (Labov 1990: 213). The tendency of males to use more unprestigious and, hence, more stigmatized variants can be interpreted in respect to possible positive values such as virtues of masculinity and toughness that the use of non-standard speech may have for a male (cf. Trudgill 1983: 172). Thus, when explicitly asked, men often claim to make much more use of unprestigious forms than they usually do and women, in turn, say they use more of the prestigious forms than they actually do (cf. Wolfram&Fasold 1974: 93ff).

Principle 2 refers to the emergence of language change: “Principle 2: In change from below, women are most often the innovators” (Labov 1990: 215). This principle is supported by the fact that studies of linguistic change occurring across the United States indicate the common sociolinguistic tendency of females “to be in the vanguard of language change” (Gordon 2008: 84) and often being responsible “for the initial adoption of new prestige variants” (Wolfram&Fasold 1974: 94).

Principle 3 states that “in linguistic change from below, women use higher frequencies of innovative forms than men do” (Labov 2001: 292). This principle is also supported by the fact that women in the U.S. and England have been shown to “correct more than men in formal styles in response to social pressure” (Labov, Yaeger&Steiner 1972: 54). This shows that females try to express themselves as prestigiously as possible in formal speech. These aspects indicate that linguistic behavior is closely connected to social acceptance and a socially ascribed gender-role.

However, as Gordon and Heath (1998) criticize, Labov’s principles may summarize findings stated by various researchers, but they “do not have much explanatory value” (1998: 425) and even produce contradictions. While principles 1 and 1a imply or predict that women will not acquire any innovation that does not carry prestige, the changes which are initiated by women mentioned in principle 2 cannot be realized as prestigious ones, as they are changes from below. Hence, according to these principles, females may abandon a standard, prestige form in order to take over a nonprestigious one they have innovated (cf. Gordon&Heath 1998: 425).

Nevertheless, Labov’s principles point out the differing roles of males and females in language change and establish the dominant approach to sex-asymmetric linguistic behavior, i.e. the “Gender Paradox: Women conform more closely than men to sociolinguistic norms that are overtly prescribed, but conform less than men when they are not.” (2001: 292f). Since there is no obvious linguistic reason for the sex-asymmetrical gender-paradox, “scholars have tended to explain [...] [it] by making general claims about the privileged role of women in social-symbolic change” (Gordon&Heath 1998: 422).

When it comes to sound changes, Labov, Yaeger and Steiner (cf. 1972: 54) found that women are considerably more advanced, as well. The same applies for Labov’s analysis of data taken from the Atlas of North American English (ANAE cf. chapter 2), which established that women lead in the Northern Cities Shift (NCS)(cf. 2001: 279ff). This observation matches the findings already made in 1905 in Charmey, a small village in Switzerland by Louis Gauchat, who found “that for each sound change women were further advanced than men by as much as a full generation” (Labov, Yaeger&Steiner 1972: 54).

However, it is not only the progress of language change in which females tend to lead. They also seem to focus on other features of change than males do. As a specific female feature of sound change, Labov, Yaeger and Steiner, for instance, state that typically “women will lower tense vowels drastically without affecting their peripheral character, shifting F1 but not F2. The vowel systems of men do not show such sharp differences” (1972: 51). However, they do not give an explanation for this phenomenon. In contrast, Gordon and Heath (1998: 421) attempt to explain the extreme fronting and raising of /ɪ/ in the NCS by women and the changes in /ɔ/ which are rather performed by men with the fact that “females and males are attracted asymmetrically to different subjective poles of the vowel system” (1998: 421). In this case, women are attracted to the high front unrounded vowel quality of /ɪ/ while men prototypically prefer back vowels, rounded or not, i.e. /ɑ/, /ɔ/, /ʊ/ etc. (cf. Gordon&Heath 1998: 423).

It is, however, not clear whether there are differences in the female and male perception of vowel qualities, why each gender prefers other sounds, by which sounds men or women are especially attracted or why they change the sound qualities the way they do. As Labov (1990: 219) puts it: “It would be quite satisfying if we could arrive at a straightforward grouping of male- and female-dominated changes in their phonetic character. [...] Correlated with sex, there must be some causal link between the phonetic substance in question and (biological) sex or (constructed) gender.”

3.1.2.3 Age

Although sound change is generally analyzed diachronically, i.e. as a change over time, a linear view of it may help to identify and compare the relative movements of different variables. When the movements of sounds in current time are set into relation within a continuous dimension of age, their analysis can help to determine a trend or indicate the general direction in which a language change develops (cf. Labov 1994: 60).

Since an individual’s speech is formed during his or her adolescence, about 80 years of language history can be observed in a single speech community by analyzing subjects from different age groups. As Labov points out, the differences between the various age groups in the continuum do not increase steadily, “but rather appear to move by generations, for the adults at least: a younger generation 20-49 and older generation 50 and above” (Labov 1994: 60). The exact ages of the groups are less important, the focus of an empirical study should lie on the fact that the two groups examined belong to two clearly different generations in order to ensure a relevant contrast.

Considering the changes of the NCS, for instance, older speakers analyzed by Labov, Yaeger and Steiner (1972) show apparent fronting and raising of /æ/ while /ɑ/ and /ɔ/ seem to be more realized in a more conservative manner. Furthermore some middle-aged speakers with almost the same conservative positions of /ɑ/ and /ɔ/ show lowering and/or centralization of /ɪ/ and /ɛ/, while younger speakers who show shifted /ɑ/ and /ɔ/ positions, do not reveal movement in /ɪ/ or /ɛ/. These findings show that there are clear age-differences in language use, but they also bring up difficulties in verifying the chronological order of the NCS changes which is additionally impaired by limited linguistic evidence.

When studying language change with regard to age, Labov establishes four different scenarios:

[Stability], where both [the individual and the community] remain constant; agegrading, where the individual changes but the community remains constant; generational change, where the individual preserves his or her earlier pattern, but the community as a whole changes; and communal change, where individuals and the community change together (Labov 2001: 76).

Age and linguistic change cannot only be analyzed with regard to different age groups or generations, they are also connected to social class. Labov points out that an individual’s alteration of speech patterns over the years may be linked to his or her social status. While the upper-middle class, for instance, adopts one, often prestigious language style very early in life and mainly preserves it, especially when getting older, the lower-middle class, in contrast, aims for more prestigious speech forms and tends to adopt them from younger speakers from the upper-middle class (cf. 1972: 134). This linguistic insecurity of the lower-middle class may be due to the low density of their social network, which is not cohesive and integrating enough to establish and maintain a self-confident language variable of its own. Hence, its members are, as mentioned above, more prone to be affected by language change. In addition, as e.g. Eckert (1980: 190ff) points out, especially young persons have the need to belong to a social group and to position themselves within the social system. The acquisition of a specific language style may help in this attempt. Hence, the eagerness of individuals to identify with social choice may, with respect to their age, an important source of social energy in the progression of language change (cf. Eckert 1980: 208).

Age, however, is rarely the only impact influencing language change. It should be analyzed and interpreted in a more complex relation with regard to other sociolinguistic factors such as gender or social class (cf. Labov 1994: 60).

3.1.2.4 Ethnicity

Ethnicity can be a key component to the variation of the English language, since ethnic groups usually form subcultures or parallel cultures within the mainstream and may define their distinctiveness or strengthen their separate identity through the use of diverging linguistic forms (cf. Trudgill 1986: 191).

As Labov points out, especially Euro-American speakers participate in present sound changes, while African Americans, Latinos, Native Americans and Asians refrain from incorporating new sound features. Instead they have developed rules of their own, which is the case especially for African American Vernacular English (AAVE), which shows present rules which are very distinct from the current trend of sound changes observed (Labov 2001: 506f). In addition, “the role of linguistic distinctiveness varies greatly from subculture to subculture”, resulting in each subculture having a different kind of language and sound variation (cf. Trudgill 1986: 191). Nevertheless, generally speaking all ethnic subgroups try to isolate themselves linguistically from the language and linguistic trends of the leading group of Caucasians (cf. Labov 2001: 508). Hence, “[the] extent to which ethnicity correlates with linguistic diversity is a function of the distance between particular ethnic groups” (Wolfram&Fasold1974: 94). As a consequence, one might hypothesize that the smaller and the more isolated an ethnic, the more isolated and distinctive their features of speech will be (cf. Trudgill 1986: 191).

Although there is no doubt that linguistic variation and ethnicity correlate, the precise impact of ethnic group membership on the overall change of dialect or ethnolect is not easy to identify (cf. Trudgill 1986: 190), since it does not stand alone in its correlation with language change; Moreover, the sociocultural factors that define and constitute ethnicity itself are difficult to identify, as well. The simplisticassumption that the ethnicity of a subgroup changes their speech variables and the more they are isolated, the more distinctive their speech patterns are, hardly works (cf. Trudgill 1986: 190f). What is commonly identified as ethnicity ^[12] may often be difficult to distinguish or separate from other influencing social factors, “such as region, social class or language background” (Trudgill 1986: 190). The popular notion of Jewish English, for instance, has a strong regional connection to New York City English (cf. Labov 1966) while African American Vernacular English is strongly linked to social status, age, and Southern regional English and Chicano English are often linked with bilingualism (cf. Trudgill 1986: 190). Accordingly, ethnicity “interacts with a wide array of other social and sociopsychological relationships, processes, identities” (Trudgill 1986: 191).

3.1.3 Chain Shifts

In general, chain shifts are linguistic sound changes in which “the movement of one vowel triggers movement in another, which in turn may trigger others in a sort of chain reaction” (Gordon 2001: 1). Aitchison (cf. 1981: 185) compares this phenomenon to a musical-chair movement, since the changes are related to each other.

The most prominent sound change in the history of English is the Great Vowel Shift (GVS), which took place between the 15th and 17th century and in which long vowels’ realizations remarkably changed places in the oral cavity (cf. Wells 1982: 184ff; Aitchison 1981: 185). This vowel movement finally led to the pronunciation of English as it is known today.

In chain shifts, any given vowel may move back or forth, up or down owing to a number of conditioning factors (c.f. Labov, Yaeger&Steiner 1972: 265). However, chain shifts do not proceed randomly. First, long vowels are more likely to fluctuate; In addition, vowels cannot combine in any possible way when shifting around, but follow certain principles, which can be observed across different languages. These principles are independent from each other and can combine freely (cf. Labov 1994: 121). They were first identified by Henry Sweet (1888), established by Labov, Yaeger and Steiner (cf. 1972: 106, 228) and later refined by Labov (cf. 1994: 176ff.). The three basic and universal constraints on vowel chain shifts are the following:

(I) In chain shifts, tense nuclei rise along a peripheral track [in the oral cavity].

(II) In chain shifts, lax nuclei fall along a nonperipheral track [in the oral cavity].

(Labov 1994: 176)

(III) In chain shifts, tense vowels move to the front along peripheral paths, and lax vowels move to the back along nonperipheral paths. (Labov 1994: 116).

Note, however, that these principles are not uncontested and are constantly being revised or augmented as further chain shifts are analyzed.

“The biggest problem, with any chain, is finding out where it starts” (Aitchison 1981: 186). Not only is thereconsiderable controversy as to which vowel [...] started the GVS (cf. Aitchison 1981: 185), but every chain shift also brings up the question whether most of its sounds were pushed, dragged or both. The terms drag chain and push chain (chaîne de traction and chaîne de propulsion, respectively) were coined by the famous French linguist André Martinet in his book Économie des changements phonétiques in 1955.

According to Martinet (cf. 1955: 55f), a drag-chainshift occurs when the movement of a certain vowel triggers the movement of another one, in the sense that the space vacated by the first sound is filled by the second as the latter follows and moves towards the empty space. This kind of movement was defined by Labov as a chain-shifting principle:

When the phonetic space between two members of a subsystem is increased by the shifting of one member (the leaving element), the other member will shift its phonetic position to fill that space (the entering element). (1994: 184)

The opposite of a drag-chainshift is a push-chain shift, i.e. a vowel moves away from an approaching vowel and finds itself a new space, which again forces another vowel to leave its position and move on. However, “a language never allows disruptive changes to destroy the system” (Aitchison 1981: 197) it will always try to ensure a clear distinction between sounds. Therefore, vowels constantly try to keep up a safety space, putting pressure on each other to keep this distance maintained. As a result, it is often difficult to determine which of the two chain reactions has taken place or whether both have taken place at the same time and in which order.

It becomes clear that ‘disruptions’ resulting from language change set off counter-movements or, as Aitchison calls them, “therapeutic changes [which] are likely to intervene and restore the broken patterns - though in certain circumstances therapeutic changes can themselves cause further disruptions by setting off a chain of changes which may last for centuries” (1981: 197).

3.2 The Northern Cities Shift

The phenomenon designated by the term Northern Cities Chain Shift or simply Northern Cities Shift (NCS) is “one of the most widely known examples of sound change currently in progress” (Gordon 2001: 1). It is a fairly recent and very uniform addition to the speech of a large section of the northern United States (cf. Labov 2007: 55). It developed simultaneously in all areas of the Inland North and “operates with a high degree of consistency” (Labov 2007: 55).

The NCS is a vowel shift in which as many as six short vowels of the American English sound system are involved and changed. The directions in which the sounds at issue move suggest a coordinated pattern comparable to a domino effect, which links the movements of these six vowels in an overall rotation (cf. Labov 2007: 55). Hence, following the linguistic tradition, this sound change has been labeled a chain shift.

In the Northern Cities Shift all three basic chain shift principles stated by Labov are realized, resulting in a nearly clockwise rotation of 6 short vowels linked into a circuit (cf. Gordon 2001: 1), as seen in figure3.2.1.

The chronological order of the sound changes involved is still a big issue (cf. Gordon 2001: 33). Most likely, the movement was initiated with the fronting and raising of the TRAP-vowel^[13] /æ/, also known as shorta, to a mid or high position, often produced with an inglide, i.e. [ɛə] or [ɪə] (cf. Gordon 2008: 82). This is usually described as the first phase of the NCS (cf. Labov, Yaeger&Steiner 1972; Wells 1982; Gordon 2001; Dinkin 2009). The vacant space at the low front vowel position in the oral cavity, i.e. the former place of articulation of /æ/, triggered the movement of other vowels, i.e. the fronting of the LOT-vowel /ɑ/ and the lowering and/or fronting of the CLOTH-vowel /ɔ/ to something near [ɑ] (cf. Gordon 2008: 82). Note that the latter movement maintaines a clear distinction between /ɑ/ and /ɔ/ and, therefore, excludes the occurrence of a pattern which is commonly referred to as the caught - cot merger and was first described by Kurath (1939) in Eastern New England and Kurath&McDavid (1961) in Western Pennsylvania.^[14]

illustration not visible in this excerpt

Fig. 3.2.1: A view of the Northern Cities Shift.Source: Based on Labov (1994: 191; 2007: 39)

In addition, the absence of /æ/ in its usual vowel space resulted in the DRESS-vowel /ɛ/ shifting downward. In later developments, /ɛ/ also shifted back towards the STRUT-vowel /ʌ/, which in turn moved backward to the position vacated by /ɔ/, while the KIT-vowel /ɪ/ moved down and back (cf. Labov 2007: 39).The description given above as well as its depiction in figure3.2.2, are not uncontested, cf. the alternative model given by Gordon in figure 3.2.2.

Since these sound changes appear the form a drag chainshift, i.e. a shift where a vowel moves into an empty space already vacated by a neighboring vowel, “their chronology is of great theoretical interest” (Gordon 2008: 83). The order of the occurrence of each movement is debatable, but most linguists agree on the fact that the changes involving /æ/, /ɑ/ and /ɔ/ are older than the others.However, some researchers, e.g. McCarthy (2008), argue that the fronting of /ɑ/ preceded the movement of /æ/, which would make the beginning of the NCS rather a push chain. The current view of the NCS situation is that the movement of /æ/ is described as “nearing completion”, /ɑ/ and /ɔ/ are “midrange changes” and the others are “new and vigorous changes” (Labov 1994: 195). It is considered partly a push chain, and partly a drag chain (cf. Aitchison 1981: 194).

illustration not visible in this excerpt

Fig. 3.2.2: A different view of the Northern Cities Shift.Source: Based on Gordon (2008: 255)

“Changes associated with the NCS operate unconditionally in the sense that the vowels may be shifted in any phonological context” (Gordon 2008: 83). Nevertheless, phonetic environment does play a role, since some favor and disfavor the change. In the case of the NCS, the most striking fact established already by Labov, Yaeger&Steiner (1972) is the favoring influence of nasals on /æ/-raising. In addition, when /æ/ is followed by a nasal it is more peripheral than in other environments, i.e. the values of the two formants place it near the front edge of the vowel envelope of /æ/. Furthermore, one of the most disfavoring phonetic contexts is /æ/ followed by /l/ (cf. Labov, Yaeger&Steiner 1972: 60). However, this does not mean that raised forms never appear in the latter mentioned phonetic environment, but that the raising of the TRAP-vowel before /l/ is less common or less advanced.

Although chain shifts should maintain “heterogeneity in synchrony” (Gordon 2001: 4), i.e. the distinctions between the changed sounds should be preserved “so that listeners can recognize them as different” (Wolfram&Schilling-Estes 1998: 49); the new vowel realizations cannot be adopted overnight. As a result, the old and the new realizations first coexist as competing variants for a certain amount of time until one of the forms, usually the older one, is completely replaced by the new one and disappears from use (cf. Gordon 2001: 3f). Notwithstanding, the Northern Cities Chain Shift apparently preserves some distinctions by vowel movement, while endangering others at the same time. Gordon provides the example of the “the low back corner of vowel space where the alternative paths taken by ɔ and ɑ appear to lead them directly into the path of ʌ as it is backed and lowered” (2002: 255). It is not clear yet what the effect of this possible collision may be or how distinctions between neighboring vowels which overlap because of the NCS will be preserved.

Although the NCS is so widely known, our knowledge about it is based on a limited scope of studies with “rather restricted sets of data collected by only a handful of researchers (Gordon 2001: 13). In addition, many studies treat the NCS as a byproduct, as Gordon complains, and only a few linguists have conducted primary research with empirical case studies on it. As a result, there are many aspects of this sound change that are not well documented, which leads to a considerable amount of open questions and unexplained issues (cf. Gordon 2001: 13f).

When it comes to the movement of vowels, not all environments that favor or disfavor a certain change are clear. Labov, Yaeger and Steiner, for instance, found “a following velar stop to be disfavoring context for /æ/ raising in Detroit and Buffalo whereas it seemed to have the opposite effect in Chicago” (1972: 265). Furthermore, Gordon (2001) identified following /l/ as a leading promoter of /æ/ raising, which is a finding that runs counter to the effects reported by other studies of urban speakers. Such contradictory data are not uncommon, but there is enough representative evidence to decide for which assumption is correct or whether the phonetic environments influencing the shift differ from city to city. Yet the NCS has developed simultaneously and uniformly in the Inland North, a fact that excludes the last assumption. In addition to factors whose influences on the sound change are not clear, there are patterns caused by reasons still unknown. The rising of long, tense or peripheral vowels, for instance, is owing to a factor, which remains unexplained at present (Labov, Yaeger&Steiner 1972: 265).

3.3 Origin and Occurrence at Present

3.3.1 The Origin and Historical Spread of the NCS

“Fixing the origin of a sound change in time is a notoriously difficult task in historical linguistics” (Gordon 2001: 24), since texts from a specific period may offer some evidence that a variant was prevalent at a certain point in time, but they cannot tell how long a particular pronunciation had been around before it was noticed by e.g. the orthoepists (cf. Gordon 2001).

“The earliest records we have of the chain shift of /ae/, /o/ and /oh/ date from the 1960s. Yet there is reason to think that the initiating event of the NCS took place a hundred years earlier with the construction of the Eire Canal in Western New York State” (Labov 2007: 42) (for map depicting the location of the Erie Canal see fig. 3.3.1.1 below).

In the 19th century Northern America had to face severe infrastructural issues^[15], to which the Erie Canal, whose original construction began in 1817 and was completed in 1825, proved to be the key. It brought social and economic changes to the region, attracted American settlers and led to a first great westward movement, provided easier access to the resources west of the Appalachians and finally made New York City an outstanding commercial center in the U.S. (cf. Online Sources The New York State Canal Corporation 1998). Moreover, it pushed and eased the spread of the NCS, since

with the exception of Binghamton and Elmira, every major city in New York falls along the trade route established by the Erie Canal, from New York City to Albany, through Schenectady, Utica and Syracuse, to Rochester and Buffalo. Nearly 80% of upstate New York's population lives within 25 miles of the Erie Canal. (Online Sources The New York State Canal Corporation 1998)

The distribution of the northern dialect and later the NCS is closely connected to the patterns of westward migration. The expansion movement to the west was described by Kniffen and Glassie (1966) who analyzed the former settlers’ building material found in the specific regions. Their findings allow the depiction of the different settlement streams and their directions, showing that the movement in question, i.e. the New England stream, expanded from the east coast mainly along the Erie Canal (cf. fig. 3.3.1.2), which to some extend might explain the striking spread of the NCS in this region.

illustration not visible in this excerpt

Fig. 3.3.1.1: The Erie Canal, constructed 1817-1825.Source: Online Sources Labov, William 2010b. Yankee cultural imperialism and the Northern Cities Shift.Paper given at Modern Language Association Annual Meeting, Philadelphia, transparency 31

illustration not visible in this excerpt

Fig. 3.3.1.2: Westward settlement streams as shown by building material.Source: Kniffen&Glassie 1966: Fig.27

In addition, “the New England stream was a community movement on a large scale” (Online Sources Labov 2010a: 7), since mass migration of sometimes entire communities was a Puritan tradition, while the new settlers of the Midlands were rather individual families. Hence, the speech communities of the Northern Inland States were not only connected geographically by the Erie Canal, but they also shared ideologies and common values, such as religious or political orientations. As Labov, for instance, points out, the early abolition of death penalty and slavery in these states is notable (Online Sources Labov 2010a: 28ff) and may represent the existence of a common mentality.

Such a strong connection, as it was present in the northern region thanks to geographical and ideological ties between the subjects of the speech community, brings the broad spread of a certain dialect feature such as the NCS forward (cf. Labov 2010a:7f).

Finally, when migrants and new working settlers “from all over the northeast were integrated into the rapidly expanding cities of Rochester, Syracuse and Buffalo” (Labov 2007: 42), their various and complex realizations of /æ/ were tensed at a certain point in time, although none of the input dialects had had /æ/-raising before. Nevertheless, this raising stayed consistent in all speech communities in central and western New York State and all speech communities created by the westward expansion from it.

However, the most important question has not been answered yet: Why and how did the Northern Cities Shift start at all? As with all linguistic changes, no satisfactory explanation has been found, yet. Most likely, its genesis and also its further development involve several elements. Phonetic simplification is probably not the only reason that evokes and supports this sound change. Labov, for instance, points out that “any explanation of the fluctuating course of sound change must involve the continual fluctuations that take place in the structure of the society in which language is used” (1980: 252), which, in regard to New York City, will be included in chapter 3.4.

3.3.2 Present Geographical Spread

Although the speech areas now affected by the NCS were first delimited by Kurath as early as 1949, so far their geographic range has not been determined with any precision (cf. Gordon 2001: 19). The Atlas of North American English (ANAE) states that the NCS is dominant in the cities in western and central New York, as well as in northern Ohio, Michigan, northern Illinois, Indiana, eastern Wisconsin and the extreme northern areas of Pennsylvania (cf. ANAE: 191; Gordon 2001: 19). More advanced forms of the shift are heard in the cities near the Great Lakes, i.e. Rochester, Buffalo, Cleveland, Detroit, Chicago, and Milwaukee as depicted, for instance by Aitchison (1981) in fig. 3.3.2.1 below.

illustration not visible in this excerpt

Fig. 3.3.2.1: The Great Lakes region with its major cities, which show advanced forms of the NCS.

Source: Aitchison 1981: 194 fig. 13.15

Other cities affected by the NCS range from the White Mountains in Vermont westward, e.g. Syracuse, Toledo, Flint, Grand Rapids, Kalamazoo, Gary, Kenosha, and Madison (cf. ANAE Chs. 11, 14; Labov 1994: 185).

“The most remarkable fact about the NCS is its uniform distribution across the vast area surrounding the Great Lakes” (Labov 2007: 39f). In addition, the regions mentioned above share the designation Inland North depicted in the map below (cf. fig. 3.3.2.2), “but are separated by the NCS-free city of Erie Pennsylvania” (Dinkin 2009: 13).

illustration not visible in this excerpt

Fig. 3.3.2.2: Dialect regions defined by the Atlas of North American English, the Inland North being highlighted in dark blue. Source: Labov, Ash&Boberg 2006: 142 map 11.13^[16]

However, features of the NCS cannot only be traced throughout this area. Evidence from Laferriere’s Boston study (1977) already suggested that the raising of the TRAP-vowel /æ/ is spread even further to the east and research data from the ANAE “suggests that aspects of the NCS can be heard as far west as the Dakotas” (Gordon 2001: 19). Considering the same pattern, Callary (1975) observed /ae/-raising all the way to the Mississippi River. Furthermore, Boberg (2001) claims to have recognized some features of the shiftin the speech of Southwestern New England and “one of the Telsur corpus’s four speakers in Northwestern New England shows an NCS-like vowel system as well” (Dinkin 2009: 13).

3.3.3 General Sociolinguistic Aspects of the NCS

Besides the present geographical spread of the NCS, its general sociolinguistic distribution among speakers is of great interest. In a number of studies (e.g. Feagin 1986; Yaeger-Dror 1989), speakers from different social groupings, ages, and genders have been examined and compared in order to determine the sociolinguistic distribution and embedding in the social setting of the NCS (cf. Fasold 1990: 232). There are two main social characteristics concerning this vowel shift.

First, as with other linguistic studies on language change, research concerning the social stratification of the NCS revealed that women are leading in this change while men lag behind. Hence, “women display more advanced forms of the shift” (Gordon 2008: 84). Nevertheless, both genders are clearly affected by it.

Second, considering the fact that language change often occurs when a social subgroup sees the necessity to establish their identity (cf. Labov 1972: 178ff), it is very surprising that the NCS is not spoken by a small subgroup which needs to establish their position in society via linguistic means, but it is rather adopted by the broad population of the Inland North.

There isone restriction, however. The NCS is affected by race; it is a characteristic of white speech (cf. Labov 1989: 53; Gordon 2008: 84). Neither African Americans nor Latinos participate “in the phonological system of the white vernacular, and, with very few exceptions, they will show a radically different organization of the short a class” (Labov 1989: 53). However, while the NCS is clearly noticed by speakers from other areas, it is not recognized as a distinctive feature of their region by the white speech community affected by it. “The belief that their speech is ‘accentless’ remains very common among Northerners” (Gordon 2008: 84). Hence, the NCS is a change from below, i.e. “below the level of conscious awareness” (Fasold 1990: 228) of the speakers who move the change forward.

3.4 Previous Research

Although first signs of the Northern Cities Vowel Shift were already evident in the work of Thomas (1935-1937), whose studies of college students in the 1930s reported /æ/ raising and /ɛ/ centralization as characteristics of Upstate New York, Labov (1994: 178) points out that Fasold should properly be considered the discoverer of the NCS pattern. In an unpublished paper in 1969 the latter traced the changes involving the TRAP-vowel /æ/, the LOT-vowel /ɑ/ and the CLOTH-vowel /ɔ/ among 24 speakers who had been interviewed for the Detroit survey of Shuy, Wolfram and Riley (1967) and explicitly recognized “this extraordinary rotation of American English vowels” (Labov 1994: 178).

From that moment on, the NCS has generally been accepted as one of the most important chain shifts in progress. However, it has “attracted very little attention in the way of primary research” (Gordon 2001: 13), which, with regard to its phonological, sociolinguistic and geographical impact, is very surprising.

3.4.1 The Quantitative Study of Sound Change in Progress and other work by Labov

When it comes to the few major studies about the NCS, William Labov has clearly been the most active figure. His work together with Yaeger and Steiner (1972) can be seen as the primary empirical work with quantitative results and detailed phonetic analyses (including vowel charts depicting each speaker’s vowel ellipses) of the NCS changes^[17]. With the aim “to discover the general principles which constrain, govern and promote sound change by the direct observation of change in progress” (Labov, Yaeger&Steiner 1972: 1) they conducted seminal socioacoustic work between 1969 and 1972 and made a major contribution to the study of shifting. Labov, Yaeger and Steiner were the first linguists to describe the NCS in detail in Upstate New York, taking “social factors into account” (Labov, Yaeger&Steiner 1972: 3) and including the Great Lakes region and New York City, by examining vowel shifting patterns in samples of speakers. They analyzed recordings mainly taken in urban areas of the U.S. and Norwich, Great Britain. The NCS data were based on 25 informants from 3 locations, i.e. 12 from Detroit, 9 from the Buffalo area (including Rochester and Chili), and 4 from Chicago, which is criticized by Gordon (2001: 14) as non-representative, but no less important. Not only did they describe the Northern Cities Shift, but also provided evidence and descriptions for the Southern Shift occurring in American English (cf. Gordon 2001:13f). Furthermore, they named four patterns of vowel shifting observed across different languages and established several principles for vowel chain shifting which were later revisited by Labov (1991, 1994).

Sociolinguistic patterns (Labov 1991) deals with social motivations and mechanisms of sound change, names problems in the study of linguistic change and suggests several steps that a sound change in progress can be expected to go through. Special focus is put on the NCS in Labov’s Principles of Linguistic Change Vol. I (1994), which not only picks up and refines the principles of chain shifts given by Labov, Yaeger and Steiner (1972), but also discusses the difference between chain shifts and mergers and provides detailed descriptions of sound changes in progress, including the NCS with all its phonetic conditions. Much of Labov’s theoretical approach to chain shifting is presented there. The third Volume of Principles of Linguistic Change is to be released in October 2010 and will provide additional cognitive and cultural factors of sound change. It explicitly deals with the NCS and tries to find connections between the NCS and the historical and ideological background of the area it occurs in.

One of Labov’s most important studies to be referred to in the present thesis is The Social Stratification of English in New York City (1966). It was a sociolingustic research project studying the present patterns of New York City speech and its variation throughout social groups. In this work Labov puts special emphasis on the random selection of speakers and their speech, i.e. spontaneous speech is claimed to be more representative of a regional dialect while formal speech or word lists will influence the results negatively since, as he states in another work, “uniformity [of a speech community] will not appear to someone who approaches the community with a list of words in hand” (Labov 1989: 2). In his analysis of the speech communities of New York City he focuses on phonological features and social variables to establish language patterns. He also takes ethnic groups, social status, age and gender differences into account, in order to establish a New York City vowel system with its “wide variety of phonological shifts, and mergers (2006: 376)

The Social Stratification of English in New York City was the first large-scale study of its kind. It opened up a new branch of linguistic research, since language change had only been examined retrospectively as a historical phenomenon. Today this work serves as a representative source for the speech of the population in NYC. In addition, it is a basis for later research on language change in progress, for the analysis of speech communities, for methodology in empirical case studies. It may be seen as a guideline for any kind of sociolinguistic research, with its last edition even providing suggestions as to studies still to be conducted and questions still to be answered.

3.4.2 Other Studies

Research regarding the social profile of the NCS has been conducted by Penelope Eckert (1987, 1988, 1989, 1991, 2000), who did ethnographic studies and whose work is mainly based on two years of participant observation carried out in a suburban Detroit high school. Not only did she first note variants of /ʌ/ approaching /ʊ/, but she also found that some changes in the NCS contained gender-based differences. In addition, she explored variation associated with class-based distinction, i.e. the separation of two social groups named Jocks and Burnouts who were the two main rival groups of students in the school of interest.

A more recent study was done by Gordon (2001), who investigated the NCS in two rural communities in Michigan, located between Detroit and Chicago, i.e. Paw Paw and Chelsea. He interviewed 16 informants from each speech community, divided into two age groups, i.e. 16-18 and 39-51 (cf. Gordon 2001: 40) and analyzed how the NCS had affected their vowel systems. He labels the older changes of the NCS, i.e. the shifting of /æ/, /ɑ/ and /ɔ/, lower elements, while the more recent changes, i.e. the shifting of /ɛ/, /ɪ/ and /ʌ/, are termed upper elements (cf. Gordon 2001: 216).

Gordon provides detailed descriptions and various visualizations of the vowel movements, analyzes them with regard to phonological, social and geographical constraints and also takes gender- and age-differences into account, implying that the NCS is still in progress in the speech of the younger informants.

Besides Eckert and Gordon several previous studies, including Callary (1975), Laferriere (1977), Herndobler (1977), Ito&Preston (1998), Hillenbrand (2003) and McCarthy (2008) have dealt with the NCS in Chicago, as well as Detroit, Boston and other parts of Michigan. Herndobler (1977), for instance, investigated a working-class community in Chicago and surveyed the speech of 82 informants, covering a wide range of ages. Callary (1975) conducted a study of /ae/-raising in northern Illinois, discussed phonological conditioning in his work and reported an interesting geographic pattern, by correlating the population size of the community with the extent to which it was affected by the NCS, i.e. the larger a community was, the more advanced was their NCS adoption.

One of the latest works on the NCS is Dinkin’s dissertation (2009), in which he interviewed persons from 5 established dialect regions in upstate New York and compared their pronunciation, focusing on phonological change, especially the NCS and short- a and short- o systems.

3.5 New York City and the NCS

Shifts in language and their social significance may take place from generation to generation. As Labov (cf. 1966: 342-349) has shown for New York City, the social significance of postvocalic r, for instance, has shifted throughout the past 50 years in the way that younger speakers share the opinion that the presence of non-prevocalic /r/ is valued more highly than its absence (cf. Trudgill 1986: 185). Among young New Yorkers /r/-lessness is replaced by other incorrect forms like /æks/ for ask and /tu:f/ for tooth (cf. Newman 2005: 86). Other declines can be observed in the pronunciations of /t/ and /d/, <th> and <ir>. While /t/ and /d/ would be pronounced rather slightly dental than alveolar in the older New Yorker generations, they are now realized at the standard place of articulation, i.e. the alveolar ridge. In addition the stereotype of Brooklynese /toidi: toid/ for thirdy-third, i.e. the production of the dental fricative <th> as a voiceless alveolar plosive and the realization of the NURSE-vowel /ɜ/ as a diphthong, is rarely heard in the younger generations (cf. Newman 2005: 84f).

However, the accent of New York City is generally very constant and maintained conservatively by its speakers. It is “confined to the city itself and several neighboring cities in northeastern New Jersey^[18] [...] [and] has been stable through most of the twentieth century” (Labov 2007: 17). Nevertheless, its short vowels are verifiably affected by the Northern Cities Shift. This connection is of particular interest, since the local accent of New York City contains peculiarities in its vowel realizations that differ from General American (GA) and which collide somewhat with the changes of the NCS.

Most importantly, there are four basic types of short- a patterns in Eastern United States dialects (cf. Labov, Yaeger&Steiner 1972; Labov 1974; Payne 1980), as shown in figure 3.4.1. Every triangle in the figure below is an abstract depiction of the oral cavity, the high left-hand side being the high front position, indicated by the cardinal vowel /i:/, and the high right-hand side being the high back position, indicated by the cardinal vowel /u:/. Note that the shape of the oral cavity is rather trapezoidal than triangular and that in the former the position assigned to /ɑ/ should lie more in the back, i.e. on the low right-hand side of the vowel chart. However, this is not possible in a triangular depiction as it has been chosen by Arvilla Payne (1980: 175). In Payne’s vowel charts, the lax^[19] TRAP-vowel is designated as /æ/ while its tensed allophone is called /æh/.

illustration not visible in this excerpt

Fig. 3.4.1: Four basic short- a patterns. Source: Based on Payne (1980: 157 Fig. 7.3)

Figure 3.4.1A displays the first possible short-a pattern in Eastern American dialects in which all allophones of /æ/ have remained lax in all phonetic environments. Second, some dialects are nasal and, accordingly, their realization of /æ/ has a bimodal distribution (cf. fig. 3.4.1B), i.e. the vowel is raised and tensed when followed by a nasal and voiceless fricatives and stays lax elsewhere (cf. Payne 1980: 157). Furthermore, in the short- a class of the Northern City dialects (cf. fig. 3.4.1D) all its allophones are raised and tensed and /æ/, “if it exists at all in the system, is a front allophone of short o” (Payne 1980: 157). However, while the general raising and fronting, i.e. tensing, of /æ/ is relatively common in the northern area of the U.S., only a few areas, such as New York City or Philadelphia, show complex distributions of this variable (cf. figure 3.4.1C) governed by “complex phonological, grammatical, stylistic and lexical conditions” (Labov 2007: 14) which are not distinguished in GA. Basically, a set of phonetic environments condition tensing if they follow /æ/.

illustration not visible in this excerpt

Fig. 3.4.2: Codas that condition tensing of short-a in New York City. Source: Labov 2007: 15 Fig. 5.

Figure 3.4.2 depicts all consonants of English, the ones with a tensing impact being highlighted with a rectangle^[20]. These basic conditions were augmented by several specific conditions:

1. Function word constraint: Function words with simple codas (an, I can, had) are lax while corresponding content words are tense (tin can, hand, add), while can’t, with a complex coda, remains tense. Thispreserves the contrast of tense can’t vs. lax can in environments wherethe /t/ is elided or neutralized.
2. Open syllable constraint: Short-a is lax in open syllables, yieldingtense ham, plan, cash but lax hammer, planet, cashew).
3. Inflectional boundary closing: Syllables are closed by inflectional boundaries, so that tense forms include planning as well as plan,staffer as well as staff, There is considerable variation before voicedfricatives (magic, imagine, jazz).
4. Initial condition. Initial short-a with codas that normally producetensing are lax (aspirin, asterisk) except for the most common words(ask, after).
5. Abbreviations: Abbreviated personal names are often lax (Cass, Babs).
6. Lexical exceptions: There are a number of lexical exceptions: e.g.avenue is normally tense as opposed to lax average, savage, gavel,
7. Learned words: Many learned or late-learned words with short-a intense environments are lax: alas, carafe. (Labov 2007: 15).

This complex system has even spread over the borders of NYC.^[21] According to these rules, in New York City English (NYCE) there is typically a phonemic contrast between, for instance, /æ/ in dragon and in dragging, while had (function word) contrasts with bad and can ‘be able’ with can ‘container’. Hence, when one tries to determine phonetic environments which favor the NCS and in this case the raising of /æ/, it is necessary to keep in mind which phonetic environments tense a vowel in the accent in question. In addition, NYCE moves the TRAP-vowel in only some environments, while the NCS may affect different ones, initiate counter-movements or affect all variants and change them, so that these sound changes compete with each other.

Furthermore, the words marry, merry and Mary, for instance, are all commonly distinguished in either a two- or three-way contrast of /æ/ before /r/ by educated speakers, while they have become homophones for most speakers of GenAm (cf. Gordon 2008: 47). An educated New Yorker would utter marry with a low [æ], merry and Mary towards [ɛ] or even realize the latter distinctly either as [e] or something like [ɛə] (cf. Gordon 2008: 73). Just like the diphthongal realization of /ɛə/, NYCE also has /ɪə/ in its phonetic repertoire.Since in the NCS “[æ] is giving way to closer, longer, diphthongal qualities in an increasing number of trap and bath words” (Wells 1982b: 510), it may collide with the two phonemes from New York City English mentioned before (cf. Wells 1982b: 510).

As another feature, Babbitt in 1896 already considered the centralization of /ɪ/ as common for New York City English (cf. Babbitt 1896: 461). Surprisingly, one of the sound changes of the NCS also includes the centralization, or lowering and backward movement of /ɪ/. Hence, when speakers in the present thesis show a shifted vowel position of the KIT-vowel, this feature does not have to indicate affection by the NCS, but may simply belong to the accent of the Big Apple.

As in other American dialects, speakers from NYC mostly realize the LOT-vowelas [ɑ]. However, when /ɑ/ occurs before a word-final voiced stop, /dʒ/ or /m/ (as in cob, cod, cog, lodge or bomb) there is a more lengthened and diphthongized variant, i.e. [ɑə], in New York City English. The same feature variably appears before voiced fricatives (e.g. bother), the unvoiced fricative /ʃ/ and in the words on, John and doll, i.e. before another nasal and a lateral (cf. Wells 1982b: 514).

Interestingly, in the highly variable class of short o words with /g/, New Yorkers tend to have /ɑ/ in hog, frog, fog, and log, but /ɔ/ in dog (cf. Gordon 2008: 70). The latter involves a raising shift in NYCE which varies on a scale from [ɔ] to [ʊ] (Kurath&Mc David 1961 note the pronunciation of won’t with /u/ as a feature of NYC, but this form is less common today, according to Gordon 2002: 70). An inglide of the vowel often includes diphthongization to higher variants such as [oə] or [ʊə] (Gordon 2002: 71f). The latter pattern occurs quite often in preconsonantal environments, especially before intervocalic consonants in polysyllables (cf. Wells 1982b: 513). Labov suggests that this pattern is part of the New York City Shift, a chain shift that includes the raising and backing of the PALM-vowel and the CLOTH-vowel. This may evoke a fronting of /ʊ/ and /u:/ since they cannot move upwards anymore. However, the forms stay clearly distinguishable and do not produce a merger. The raising of /ɔ/ is “common in the casual style of speech of the middle class and also in the careful reading style of the lower class” (Wells 1982: 513). It is less consistent than the tensing and raising of /æ/, but still these two patterns show similarities, which suggests that they may arise for the same reasons or as a kind of parallelism.

The positions and movements of /æ/, /ɑ/ and /ɔ/ in New York City English are tied up with the NCS, i.e. the earliest sounds changes of the NCS (cf. Dinkin 2009: 14). As NYCE is also affected by the NCS, it will be interesting to see how this shift influences the specific sound patterns described above, since /æ/ already has a raised position in NYCE, but the NCS may raise it in different phonetic environments. The same applies for the fronting of /ɑ/, which already shows movement, but may be pushed further in the direction it is already taking. The most interesting process involving these three vowels is the movement of /ɔ/ which has a relatively high position in NYCE and even moves into the direction of /ʊ/ or /ə/ whereas in the NCS it is dragged downward by the movement of /ɑ/.

4 Research Design and Techniques

4.1 Interview Methodology

In the present empirical study of the Northern Cities Vowel Shift in New York City all recordings were taken in semi-anonymous in-person interviews ranging from 10-60 minutes during a 5-week-stay in New York City. Informants were either chosen randomly in publicly accessible places like parks, libraries, cafés, bars and shops or had been introduced to me as people considered suitable, helpful and interested in participating in the interview.

In order to gain knowledge about the vowel movements in the pronunciation of the subjects and the development of the NCS over at least one generation, speakers were made sure to be either younger than 25 years or older than 40. In addition, special care was taken to ensure that every speaker had been born and raised in New York City, so that their native dialect was New York City English, i.e. the subject matter of the present thesis. Since the NCS is affected by race, all informants had to be Caucasian, which also made them a more homogenous group, easier to compare. Fulfilling these two criteria, i.e. finding persons with this specific origin and ethnicity, turned out to be a challenging task, considering the figures given in chapter two and the fact that white non-foreign-born New Yorkers mostly originate from some other places in the U.S. and have moved to New York City. Owing to the fact that interviews were conducted throughout Manhattan, regardless of the neighborhoods, the speakers’ socioeconomic class was neglected and will not be taken into consideration.

When looking for potential interviewees, persons of both genders with the approximately suitable age and ethnicity were asked whether they were born and raised in New York City and, if they answered in the affirmative, whether they were willing to participate in a short interview for a research project on cultural studies about New Yorkers. If a speaker gave his or her consent and the permission to be recorded, the interview began. All subjects started with a short introduction of their demographic information, i.e. name and age^[22], place of birth and childhood, marital status, education, career and occupation. However, no personally identifying information was requested such as addresses or telephone numbers. Additionally, to guarantee the speakers’ anonymity, all names in thepresent thesis are pseudonyms with the subject’s age and gender given in brackets.

The next questions were aimed at family, i.e. where the subjects’ parents were from, whether they had siblings or children. All participants were asked where they lived at the moment, to describe their neighborhood in terms of social class and ethnicity and to state whether they had ever lived somewhere else than in New York City. This was usually followed by questions about hobbies, favorite sports, pets, passions, interests and, if male, any team they were a fan of or they supported. Furthermore, they were asked to comment on their religion and spirituality. After these rather personal topics had been covered, subjects were asked about their opinion of the most severe issues in the U.S. at the time of the interview. Some of the informants were additionally asked how they had experienced 9/11 or requested to comment on the, at that time recently carried health care bill and the government deficit resulting from the economic crisis.

Since some interviews were ordinary conversations in cafés or bars, which is why two participants might have been recorded at the same time, only necessary demographic information was requested, while the topics mentioned above were only brought up by the interviewer to assure that the conversation was continued. However, the majority of them were covered by every speaker. Subjects were free to bring up topics of their own at all times and were asked to elaborate on them whenever possible.

Although the phonetic features of the NCS, have been found “not to be substantially influenced by the speaker’s relative degree of carefulness or casualness of speech” (Dinkin 2009: 37), at the end of each interview the informants were asked to read out a short word list (cf. appendix A, fig. 4) with items containing the vowels of interest, so that what was considered as careful and correct pronunciation could be compared with the earlier parts of the interviews. After this last task, the purpose of the interview was revealed to every speaker.

Only one interview per person was conducted and besides the slight difference in subject matters, the same interview methodology was employed with all speakers, regardless of how they had been found. In all cases, in-person interviews were recorded directly in .wav (Wave, uncompressed audio) format using a laptop and the freeware No23 Recorder (cf. appendixA, fig. 5) with a lavaliere microphone clipped to the collar of each speaker’s sweater.

Afterwards, when listening to the recordings again, the eight participants considered the most comparable ones were selected. Attention was paid on a balance between male and female speakers, as well as the two different age groups. Hence, two females younger than 25, two males younger than 25, two females older than 40 and two males older than 40 were chosen for further analysis and comparison. To ensure the comparability of their pronunciations, only subjects born and raised in Manhattan were selected. One participant was raised in Brooklyn, but was nevertheless considered suitable, since he had attended a Manhattan school and College and had mainly been influenced by the speech of Manhattan, where he had also moved to later on. Another informant was born and raised in Manhattan, but has moved to Brooklyn after marriage. All other subjects still live in Manhattan.

4.2 Data Analysis with Praat

Thanks to computer software, phonological change can be examined through the measurement of speech sounds, which is conducted via sound spectrography (cf. Fasold 1990: 229), a technique of “dispersing radiation (as electromagnetic radiation or sound waves) into a spectrum and recording or mapping the spectrum“ (Merriam-Webster 2003: s.v. spectrography). This method visualizes the formants^[23], i.e. the “acoustic characteristics” (Fasold 1990: 229) of a specific sound, and, thus, is a useful means for vowel analysis. In order to analyze the principal phonetic feature being studied in the present research, i.e. the advancement of the NCS which is resulting from vowel movements, the vowels’ place of articulation in each speaker’s oral cavity needs to be located. For that purpose, formant 1 (F1), indicating the openness of the mouth, and formant 2 (F2), specifying the backing or fronting of the vowel, have to be extracted from the spectrogram (cf. Fasold 1990: 229).

Among the range of computer software for the production of spectrograms, the most solidly established one in the fields of linguistics and communication science is Praat (Dutch: speech, talk), an open-source program developed by the Dutch linguists Paul Boersma and David Weenink at the Institute of Phonetic Sciences of the University of Amsterdam. Praat ^[24] is a modern, versatile program with numerous options for phonetic analysis, synthesization and modification. Not only can it be used for creating spectrograms, labeling them with speaker and sound underneath via the TextGrid function and extracting formants, but it also offers more complex features, such as drawing sounds. It has been given a user-friendly layout and has well-arranged, self-explanatory features, but also offers a helpful manual and has an active user group^[25]. It is constantly being improved and expanded, with approximately 10-20 new upgrades available every year.

For each of the selected 8 speakers, measurements of F1 and F2 were extracted for the vowels using Praat 5.1.37. Since the program offers only limited options when analyzing a long sound file, all recordings first had to be cut with Audacity freeware (cf. appendix A, fig. 6) first into pieces not exceeding the length of one minute. When analyzing a speech sample with Praat, one selects the ‘Read’ button, then ‘Read from File’ and chooses the short sound of interest, whereupon the file is inserted in the left Praat window, i.e. the list of ‘Objects’ selected to work with. In order to visualize the sounds and work with them, one clicks the ‘Edit’ button, whereupon another window is opened, showing the oscillogram in the upper part and the spectrogram of the speech sample below (cf. fig. 4.2.1). Formants of a vowel are indicated in the dark vowel representation, highlighted as red dots in the spectrogram. To analyze a specific sound, one has to select a section with the cursor, zoom in and extract the formants. Every operation can be conducted by clicking the particular button in the left-hand corner of the Praat window, or by using the key combinations Ctrl+i to zoom in or Ctrl+o to zoom out and Tab to play and stop the sound file.

As suggested in the Praat manual (cf. Online Sources Mayer 2009: Intro 3.2.), the spectrogram was viewed in the range of 0-5000 Hz. All adjustments concerning the formants were left at the convenient default settings, i.e. the suggested number of five formants to extract (cf. Online Sources Mayer 2009: Intro 5), the maximum formant having 5500 Hz, the window length set at 25ms and the dynamic range lying at 30dB.

When examining the recordings with Praat, “each vowel token was measured at a single point selected by hand as being characteristic of the central tendency of the vowel nucleus” (Dinkin 2009: 52). Before extracting the formants, each selected point was checked by ear, making sure that neither the syllable onset nor the syllable coda was audible, to guarantee that it was definitely located within the vowel nucleus (cf. Dinkin 2009: 52f). After having selected the vowel nucleus with the cursor, the command for the ‘Formant Listing’ was given, whereupon a window showing five formants was opened. Now F1 and F2, which are given in Herz (Hz), were copied and collected in a Microsoft® Excel spreadsheet and completed with the phonetic environment in which the specific vowel occurred and which word it was taken out of. The vowel surrounding was not performed by Plotnik but taken down by hand, which made the creation of a Praat TextGrid unnecessary. Besides the six vowels possibly affected by the NCS, i.e. /æ/, /ɛ/, /ɪ/, /ʌ/, /ɔ/ and /ɑ/, the TRAP-vowel, i.e. /æ/, occurring in the environment of a nasal (/æ+N/) was measured separately and entered into an individual Excel table.

In order to identify ambiguous or unclear vowels irrespective of a shift, Daniel Jones’ English Pronouncing Dictionary was consulted. If Jones offered two possible pronunciations, especially both /ɑ/ (LOT-vowel) and /ɔ/ (CLOTH-vowel) in words like dawn or thought, the speaker’s tendency towards one of the vowels was checked by ear. In addition, the formants in question were compared with already existing results of the informant and assigned to the specific Excel table. Moreover, when it was not clear whether the speaker had pronounced a /ə/ or a different short vowel, the sound was excluded from the analysis.

illustration not visible in this excerpt

Fig. 4.2.1: Screenshot of the Praat window, showing a zoomed-in TRAP-vowel, highlighted with the cursor, and its formants given in the Info-window on the left hand side

4.3 Organization and Visualization with Microsoft® Office Excel

In order to arrange and analyze the results, a Microsoft® Excel workbook was created for every speaker, containing a worksheet for each vowel possibly affected by the NCS, i.e. the TRAP-vowel (/æ/), the TRAP-vowel plus nasal (/æ+N/), the DRESS-vowel (/ɛ/), the KIT-vowel (/ɪ/), the STRUT-vowel (/ʌ/), the CLOTH-vowel (/ɔ/) and the LOT-vowel (/ɑ/). Each vowel-worksheet contained a chart with four columns, column A showing the word a vowel was extracted from, column B F1, column C F2, and column D the phonetic environment^[26]. Then all collected data were entered into the specific Excel table, with F1 and F2 both being rounded to one decimal digit. These detailed figures allow a thorough analysis and deliver “even fine nuances when similar realizations of vowels are compared” (Bause 2010: chapter 4.2). Words read out by the informants from the word lists were put at the end of each table and were marked with an asterisk (*). In addition, each speaker was given a worksheet in which the means of all vowels were entered.

To visualize the collected data of F1 and F2, it is possible to create a grid that looks like a traditional vowel chart in phonetics books. In order to create such a grid with Excel, an XY scatter-chart was selected. It shows F1 and F2 in relation to each other, F1 being depicted on the vertical axis, with lower frequencies at the top and higher frequencies at the bottom and F2 being depicted on the horizontal axis with lower frequencies to the right and higher ones on the left (cf. Fasold 1990: 229). Since the values are taken in the reverse order, F1, which was set to range from 300 to 1100 Hz, had 300 Hz as the highest point of the chart and cut the x-axis at 1100 Hz. F2 ranged from 1000 to 2500 and was also depicted in reverse order, i.e. the coordinate 2500|1000 signifies the crossing of the axes. For an adequate vowel-depiction both axis-ranges were adjusted if needed. If this is done, the scatter-chart resembles an abstract realization of each person’s oral cavity in its profile, while the left-hand side of the grid shows the front and the right-hand side the back of the mouth^[27]. Hence, when the F1/F2 pair of a vowel is entered into the grid, its place of articulation in the oral cavity is portrayed.

After all two-dimensional displays of F1 and F2 belonging to the same specific vowel had been entered into the Excel grid, an ellipsis was drawn as narrow as possible around all extreme values of the given sound. However, owing to the predispositions of Microsoft® Excel, the editing of the vowel envelopes was only allowed tocertain extends. Regardless of how many F1/F2 pairs of how many vowels are depicted in one scatter-chart, one elliptical envelope always includes all displays of only one specific vowel (cf.Fasold 1990: 330). However, whenever the formant value of one single vowel was situated in a completely different position than its allophones, it was neglected as a rare exception and not includes in the vowel envelope. Every ellipsis shows the considerable variation in the realization of a vowel and its shape gives information on a general tendency in which direction the sound is moving. Moreover, “the envelopes for various vowels overlap each other” (Fasold 1990: 330), which may provide conclusions about which vowels change place and how they are replaced by others.

If any vowel token lay conspicuously outside the ellipsis of other realizations of the same phoneme, Praat was used again to re-measure the vowels in question and to check for possible errors in the measurement. If the originally recorded formant values were reconfirmed, they stayed in the chart. If they turned out to be substantially different, they were replaced (cf. Dinkin 2009: 56).

This method of visualization differs from the one known from traditional vowels charts, since in the latter, sounds “are positioned at a particular point on the chart” (Fasold 1990: 230) and do not show the speaker’s variation in pronunciation. However, each workbook also includes one Excel grid without envelopes, i.e. one depicting all calculated mean vowels of a speaker.

In order to identify the phonetic surroundings of the F1/F2 pairs in the charts, Column D is taken into consideration. It contains the list of phonetic environments, which had been collected by hand when extracting the formants and suggests how the vowel in question is influenced by neighboring phonemes. Every respective point depicted in the vowel chart can be labeled with the corresponding phonetic realization of a vowel and its environment by using the add-on XY Chart Labels, available for download on the internet.

Since some of the recordings were of considerable length and some vowels therefore occurred remarkably often, the numbers of visualized and labeled F1/F2 pairs produce a relatively confusing Excel grid. With the objective of simplification, in vowel charts where too many results were available, only vowels with first or secondary stress were taken into consideration. In addition, in these cases, recurring vowel environments were eliminated after their third occurrence. As a result, a more general overview of the vowel spread is given (cf. Bause 2010: chapter 4.2). However, as long as the grid stayed convenient for analysis, as many results as possible were included and depicted.

After the visualization and separate analysis of each of the seven vowels, an Excel grid that contains all vowel clusters of one speaker was created. It shows the major overlaps of all sounds and depicts the relation of all the vowels to each other. This chart provides an overview how all sound realizations are located in a speaker’s oral cavity. It can be used in order to compare the speakers’ pronunciation with the vowel realization in General American (GA), or to compare the speakers with each other, i.e. the shape and size of their vowel envelopes, since the mere formant values are not comparable owing to age, sex, pitch and shape of the mouth. In addition such a cluster overview is helpful in the analysis of a possible chain shift, showing which vowels move, how and where they overlap or change places, etc.

The overarching vowel grid which depicts all seven vowels was then used in order to ‘zoom’ into a specific area of the oral cavity. It was used to check which vowels overlap and then one cluster set was set into relation to its neighboring vowels, e.g. /æ/ is overlapped or overlaps /æ+N/, /ɪ/ and /ɛ/. Hence, the envelopes of these overlapping vowels were drawn again, labeled with the respective phonetic symbol and only the phonetic environments of the vowels in overlapping areas were added. Taking the influence of certain phonetic environments into account, charts of this type provide a more detailed insight into possible vowel movements, e.g. their direction and their effect on other sounds, which finally results in a chain shift.

The Excel grids which depict the mean vowels, i.e. the focal points of the seven vowel realizations in question, additionally provide a more simplified and clearer view of the relation of the sounds towards each other, irrespective of their spread and extent. As mentioned above, traditional GA vowel charts only show sound positions at one particular point and are rarely presented in clusters. Therefore, the depiction of the speaker’s mean vowels allows a more decisive comparison between the abstract GA vowel realization and that of a particular informant. In addition, these measurements can be set in relation to the positions of individual realizations of a specific vowel in order to compare their distance to the speaker’s average pronunciation. This comparison may provide further useful information about the influence of the phonetic environment on the movement of its neighboring vowel.

4.4 Vowel Charts

When it comes to comparing of the speakers’ results with each other, two general issues need to be solved.

First of all, an ideal model of the vowel realization, i.e. their basic place of articulation in the oral cavity, must be established, in order to show how much and in what way the individual sound pronunciations of each informant differ from it. Since the NCS affects short vowels in the speech system and all speakers were born and raised in New York City, U.S.A., a vowel chart depicting the position of the short vowels in General American (GA), i.e. the Standard American pronunciation, is needed. The following vowel diagram (cf. fig. 4.4.1), published online by the University of Arizona, portrays all monophthongs in the pronunciation of GA, the short/lax ones being positioned in the area which is highlighted in gray. It is important to note that the low back vowel that is depicted as /a/ will be referred to as /ɑ/,as it was done throughout the whole thesis.

illustration not visible in this excerpt

Fig. 4.4.1: Vowel chart depicting the monophthongs of GA. Published by the University of Arizona.

Source: Online Sources IC Arizona Vowel Chart

The second issue that arises is the difficulty of comparing the pronunciation of eight different persons of both genders and various ages with each other.Kellermann (2001) transferred all F1/F2 values which were given in Hz into Bark units, employing a formula established by Traunmüller (1983) which she quotes from Bladon (1986: 5), where f represents the Hz frequency:

illustration not visible in this excerpt

Fig. 4.4.2: Formula to transfer Hz frequencies into barks. Source: Kellermann 2001: 324

Barks divide a speaker’s oral cavity into small rectangles instead of putting it into a linear Hz frequency scale (cf. fig. 4.4.3). Kellermann argues that “Bark scales modify the frequency scale in such a way that it represents perceptionally distinctive categories. In this way, Bark-sized circles, marking F1/F2 values, virtually simulate the phonetic distance” (2001: 324). Hence, the distance between a speaker’s vowels can be compared with the vowel distances of other subjects, since the parceling out of speakers’ vowel systems into fixed intervals allows an easier comparison.

However, Kellermann’s procedure compares speakers’ vowel realizations and places of articulation as if each individual’s oral cavity did not have a unique shape and an identical F1/F2 constellation. The phonetic distance of, for instance, one Bark may be a small distance for a person with a wide F1/F2 range, or a larger distance for a person with a smaller oral cavity. These physical predispositions depend on age and gender and vary with every individual. Therefore, Kellermann’s approach to the comparison of vowel charts constructed from formant values of different speakers is not adequate, since individual formant values cannot be compared with each other without considering the shape of each subject’s oral cavity.

Fig. 4.4.3: A speaker’s oral cavity and the position of the monophthongs depicted in Barks. Source: Kellermann 2001: 405 fig. 5.6

A more differentiated option is an approach based on the assumption “that there is one most open position for low vowels, and any movement among them is up from that position or down to it” (Labov, Yaeger&Steiner 1972: 109). Accordingly, an individual’s oral cavity has a range from F1 100% to 0%, i.e. the most open position for low vowels with the highest F1 to one most closed position for high vowels so that they can still be pronounced. Furthermore, the range of F2 also reaches from 100% to 0%, i.e. from one most front position for front vowels with the highest F2 to one most back position for back vowels. This approach was used, for instance, by Tatjana Bause (cf. 2010: 42f) who first extracted the vowel chart containing only the short vowels of GA and then surrounded it with a rectangle defined by the extremes of F1 and F2 (cf. fig. 4.4.4). Note that the chart depicted in fig. 4.4.4 has been clipped from the bottom, “which gives the lower vowels more extreme positions than in the complete vowel chart” (Bause 2010: 43). Since the long vowels /i:/ and /u:/ were excluded from measurement and vowel chart, the most extreme vowels in the chart depicted above are /ɪ/ in the high front corner, and /ɑ/ in the low back corner (cf. Bause 2010: 42).

illustration not visible in this excerpt

Fig. 4.4.4: Vowel chart depicting the short vowels of GA and the rectangle defined by the extremes of F1, F2. Source: Bause 2010: 43 fig. 4.3.2

“With these extreme values, the size, i.e. the lengths of the sides of the newly acquired rectangle, is defined, and the position of all the relevant vowels can be set into relation to the length of each side” (Bause 2010: 42). In the chart depicted in fig. 4.4.4 the GA realization of /ɑ/, for instance, has 0% divergence from the highest value of F1 and from the lowest value of F2. Using this chart, Bause (cf. 2010: 43) determines the ideal GA vowel positions of the oral cavity measured in percentages of divergence from extreme F1 and F2:

illustration not visible in this excerpt

Fig. 4.4.5: Position of GA vowels expressed in percentage of divergence from extreme F1, F2. Source: Bause 2010: 43 fig. 4.3.3

With the help of the charts given above, Bause (2010) suggests the measuring of the maximum of each individual’s oral cavity by extracting a person’s most extreme values of F1 and F2 from the collected data, which will indicate the scope of the F1/F2 realization of one person. For Annette (87f), for instance, the extreme values reached 455.77-1138.00 Hz for F1 and 715.94-2461.97 Hz for F2. Note that these are the maximum values that were measured during the interviews. The informant might produce even more extreme values, which she simply failed to utter in her interview. Hence, the maximum range of each person’s oral cavity established in the present thesis represents the maximum observed.

Furthermore, the mathematical placement of the short vowels in a person’s oral cavity via the percentages of divergence from extreme F1 and F2 serves “as a tool to define the direction and amount of movement of the analyzed vowels” (Bause 2010: 43). In addition each informant’s mean value of a particular vowel can be set in relation to his or her extreme values and the percentage of divergence can be compared with the already established ideal GA vowel positions. As suggested by Bause (cf. 2010: 44), each person’s individual percentages were calculated, through the following formula:

The percentages of F2 were calculated accordingly. In addition, “the average of all persons’ percentages was established to be able to interpret the general shifting” (Bause 2010: 44) of the vowels in the speech of the New York City informants. Note that for the simplification of the results and “since F1 high and F2 low are complementary” (Bause 2010: 44), only the percentages of divergence from F1 low will be presented in the results of the present study.

5. Speaker Analysis

The aim of the present analysis is to establish and examine the places of articulation of the articulation of each speaker and set the collected sound results in relation to the ideal vowel positions of General American. The analysis will include the description of vowel movements and will attempt to make an interpretive guess about the development and trend of the NCS in New York City. However, the present thesis includes only eight speakers and, thus, rather is a sample than a broad, representative overview. In order to expand the scope and generalize the interpretations given, further research would be needed.

In the following, each speaker is given a separate subchapter. The order of occurrence is based on the sociolinguistic variable (age), starting with the oldest participant.Each subchapter contains two vowel grids, the first one depicting the scope and positions of the ellipses of all six short vowels affected by the NCS, the second one capturing the mean values of each vowel. These two representations are followed by a table presenting the extreme values of F1/F2 for all six vowels, where the maxima, i.e. the highest and the lowest values of F1 and F2, of the particular informant’s vowel realizations are highlighted with a dark contour. These values “constitute the boundaries of the percentile scale” (Bause 2010: 45) and were used for the calculation of the percentages of divergence from the extreme F1 and F2 as described above. The mean values needed for each calculation are given in the middle of the same table^[28] and are set into relation to the extreme values. The resulting percentages describe the percentages of divergence from the boundaries, i.e. the maxima, of the vowel grid, which represents the individual oral cavity, and were added to the right hand side of the table. Each percentage is positioned in the same row as the vowel it refers to and the respective mean value it is calculated with.

After this representation of the results, each vowel is analyzed by taking the shape of its ellipsis into consideration and giving a description of its divergence and possible movement. Hereafter, it is compared to the other short vowels measured and presented in the vowel grid.

Furthermore, a very dense or interesting area of overlapping vowel clusters is captured and then analyzed with respect to phonetic environments. This depiction serves only as an example of the analysis of vowel environments. A more thorough study of more vowel overlaps and their phonetic environments would go beyond the scope of the present study. Besides the single depiction of a dense area and the vowel environments involved, the visual distribution of each vowel’s phonetic environment is presented in Appendix C, again the order of the informants is the same as in the present chapter.

5.1 Annette, 87f

illustration not visible in this excerpt

Fig. 5.1.1: Vowel charts of Annette (87f) displaying vowel envelopes and mean values

The extreme values of each vowel measured in the speech of Annette are presented below (cf. fig 5.1.2). Both, the highest and lowest F1, as well as the highest and lowest F2 values were highlighted in the chart. In the case of Annette, the lowest F1 and the highest F2 are the values for /ɪ/, the highest F1 is the value for /æ/ and the lowest F2 thatfor /ɔ/.

illustration not visible in this excerpt

Fig. 5.1.2 Extreme values, mean values and percentages of divergence of Annette (87f)

All mean values are given in the section next to their respective extreme values. Based on the four extreme values mentioned above and the respective mean value, the mean percentage of divergence of each vowel can be calculated. Consequently, the percentage of divergence from the lowest F1 value of/æ/, for instance, was calculated as follows:

Annette’s grid displaying her vowel ellipses (cf. fig. 5.1.1) shows that her vowel realizations spread over a broad range of her oral cavity, ranging from 700Hz for the lowest, i.e. most backed F2, to 2600 Hz for the highest, i.e. the most fronted F2. The positions of the mean vowels almost resemble the shape of a trapezoidal vowel chart with /ɪ/, /ɛ/, /æ/ and /ʌ/ in the ideal spots assigned by the GA vowel chart, and /ɑ/ and /ɔ/ in nearly the assigned spots of the GA vowel chart, as well.

The ellipsis of the TRAP-sound, i.e. /æ/, is flat and wide and spreads along the F1 axis, showing an upward and fronting movement. Its lowest position in the oral cavity is F1 1138Hz, which at the same time makes it the most extreme value of all vowels. Accordingly, it is positioned even slightly lower than /ɑ/, with its F1 being 155Hz higher than that of /ɑ/. The means of the two vowels, however, are nearly at the same low level in the oral cavity.

Besides its very low starting point, Annette’s /æ/ shows a raising. In consequence, the vowel ranges into the ellipsis of /ɛ/ in some environments, and in the environment of a nasal it even spreads into the range of /ɪ/. In addition to its upward movement, /æ/ is also fronted. Both, the most fronted (F2 2544.86) and most raised (F1 627.69) TRAP-vowels are followed by a nasal. The percentage of divergence of the means of /æ/ from the lowest F1 is 60.53% and from the lowest F2 67.86%. A comparison with the ideal percentages of divergence of the GA vowels shows, that /æ/ has nearly the exact ideal GA F2 position (67%), i.e. it is not yet fronted as a whole, but only in some environments. However, it is clearly raised, since its actual position (60.53% of divergence from the lowest F1) differs considerably from the ideal GA position (100%). In addition, if the measurements of /æ/ had not been separated from /æ/+N, its mean value would even lie higher and more fronted and show a percentage of divergence from the lowest F1/F2 that would differ much more from the percentages in GA. An analysis of the phonetic environment of these high values may provide information whether the raising of Annette’s /æ/ only reveals the typical New York City short- a pattern, or whether the NCS has already affected it.

The ellipsis of the LOT-vowel, i.e. /ɑ/, is very wide and long. It shows a slight lowering, but constant fronting movement along the F2 axis. In some phonetic environments it is already located in the scope of the vowel envelope of /æ/. This is the case in an area where only very few /æ/ realizations are left, i.e. in rather back and low positions. Its mean value also already shows a much more fronted position than the mean value of /ɔ/, although these two vowels are positioned on the same F2 level in the GA vowel grid. Its percentages of divergence from the lowest F1, F2 also show that /ɑ/ is more fronted and also higher than it is in GA. Thus, a forward movement is clearly visible. The overlapping of /ɑ/ and /æ/ suggests that the two vowels are already affected by the NCS, since /æ/ has started moving upwards in the oral cavity and vacates a space in to which /ɑ/ has already started to shift. The relatively high position of /ɑ/ may be an individual feature of the female informant whose lowest realization of /æ/ is also quite high already and at the same relatively low F1 level as /ɑ/. Hence, it is very likely that the speaker simply pronounces her lowest vowels at a quite high level in general.

The ellipsis of the CLOTH-vowel, i.e. /ɔ/, spreads downwards along the F1 axis in a slightly fronting movement, although its mean value still displays a very high and back position. Its lowest and most fronted values already overlap with the ellipsis of /ɑ/, which may be owing to the fact that the fronting of /ɑ/ vacates a space in the back of the oral cavity and causes /ɔ/ to fill that space.

Interestingly, in some environments /ɔ/ is placed at the same level as /ɪ/, whose parallel back vowel should be /ʊ/ and not /ɔ/. Hence /ɔ/ is has a very high spot, which is supported by the fact that its percentage of divergence of the lowest F1 is 20.47% while it is positioned 56%, i.e. more than double as low, in GA. This suggests that either /ʊ/ has an even higher position than /ɪ/ in the realizations of Annette, or that /ʊ/ and /ɔ/ may overlap in some environments, but they do not form a merger. The high position of /ɔ/ may accounted for the New York City shift suggested by Labov (cf. 1994: 203f), where /ɔ/ moves up towards /u:/. Nevertheless, the dragging evoked by the NCS is very strong, since the vowel clearly shows a lowering movement which will soon eliminate any overlaps of /ʊ/ and /ɔ/.

The ellipsis of the DRESS-vowel, i.e. /ɛ/, has a similar shape as the one of /æ/+N and spreads along the F2 axis in a parallel but notably backing manner, ranging into the ellipsis of /ʌ/. It also shows a slight lowering which, in some environments, causes overlaps with /æ/ or /æ/+N. Moreover, in some environments it overlaps with /ɪ/. However, its mean value is still at the spot which it has in the GA vowel chart, i.e. movement is already visible in some phonetic environments, but the overall realization of the vowel has not been moved to a new position, yet. This is supported by the fact that no phonetic environment could be made responsible for any direction of movement and the variation of the realization of /ɛ/ does not follow a particular ruleas yet. The percentage of divergence from the lowest F1 is slightly smaller than it is in GA, i.e. Annette’s /ɛ/-position is even higher than in the Standard American pronunciation. However, the percentage of F2 shows that it is slightly more back than in GA.

The ellipsis of the KIT-vowel, i.e. /ɪ/, shows a very small scope and a clearly assigned spot. It spreads along the F2 axis in a parallel backing movement. In some environments it overlaps with the envelopes of /ɛ/ and /æ/+N or /æ/ as they are raised and is more fronted than /æ/, but not as much as /æ/+N. Its percentage of divergence from the lowest F1, F2 shows that it still has a very front and very high position in the oral cavity. Its slightly backed position may be owing to Annette’s New York accent in which the centralization of /ɪ/ is very common.

The ellipsis of Annette’s STRUT-sound, i.e. /ʌ/, is egg-shaped and smaller than all other ellipses in the grid. It spreads along the F2 axis in a parallel manner, but its shape suggests that it does not have much movement. It is overlapped by /ɛ/ in some environments and overlaps with /ɑ/ in other, rather back areas. The ellipsis has a central position, while its mean value is placed slightly more to the right, i.e. more back. This position nearly resembles the ideal spot of /ʌ/ in the GA vowel grid, although its percentage of divergence from the lowest F1, F2 shows that it is slightly more front and higher.

Since the ellipses of /ɪ/, /ɛ/, /æ/ and /ʌ/ of Annette are a very dense area in the upper front of the informant’s oral cavity, a further examination of their phonetic environments may be helpful to understand their distribution.

illustration not visible in this excerpt

Fig. 5.1.3: Vowel chart of Annette (87f) displaying phonetic environments in a dense area (L= Lateral, N=Nasal, P=Plosive)

Figure 5.1.3 shows that most overlaps occur in different contexts. Nasals tend to front a vowel, which causes slight overlaps between /æ/ and /ʌ/. This rather general phenomenon in the realizations of American vowels and was already established by Labov, Yaeger and Steiner (cf. 1972: 60) as a favoring context for peripherality, i.e. fronting and also raising in the vowel envelopes.

Furthermore, vowels followed by both voiced and voiceless plosives are already moving forward and upwards into the range of /æ/+N. Since in New York City English voiceless plosives do not have a tensing effect on /æ/, but /æ/+P [-voice] in Annette’s speech is fronted and raised, it becomes clear that it is already affected by the NCS. In addition, the backing influence of a lateral is noticeable, as depicted above in combination with /ɪ/ and also in appendix C when following /æ/.

In sum, the pronunciation of Annette suggests that /æ/, /ɑ/ and /ɔ/ have already started shifting and, therefore, are affected by the NCS. There are indicators for other vowel changes to follow, e.g. the shapes of the vowel ellipses showing first movements into directions typical of the NCS, but they are not of considerable depth, yet. Nevertheless, the 87-old informant already shows the first features of the NCS.

5.2 Merryl, 61f

illustration not visible in this excerpt

Fig. 5.2.1: Vowel charts of Merryl (61f) displaying vowel envelopes and mean values

All extreme values and calculated figures of Merryl are as follows:

illustration not visible in this excerpt

Fig. 5.2.2 Extreme values, mean values and percentages of divergence of Merryl (61f)

As it was the case with Annette, Merryl’s oral cavity shows a broad range (cf. fig. 5.2.1). Especially the depth is notable, the F2 values starting at 500Hz in the back and reaching 2500Hz in the front, resulting in a horizontal vowel spread of 2000Hz. Hence, if vowels shift in Merryl’s speech, they are very likely to be identified easily as changing their positions, since they need to move a considerable distance to take over the vacated space of another vowel.

The first striking fact in the grid depicting the informant’s mean values is that it has lost its trapezoidal shape. Hence, some more considerable changes than in Annette’s speech must have taken place.

Merryl produces a very wide and flat /æ/-ellipsis which shows an upward movement with a slight fronting. It overlaps with more than half of the ellipsis of /ɛ/ and also reaches very high up into that of /ɪ/. Interestingly, it is strongly fronted when occurring before a nasal, but some values which range high up into /ɪ/ are also followed by other environments. Furthermore, all nasal environments do not cause raising, some are still in the average range of /æ/. Although /æ/ is not the lowest vowel in this chart (F1 high belongs to /ɑ/) and its percentage of divergence of the lowest F1 (64.50%) is much smaller than in GA (100%), its F1 mean value is still the highest, i.e. on average /æ/ is still the lowest sound, although positioned relatively high. Moreover, its percentage of divergence from the lowest F2 (68.79%) almost equals that of GA (67%) and suggests only a very slight fronting. Its diagonal upward shift, however, already clears a space in the low, back area of its ellipsis, which is taken over by the front realizations of /ɑ/.

Consequently, the ellipsis of /ɑ/, which is very wide and spreads from 958.28 to 1725.03 Hz along the F2 axis in a diagonal downward movement, overlaps the ellipsis of /æ/ in some environments owing to a drag chain shift caused by the changed position of /æ/. It also overlaps with the ellipses of /ɔ/, /ʌ/ and slightly with /ɛ/. Not only does the vowel envelope of /ɑ/ suggest considerable fronting, but also the position of its mean value and percentage ofdivergence of the lowest F2 (43.05%) supports that fact. Interestingly, /ɑ/ is also lowered to a large extend in order to take over the exact low, back position vacated by /æ/, but its F1 percentage of divergence (60.87%) shows that it is still relatively high for a low vowel.

The back of the /ɑ/ ellipsis is overlapped by the ellipsis of /ɔ/. The latter spreads parallel along the F2 axis in a fronting movement with an almost non existing lowering. Its position is very high, i.e. only slightly lower than /ɪ/. This accounts both for its ellipsis as well as for its mean value and a consequence to the New York City shift already mentioned. Despite the clear fronting which is visible in the spread of the vowel envelope, its mean value is still very far back, with a percentage of divergence from the lowest F2 of only 19.11%. Again, this movement is dragged by the shift of /ɑ/.

The ellipsis of /ɪ/ is narrow and wide. It overlaps with those of /æ/ and /ɛ/, spreads parallel along the F2 axis and shows a clear backward movement, but no lowering. This is supported by the depiction of its mean value as well as its percentage of divergence from the lowest F2 (76.85% in contrast to GA 100%), which shows that the KIT-vowel has left its ideal spot from the GA vowel chart and has shifted backwards. Nevertheless, it still is the highest (F1 low 456.54 Hz, percentage of divergence of F1 only 14.93%) and most fronted (F2 high 252.93 Hz) vowel in the chart. Interestingly, the spread of the values of /ɪ/ is more focused and dense on the right-hand side, i.e. in the back of the vowel envelope. This phenomenon may be due to Merryl’s New York City accent in which /ɪ/ is centralized, but it may also be influenced by a push chain reaction caused by the fronting and upward movement of /æ/, i.e. /ɪ/ has started to leave the space which is slowly being occupied by the shifted /æ/.

The ellipsis of /ɛ/ is wide and flat and shows backing along the F2 axis as well as slight lowering. It overlaps with the ellipses of /æ/, /ɪ/ and /ʌ/. The overlap with /ɪ/ is the result of the backing of the KIT-vowel, the overlap with /æ/ results from the raising of the TRAP-vowel and the overlap with /ʌ/ is caused by the backward movement of the DRESS-vowel itself. Not only the ellipsis, but also the depiction of the mean value of /ɛ/ suggests a clear backing of the sound. The percentage of divergence from the lowest F1 (31.89% in contrast to GA 56%) suggests that it is still positioned relatively high, but the divergence of the lowest F2 (67.22%) shows a backward movement of approximately 10% as compared to the vowel position in GA.

The ellipsis of /ʌ/ is small and egg-shaped. It lies in the middle of the speaker’s oral cavity, spreads horizontally along the F2 axis and is overlapped in the front part of its ellipsis by the shifting vowels /æ/ and /ɛ/. In addition, it merges with the ellipsis of /ɑ/, which may partly be because of the fronting and lowering of the latter and may even suggest a backward movement of /ʌ/ towards /ɔ/ and the back values of /ɑ/. However, it does not overlap with /ɔ/ as yet. Both vowels are still clearly distinct. It is worth considering the percentages of divergence of the lowest F1, F2, which show that Merryl’s realization of /ʌ/ is still higher and more fronted than it is in GA. This fact indicates that a shift of the STRUT-vowel most likely has not taken place, yet.

Since the ellipses of /æ/, /ɪ/, /ɛ/, /ʌ/ and the front part of /ɑ/ form a very dense area, a short analysis of the phonetic environment may be helpful to understand some of the movements.

illustration not visible in this excerpt

Fig. 5.2.3: Vowel chart of Merryl (61f) displaying phonetic environments in a dense area (A=Approximant, Af= Affricate, F= Fricative, N=Nasal, P=Plosive)

As mentioned above, a following nasal causes peripherality, which is why /æ/ followed by a nasal is noticeably fronted and also raised. However, /ɪ/ in the environment of a nasal still has a very front position (cf. figure 5.2.3), but it is not the highest value any longer and also ranges far back. Hence, the nasal supports, but does not guarantee fronting and raising anymore, especially not when the tendency of the whole sound is toward moving back- and/or downwards. Note that the highest values of the highest vowel, i.e. /ɪ/, are maintained by the environment of a fricative, while vowels followed by plosives in this case apparently are prone to backing, as can be seen in the case of /ɪ/ and /ʌ/.

Furthermore, /ʌ/ in the environment of an affricate tends to be realized rather in the back. Second, environments supporting or impeding the backing of /ɛ/ in the NCS could still not be established. However, the most back value of /ɛ/ to be measured is followed by a lateral whose backing influence has already been discussed and confirmed above. Finally, note that in the chart of Merryl /æ/ is not only raised in the environment of a nasal, but also when being followed by an approximant or even a fricative, as shown by the combination of æ/ followed by /s/. Interestingly, the highest value of /æ/ is not one before a nasal, but followed by an approximant, i.e. /r/, and even lies higher than /ɪ/+N. In addition, /æ/ followed by voiceless fricatives is also tensed, as it is the rule for the New York short- a pattern. In addition, a following plosive, voiced or unvoiced, causes a fronting in Merryl’s realization of /æ/. This was already the case in Annette’s speech and has been shown to be untypical of the New York accent and, hence, a feature of the NCS.

In sum, Merryl’s vowel grid shows a further stage of the NCS, since not only /æ/, /ɑ/ and /ɔ/ are already being shifted, but also /ɪ/ and /ɛ/ indicate movement that has the characteristic of the NCS.

5.3 Tim, 48m

illustration not visible in this excerpt

Fig. 5.3.1: Vowel charts of Tim (48m) displaying vowel envelopes and mean values

All extreme values and calculated figures of Tim are as follows:

illustration not visible in this excerpt

Fig. 5.3.2 Extreme values, mean values and percentages of divergence of Tim (48m)

Tim’s vowel grid does not show an oral cavity as broad as the two females analyzed above. Nevertheless, his vowel realizations spread over an area of over 1500Hz along the F2 axis and over 600 Hz along F1 (cf. fig. 5.3.1). This space would allow each vowel to maintain a security distance to the surrounding sounds. The vowel ellipses give the overall impression to be still relatively distinct, depending on the specific sounds. Tim’s mean values still resemble the shape of a trapezoid, although some significant changes have taken place.

The ellipsis of the TRAP-vowel is egg-shaped and vertical. It spreads along the F1 axis in a parallel upward movement and overlaps with that of /ɛ/. In the environment of a nasal, /æ/ is fronted and shifts even higher, so that it overlaps with more than half of the vowel envelope of /ɪ/. Beside the combination with a nasal, no other environment causes an overlap of /æ/ and the KIT-sound. The mean value of /æ/ is placed relatively high, which is confirmed by the percentage of divergence from the lowest F1 (68.81%) in comparison to GA (100%). An analysis of the phonetic environment may provide further information concerning the question whether the position of Tim’s /æ/ simply resembles the short- a pattern of New York City and is only shifted in specific environmens or whether it has already left its orinigal spot and is moving upwards, the environment of a nasal dragging the rest of the vowel.

The assumption that /æ/ must have already left the space it ideally has in the GA vowel chart is supported by the shape and movement of the ellipsis of /ɑ/. Tim’s LOT-vowel has a long and rather narrow envelope which shifts down and forward in a strong diagonal movement. The spot it tries to reach is already vacated by /æ/. Thus, the ellipsis of /ɑ/ has such a strong direction and fast shift, that it overlaps with /ɔ/ only marginally and does not allow the formation of a cot/caught merger in Tim’s New York City English, which is one of the most essential features of the NCS. Its percentage of divergence is 34.75% for F2, which in comparison to GA (0%), means that there is considerable fronting. In addition, its mean value has a more fronted position than /ʌ/ and its envelope even includes the ellipsis of /ʌ/ completely. Despite its very high F2 and strong fronting movement, however, it hardly overlaps with /æ/ either. This may be because /æ/ has already shifted upwards and /ɑ/ has just started, so that the two vowels will either overlap in later stages, when /ɑ/ has been shifted more towards /æ/, or the drag chain of Tim’s short vowels preserves heterogeneity that much, that /æ/ moves even more upwards as /ɑ/ approaches it and then /ɔ/ follows the dragging of /ɑ/ without touching it. However, the latter scenario is rather unlikely, since, as mentioned in chapter 3.2, competing variants tend to coexist for a while before the older one gets lost or gives place for the new one (Gordon 2001: 3f).

The cluster of /ɔ/ is circle-shaped and situated in the back of the oral cavity. It has an almost perfect GA position, with the F1 percentage of divergence being 42.17%, i.e. it is located even slightly higher than in the Standard American pronunciation (GA has 56% F1 divergence), but not as high as in the speech of the female informants before. In Tim’s speech the New York City shift apparently is not as developed as in the speech of the former informants. Furthermore, its F2 divergence is 10.17%, i.e. in this case the CLOTH-vowel is realized very far back. Its round envelope, mean value position and percentage of divergence suggest that is has not shifted, yet, or that is has only slightly moved downwards. Surprisingly, although /ɔ/ seems not to have shifted much, yet, further and younger movements of the NCS are identifiable in Tim’s pronunciation, i.e. starting shifts of /ɪ/ and /ɛ/.

The ellipsis of /ɪ/ has almost the same shape as /æ/ and is only a bit longer. It overlaps with /æ/+N and /ɛ/ and spreads horizontally along the F2 axis and indicates a backward as well as a slight downward movement, which is the reason why it merges with the ellipsis of /ɛ/. The overlap with /æ/+N is owing to both the upward movement of /æ/ and the lowering of /ɪ/. Its mean value, however, is still at the spot where it is ideally placed in GA vowel charts. The percentage of divergence (F1 22.44% in comparison to GA 0% and F2 74.85% in comparison to GA 100%) indicates that it is already a bit lower and more backed than it should be ideally and, thus, is affected by the NCS. However, its shifting apparently has not evolved much yet, since the movements indicated by the ellipsis have not yet affected the whole vowel in all environments, which is why the position of the mean value still resembles the one in the GA chart. Hence, the position of /ɪ/ may be traced back to the New Yorker centralization of it and not only to the NCS.

The ellipsis of /ɛ/ has an almost round shape which shows a backing and a slightly stronger lowering development as it spreads rather vertically along the F1 axis. However, both movements, i.e. backing and lowering, as they appear in the NCS, are idendifiable. Because of the raising of /æ/ and its own lowering, the ellipsis of the DRESS-vowel overlaps with /æ/. However, it clearly distinguishes from /ʌ/ with a considerable gap between these two sounds. The depiction of its mean value and the percentage of divergence of F1 (40.15% in comparison to GA 56%) show that the location of /ɛ/ is even a bit too high in comparison to its position in the GA vowel chart. In addition, the mean value and the percentage of divergence of the lowest F2 (65.04%) show, in comparison with the Standard American 77% divergence, that the realization of /ɛ/ has already been slightly moved backwards and that this shift does not happen in specific environments anymore but with the DRESS-vowel in general.

The vowel envelope of /ʌ/ has a perfectly round shape and completely overlaps with /ɑ/. Its percentages of divergence from the lowest F1, F2 show, however, that it is even positioned a bit too high (51.93%) in comparison to GA (56%) and a bit too fronted (31.14% while it is 27% in GA). Hence, the collision between /ɑ/ and /ʌ/ or, the complete incorporation of /ʌ/ by /ɑ/ is owing to the strong shift of /ɑ/. When the latter completes its fronting and lowering shift, it will vacate the space for the backing of /ʌ/ and also the fronting and lowering of /ɔ/. The latter two movements may result in another collision, which is already suggested by the slight overlap of the ellipses of /ʌ/ and /ɔ/.

Since the back area of Tim’s oral cavity still has a clear distinction and is only dominated by the shift of /ɑ/, the informant’s high- and mid-front area shall briefly be analyzed in terms of phonetic environment.

illustration not visible in this excerpt

Fig. 5.3.3: Vowel chart of Tim (48m) displaying phonetic environments in a dense area (N=Nasal, P=Plosive)

The vowel grid above (cf. figure 5.3.3) depicts the highest point of /æ/, as well as the highest and most fronted points of the ellipses of /ɪ/ and /ɛ/. Tim’s distribution of phonetic environments first of all shows the basic environmental influence that has already been stated, i.e. the relatively high and front position of vowels which are followed by a nasal. This causes an overlap of the three mentioned vowels, since /æ/ moves upwards owing to the influence of a nasal environment, while /ɪ/ and /ɛ/ move backwards, but still maintain some values in a very high and front position, i.e. the ones which are followed by nasals. Furthermore, the environment of a plosive seems to influence the position of Tim’s vowels. As depicted above, /ɪ/+P also has a relatively high and front position while the highest values of /æ/ besides the ones followed by nasals are vowels that are followed by plosives or, in one single case, by an approximant. The fact that the plosives involved are both voiced and unvoiced suggests that Tim’s /æ/ is clearly affected by the NCS, since the New York City short- a tensing is not favored by voiceless plosives.

Apparently, the NCS in Tim’s pronunciation has not developed as far as in the speech of the two females in the chapters above. His realization of the six short vowels affected by the NCS shows that the first stages, i.e. the shifting of /æ/ and /ɑ/ are already in progress and quite developed, while others, i.e. the shifting of /ɪ/, /ɛ/ and /ɔ/ have just started.

5.4 Philipp, 48m

illustration not visible in this excerpt

Fig. 5.4.1: Vowel charts of Philipp (48m) displaying vowel envelopes and mean values

All extreme values and calculated figures of Philipp are as follows:

illustration not visible in this excerpt

Fig. 5.4.2 Extreme values, mean values and percentages of divergence of Philipp (48m)

With a horizontal spread of only 1200 Hz and an F1 from 400 to 800 Hz (cf. fig. 5.4.1), Philipp’s oral cavity only shows a small range for the realization of vowels. The positions of his mean values also support this fact; they are located very close to each other while the majority of them show a centralizing tendency. The vowel trapezoid is lost completely and now rather resembles a circular movement as the one depicted in chapter 3.2. This suggests that several sound changes are already in progress.

Philipp’s realization of /æ/ is a vertical, almost round ellipsis which spreads along the F1 axis in an upward movement. It still has the highest F1 value; so far it has not left its former position completely, but it shows a strong tendency to the left hand corner of the vowel chart, especially when followed by a nasal. Therefore its lowest F1 values overlap with the ellipsis of /ɪ/. The percentage of divergence from the lowest F1 (54.99%) suggests that /æ/ has already undergone an upward movement of over 50% in comparison to the average 100%. Furthermore, its F2 divergence is 71.74%; it has also moved forward compared to its average position in GA (67%). In combination with the percentages of /æ/+N, it is located even higher and more fronted. Interestingly, a few values of /æ/ are located very far back, but they will probably be fronted and dragged upwards, as /æ/ rises and /ɑ/ approaches from the back.

With an F2 range of over 600Hz the ellipsis of /ɑ/ is comparatively wide. It overlaps slightly with the ellipsis of /ɔ/ in the back and overlaps more than half of the ellipses of /ʌ/, /ɛ/ and /æ/. Only a few values are already located in the range of the vowel ellipsis of /æ/. Nevertheless, this is an indicator for the drag chain known as NCS, where /ɑ/ gradually takes over the vowel space vacated by /æ/. Moreover, the length of the ellipsis of /ɑ/ suggests a strong forward movement to the newly assigned spot. However, the majority of its values are located in the area of the /ʌ/-ellipsis, causing that the mean values of these two vowels nearly have the same position in the informant’s mouth with /ɑ/ moving further forward and most likely passing /ʌ/ as the NCS develops further. So far, it is located 45.81% more fronted than the average percentage of divergence of the lowest F2.

The cluster of /ɔ/ is egg-shaped and spreads vertically along the F1 axis, indicating a clear lowering which is probably dragged by the movement of /ɑ/ with /ɔ/ slowly filling the empty space. Its mean value is that much lowered (F1 720.21 Hz) that it is nearly on the level of /ʌ/ (F1 698.87 Hz). It only marginally overlaps with the ellipsis of /ɑ/ which has approached too much forward to merge with /ɔ/. It is also worth pointing out that the CLOTH-vowel has a central back position and is not located in the range of /ʊ/ or /u:/ anymore. Hence, either Philipp’s /ɔ/ was never affected by the New York City shift and has never moved upwards, or the NCS has developed that far that /ɔ/ has already been lowered considerably to its present position.

The vowel envelope of /ɪ/ is small and rather narrow and spreads horizontally along the F2 axis. Its percentages of divergence indicate a still high and front position of its mean value; its vowel envelope, however, clearly shows a backing with slight lowering in a few phonetic environments. This movement evokes that /ɪ/ overlaps the ellipsis of /ɛ/ and is overlapped by the rising /æ/-envelope.

The ellipsis of /ɛ/ is long and resembles the ellipsis of /ɑ/. It shows a strong, diagonal downward movement along the F1 axis towards the center of the chart, although its mean F1 value does not show a general lowering yet. In fact, its percentage of divergence shows that /ɛ/ even is positioned higher (38.62%) than the average (56%). This is owing to the fact that half of the values of /ɛ/ are still located in the upper front corner, while the other half has already been lowered and backed considerably. Hence, not the whole vowel has changed its position, yet, but only performs the shift assigned to the NCS in some specific phonetic environments. Nevertheless, it has clearly shifted backwards already and is gradually lowered as the NCS approaches. This shifting in addition to the fronting of /ɑ/ and raising of /æ/ causes a massive overlap of the ellipses of /ɛ/, /æ/ and /ɑ/ and also a slight collision with the front part of the /ʌ/-ellipsis.

The ellipsis of /ʌ/ is one of the smallest in the vowel chart. It is egg-shaped and tilted to the side, showing a strong diagonal movement in the low back corner, which indicates lowering and backing. Half of its values maintain the original vowel spot and merge with the backed and lowered values of /ɛ/ and the other half is strongly shifted. The backing of the latter evokes that the ellipses of /ʌ/ and /ɔ/ touch each other, but do not overlap. Their lowering however, results in the complete collision of /ʌ/ and /ɑ/ as it was described by Gordon (2002: 255). However, the mean value and percentages of divergence of /ʌ/ do not indicate a strong lowering or backward movement, yet. Nevertheless this shift is clearly visible in Philipp’s vowel chart. Therefore, as it is the case with the other vowels, the shift of /ʌ/ has started in some environments, but has not affected all of them. The positions of the informant’s mean vowels and the directions of his ellipses suggest the following movement:

illustration not visible in this excerpt

Fig. 5.4.3: Mean values of Philipp (48m) and their movement tendency.

Figure 5.4.3 depicts how every vowel’s movement is connected to the movement of another, indicating a clear chain shift. Furthermore the vowels’ positions and their directions of shifting resemble the ones assigned to the NCS, as they are depicted in figure 3.2.1 in the present thesis. Philipp’s vowel realization suggests that all six short vowels have already been affected by the NCS and are moving towards their new positions.

However, none of the sounds has reached its newly assigned spot as yet. At the moment all vowels only shift in specific environments. This causes a considerable overlap in dense areas. A simplified overview of the phonetic environments in the specific area may help to identify the consonants favoring or impeding the shift.

The vowel chart below (cf. figure 5.4.4) displays the phonetic environments that overlap and the most extreme values of the vowels in the dense area. First, the influence of a nasal environment needs to be pointed out, again. The most fronted values of /ɪ/, /æ/ and also the highest and most fronted points of /ʌ/ are the ones which are followed by a nasal. Note the outstanding position of /kæn/ in the high front corner. This value was extracted from an auxiliary verb which, according to Labov’s New York short- a pattern rules should be lax (cf. 2007: 15). The fact that it is raised anyway may suggest that all values of /æ/+N might move towards that extreme position owing to the NCS.

illustration not visible in this excerpt

Fig. 5.4.4: Vowel chart of Philipp (48m) displaying phonetic environments in a dense area (F=Fricative, L= Lateral, N=Nasal, P=Plosive)

Furthermore, it is interesting to consider that the environment of a plosive maintains the average F2 position of /æ/, but is already reaching very high up, while it is backed when combined with /ʌ/ and fronted when following /ɑ/. Accordingly, Philipp’s short-a pattern is already influenced by the NCS, since his TRAP-vowel even is raised in combination with voiceless plosives. Also note the fronting and raising influence of fricatives on the realization of /æ/.

The highest value of /ɑ/ is followed by an affricate and is already located in the middle of the vowel chart. Furthermore, /ɑ/ in combination with a plosive merges with /æ/+P and /æ/+F. Finally, the most backed values of /ɛ/ turn out to be followed by a lateral and also a fricative. The backing influence of the former has already been established by the values of the speakers analyzed above. The latter, however, may have a backing F2 and centralizing F1 tendency, since not only one of the most backed and centralized values of /ɛ/ is followed by a fricative, but also some of the lowest F2 values of /æ/ occur in the environment of fricatives.

As it was shown in figure 5.4.3, Philipp’s vowel chart shows the affection by the NCS of all six short vowels known to be involved in it. However, in the informant’s vowel realization none of the shift-stages are completed, yet.

5.5 Jake, 24m

illustration not visible in this excerpt

Fig. 5.5.1: Vowel charts of Jake (24m) displaying vowel envelopes and mean values

All extreme values and calculated figures of Jake are as follows:

illustration not visible in this excerpt

Fig. 5.5.2 Extreme values, mean values and percentages of divergence of Jake (24m)

Jake’s vowel realization has a vertical spread of 450Hz and only 1100Hz on the horizontal level (cf. fig. 5.5.1), which makes his oral cavity the smallest of all eight informants. As a consequence, the distribution of the six short vowels possibly affected by the NCS may not be as structured and distinctive as in the ideal GA vowel chart. This assumption is confirmed by the unusual adjustment of the mean values which do not form the common trapezoidal shape anymore.

The shape of the /æ/-cluster is egg-shaped and wider than the ones of the other male speakers, but shorter than the female TRAP-ellipses. It has a clear diagonal upward movement towards the high front corner and overlaps with the vowel envelope of /ɪ/. Interestingly, it is located almost as high as /æ/+N which is only slightly more fronted and raised and also overlaps severely with the ellipsis of /ɪ/. The percentage of divergence of the lowest F1 amounts up to 57.66% for /æ/ and 37.42% for /æ/+N. Thus, the TRAP-vowel is located between 40% to over 60% higher than the average in GA. Besides a few values in specific environments, it has almost left its former low front position completely. In addition, the space it has cleared is already taken over by some values of /ɑ/. Hence, the upward shift of /æ/ as it occurs in the NCS is almost completed.

Although the shift of /ɑ/ is the second movement of the NCS and all other vowels have started shifting already, in Jake’s speech it is the one that is still performing the most movement. It is dragged forward by the shift of /æ/ and spreads over the whole lower part of the oral cavity, overlapping the lower cluster-parts of /ɔ/, /ɛ/, /ʌ/, /æ/ and even the lowest part of /æ/+N. Its highest F1value states the most extreme F1 of the whole vowel system, i.e. it has the lowest position. Furthermore, its highest F2 is very far front (1811.61Hz) and almost at the level of the most fronted value of /æ/ (1812.80Hz). This strong forward movement results in its mean value being located 40% more fronted than average (0%) and over 30% higher. Since not all values of /æ/ have shifted upwards, yet, the lowest values of /æ/ and the front values of /ɑ/ merge. Moreover, /ɑ/ overlaps with /ɔ/ because its own movement drags /ɔ/ down and forward into the vowel space that currently is being vacated by /ɑ/. However, the LOT-sound has not left its old position completely, so that its lowest F2 values merge with the front and low values of /ɔ/.

The fronting and lowering of /ɔ/ has formed its shape into a horizontal ellipsis that spreads along the F2 axis. The position of its mean value and percentages of divergence (38.22%) do not indicate a visible movement so far and rather indicate a relatively high position still owing to the New York City chain shift, but the shape and direction of its ellipsis clearly suggest that it is in the progress of shifting into a different direction. This shift does not only evoke an overlap with the upper back part of the /ɑ/-cluster, but also a collision with the backed values of /ɛ/ and /ʌ/.

The next stage of the NCS after the lowering and fronting of the back vowels is the backing and lowering of high- and mid-front sounds. This phenomenon is already visible in the ellipsis of /ɪ/, which is long and a little narrower than the ellipses of /æ/ and /æ/+N and spreads horizontally along the F2 axis in a backward movement. A few values in specific environments are not only backed, but also considerably lowered. However, the majority is still in a very high position. This distribution affects the mean value of /ɪ/, whose percentage of divergence proves that it has already moved backward (divergence from the lowest F2 77.11% in comparison to the average 100%), but is still very high (divergence from the lowest F1 14.28%). The cluster’s lowest values merge with the high, raised values of /æ/ and its low, back values overlap with the front part of the vowel envelope of /ɛ/. Hence, the KIT-vowel is already affected by the NCS.

The same fact applies for Jake’s realization of /ɛ/, whose horizontal ellipsis is very wide, but not as long as the one of /ɑ/. It shows a strong backward shift towards /ʌ/ and /ɔ/, its mean value having a percentage of divergence from the lowest F2 of only 58.58%, i.e. it is located 20% further back than the average (77%), flouting the trapezoidal shape of the vowel chart. The front values of /ɛ/ which were not shifted, yet, merge with the shifted values of /æ/, /æ/+N and /ɪ/. In addition, its lowered and backed values merge with the already fronted values of /ɑ/ and /ɔ/, as well as with the front values of /ʌ/ which have not been moved as yet.

As the lowering and backward movement of /ɪ/ has pushed /ɛ/ backwards, the latter now forces /ʌ/ to move backwards, as well. As mentioned before, this part of the NCS is clearly a push- rather than a drag-chain shift.

As /ʌ/ is pushed backwards, its ellipsis becomes wider and flatter and spreads horizontally along the F2 axis with a slight diagonal lowering in the back part. This shift evokes an overlap with the vowel envelopes of the fronted /ɔ/ and /ɑ/, while the unshifted values of /ʌ/ are in the way of the backed values of /ɛ/.

The speech of Jake shows all movements associated with the NCS. The depiction of the mean values resembles the positions and movements of Philipp’s mean vowels (cf. fig. 5.4.3) and, thus, also the general model of the NCS movements in figure 3.2.1. However the majority of the sound shifts in Jake’s vowel realization are not nearing completion yet. Accordingly, only specific phonetic environments are involved in the shift while others still stay at their former vowel spots. An analysis of the environments in overlapping areas may clarify some shifting tendencies. The densest accumulation of vowels is situated in the upper (high) front of the vowel grid.

illustration not visible in this excerpt

Fig. 5.5.3: Vowel chart of Jake (24m) displaying phonetic environments in a dense area (A=Affricate, F=Fricative, L= Lateral, N=Nasal, P=Plosive)

The distribution of phonetic environments is the most dense and complicated so far. As a basic fact, the ellipsis of /ɛ/ merges with every other vowel, but always in a different phonetic environment (cf. figure 5.5.3). Since the overlaps are too dense to identify the influences of particular vowel-consonant-combinations, they are extracted and depicted in individual diagrams with regard to their place of articulation.

The most characteristic feature of the NCS is the fronting and also raising influence of nasals. The chart above (cf. fig. 5.5.4) shows that the most fronted values of /ɪ/, /æ/ and especially /ɑ/ are the ones which are followed by a nasal. It is important to note how far the LOT-vowel has moved to the front, as its extreme values reach into the F2 range of /æ/. Interestingly, the most fronted values of /ɛ/ are not the ones which are followed by a nasal, but rather values in the environment of /r/, i.e. an approximant. The same fronting and also raising influence of /r/ can be seen in the distribution of /ɪ/, where the realization of /spɪr/ is one of the highest and most fronted values of the KIT-vowel. Furthermore one should note the collision of /ɛ/+N with one of the most fronted values of /ʌ/ which, surprisingly, is followed by a lateral that usually should have a backing influence, instead of fronting. The next phenomenon to be analyzed is the influence of plosives.

illustration not visible in this excerpt

Fig. 5.5.4: Distribution of vowels in the environment of a nasal in a dense area

All shape clusters of plosive environments in the chart below (cf. fig. 5.5.5) show a rather broad spread on the right-hand side of the cluster, i.e. in the back, and become relatively narrow as they approach to the front. Or, to formulate their shape the other way round, the phonetic environment of plosives is rather rare in the front and shows a broader spread as the vowels move backwards. As the front and central vowels, i.e. /ɪ/, /ɛ/ and /ʌ/ shift backwards, the density of the plosive environments is increased. In this context it is worth to consider that /ʌ/+P and /ɑ/+P almost touch the values of /ɔ/. At the same time, as the back vowel /ɑ/ approaches the front and /æ/ is being raised and fronted, the phonetic environments become less with increasing F2 values. This suggests that in the case of Jake plosives have a backing influence on the sounds they are following, i.e. they are dragging front sounds backwards or often keep some values of fronted sounds in the back. This tendency was already pointed out in the vowel grids of Merryl and Philipp. This suggests that the backing influence of plosives is not an individual variation, but may rather be seen as a general tendency which needs to be proved or disproved by further research and analyses.

illustration not visible in this excerpt

Fig. 5.5.5: Distribution of vowels in the environment of a plosive in a dense area

illustration not visible in this excerpt

Fig. 5.5.6: Distribution of vowels in other phonetic environments in a dense area (A=Affricate, F=Fricative, L=Lateral)

As clearly displayed in figure 5.5.6above, in Jake’s pronunciation the phonetic environment of a fricative has a backing influence on the front and central vowels /ɪ/, /ɛ/ and /ʌ/. Although the most fronted values of /ɛ/, i.e. the ones that are followed by approximants, lie in the range of /ɛ/+F, some values that are followed by a fricative already merge with the ellipsis of /ʌ/. The latter shows a clear backing movement in combination with fricatives, as well.

The only phonetic environment that has an even stronger backing tendency is the one of laterals, which causes /ɛ/ to move backwards and impedes /ɑ/ to shift forward. In contrast, note that the most fronted value of /ɔ/ is followed by a fricative as well, which would disprove the argument of fricatives as backing influences. The same contradiction occurs with affricates, as the values of /æ/+A are situated rather remote from the front, i.e. already in the range of the backed and lowered /ɛ/, while the most front value of /ʌ/ is followed by an affricate. Unfortunately, a distinct position of /æ/+F could not be established. Thus, a clear distinction cannot be made and requires further data collection and analysis.

In sum, Jake’s short vowel realization clearly reveals a considerable amount of features assigned to the NCS and, thus, is indisputably affected by it. Since all six short vowels are involved, the distribution of phonetic environments which favor or disfavor the shift is more advanced than with the speakers analyzed in the previous chapters. However, not all phonetic environments’ influences can be identified as the sound change is not yet completed.

5.6 Patrick, 18m

illustration not visible in this excerpt

Fig. 5.6.1: Vowel charts of Patrick (18m) displaying vowel envelopes and mean values

All extreme values and calculated figures of Patrick are as follows:

illustration not visible in this excerpt

Fig. 5.6.2 Extreme values, mean values and percentages of divergence of Patrick (18m)

In contrast to the other three male participants, Patrick’s oral cavity (cf. fig. 5.6.1) shows a slightly wider shape with an F2 range of approximately 1600Hz (ranging from 734.23Hz to 2392.23Hz). However, his F1 values spread on a scope of only 400Hz, suggesting that high and low vowels most likely are situated very closely to each other. This assumption is supported by the shape of the mean value chart. It is not trapezoidal, but rather strongly resembles the mean vowel charts of Philipp and Jake, i.e. it looks like the model depicting the circular vowel movement of the NCS (cf. fig. 3.2.1). Hence, all six short vowels displayed in the grids in figure 5.6.1 are obviously affected by the NCS. However, the individual vowel clusters show some peculiarities which have not occurred with the other speakers, yet.

As the majority of the informants, Patrick produces an egg-shaped ellipsis of /æ/. It spreads along the F1 axis in a diagonal upward movement. This leads to overlaps with the vowels envelopes of /ɪ/ and /ɛ/ which are pushed by the shift of /æ/, but also to marginal overlaps with /ʌ/ and /ɑ/ since the latter is dragged forward into the vowel space that is left by /æ/. The percentage of divergence of the lowest F1 for /æ/ is 71.13% in comparison to the GA average (100%), indicating that it may already have moved upwards, if other phonetic environments than the ones occurring in the New Yorker short- a pattern are involved. Furthermore the values of /æ/+N are situated even higher and overlap even more with the ellipsis of /ɪ/.

Owing to the NCS, the shift of /æ/ drags the ellipsis of /ɑ/ forward. It starts in the low back corner of the vowel chart and changes its position towards the low front corner in a horizontal movement along the F2 axis without any diagonal tendency. However, the scope of the ellipsis resembles the one of /æ/ in the sense that although a clear direction of the shift is visible, some values are also situated relatively high in the central area of the vowel chart. Nevertheless, its percentages of divergence from the lowest F2 (39.33%) indicates that it considerably has been dragged forward already.

As /ɑ/ is fronted, it vacates a vowel space that is taken over by /ɔ/. The ellipsis of Patrick’sCLOTH-sound is relatively small and egg-shaped. In contrast to all other participants’ /ɔ/ clusters, it shows a clear lowering with a diagonal backward movement instead of a fronting. Thus, it indicates the tendency to fill the vacated vowel spot instead of following /ɑ/ to the front.

The ellipsis of /ɪ/ is the first one in the present study to be of a very wide and relatively narrow shape. It indicates, however, the further tendency, as it will become clear in the analyses of the next two informants. The vowel cluster shows a strong fronting while it is also located far back, which causes its wide spread. Its percentage of divergence from the lowest F2 indicates a backwards movement of over 30% in comparison to the GA average, but its F1 divergence still testifies a very high position. Owing to the raising of /æ/, /ɪ/ overlaps with the highest values of the former. In addition its own backward movement and slight lowering lead to a collision with the highest values of /ɛ/.

The ellipsis of /ɛ/ is pushed away from its former position because of the upward shift of /æ/. It has a very wide and even flatter shape than the ellipsis of /ɪ/ and shows a horizontal backward movement with some lowering. Interestingly, it moves so far back that it already marginally merges with the ellipsis of /ɔ/ and on its way towards the back incorporates all realizations of /ʌ/. In some informants’ speech the latter sometimes has been incorporated by /ɑ/, but never by /ɛ/ before. The percentage of divergence of the lowest F2 for /ɛ/ (56.11% in comparison to the GA average of 77%) numerically confirms the suggestion that the vowel has strongly shifted backwards.

The ellipsis of /ʌ/ is small and almost round, indicating that the slightly visible backward movement has only started and at the same time was passed and overtaken by /ɛ/. It also merges with the high front realizations of /ɑ/ as it moves forward and /ʌ/ is lowered while moving backwards. Interestingly, Patrick’s realization of the STRUT-vowel is situated relatively high, since its highest and most fronted values range into the ellipsis of /ɪ/ and its percentages of divergence from the lowest F1 (51.28% suggest the same). This collision will be eliminated as /ʌ/ shifts further back. However, it is interesting to see where exactly it will shift since the central position is already taken over by /ɛ/ and the mid-back is still occupied by /ɔ/, but since the latter already indicates a lowering, its movement may drag /ʌ/ backwards in order to fill the vacated space.

As it is the case with the majority of the other speakers analyzed in the present thesis, Patrick’s vowels mainly accumulate in a central area which makes the analysis of the overlapping vowel environments interesting.

illustration not visible in this excerpt

Fig. 5.6.3: Vowel chart of Patrick (18m) displaying phonetic environments in a dense area (F=Fricative, N=Nasal, P=Plosive)

Patrick’s distribution of phonetic environments (cf. fig. 5.6.3) represents the general tendency which was already pointed out in the other speakers’ analyses, i.e. the fronting and raising influence of nasals, the raising tendency of /r/ and fricatives, and the tensing influence of plosives unusual for the New York short- a pattern.Note that the vowel environments of /ʌ/ are not depicted since the vowel does not show enough movement, yet to provide a clear distribution of vowel environments which favor or disfavor its shift.

In sum, Patrick’s vowel grid indicates that his speech is clearly affected by the NCS and already shows all features of it, although the shifting of /ʌ/ has just started and all other movements are still in progress. Nevertheless, the depictions of the informant’s mean values (cf. fig. 5.6.1) clearly resemble the NCS model with its vowel positions and shifting directions (cf. fig. 3.2.1). Interestingly, this shape of the mean vowel chart is only found in the realizations of the male participants while the female informants’ vowel grids do not resemble the NCS model yet or not anymore, as it will be seen below in the analyses of the two younger female speakers.

5.7 Samantha, 17f

illustration not visible in this excerpt

Fig. 5.7.1: Vowel charts of Samantha (17f) displaying vowel envelopes and mean values

All extreme values and calculated figures of Samantha are as follows:

illustration not visible in this excerpt

Fig. 5.7.2 Extreme values, mean values and percentages of divergence of Samantha (17f)

Just like the older female participants, Samantha’s oral cavity has a wide scope (cf. figure 5.7.1), the range of F2 spreading over approximately 2200Hz (between 661.19-2628.23Hz) and the range of F1 opening up to almost 1100Hz (between 231.99-1308.86Hz). Hence, the distances between her vowels are likely to be very long, which makes vowel shifts easy to identify as the movements have to be performed over a wide space. As with all other younger participants whose speech was affected by the vowel chain shift, Samantha’s mean value depiction has lost its standard trapezoidal shape. However, it does not resemble the NCS vowel positions and movements, either. An analysis of her vowel clusters may provide explanations for this phenomenon.

Samantha produces an ellipsis for /æ/ that is very wide and relatively flat. It spreads along the F1 axis in a diagonal upward movement from a mid-low position to the high front corner. It slightly overlaps with the ellipses of /ɑ/ and /ʌ/ in its low back corner and with the ellipsis of /ɪ/ in the high front region. Moreover, it overlaps with more than half of the ellipsis of /ɛ/. This phenomenon will be explained when the latter vowel is analyzed. Interestingly, the majority of the values of /æ/ shows an upward movement from the bottom to a central F1 position; there is a space in the upper middle where no value is produced and then they occur again in the upper area of the vowel chart and collide with the ellipsis of /ɪ/. Except for one value that is placed in the space where no other /æ/ vowels are produced, all realizations of /æ/ in the phonetic environment of a nasal occur in the high front part of the oral cavity and overlap with /ɪ/. Besides their raising, they also show a strong fronting. /æ/ shows 69.66% (average 100%) percentage of divergence from the lowest F1 and 59.74% (average 67%) from the lowest F2. These figures support the fact that the vowel position of /æ/ has been fronted and raised owing to the NCS. As it shifts upwards, it pushes the /ɛ/ and /ɪ/ forward, which is why these two vowels shall be analyzed next.

The ellipsis of /ɪ/ is very long and the narrowest of all /ɪ/ ellipses produced by the eight participants. It merges with /æ/+N and /æ/ in some other phonetic environments in the upper left-hand side (i.e. high front) of the vowel chart and, hence, as is the case with every /ɪ/ affected by the NCS, has started shifting backwards, in order to maintain its vowel distinction. Its percentage of divergence from the lowest F2 (73.69%) confirms the existence of a backward movement andits percentage of divergence from the lowest F1 26.76% indicates that /ɪ/ has already been slightly lowered.

The ellipsis of /ɪ/ slightly overlaps in its lower area with some high values of /ɛ/ whose ellipsis resembles the one of /æ/. The /ɛ/ ellipsis starts in the upper front corner in the range of /ɪ/ and then moves down and backwards in a strong diagonal movement. As it is lowered, it merges with the rising ellipsis of /æ/ and at its lowest end point overlaps with /ʌ/ and /ɑ/. The position of its mean value and the percentages of divergence, i.e. 48.28% from the lowest F1 and 61.37% from the lowest F2, prove that it has moved backwards considerably and was also lowered.

The shifts of /æ/, /ɪ/ and /ɛ/ described can clearly be traced back to the NCS. A more difficult situation arises with the other three short vowels possibly affected by the shift.

The ellipsis of /ɔ/, which is relatively short and narrow, starts at the upper right side of the vowel chart and spreads along the F1 axis in a horizontal lowering and fronting movement towards /ɑ/ and /ʌ/. Its starting point is situated very high in the vowel grid, i.e. on the F1 level of /ɪ/ and its mean value is still higher than the average, which is proved by the relatively low percentage of divergence from the lowest F1 (42.40%) in comparison to the GA average (56%). Nevertheless, it performs a clear fronting and lowering movement, which causes /ɔ/ to overlap with /ʌ/ and /ɑ/ and almost touch the displaced ellipsis of /ɛ/. The shift of the /ɔ/ vowel cluster is most likely to have been dragged by the fronting movement of /ɑ/ which, in turn, had been dragged by the movement of /æ/ before. This chain reaction describes the first two stages of the NCS and is clearly noticeable in the movement of the vowel clusters. However, although /æ/ vacates a space which should drag /ɑ/ forward into the low left-hand corner of the chart and although /ɔ/ is clearly dragged by the movement of /ɑ/, the shift of Samantha’s LOT-vowel has taken a surprising direction. The wide, almost round ellipsis and the mean value of /ɑ/ clearly indicate a fronting. However, they also display a considerable raising, which causes a merging of /ɑ/ with the lower back parts of /ɛ/ and the lower front parts of /ɔ/ and an almost complete incorporation of the ellipsis of /ʌ/ by the ellipsis of /ɑ/. Furthermore, it overlaps with the ellipsis of /æ/ marginally. Its percentages of divergence, i.e. 67.62% from the lowest F1 and 39.49% from the lowest F2 also prove the fronting of /ɑ/ as it occurs in the NCS, but at the same time indicate its considerable raising. Hence, instead of taking over the vacated space of /æ/, /ɑ/ moves upwards to the position of /ʌ/.

At the same time, the STRUT-vowel now loses its vowel space because of the shifts of /ɛ/, /ɔ/ and /ɑ/ towards its position. Since /ɑ/ completely invades the vowel position of /ʌ/, the latter is forced to move away in order to stay distinct from /ɑ/. As the narrow and wide ellipsis of /ʌ/ shows, it moves to the front. Thus, instead of performing the usual NCS direction towards /ɔ/, which would only lead to a collision, /ʌ/ also avoids merging with the raised and fronted /ɑ/ by taking over the shifting features of /ɑ/ and moving towards the empty vowel space vacated by /æ/. This theory is supported by the position of the mean values of /ɑ/ and /ʌ/. They clearly depict the position of /ɑ/ as being in the exact space of GA /ʌ/. It is situated even higher than the mean value of the lowered /ɛ/. The mean values of /ɑ/ show a percentage of divergence from the lowest F1 of 67.62% and 39.49% from the lowest F2, in comparison to the average GA percentage of divergence of /ʌ/, i.e. F1 56% and F2 27% and the average GA percentage of divergence of /ɑ/, i.e. F1 100% and F2 0%. Consequently, Samantha’s /ɑ/ is located much closer to the GA position of /ʌ/ than to its own average. Apparently, /ʌ/ and /ɑ/ have not only switched positions, but also have taken over the shifting characteristics of each other. If this is the case, Samantha’s pronunciation shows all features of the NCS, only that /ʌ/ and /ɑ/ have inverted roles.

In the following, a brief analysis of the phonetic environments in the dense areas shall be given. As it becomes clear in the vowel chart below (cf. fig. 5.7.3), the environments overlap too much and need to be separated from each other in order to depict a clear picture. Hence, first the environment of nasals was extracted.

The environment of a nasal (cf. fig. 5.7.4) shows a clear fronting and raising tendency in combination with /æ/, /ɪ/ and /ɛ/. It should be noted that one of the most fronted values of /ɔ/ and /ɑ/, respectively, are followed by a nasal. Usually phonemes overlap in different allophonic distribution, which still allows a clear distinction. However, as depicted in figure 5.7.4 below, the realizations of Samantha’s /æ/, /ɪ/ and /ɛ/ overlap in the same allophonic distribution, i.e. in the environment of a nasal, which brings up the question how the three vowels can be distinguished in this particular position. This issue will be discussed in chapter 6, where the developments of the specific vowels over time are being discussed. In contrast to the other participants of the present study, plosives do not favor tensing in Samantha’s speech. It should be noted, for instance, that /ɛ/+P lies as far back as the values which are followed by a lateral, which clearly favors backward shifting.

illustration not visible in this excerpt

Fig. 5.7.3: Vowel chart of Samantha (17f) displaying phonetic environments in a dense area (Ap=Approximant, F=Fricative, N=Nasal, P=Plosive)

Furthermore, /ɛ/ has shifted that far that it overlaps with the most fronted values of /ʌ/ which occur in the environments of fricatives or approximants. This brings up the question how these two environments are distributed in the speech of Samantha. Interestingly, some fronted values of /ɔ/ are also followed by a plosive, which again suggests that plosives have a centralizing tendencyat least.

illustration not visible in this excerpt

Fig. 5.7.4: Distribution of vowels in the environment of a nasal in a dense area

illustration not visible in this excerpt

Fig. 5.7.5: Distribution of vowels in the environment of a plosive in a dense area

illustration not visible in this excerpt

Fig. 5.7.6: Distribution of vowels in other phonetic environments in a dense area (Ap=Approximant, F=Fricative)

As mentioned above, an approximant, i.e. /r/ clearly favors raising and fronting. This can be seen in the chart above (cf. fig. 5.7.6) when the positions of /ɑ/, /ɛ/ and /ɔ/ in the environment of /r/ are analyzed. It is important to note that one of the highest and most fronted values of /ʌ/ is followed by /r/ and that the cluster of /æ/+Ap with the approximant always being /r/ depicts the only 4 values of /æ/ which merge with the ellipsis of /ɪ/. Apparently /r/ favors raising and fronting. Again, laterals, as depicted above in the combination with /ɛ/, have a backing influence. The same seems to be true for fricatives which, however, do not initiate strong backward movements but rather have a centralizing tendency.

In sum, Samantha’s vowel realization shows all vowel changes of the NCS if the theory of /ʌ/ and /ɑ/ switching positions and characteristics is taken into consideration. Furthermore, the basic influential tendencies of certain vowel environments, such as nasals, plosives, laterals and specific approximants, can clearly be recognized.

5.8 Allie, 17f

illustration not visible in this excerpt

Fig. 5.8.1: Vowel charts of Allie (17f) displaying vowel envelopes and mean values

All extreme values and calculated figures of Allie are as follows:

illustration not visible in this excerpt

Fig. 5.8.2 Extreme values, mean values and percentages of divergence of Allie (17f)

The scope of Allie’s oral cavity resembles the one of the other female informants. It is much wider than the oral cavities of the four males analyzed, its F2 having a considerable depth of 2200Hz (F2 between 995.01 and 2938.37 Hz) and a height ranging from F1 391.69Hz to 1246.96Hz, i.e. spreading over a scope of almost 900Hz. While the other three females had an F2 maximum between 2461.97Hz (Annette) and 2628.23Hz (Samantha), Allie’s highest F2 lies at almost 3000Hz (2938.37Hz), i.e. extremely far front. Apparently, the scope of the oral cavity’s area used for pronunciation has not changed much over the years, but the places of articulation have been moved slightly forward, in general. Interestingly, the depiction of Allie’s mean values almost resembles the trapezoidal shape of a GA vowel chart. However, as the analyses of each single vowel cluster will show, some astonishing changes are in progress.

Allie produces an /æ/-vowel, whose ellipsis is the smallest /æ/-realization for all female informants and resembles the shape of Jake’s ellipsis, which is the widest of all male participants. The ellipsis is egg-shaped and stretched, shows a slight diagonal adjustment and spreads vertically along the F1 axis. Interestingly, it does not overlap with /ɪ/ anymore, and only marginally collides with /ɑ/ and /ʌ/. However, as was the case with Samantha, more than half of the ellipsis overlaps almost the whole vowel envelope of /ɛ/, which will be discussed in detail when dealing with the latter. Moreover, the majority of the values of /æ/+N are situated exactly in the ellipsis of /æ/, except for three single realizations. These realizations are the pronunciations of thank you. They reach very far front (F2 maximum 2938.37Hz) and very high up (F1 maximum 445.33Hz), which is why they merge with the ellipsis of /ɪ/, whose extreme values are only slightly more fronted and bit higher than the three realizations of /æ/+N.

Surprisingly, although Allie is the youngest participant in the present study and her NCS is very likely to have reached another stage or developed further, e.g. her /æ/ may have left its low position completely, the mean value of /æ/ is the lowest, but at the same time most backed one of all eight informants. This is supported numerically by the percentages of divergence which are 70.27% from the lowest position of F1 (100%) and 41.51% from the most fronted position of F2. This means that Allie’s realization of the TRAP-vowel is situated lower and farther back than for all the other participants. This also applies for /æ/+N, which, except for the three extremely high and front realizations of /æɳ/, is located almost as low as /æ/ in other environments and does not show much fronting, either. Because of the extremely wide spread of the /æ/+N ellipsis, its mean value is situated at a position that once belonged to /æ/ in the vowel trapezoid. It is neither clear, whether /æ/ is still moving upward nor why it is so much in the center nor why only some of the /æ/+N-values have spread that high and front. However, the new position of /æ/ suggests that the New York City short- a pattern is omitted or no longer existent.

In order to give a more structured analysis of the complicated changes that are in progress in Allie’s speech, the order of the vowels analyzed is changed in the following. Allie’s ellipsis of /æ/ strongly overlaps with the ellipsis of /ɛ/. The latter has a similar shape as the former; it only is slightly more narrowed, and spreads vertically along the F1 axis without any diagonal tendencies. Interestingly, its highest F2 values are more fronted in the vowel chart than the values of /æ/ and /æ/+N, when the three high front realizations of /æɳ/ are not taken into consideration. /ɛ/ overlaps with /ʌ/ only marginally. Furthermore, it is strongly lowered, which leads to /ɑ/ marginally overlapping with /ɪ/.The percentage of divergence of /ɛ/ shows that it is located even slightly higher (48.63%) than the GA average (56%), but also considerably more back (45.11% in comparison to the GA average of 77%). Hence, owing to the NCSs, /ɛ/ has moved backwards and now performs a strong lowering movement towards the bottom of the vowel chart. It has shifted so far that it only overlaps with the lowest and most back values of /ɪ/, but surprisingly does not really merge with /ʌ/, although it has moved very close to it.

The vowel cluster of /ɪ/ is very wide and adjusted horizontally, but not as narrow as Samantha’s ellipsis. It is overlapped by the high front values of /æ/+N from its high front part to its middle. Its back values overlap with the highest ones of /ɛ/. In addition, its percentage of divergence shows that it is situated very high (F1 17.09%), but its mean value has shifted considerably backwards (F2 66.35% in comparison to the average 100% divergence from the lowest F2). Its position is not unusual and resembles the general positions of /ɪ/ in the other seven informants’ pronunciation. Nevertheless, its vowel envelope shows a strong backward movement and nearly overlaps with the highest and most fronted values of /ʌ/ and /ɔ/ while it has the most fronted values of all eight informants at the same time. This tremendous spread is very peculiar. Apparently, /ɪ/ has not yet been lowered much by the NCS and only shows a slight tendency towards it. However, while it is still extraordinarily fronted a tendency to shift backwards is apparent already.

As mentioned above, the lowest back values of /ɪ/ almost overlap with the highest front values of /ʌ/. The ellipsis of the latter has an egg-shaped ellipsis that indicates strong lowering and a slight backward movement. The majority of its values are situated in its front part, i.e. very central. Although /ɛ/ has already shifted backwards, /ʌ/ only overlaps with it marginally. Hence, /ʌ/ might have shifted backwards, as well, in order to avoid the collision with /ɛ/. This push-chain movement is one of the last steps known from the NCS. However, the percentage of divergence of /ʌ/ shows that the vowel is situated almost at its ideal standard American spot. Hence, /ɛ/ has shifted backwards, but only to the extent that it has eliminated the distance between the DRESS- and the STRUT-vowels and does not overlap with the latter. The backward tendency of the /ʌ/ cluster may indicate a further backward movement to come, which would allow /ɛ/ to shift even further back. However, so far /ʌ/ does not seem to have moved much. Another suggestion is that maybe /ʌ/ was pushed backwards by the movement of /ɛ/, then collided with /ɔ/ and /ɑ/ and moved downwards to avoid s further merging. During this last fronting, it did not move much further than to its original spot and /ɛ/ was adjusted as far back and as close to /ʌ/ as possible, but as distinctive as possible at the same time. This theory suggests that the lowest F2 value of /ʌ/ does not drag the vowel backwards, but is rather dragged forward again, because it collided with the fronted /ɔ/.

As mentioned above, the horizontally placed ellipsis of /ɔ/ shows a fronting towards /ʌ/. Furthermore, it is situated relatively high, its percentage of divergence from the lowest F1 being only 29.84% in comparison to the GA average of 56%. Its mean value still has a very backward position, but the ellipsis tends to a clear fronting and lowering. Considered that the majority of the /ʌ/ values are situated in an area, where they overlap neither with /ɛ/ nor with /ɔ/, suggests that /ʌ/ tries to avoid collision with either vowel, while both are shifting into its direction. Hence, /ʌ/ may soon be incorporated by them, which might result in merging or a continued shift, where /ʌ/ needs to find a new vowel spot, since it also overlaps with /ɑ/ at the bottom of its ellipsis.

The vowel cluster of /ɑ/ is relatively small and spreads horizontally along the F2 axis. It only marginally overlaps with /ɔ/, which is too high to fully merge with /ɑ/. Since the ellipsis of /ɑ/ shows a forward movement, it must have been dragged by the fronting and raising of /æ/. Its percentage of divergence from the lowest F2 is 22.35% more fronted than the average (0%), but its F1 divergence of 64.76% indicates that it is situated considerably higher than the GA average (100%). This extremely high position leads to a collision with /ʌ/ as /ɑ/ is being fronted.

So far it has become obvious that all vowels except /ɪ/ and some values of /æ/+N have shifted in a centralizing manner and have started to incorporate all realizations of /ʌ/, instead of shifting in a circular movement as it is otherwise known from the NCS. So far, Allie’s vowel distribution does not show a regular pattern. An analysis of the phonetic environments in the central area may provide some answers.

illustration not visible in this excerpt

Fig. 5.8.3: Vowel chart of Allie (17f) displaying phonetic environments in a dense area (Ap=Approximant, F=Fricative, L=Lateral, N=Nasal, P=Plosive)

The influence of a nasal environment (cf. fig. 5.8.4) is clear in the case of /æ/+N, which shows a strong raising and fronting, /ɪ/, which is situated very high up and fronted when followed by nasal, and /ɑ/, whose most fronted values are the ones which are among other followed by nasals. A more striking phenomenon occurs with /ɛ/, whose highest F2 values lie in the area of the highest F2 values of /æ/, but not when in a nasal environment. Quite the contrary, in the phonetic environment of a nasal, /ɛ/ is advanced even more backward even further than in the environment of a lateral which, as mentioned in the chapters above, has a backward influence. Furthermore, /ɛ/+N even merges with the low front values of /ʌ/ and is closely situated to /ɑ/. Apparently, the environment of a nasal does not necessarily have a fronting influence in Allie’s pronunciation. However, it still favors raising, since some of the highest F1 values of /ɛ/ are followed by a nasal.

illustration not visible in this excerpt

Fig. 5.8.4: Distribution of vowels in the environment of a nasal in a dense area

illustration not visible in this excerpt

Fig. 5.8.5: Distribution of vowels in the environment of a plosive in a dense area

As already pointed out in the analyses of other speakers’ phonetic environments, plosives often favor raising or tensing of a vowel. However, Allie’s values of /æ/ show a central but vertical position when followed by a plosive, so that they collide with some front values of /ɑ/. The same applies for /ɪ/ and /ɛ/, which are rather positioned in a central F2 position when followed by a plosive, so that they then collide with /ʌ/. /ɛ/+P is located even further back than /ɛ/ followed by a lateral. However, /ɑ/+P is situated in a relatively central position compared to the lowest F2 value, which is followed by a lateral.

illustration not visible in this excerpt

Fig. 5.8.6: Distribution of vowels in other phonetic environments in a dense area (Ap=Approximant, F=Fricative)

Figure 5.8.6 shows that in Allie’s pronunciation vowels which are followed by a fricative have a central position with a slight fronting and raising tendency.Apparently, the approximant /r/ has a fronting influence again, as can be seen in combination with /æ/ and /ɑ/ in the chart above.

In Allie’s speech, only the basic features of the NCS could be established, i.e. the absence of mergers and the raising and fronting influence of a nasal environment. Moreover, the assumption that plosives favor tensing or at least centralizing movements was supported. Surprisingly, the depiction of the informant’s mean values resembles the GA trapezoidal vowel chart with a slightly centralized /ɛ/ and /æ/+N being located at the former position of /ɛ/ rather than the NCS model. Since Allie is the youngest speaker analyzed in the present thesis, it is assumed that her vowel changes are the most advanced ones. Her realization of /æ/ does not reveal the New York City short- a pattern anymore and also no raising in other phonetic environments that might have been driven by the NCS. In addition to this centralization, the young interviewee reveals an affinity to high vowels, which may be the reason why /ɪ/, /ʌ/ and /ɔ/ are positioned unusually high up.

Nevertheless, the changes in Allie’s speech are too severe and still too much in progress to establish regularity in her vowel movements or in the distribution of her phonetic environments.

6 Vowel Analysis

After all eight speakers have been analyzed separately with respect to their realizations of the vowels /æ/, /ɑ/, /ɔ/, /ɪ/, /ɛ/ and /ʌ/, the results from the last chapter shall be compared in order to discuss their coherence as well as a general movement tendency. Note that because of the close relation of all six short vowels and /æ/+N, the traditional distinction between the lower elements, i.e. /æ/, /ɑ/ and /ɔ/ and the upper elements, i.e. /ɪ/, /ɛ/ and /ʌ/, was not made. Hence, they are not treated as individual chain shifts with possible links between their respective elements, but rather as members of an entity, i.e. a speaker’s vowel system.

6.1 The TRAP-Vowel

The TRAP-vowel, i.e. the sound assumed to have initiated the NCS, has a complex realization in the speech of the eight speakers analyzed in the present study. Not only does it have a very complex allophonic distribution in New York City English, but it also is affected by the NCS, which makes a distinction between dialect features and NCS influences a challenging task. Since the short- a pattern of New York City has already been described in chapter 3.5, it shall now be analyzed how /æ/ is realized by each of the eight informants and in which direction it generally tends to shift. Moreover, it shall be established which phonetic environments might favor tensing, i.e. stress, raising and/or fronting and whether the impact of a specific environment can be traced back to the accent of NYC or to the influence of the NCS. In addition, a distinction between /æ/ and /æ/+N was drawn because of the strong raising and fronting influence of nasals on the TRAP-vowel (cf. e.g. Oomen 1982: 108). Therefore, the values of /æ/+N were extracted separately, and their mean values as well as their percentage of divergence were calculated, although GA does not provide an ideal position of /æ/+N in the vowel chart. The results, their analysis and the discussion of their relation to /æ/ in other environments are presented in a subchapter.

6.1.1 The TRAP-Vowel Preceding a Non-Nasal

In General American (GA), the TRAP-vowel is situated in the low front corner of the trapezoidal vowel chart (cf. fig. 4.4.4) and according to this position is given the following percentages of divergence:

F1 low – 100%, F2 low – 67% (cf. fig. 4.4.5)

Compared to these standard positions, the speakers’ values given in figure 6.1.1.1 and their vowel clusters depicted in figure 6.1.1.2 show a considerable divergence.

illustration not visible in this excerpt

Fig. 6.1.1.1: Percentages of divergence of /æ/

illustration not visible in this excerpt

Fig. 6.1.1.2: Ellipses of all speakers covering the phones of /æ/ based on their percentages of divergence

First, the average percentage of divergence from the lowest F1 (64.68%) indicates that all eight speakers pronounce /æ/ considerably higher than in GA (100%). The range of only 16.14%, the lowest F1 position in the oral cavity being Patrick’s 71.13% and the highest F1 position being Philipp’s 54.99%, indicates an unconscious agreement among the speakers. This common sense is their New York City accent with its complex short- a pattern, which raises /æ/ in a number of phonetic environments (cf. chapter 3.5) and leads to an overall high percentage of divergence of the TRAP-vowel from the lowest point in the oral cavity. Interestingly, a male person (Philipp) raises /æ/ the most. In addition, the female average shows a higher percentage of divergence from the lowest F1, i.e. less /æ/ raising then the male participants. This fact contradicts the common linguistic assumption that generally “vowels are farther from the center of their respective vowel space in the women’s data than they are in the men’s” (Gordon&Heath 1998: 431). Nevertheless, the least raising is also found in the speech of a male speaker (Patrick). Another striking fact is that the percentages of divergence from the lowest F1 increase as the informants get younger. The older speakers’ realization of /æ/ is relatively high (62.21% divergence) and the younger speakers’ pronunciations indicate a lower position of /æ/. Note that Jake has one of the lowest percentages of divergence and rather pronounces his TRAP-vowel as the generation over 40. All teenagers, however, show a considerable lowering (between 69.66% and 71.13% of divergence from the lowest F1). This is an unexpected phenomenon, since the younger generation should not only produce the New York short- a pattern, but it should also show further developed features of the NCS and, thus, a much more raised /æ/. Instead, this generation has lower /æ/ positions than the older speakers. This indicates that the TRAP-vowel has been lowered instead of raised.

illustration not visible in this excerpt

The speakers’ average percentage of divergence from the lowest F2 (62.61%) indicates that it is pronounced slightly more back than in GA (67%). However, the F2 range of 30.23%, with the most backed F2 position of /æ/being Allie’s 41.51% and the most fronted being Philipp’s 71.74%, indicates that speakers’ F2 realizations differ considerably. Interestingly, the male informants of the present study pronounce /æ/ slightly too fronted (65.75%) while females realize it slightly too far in the back (59.48%) of their oral cavity. Furthermore, the younger the participants are, the more the percentage of divergence from the lowest F2 is decreased, i.e. the more /æ/ is moved backwards. As a consequence of the gender and age averages, the most backed F2 position of /æ/ belongs to the youngest female speaker (Allie). The wide range of F2 divergence suggests an insecurity concerning the horizontal position of /æ/ among the speakers.

For better comparability, the F1/F2 values of each phone were transformed from Hz into percentage of divergence analogous to the values in figure 6.1.1.1. Accordingly, figure 6.1.1.2 shows all informants’ ellipses of /æ/. Considering the shapes of the vowel envelopes, note that their shapes and spreads are gender-specific rather than age-related. The males realize /æ/ slightly too fronted and a litte too raised. Their vowel envelopes cover a relatively small range and sometimes even have a round shape. This indicates a consistency in the fronting and raising of /æ/. In contrast, the females’ clusters are much wider and cover a broader range of the oral cavity, their vowel realizations spreading from very high front to low central positions.

The phonetic environment of /æ/ is very complex and can only be described in approximation. First, the generation over 40 clearly shows the presence of the New York City short- a pattern, i.e. with nasals (cf. chapter 6.1.2), voiceless fricatives and voiced plosives favoring tensing. However, some high front values are already followed by voiceless plosives, /r/ or voiced fricatives. A vowel chart representative for these facts was selected from the eight speakers and is given in figure 6.1.1.3 below.

illustration not visible in this excerpt

Fig. 6.1.1.3: Ellipsis of /æ/ including its phonetic environments (Philipp, 48m)^[29]

The allophonic distribution of /æ/ shows clear differences in the speech of the younger generation. In the pronunciation of the speakers aged between 17 and 24 most phonetic environments have started moving backwards, leaving nasals (cf. chapter 6.1.2) and /r/ as consonants favoring raising the most. Some high values still show the environment of voiced fricatives and voiceless plosives. However, the tensing influence of voiceless fricatives as a feature of the New York City short- a pattern is almost non-existent. Quite the contrary, the younger the speakers are, the more voiceless fricatives seem to be placed in the back part of the vowel ellipsis.

The spread of the vowel ellipses of /æ/ indicate that women apparently share a higher flexibility concerning the position of the vowel. Men raise the vowel in almost every environment, indicating that they do not only have the short- a pattern, but are also affected by the NCS. Since all female speakers raise /æ/ in environments which do not follow the tensing rule in the New York short- a pattern, they most likely are affected by the NCS, as well. It is not clear whether the wide range of their vowel envelopes can be traced back to linguistic insecurity or hypercorrection, i.e. the exaggerated lowering and backing of vowels in some environments in order to avoid too much tensing evoked by the short- a pattern which may be considered unprestigious by female speakers. This would suggest that the New York City short- a pattern takes place at a conscious linguistic level. Interestingly, the younger the informants, the lower and more backed their /æ/ realizations are. However, there are two outstanding exceptions. The TRAP-vowel is still raised in the environment of nasals and, surprisingly, when it is followed by /r/. Note that the tensingrule of New York City short- a pattern excludes /r/ (cf. Labov, Yaeger&Steiner 1972: 66). Hence, the raising in the environment of /r/ must be a feature of the NCS. If the analysis of the nasal environments in the next chapter reveals that the younger informants’ /æ/+N realizations are especially raised when followed by a velar nasal (/ɳ/) which also does not favor tensing in the NYC short- a pattern, it is occurring as a result of the NCS.

6.1.2 The TRAP-Vowel Preceding a Nasal

Since GA does not provide an average percentage of divergence for /æ/+N, the speakers’ results can only be set in relation to each other and the standard position of /æ/ in the GA vowel chart. Without doubt, the position of /æ/+N will be situated considerably higher than /æ/ in other environments, for nasals favor raising and fronting in both the NCS and the short- a pattern of New York City. The raising may occur to the extent that /æ/ is diphthongized in order to maintain a distinction between the KIT-vowel and the TRAP-vowel. Kurath and McDavid, whose results are based on the influence of New England folk speech, describe the realization of /æ/ plus nasal in can’t, stamp and hammer without any distinction from /æ/ in other environments. According totheir results, /æ+N/ moves toward [ʌ] or [ɑ], and in the case of can’t it isalso claimed to move in the direction of [ɛ]. Although /æ/ shows a clear upward movement in their data, Kurath and McDavid do not mention a diphthongization to forms like [əæ] or [ɪæ] (cf. 1969: 136ff).

The informants’ percentages of divergence were calculated as given in figure 6.1.2.1. The average percentage of divergence of F1 (40.94%) of all eight informants indicates a raising of /æ/+N of nearly 16% from the assigned spot for /æ/ and 76.15% fronting. Interestingly, males and females nearly have the same average percentage of divergence from the lowest F1 and a brief look at the single speakers reveals that in all age-groups and genders there are persons who have relatively high (Merryl, Tim, Jake, Samantha) and relatively low (Annette, Philipp, Patrick, Allie) positions of /æ/+N. In addition, although both age-groups nearly have the same percentage of divergence of F1, the highest and the lowest F1 position can be found in the younger generation (Samantha and Patrick, respectively), resulting in a range of 21%. For comparison, note that the highest F1 value of /æ/ preceding a non-nasal had a percentage of divergence of 54.99%, meaning that all percentages of divergence from the lowest F1 are lower for /æ/+N, i.e. all speakers show a clear raising in the environment of a nasal.

illustration not visible in this excerpt

Fig. 6.1.2.1: Percentages of divergence of /æ/+N

The average percentage of divergence of F2 (76.15%) of all eight informants also indicates a clear fronting of /æ/+N in comparison to the assigned spot for /æ/ (67%) in GA. There is almost no difference in gender, but there are considerable differences in age. First of all, in the generation over 40, each speaker has an F2 value around the generation average of 80.19%, i.e. all older speakers show a strong fronting and also raising of /æ/ in the environment of a nasal. In the younger generation, however, there are considerable differences in the F2 values. Among all eight speakers, the two young female participants have both the highest (Samantha: 85.53%) and the lowest (Allie: 59.87%) percentage of divergence from the lowest F2 with a range of 25.66%. The two young males’ percentages lie in between these two extreme values, so that the young participants’ percentages of divergence are staggered in gaps of 6-10% to the next speaker.

The depiction of the informants’ vowel clusters shows that /æ/+N covers the front part of the oral cavity. The ellipses of all male speakers again show a rather small range. The older participant’s vowel envelopes are situated relatively high in the front and indicate a diagonal upward and fronting movement. The younger male informants’ clusters are located in the central front, spreading horizontally along the F2 axis and showing a slightly backing tendency. In contrast, of all eight informants’ clusters the girls’ envelopes both start at the highest and most fronted points and then are lowered in a diagonal backing movement, Samantha’s ranging to the central front and Allie’s reaching to a very low central position.

illustration not visible in this excerpt

Fig. 6.1.2.2: Ellipses of all speakers covering the phones of /æ/+N based on their percentages of divergence

On the gender level, the ellipses of all female speakers are much wider and longer than the ones of the male participants. Their position is also more fronted. All four female clusters show a diagonal upward movement toward the left-hand (high front) corner of the vowel grid. The vowel envelopes belonging to the women from the generation over 40 are much smaller and less fronted and raised than the ones belonging to the two young girls. Despite the differences in size, fronting and raising, the vowel envelopes of all women nearly have the same lowest F2 value, i.e. Samantha’s envelope ends in the same location as Merryl’s on the F2 axis, whereas Allie’s vowel cluster ranges far more back, and so does Annette’s. The high position of the vowel envelopes and theirpercentage of divergence indicate that /æ/+N clearlyranges into the cluster of /ɪ/.

illustration not visible in this excerpt

Although speakers ofNew York City English tend to centralize /ɪ/, someallophones of both vowels overlap and make a distinction difficult. However, /æ/ and /ɪ/ are mostly distinguishable since /ɪ/ is realized as a monophthong,while /æ/+N is diphthongized to avoid homophony. In order to prove the diphthongization of /æ/ in the environment of a nasal, all F1/F2 formant values within the interval of /æ/+N in the pronunciation of seven speakers were extracted via Praat. Later, their curve was depicted in an Excel Grid in order to display their movement (cf. Appendix D). For a better comparison, it was made sure that the pronunciation of the same word was measured. The selected utterance was thank you, since it states the highest or one of the highest values of the younger participants and, therefore, lies deeply in the range of /ɪ/. Figure 6.1.2.3 shows the diphthongized pronunciation of /ʘæɳk/ in thank you of Philipp.

illustration not visible in this excerpt

Fig. 6.1.2.3: Diphthongized pronunciation of the /æ/ in thank you by Philipp (48m)

Since each speaker uttered the word only once, no polynomial regression was needed to calculate the tendency of the sound development curve. The depiction of one specific diphthong is not meant to be representative, but it shall rather provide an idea how the speakers’ diphthongization of /æ/+N develops when it is uttered.

With an approximate range of 1000Hz on the F2 axis, the stretch of the vowel is relatively similar throughout the majority of the speakers. The only exception is Samantha whose diphthong starts in the very front of the vowel chart, stretches backwards over a range of nearly 2000Hz and in the back even collides with the highest values of /ɔ/. The other informants’ realizations of /æ/+N also start relatively far in the front of the oral cavity and then are backed, overlapping the range of /ɪ/, /ɛ/ and sometimes /ʌ/. On the F1 level the diphthongs show greater differences, since some speakers (e.g. Philipp) start their vowel at an F1 level of 300Hz, then lower it, move it backwards and sometimes raise it again in the back, while others (e.g. Merryl) start low F1 position such as 800Hz and slightly raise their vowel during its backward movement. Speakers starting in between these extreme values often have a slight lowering as their vowel starts moving backwards and as it approaches the center of their oral cavity it is somewhat raised again.

Depending on the movement of their diphthong, each speakers ellipsis of /æ/+N is positioned and related differently to the respective ellipsis of /æ/, as depicted in figure 6.1.2.3. Cf., for instance, Samantha’s ellipsis of /æ/+N, which stretches horizontally along the F2 axis in a backward movement and is only overlapped by the lowest F1 values of /æ/. Its position and direction of movement suggests that it is clearly located in the range of the vowel envelope of /ɪ/ and that the /æ/+N diphthong also might move towards the KIT-vowel. This assumption is supported by the fact that the recording of Samantha (17f) uttering thank you is almost homophonic with * think you.

illustration not visible in this excerpt

Fig. 6.1.2.3: Overlapping ellipses of /æ/ and /æ/+N

When comparing the cluster overlaps and their movements, again note the small shapes of the male speakers’ ellipses. Moreover it is important to point out that the vowel envelopes of neither the older nor the younger generation show a discrepancy or clear distinction between the realizations of /æ/ and /æ/+N. It is difficult to determine whether these overlaps are consequences of the NCS only by analyzing the overlaps of the vowel envelopes, since the complex short- a pattern of New York City may cause these vowel overlaps, as well. Hence, cluster shapes and overlaps are not significant enough to understand the development of the NCS in New York City English and phonetic environment must be taken into consideration. The speech of all participants still shows a clear raising influence of a nasal on the TRAP-vowel. The male participants raise it in the environment of every nasal, even if it is velar, i.e. does not favor the tensing rule in the short- a pattern of New York City. Thus,the NCS has also affected their speech. Annette shows the most raising in the phonetic environment of /n/ and /m/, i.e. the two nasals favoring the tensing of short ɑ in the accent of New York City, while /ɳ/ rather is one of the lower values (cf. Appendix C). In contrast, Merryl raises and fronts all TRAP-vowels in the environment of every nasal. Moreover, the most fronted values are the ones followed by the velar /ɳ/, i.e. the consonant that falls into the tensing rule of the NYC short- a pattern. The same fact applies for all four male speakers. Hence, Merryl and all male informants are clearly affected by the NCS. However, in the speech of the younger generation, a strong separation of /æ/+N[+velar] from /æ/+N[-velar] can be identified. This process is already indicated in the speech of Jake (24m), Patrick (18m) and Samantha and is clearly developed further in the speech of Allie whose realizations of /æɳ/ lie considerably higher than all other /æ/+N. The strong raising and fronting influence of the velar nasal confirms that the NCS is clearly present in the speech of the younger generation. However, the environments of other nasals which favor tensing both in the NCS and in the New Yorker short- a pattern, are still fronted, but situated lower and further back than phonetic environments that do not favor the tensingrule of short- a, i.e. especially /r/ and /ɳ/. Apparently, the NCS has a stronger impact on the raising of /æ/ than the short- a pattern and especially affects environments which are strictly excluded by the tensing rule of short- a.

6.2. The LOT-Vowel

In GA, the LOT-Vowel is located in the low back corner of the oral cavity. However, the influence of the NCS has evoked a shift of /ɑ/ towards /æ/ in order to fill the vowel space vacated by the latter. In the standard GA vowel chart /ɑ/ has the following percentages of divergence:

F1 low – 100%, F2 low – 0% (cf. fig. 4.4.5)

The percentages of divergence of each speaker are given in figure 6.2.1 below. The average percentage of the lowest F1 divergence (65.88%) indicates that all eight informants realize /ɑ/ too high. The range between the speakers being only 16.55% for F1, the common tendency towardsa raised /ɑ/ becomes clear. This phenomenon may be connected with the changes of the New York City shift, where the back vowels are raised. On average, the female participants’ realization of /ɑ/ is 5% higher than the males’. The lowest percentages of divergence belong to Annette and Merryl and are then followed by the two young female participants. Surprisingly, Philipp’s percentage of divergence from the lowest F1 (59.18%) is almost at the same level as Annette’s (59.12%), i.e. his /ɑ/ is situated as high as the female vowel realization. Except Philipp’s high vowel position, all other male participants show much higher percentages of divergence from the lowest F1 (between 67.09% and 75.67%), i.e. their /ɑ/ is situated lower than the female realizations. In addition, note that among the eight participants the vowels of the young males have the lowest position in the oral cavity. Hence, the males in the younger generation are the ones that show the most lowering; females of both age-groups favor a higher position of /ɑ/.

illustration not visible in this excerpt

Fig. 6.2.1: Percentages of divergence of /ɑ/

The average percentage of divergence from the lowest F2 (37.69%) indicates that /ɑ/ has also been considerably fronted. With a discrepancy of less than 5 percent between the two age groups and genders, all informant groups show a general tendency to realize /ɑ/ at a more fronted position than it should be, while men even show a slightly further developed fronting and the older generation has a stronger fronting tendency. The most fronted vowel is produced by Philipp with a percentage of divergence of 45.81%, whereas the lowest percentage of divergence of F2low (22.35%) belongs to Allie.

Although the females have average positions of /ɑ/ which are situated relatively high in the oral cavity and their percentages of divergence from the lowest F2 indicate that they realize the vowel further in the back than men do, their vowel ellipses are the widest ones. Especially the two women in the generation over 40 show a broad vowel spread, horizontally or diagonally along the F2 axis towards the left-hand (low front) corner of the oral cavity. However, the shapes of the male ellipses are not small, either. They are narrower but also slightly longer than the ones produced by the females from the younger generation. Thus, males lead both in lowering and in fronting. All speakers’ ellipses horizontally or diagonally spread along the F2 axis in general and all show a clear fronting movement. Interestingly, the ellipses which reach far to the back belong to two females from both age groups.

illustration not visible in this excerpt

Fig. 6.2.2: Ellipses of all speakers covering the phones of /ɑ/ based on their percentages of divergence

Gordon (2002: 256f) provides a number of phonetic environments with influence on the shifting of /ɑ/. His claim that following interdentals and voiceless plosives favor its fronting is confirmed by the present data. However, in the present study the occurrence of /l/ clearly disfavors fronting and following voiced obstruents could not be recognized to generally favor tensing, which are both contradictions to Gordon. Moreover, he claims that /æ/ and /ɑ/ show a high degree of consistency when it comes to phonetic environments favoring their movements. As with /æ/, voiceless plosives also encourage the fronting of /ɑ/ and /r/ sometimes also favors raising but not fronting, and /l/ disfavors fronting, as it can be seen in figure 6.2.3. Furthermore, the influence of nasals shows a fronting of /ɑ/, but only a slight raising. In the younger speakers’ vowel realization, /ɑ/ has already started moving with all its environments except of /l/. Annette’s and Merryl’s ellipses did not show that much fronting, however, which brings up the question which phonetic environments have already started shifting and which did not. In the pronunciation of /ɑ/ of the two females of the generation over 40, especially voiceless plosives are being shifted towards the vowel space of /æ/. Interestingly, voiced fricatives do not show any particular influence, while nasals, the other consonant group that favors fronting and raising in the NCS, only show a very slight fronting of /ɑ/. Moreover, again the environment of /l/ disfavors the fronting movement of the older participants’ vowels, which, surprisingly, also applies for /r/. It still has a raising influence on the vowel in questions, its values being situated at the high border of the vowel envelope, but in the speech of Annette and Merryl it also sticks to the backward position of /ɑ/.

illustration not visible in this excerpt

Fig. 6.2.3: Ellipsis of /ɑ/ including its phonetic environments (Patrick, 18m)

Note that while the participants’ realizations of /ɑ/ only slightly overlap with /ɔ/, the LOT-vowel rather shows a strong centralizing and slightly lowering movement, which favors the proximity and overlap with /ʌ/.

6.3 The CLOTH-Vowel

In GA, the cloth vowel is situated in the central back ofthe oral cavity and is given the following percentages of divergence:

F1 low – 56%, F2 low – 0% (cf. fig. 4.4.5)

However, it is necessary to bear in mind that the speakers’ realizations of /ɔ/ is very likely to be higher than the standard pronunciation, since their speech is also affected by the New York City shift which triggers an upward movement of /ɔ/ towards /ʊ/ and /u:/, resulting in a diphthong between these sounds. This assumption is confirmed by the percentages of divergence of all eight speakers (cf. fig. 6.3.1) which show that /ɔ/ is realized in the very back of the oral cavity by all participants (the average F2 divergence being 15.61%), but that it is also located much higher (percentage of divergence from the lowest F1 being only 36.08%) than in GA.

illustration not visible in this excerpt

Fig. 6.3.1: Percentages of divergence of /ɔ/

Apparently all participants have not fronted their pronunciation of the CLOTH-vowel too much. Interestingly, Samantha has the most fronted position of /ɔ/ while her friend Allie has such an F2 (percentage of divergence only 10.75%) that it almost resembles the pronunciation of a male person over the age of 40 (cf. Tim’s 10.17% of divergence from the lowest F2). Moreover, there is a considerable gender discrepancy in the values of F1. With a range of 29.68% and a gender difference of almost 15%, the position of /ɔ/ is much higher in the pronunciation of the female participants. In addition, the younger participants lower the vowel more than the older generation. These facts suggest two different conclusions. First, women tend to be attracted by higher vowels, which additionally is confirmed by their high positions of /ɑ/, and, therefore, have raised their /ɔ/ during the New York City shift while men prefer to stick with lower and darker sounding vowel realizations which may suggest more masculinity. Second, the younger generation, although it may be affected by the New York City shift, shows a stronger lowering of /ɔ/ because of the NCS in which the CLOTH-vowel is dragged downwards and slightly forward by the shifting /ɑ/. This suggestion also explains why the younger females’ percentages of divergence from the lowest F1 are lower than the males’, but still higher than the ones of the older females, since they have lowered their highly positioned /ɔ/ because of the NCS, but they still have maintained a relatively high position of the vowel, which was evoked by the New York City shift, since they are attracted to it.

illustration not visible in this excerpt

Fig. 6.3.2: Ellipses of all speakers covering the phones of /ɔ/ based on their percentages of divergence

The suggestions above are confirmed by the depiction of the speakers’ vowel clusters. First, the relatively high position of /ɔ/ and its fronting and lowering movement in all participants’ pronunciations is displayed. Interestingly, the males’ vowel clusters are much smaller than the females’ and they also show only slight movement, while the ellipses belonging to the female participants show very strong lowering and fronting. As a consequence, the position of /ɔ/ is being shifted owing to the NCS, but appears much more stable in the speech of the male subjects while in the pronunciation of the female participants it is moved considerably. This fact supports the already mentioned thesis that women tend to lead in linguistic change, since the realization of /ɔ/ of the women in the present thesis is influenced much more by the NCS than the male realization is. Furthermore, it suggests that men are attracted to low and central backed vowels since they indicate masculinity and, thus, try to preserve their dark character as long as possible. This backed position is especially maintained by /r/ and disfavored by voiceless fricatives which encourage fronting, as it can be seen, for instance, in the allophonic distribution of Jake in figure 6.3.3 below.

illustration not visible in this excerpt

Fig. 6.3.3: Ellipsis of /ɔ/ including its phonetic environments (Jake, 24m)

When listening to the recording of the male participants, it becomes clear that their realization of /ɔ/ hardly overlaps with /ɑ/ but rather indicates the still present New York City shift by some diphthongized pronunciations as in dog or coffee. This phenomenon becomes less as the informants get younger. In contrast, the female subjects of the present study do not show the upward tendency towards /u:/ and do not form diphthongs, but rather shift their /ɔ/ towards /ɑ/ and /ʌ/. However, since they are affected by a chain shift, i.e. the NCS, no merger occurs.

6.4 The KIT-Vowel

The KIT-vowel is situated in the high front part of the oral cavity and, since the long vowel /i:/ is excluded from the present analysis, serves as the upper left corner of the fitted GA vowel chart (cf. fig. 4.4.4). Accordingly, it is given the following percentages of divergencefrom the extreme values of F1 and F2:

F1 low – 0%, F2 low – 100% (cf. fig. 4.4.5)

Concerning its shift evoked by the NCS, Kurath and McDavid (cf. 1961:130) describe a movement of /ɪ/ towards /ʌ/ in words like bristle, limited to New England and Upstate New York, and towards /ɛ/ in words such as rinse. Moreover, they claim the relationship of /ɛ/ and /ɪ/ to be interchangeable. The percentages of divergence of the eight informants for the KIT-vowel are the following:

illustration not visible in this excerpt

Fig. 6.4.1: Percentages of divergence of /ɪ/

The average percentage of divergence from the lowest F1 for all eight speakers indicates that generally /ɪ/ is realized in a relatively high position in the oral cavity, but nevertheless has been lowered slightly already. With a range of 15.23% on the F1 level, all informants appear to have a clear idea of where the KIT-vowel should be pronounced. However, there is a discrepancy of almost 10% noticeable in the pronunciation of one half of the participants. Surprisingly, no clear line can be drawn between genders or generations, since the difference of the F1 percentage of divergence is almost none between genders and less than 5% between generations. Note that Annette, Merryl, Philipp, Jake and Allie have a much lower percentage of divergence from the lowest F1, meaning that their /ɪ/ is realized at a very high position in the oral cavity, whereas Tim, Patrick and Samantha already show a clear lowering. Since in both generations and genders different positions of the vowel are possible, the assumption of linguistic insecurity as where to pronounce the sound arises.

On the F2 level, the average percentage of divergence of all eight speakers indicates a clear backed position of /ɪ/. The fact that the younger participants almost have 7% less divergence from the lowest F2 than the generation over 40 proves their leading position in the backing of /ɪ/ and hence, in the NCS, in general. This is confirmed by the individual percentages, the lowest being Allie’s 66.35% which indicate an almost central position of the vowel in the speech of the youngest participant, while the oldest informant has a divergence of 82.57%. As with the other vowels, note Philipp’s’ extraordinary percentages of divergence which again are almost resembling the ones of a female. In the case of /ɪ/, Philipp even has the most fronted realization of the vowel (83.85%) and raises it only 2% less than the highest value of all eight participants belonging to a woman.

illustration not visible in this excerpt

Fig. 6.4.2: Ellipses of all speakers covering the phones of /ɪ/ based on their percentages of divergence

The depiction of the vowel clusters based on their percentages of divergence confirm the descriptions established by the figures above. They clearly indicate that /ɪ/ is generally located in the high front of the oral cavity and has started to move down and backwards. Besides the fact that the females’ ellipses are much wider than the males’, the vowel envelopes belonging to the girls have an even larger size and a stronger lowering movement than the ones belonging to Annette and Merryl. They clearly reach into the central area of the vowel chart which, in addition to their lowering, suggests that they may overlap with other vowels, especially /ɛ/ and /ʌ/. Surprisingly, the only vowel realization that clearly does not overlap with these other ellipses is Philipp’s which is very small and very high front.

When it comes to the phonetic environments, no clear distinctions can be made. Annette’s most backed values are followed by a lateral while Merryl’s are followed by plosives. The male participants do not show any rule-governed movements. The younger generation’s pronunciation reveals only one clear distinction. It generally has a clear backward and lowering movement of /ɪ/ except in the environment of nasals, which disfavor backward shifting and keep the vowel in its high front position (cf. fig. 6.4.3).

illustration not visible in this excerpt

Fig. 6.4.3: Ellipsis of /ɪ/ including its phonetic environments (Jake, 24m)

In sum, the KIT-vowel has a relatively stable position among the eight informants. It indicates lowering and backing among all generations. Since the features of NCS may not be developed enough to shift /ɪ/ backwards in the speech of the participants over the age of 40, it is most likely that its backed position is owing to their New York City accent. However, the clear lowering and even further backing that is visible in the movement of the younger speakers’ ellipses indicates the presence of the NCS in this age group.

6.5 The DRESS-Vowel

The DRESS-vowel has a central front position in the GA vowel chart and has been observed to move backwards as a result of the NCS. Accordingly, its percentages of divergence are the following:

F1 low – 56%, F2 low – 77% (cf. fig. 4.4.5)

Note that although /ɛ/belongs to the front vowels, its mean F2 percentage is 77%, “since thepercentages are based on a rectangular box” (Bause 2010: 97) surrounding the trapezoidal vowel chart. The average percentages of divergence of all eight speakers indicate a position of /ɛ/ that is higher than the standard and about 15% backed (cf. fig. 6.5.1).

illustration not visible in this excerpt

Fig. 6.5.1: Percentages of divergence of /ɛ/

illustration not visible in this excerpt

Fig. 6.5.2: Ellipses of all speakers covering the phones of /ɪ/ based on their percentages of divergence

With a discrepancy of only 0.2% there is almost no gender difference in the F1 position of /ɛ/. However, younger speakers (46.53%) clearly realize the vowel at a lower position than the older generation (37.25%). The range of 16.88% is the discrepancy between two women, i.e. Merryl’s highest (31.89%) and Allie’s (48.63%) lowest vowel position. In addition, the females from the older generation produce their DRESS-vowels at the highest position of all eight speakers, while the two young girls have the lowest location besides Patrick.

illustration not visible in this excerpt

The average percentage of divergence from the lowest F2 (28.63%) already indicates a backward movement in the realization of /ɛ/. However, the range of 28.63% suggests a considerable alternation in the positions of /ɛ/ along the F2 axis. Again, there is no relevant gender difference in the vowel movement, but a clear age difference of 14%. While the older generations’ pronunciations of the DRESS-vowel do not get lower than 65% divergence from the lowest F2, while the highest divergence of the younger generation is 61.37% (Samantha). Hence, the two generations do not even overlap in their F2 positions, but have two different directions. Note that the extreme values again belong to two females, with Annette having the most fronted realization (73.74%) and Allie having a very far backed one (45.11%).

The depiction of the speakers’ vowel envelopes in figure 6.5.2 confirms the suggestions made above. The ellipses of the generation over 40 are situated further in the front while the younger generation’s clusters already lie in the central area of the oral cavity. Nevertheless, all eight ellipses have wide shapes, the women’s being even larger than the men’s, and indicate a strong centralizing tendency. Merryl’s and Annette’s vowel clusters are the most fronted ones, but range into the center of the young males’ ellipses. All other vowel clusters do not reach as far front and, hence, do not have such a long shape. The four males’ ellipses are considerably shorter than the female’s, while the boys’ start further in the center and also reach further back than the ones belonging to the older generation males. Moreover, all vowel clusters show a nearly horizontal movement towards the back, while the two young girls additionally show a strong lowering tendency. However, the extreme values, i.e. the most backed allophones of /ɛ/ belong to the two young males, suggesting that men generally lead in backing and lowering while women may rather favor raising and fronting.

When it comes to allophonic distribution, no distinctive pattern can be established in the speech of Annette. However, the backing influence of /l/ is already recognizable in the speech of Merryl and becomes clearer as the participants get younger. Furthermore, the generation under 40 indicates the fronting character of /r/, as it was already seen in combination with /æ/ (cf. fig. 6.5.3).

In sum, the eight speakers’ realization of /ɛ/ clearly indicates the presence of the NCS, since the vowel is clearly moved backwards and even lowered by some young informants. It changes its position from mid-front to mid-central or even low-central, so that it collides with some realizations of /ɑ/ as its fronted and severely overlaps with the ellipsis of the central STRUT-vowel /ʌ/. This movement may force a push-chain shift of the latter as it will try to maintain its distinctive features.

illustration not visible in this excerpt

Fig. 6.5.3: Ellipsis of /ɛ/ including its phonetic environments (Patrick, 18m)

6.6 The STRUT-Vowel

The STRUT-vowel is the most centering vowel in the present thesis. In the GA vowel chart published by the University of Arizona (cf. fig. 4.4.1), it is given a central F1 position and is located in the back half of the oral cavity on the F2 level. “The corresponding standard British English sound […] is considerably lower and somewhat farther advanced than the American sound” (Kurath 1973: 127f). Although some handbooks on American English still place it further in the front, the present thesis works with the position of /ʌ/ as given in figure 4.4.1, which results in the following percentages of divergence from the lowest extreme F1/F2 value:

F1 low – 55%, F2 low – 27% (cf. fig. 4.3.3)

The percentages of divergence calculated for the eight interviewees are given in figure 6.6.1 below. Their averages indicate that /ʌ/ generally is pronounced slightly too high and far more centralized than the GA average on the F2 level. With 48.92%/41.64% as the F1/F2 average of all eight speakers, it has an almost perfect central position with only a slight backward movement.

illustration not visible in this excerpt

Fig. 6.6.1: Percentages of divergence of /ʌ/

illustration not visible in this excerpt

Fig. 6.6.2: Ellipses of all speakers covering the phones of /ʌ/ based on their percentages of divergence

Interestingly, there is almost no gender- or age-discrepancy in the realization of /ʌ/. Males show an insignificant lowering tendency and the generation under 40 leads in lowering and backing with approximately 2%. However, no clear difference can be drawn since informants from both genders and generations show relatively high F1 values (Tim, Patrick, and Samantha) or relatively low F2 values (Tim, Jake and Allie). The range of 20.09% for F1 and 24.85% for F2 between the speakers’ highest and lowest values suggests an already present vowel movement as a result of the NCS and possibly a slight linguistic insecurity concerning its exact position. Surprisingly, the two young female participants have both the lowest and the highest F1 value, i.e. Samantha realizes the STRUT-vowel even lower in the oral cavity (62.16%) than the GA average, while Allie produces it 20% higher (42.07%). However, Allie’s percentage of divergence from the lowest F2 indicates an extreme back position of /ʌ/ (28.27%) while Samantha’s (44.99%) is lies even more in the front than the eight speakers’ average. The person with both, a relatively low and backed vowel is Tim (F1 51.93%, F2 31.14%).

illustration not visible in this excerpt

Although the average percentages of divergence from the extreme F1/F2 values indicate that /ʌ/ is realized at an almost central position, the shapes of all speakers’ vowel envelopes clearly show a backward movement. As it wasthe case with the realizations of the other short vowels, the women’s ellipses are wider than the men’s and cover a greater range of the oral cavity as the participants are younger.

illustration not visible in this excerpt

Fig. 6.6.3: Ellipsis of /ʌ/ including its phonetic environments (Allie, 17f)

Interestingly, Tim’s ellipsis is not only one of the most backed ones, but it is also very small and only shows a slight horizontal movement. This suggests that Tim either realizes the STRUT-vowel in the back half of his oral cavity in general or that it was already influenced by NCS which evokes a backward movement of /ʌ/. A combination of both reasons is very likely, since the movement of /ʌ/ is the latest phenomenon of the NCS and, thus, it probably has not been shifted from a central to such a backed position in the speech of a person over 40. Moreover, the other males’ and Merryl’s ellipses show a backward movement with a slight lowering, while Annette’s, Samantha’s and Allie’s indicate a strong raising tendency towards the vowel space that is vacated by /ɔ/. All vowel envelopes overlap in an area that is determined as the position of /ʌ/ in the GA vowelchart in figure 4.4.1.

Difficulties arise when it comes to the distribution of phonetic environments (cf. fig. 6.6.3). The older generation only shows tendencies of the influence of specific consonants, while the younger speakers’ vowel realizations of /ʌ/ indicate a backing or at least centralizing tendency of nasals and laterals. Furthermore, fricatives accumulate in the center of the vowel cluster while plosives are situated in the front of it and approach /ɛ/.

Apparently the STRUT-vowel traditionally has an almost central position in New York City English. As /ɛ/ is shifted towards /ʌ/ as a result of the NCS, the latter moves backwards to avoid collision and first approaches a space that is not occupied by another vowel. On average, this is the vowel spot it usually has in GA. However, in the speech of some participants, especially the younger females’, it moves even further since the advanced shift of /ɛ/ probably forces its movement into the back of the oral cavity where a vowel space is slowly vacated by the downward shift of /ɔ/.

6.7 Vowel Systems

After all speakers and all vowels have been analyzed individually, they now need to be set in relation to each other. For comparison, each speaker’s mean values of all vowels were incorporated into one vowel grid (cf. fig. 6.7.1). Since the informants’ percentages of divergence “even out differencesbetween the speakers in F1 and F2” (Bause 2010: 104), they were taken into consideration and used as a scale. The depiction of all speakers’ mean vowels based on their percentages of divergence indicates from which tongue position the eight participants from New York City articulate each vowel and in which directions they tend to shift.

The NCS has been observed to have its starting point in the raising and fronting of /æ/. Taking both vowel clusters into consideration, i.e. the /æ/ and the /æ/+N ellipses, the vowel shows a clear upward movement towards /ɪ/. It has slightly left its vowel space already, but besides in the environment of a nasal it has not moved too much, yet. The shift of the TRAP-vowel triggers the fronting of /ɑ/ which has clearly left its back position and approaches the vowel space that is being left by /æ/. As confirmed by the vowel chart above, the CLOTH-vowel is situated relatively high in New York City English. The shift of the LOT-vowel drags /ɔ/ downwards. However, /ɔ/ moves back to its former vowel spot, i.e. the mid-back position which it has left as a result of the New York City shift.

The KIT-vowel still has a very high front position in the oral cavity of all eight informants. It indicates a slight backward movement, but does not seem to have shifted much, yet. The mean values of /ɪ/ do not overlap with /ɛ/, indicating that no significant lowering has taken place so far, either. The relatively stable position of /ɪ/ is owing to the fact that /æ/ has not been raised enough to compete with /ɪ/. /æ/ does not need its vowel space, yet and hence /ɪ/ has not shifted.

illustration not visible in this excerpt

Figure 6.7.1: Percentages of divergence of all speakers and of all vowels

However, the upward movement of /æ/, especially /æ/+N clearly pushes /ɛ/ into the center of the oral cavity. Both ellipses show a considerable overlap which is why /ɛ/ has started shifting towards /ʌ/. As a consequence, the vowel envelopes of /ɛ/ and /ʌ/ overlap, as well, which is why the latter indicates a backing tendency in order to avoid the collision. It does not yet overlap with /ɔ/ which is just shifting back to its original central back position, but it marginally collides with /ɑ/ as it performs a fronting movement.

Note that for the back vowels, F1 and F2 values differ much less among the speakers, “which may be explained by the fact, that F1 and F2 are much closer in this region, which means that there is less of a distinction upon which to build” (Gordon&Heath 1998: 431). Labov, Yaeger and Steiner (1972: 265) point out that the explanation for the fronting of back vowels was attempted by Martinet (1981). He reasoned that this phenomenon was due to the asymmetrical shape of the oral cavity, which allows much more distinctions in the front than in the back and, thus, may trigger back vowels to the front in order to maintain a clear distinction between certain sounds. This may be a fact that favors the fronting of /ɑ/ dragged by the movement of /æ/.

Although most of the vowels show a tendency to move in directions indicating the influence of the NCS, none of the chain shifts is completed. Moreover, some vowels have a specific, rule governed realization pattern in New York City English which already has a strong influence of a particular vowel and stabilizes its position. This is also confirmed by the depiction of the mean values of percentages of divergence of all speakers in figure 6.7.2. It indicates that the mean value of /æ/ itself is only slightly raised and only /æ/+N shows considerable raising. However, this does not have to be a result of the NCS but may rather be owing to the short- a pattern of New York City English. Nevertheless, it has been shown in the present thesis that phonetic environments which do not favor raising in the short- a pattern are often especially raised.

Furthermore, the depiction of the mean values has lost its trapezoidal shape as it is known from the GA vowel chart. While /ɔ/ has almost reached its former vowel position in the mid back, the means of /ɑ/ have clearly been fronted and /ɛ/ has a strong centralizing tendency.

illustration not visible in this excerpt

Figure 6.7.2: Mean values of percentages of divergence of all speakers (The mean values of the ellipses in figure 6.7.1)

Considering the social variable (gender), the present thesis indicates that the F1/F2 divergence is greater among women and they use much more space of their oral cavity, although anatomically it should be smaller than the men’s. All women’s F1 and especially F2 scales had to be expanded when creating their vowel charts, suggesting that females articulate their vowels on a much wider scope then males. In addition, women show much greater ranges of their ellipses while men have considerably smaller ones that often resemble and overlap with the envelopes of a particular vowel of the male participants. This suggests that women vary much more in their vowel realizations while men rather maintain a specific standard. Note that Patrick and Jake are at almost the same NCS development level as Merryl, which confirms that women lead in the shift and precede men to almost one generation. Furthermore, the present analysis reveals that generally women lead in fronting and raising while men prefer backing and lowering. This fact leads to the assumption that a certain sound symbolism exists among genders (cf. Gordon&Heath 1998: 428f). Women are attracted to high front vowels while men prefer dark sounds that indicate toughness and masculinity. The most striking fact is that thank you has the highest and most fronted values in the speech of young females and was observed with different persons many times during the interviews conducted for the present thesis. Its raising until it almost sounds like /ɪ/ is considered sweet and girl-like, suggesting weakness and innocence. It is supposed to attract men as traditional role models are activated and their masculinity and strength are addressed.

The statement that the NCS has little social awareness (cf. Gordon 2008: 84) was not always confirmed in the interviews conducted for the present thesis. A young female from Manhattan that was interviewed, but not analyzed in the present thesis, after being informed about the interviews’ real purpose i.e. the vowel chain shifting explained with some examples, asked “Do we suffer from this?”. This linguistic unawareness was already reported by Labov in his studies on New York City English: In the conscious report of their own usage […] New York respondents are very inaccurate. [...] most of the respondents seemed to perceive their own speech in terms of the norms at which they were aiming rather than the sound actually produced (1966: 455). However, participants from the generation over 40 complained about the younger generation raising their /æ/ too much. Furthermore, a young actress instantly stopped raising and fronting her TRAP-vowels on hearing that the reason of the recording were vowel changes. To a certain extent, this suggests awareness of the sound changes taking place.

7 Conclusion

The present thesis offers a number of results. First, the depictions of the speakers’ vowel ellipses display that the realizations of the respective sounds overlap and that the safety distance between vowels have almost disappeared as a result of sound shifts. However, overlaps only occur in specific environments. Since two sounds mostly do not collide in the same context, their allophonic distribution maintains the vowel differences.

Second, the sound shifts which lead to overlaps of the ellipses, resemble the features observed with the NCS and, thus, suggest that the NCS is present in New York City English. However, the changes could be identified as belonging to one completely linear movement, but rather suggest shifts in multiple directions, which makes the description and analysis of the effect of one vowel on another challenging. In addition, the sound changes resulting from the NCS interact and also collide with the specific pattern belonging to the accent of New York City which provides stable, rule-governed vowel positions and, hence, prevents the vowel system to shift much more, as is does in the language of other northern cities affected by the NCS.

Moreover, the most crucial features of the NCS were identified in the present analysis. These include the absence of the caught/cot-merger and the raising and fronting of /æ/, especially in the environment of nasals. The influence of other phonetic environments could only be pointed out partially, since further research data are required to establish them.

The vowel systems of the speakers are represented using raw formant frequency data of F1 and F2 which “allow only for impressionistic comparisons across the systems of different speakers. A more quantifiable approach is possible using data from index scores” (Gordon 2002: 258). Index scores measure how often a speaker uses a given variable and assign numbers to the variants on a scale ranging from conservative to innovative. “The number of points on the scale will vary depending on the sound change being measured and the ability of the researcher to reliably distinguish variants” (Gordon 2002: 258), so that higher scores demonstrate a further development of the shift while lower ones suggest the opposite (cf. Gordon 2002: 258f). However, when working with index scores, one has to pay attention to the different shapes of the speakers’ oral cavities.

Although the use of index scores exceeds the scope of the present thesis and therefore has been neglected, the shape of the participants’ oral cavities was taken into consideration and included into the analysis. For this purpose, a new method of comparing vowel realizations among speakers, i.e. the transformation of F1 and F2 values from Hz into percentages, introduced by Tatjana Bause (2010) was tested. Although this method is very effective and helpful in the description of the data, it “still shows a number of weaknesses, e.g. it is not possible to display vowelmovement outside the boundaries set” (Bause 2010: 106). Furthermore, the percentages of divergence of the six short vowels in GA are only estimations and do not represent an exact position. These disadvantages require improvement.

In addition, further data are necessary to establish the social stratification of the NCS, especially gender differences. The same applies for age, as only the analysis of teenage speakers will reveal changes and shifting tendencies in language. Moreover, there is no doubt that the NCS may have a certain prestige or symbolic meaning for high school youth in suburban Detroit as established by Eckert (cf. 1999), “but that does not help us understand how and why it proceeds in the same way for 34,000,000 people across the entire Inland North” (Labov 1966: 403). Consequently, the structures and values of the society affected by the NCS have to be examined in a much wider scope than they were before in order to understand such a large-scale and surprisingly uniform phenomenon as the NCS (cf. Labov 1966: 403).

In sum, the present study has documented features of the NCS in the speech of individuals who were born and raised in New York City and described how their regional dialect interacts with sound changes evoked by the NCS. Moreover, it has made assumptions about shifting tendencies and brought up new questions calling for more comprehensive research.

8 Bibliography

Aitchison, Jean 1981. Language change: Progress or Decay ? Cambridge - New York: Cambridge University Press (Sixth edition 2007).

Allen, Irving L. 1993. The City in Slang: New York Life and Popular Speech. New York: Oxford University Press.

Babbitt, E. H. 1896. “The English of the Lower Classes in New York City and Vicinity.” Dialect Notes 1, 457-464.

Bause, Tatjana 2010: The Northern Cities Chain Shift in Amherst/Massachusetts: An empirical case study. Heidelberg.

Becker, Kara & Elizabeth L. Coggshall 2009. “The Sociolinguistics of Ethnicity in New York City.” Language and Linguistics Compass 3.3, 751-766.

Berrol, Selma 1997. The Empire City: New York and its People 1624-1996. Westport, Connecticut - London: Praeger.

Bladon, Anthony 1986. “Phonetics for hearers.” In G. McGregor Language for Hearers. Oxfod: Croom Helm: 1-27.

Bloomfield, Leonard 1933. Language. New York: Holt.

Boberg, Charles 2001. “The phonological status of Western New England.” American Speech 76, 3-29.

Bronstein, Arthur J. 1962. “Let’s take another look at New York City speech”. American Speech 37.1, 13-26.

Callary, Robert E. 1975. “Phonological change and the development of an urban dialect in Illinois.” Language in Society, 4, 155-169.

Chambers, Jack K., Peter Trudgill & Natalie Schilling-Estes (eds.) 2002. The Handbook of Language Variation and Change. Oxford: Blackwell.

Department of City Planning 2004. The Newest New Yorkers. Immigrant New York in the New Millennium. Briefing Booklet. New York, NY: City of New York.

Dinkin, Aaron. 2009. Dialect Boundaries and Phonological Change in Upstate New York.Doctoral Dissertation, University of Pennsylvania.

Eckert, Penelope 1987. “Relative Values of Variables.” In K. M. Denning, S. Inkelas, F. C. McNair–Knox & J. R. Rickford (eds.) Variation in Language: NWAVE XV at Stanford . Proceedings of the fifteenth annual conference on new ways of analyzing variation. Stanford, Calif.: Stanford University, Department of Linguistics, 101–110.

Eckert, Penelope 1988. “Adolescent social structure and the spread of linguistic change”. Language in Society 17:183-208.

Eckert, Penelope 1989. Jocks and Burnouts: Social Categories and Identity in the High School. New York: Teacher College Press.

Eckert, Penelope 1991 (ed.). New Ways of Analyzing Sound Change. San Diego, Calif.: Academic Press.

Eckert, Penelope 1991. “Social polarization and the choice of linguistic variants.” In P. Eckert (ed.) New Ways of Analyzing Sound Change. San Diego, Calif.: Academic Press, 213-232

Eckert, P. 1996. „(ay) goes to the City: reminiscences of Martha’s Vineyard.” In J. Baugh, G. Guy & D. Schiffrin (eds.) Festschrift for William Labov. Amsterdam -

Philadelphia: John Benjamins, 47-68.

Eckert, P. 2000. Linguistic Variation and Social Practice. Oxford: Blackwell.

Encyclopædia Britannica 2009. Encyclopædia Britannica 2009 Deluxe Edition. CD-ROM. Chicago: Encyclopædia Britannica.

Fasold, Ralph 1969. A sociolinguistic study of the pronunciation of three vowels in Detroit speech. Georgetown University mimeograph.

Fasold, Ralph & Deborah Schiffrin (eds.) 1989. Language Change and Variation. Philadelphia: John Benjamins.

Fasold, Ralph 1990. The Sociolinguistics of Language. Oxford: Basil Blackwell.

Feagin, Crawford 1979. Variation and Change in Alabama English. Washington, D.C.: Georgetown University Press.

Federal Writer’s Project 1938. New York Panorama: A Comprehensive View of the Metropolis. New York, NY: Random House.

Garcia, Ofelia & Joshua A. Fishman (eds.) 1997. The Multilingual Apple: Languages in New York City. Berlin - New York: Mouton de Gruyter.

Garcia, Ofelia 1997. “New York’s Multilingualism: World Languages and their Role in a U.S. City.” In O. Garcia & J. A. Fishman (eds.) The Multilingual Apple: Languages in New York City. Berlin - New York: Mouton de Gruyter, 3-48.

Gauchat, Louis 1905. "L'unité phonétique dans le patois d'une commune". In Aus romanischen Sprachen und Literaturen. Festschrift für Heinrich Morf. Tübingen: Max Niemeyer Verlag, 175-232.

Godfrey, Brian J. 1995. Restructuring and Decentralization in a World City. Geographical Review 85.4, 436-457.

Gold, David L. 1981. “Three New-York-Cityisms: Sliding Pond, Potsy, and Akey.” American Speech 56.1, 17-32.

Gordon, Matthew J. 2001. Small-Town Values and Big-City Vowels: A Study of the Northern Cities Shift in Michigan. Durham, N.C.: Duke University Press.

Gordon, M. J. 2002. “Investigating Chain Shifts and Mergers.” In J.K. Chambers, P. Trudgill & N. Schilling-Estes The Handbook of Language Variation and Change. Oxford: Blackwell, 244-266.

Gordon, Matthew J. 2008. “New York, Philadelphia, and other northern cities: phonology.” In E. W. Schneider (ed.) Varieties of English 2. The Americas and the Caribbean. Berlin - New York: Mouton de Gruyter, 67-86.

Herndobler, Robin 1977. White Working-class Speech: The East Side of Chicago. Chicago: Doctoral Dissertation,University of Chicago.

Hillenbrand, James M. 2003. “American English: Southern Michigan.” Journal of the International Phonetic Association 33 . 1, 121-126.

Hubbell, Allan F. 1950. The Pronunciation of English in New York City. Consonants and Vowels. New York: King’s Crown Press, Columbia University.

Ito, Rika & Dennis R. Preston 1998. “Identity, discourse, and language variation.” Journal of Language and Social Psychology 17.4, 465-483.

Jones, Daniel 2006. English Pronouncing Dictionary. Cambridge: Cambridge University Press (Fourth Edition 2009).

Kellermann, Anja 2001. A New New English: Language, Politics, and Identity in Gibraltar. Norderstedt: Books on Demand.

Keogan, Kevin 2002. “A Sense of Place: The Politics of Immigration and the Symbolic Construction of Identity in Southern California and the New York Metropolitan area.” Sociological Forum 17.2, 223-253.

Kniffen, Fred and Henry Glassie 1966. „Building in Wood in the Eastern United States: a Time-Place Perspective.” Geographical Review 56.1, 40-66.

Kurath, Hans 1939. Handbook of the Linguistic Geography of New England. New

York: AMS Press (Second edition 1972).

Kurath, Hans 1949. A word geography of the eastern United States. Ann Arbor, Mich.: University of Michigan Press.

Kurath, Hans & Raven I. McDavid, Jr. 1961. The pronunciation of English in the Atlantic States. Ann Arbor, Mich.: University of Michigan Press.

Labov, William 1963. “The social motivation of a sound change.” Word 19, 273–309.

Labov, William 1966. The Social Stratification of English in New York City. Washington, D.C., Center for Applied Linguistics: Cambridge University Press (Second edition 2006).

Labov, William 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press (Eleventh edition 1991).

Labov, William, Malcah Yaeger & Richard Steiner. 1972. A Quantitative Study of Sound Change in Progress. Philadelphia: U. S. Regional Survey.

Labov, William 1980. “The social origins of sound change.” In W. Labov (ed.) Locating language in time and space. New York: Academic Press, 251-266.

Labov, William 1981.“Resolving the Neogrammarian Controversy.” Language 57, 267-308.

Labov, William 1989. “The exact description of a speech community: Short a in Philadelphia”. In R. Fasold & D. Schiffrin (eds.) Language change and variation. Amsterdam: Benjamins, 1-57.

Labov, W. 1990. “The intersection of sex and social class in the course of linguistic change.” Language Variation and Change 2. Cambridge: Cambridge University Press: 205-254.

Labov, William 1994. Principles of Linguistic Change . Volume 1: Internal factors. Oxford: Blackwell Publishers.

Labov, William 2001. Principles of Linguistic Change. Volume 2: Social Factors.

Oxford: Blackwell.

Labov, William, Sharon Ash & Charles Boberg 2006. The Atlas of North American English: Phonetics, phonology, and sound change. Berlin: Mouton de Gruyter.

Labov, William 2007. “Transmission and Diffusion.” Language 83, 344-387.

Labov, William 2010. Principles of Linguistic Change. Volume 3: Cognitive and Cultural Factors. Oxford: Blackwell Publishers.

Laferriere, Martha 1977. „Boston short a: Social variation as historical residue.“ In R. W. Fasold and R. W. Shuy (eds.) Language attitudes: current trends and prospects. Washington, D.C.: Georgetown University Press, 100-107.

Lankevich, George J. 1998. American Metropolis: A History of New York City. New York - London: New York University Press.

Lenz, Günther H., Friedrich Ulfers & Antje Dallmann 2006. Toward a New Metropolitanism: Reconstructing Public Culture, Urban Citizenship, and the Multicultural Imaginary in New York and Berlin. Heidelberg: Universitätsverlag Winter.

Martinet, André. 1955. Économie des changements phonétiques: Traité de Phonologie Diachronique. Bern: Francke Verlag.

McCarthy, Corrine 2008. “The Northern Cities Shift in real- and apparent-time: Evidence from Chicago.” Paper presented at NWAV 37: Houston.

McDavid, Raven I. Jr. 1958. “American English Dialects.” In The Structure of American English. The Ronald Press Company: New York, 480-543.

Merriam-Webster Incorporation 2003. Merriam-Webster’s Collegiate Dictionary. Springfield: Merriam-Webster (Eleventh Edition).

Milroy, James & Lesley Milroy 1978. “Belfast. Change and Variation in an urban vernacular.” In P. Trudgill (ed.) Sociolinguistic patterns in British English. London: Edward Arnold: 19-36.

Milroy, Lesley 2002. “Social Networks.” In J. K. Chambers, P. Trudgill & N. Schilling-Estes (eds.) The Handbook of Language Variation and Change. Oxford: Blackwell: 549-572.

Oomen, Ursula 1982. Die englische Sprache in den USA: Variation und Struktur,

Teil 1. Tübingen: Max Niemeyer Verlag.

Payne, Arvilla C. 1980. “Factors controlling the acquisition of the Philadelphia dialect by out-of-state children.” In W. Labov (ed.) Locating language in time and space. New York: Academic Press, 143-178.

Schneider, E.W. (ed.) 2008. Varieties of English 2. The Americas and the Caribbean. Berlin - New York: Mouton de Gruyter

Shuy, Roger, Walt Wolfram & William K. Riley 1967. A Study of Social Dialects in Detroit. Final Report, Project 6-1347. Washington, D.C.: Office of Education.

Strom, Elizabeth & John Mollenkopf 2006. “Talking and Doing: Discourse and Development in New York and Berlin.” In G. H. Lenz, F. Ulfers & A. Dallmann Toward a New Metropolitanism: Reconstructing Public Culture, Urban Citizenship, and the Multicultural Imaginary in New York and Berlin. Heidelberg: Universitätsverlag Winter, 139-156.

Sweet, Henry 1888. A history of English sounds: from the earliest period with full word-lists. Oxford : At the Clarendon Press.

Thomas, C. K. 1935-1937: “Pronunciation in Upstate New York”: American Speech 10, 107-112, 208-212, 292-297. 11, 68-77, 142-144, 307-313. 12, 122-127.

Trudgill, Peter 1986. Dialects in Contact. New York: Basil Blackwell.

Trudgill, Peter (ed.) 1978. Sociolinguistic patterns in British English. London: Edward Arnold.

Trudgill, Peter 1983: On Dialect. Oxford: Basil Blackwell.

Trudgill, Peter 1986. Dialects in Contact. New York: Basil Blackwell.

Weinreich, Uriel, William Labov & Marvin I. Herzog 1968. „Empirical foundations for a theory of language change.” In W. Lehman & Y. Malkiel (eds.) Directions for Historical Linguistics: A Symposium. Austin - London: University of Texas Press, 97-195.

Wells, John C. 1982a. Accents of English 1. An Introduction.Cambridge: Cambridge University Press.

Wells, John C. 1982b. Accents of English 3. Beyond the British Isles. Cambridge: Cambridge University Press.

Wolfram, Walt& Fasold, Ralph W. 1974. The Study of Social Dialects inAmerican English. Englewood Cliffs, NJ: Prentice-Hall Inc.

Wolfram, Walt & Schilling-Estes, Natalie 1998. American English: Dialectsand Variation. Oxford: Blackwell.

Yaeger-Dror, Malcah 1989. “Patterned symmetry of shifting and lengthened vowels in the Montreal French Vernacular.” In R. Fasold & D. Schiffrin (eds.) Language Change and Variation. Philadelphia: John Benjamins, 59-84.

Online Sources

IC Arizona Vowel Chart. <http://www.ic.arizona.edu/~lsp/IPA/VowelChartLF.gif>.

Accessed last on August 15, 2010.

Labov, William 2010a. Principles of Linguistic Change . Volume 3: Cognitive and Cultural Factors . Chapter 10: Yankee Cultural Imperialism and the Northern Cities Shift.

<http://www.ling.upenn.edu/~wlabov/home.html>

Accessed last on August 15, 2010.

Labov, William 2010b. Yankee cultural imperialism and the Northern Cities Shift.

Paper given at Modern Language Association Annual Meeting, Philadelphia.

<http://www.ling.upenn.edu/~wlabov/PowerPoints/YCI.Yale.ppt> Accessed last on August 15, 2010.

Mayer, Jörg 2009. Phonetische Analysen mit Praat. Ein Handbuch für Ein- und Umsteiger. <http://www.ims.uni-stuttgart.de/_jmayer/resources.html> Accessed last on August 15, 2010.

The New York State Canal Corporation 1998. The Erie Canal: A Brief History <http://www.canals.ny.gov/cculture/history/>

Accessed last on August 15, 2010.

New York University 2009. About NYU. <http://www.nyu.edu/about.html> Accessed last on August 15, 2010.

TELSUR. <http://www.ling.upenn.edu/phono_atlas/home.html>. Accessed last on August 15, 2010.

US Census Bureau 2008. American Community Survey. <http://www.census.gov> Accessed last on August 15, 2010.

9 Appendices

Appendix A

illustration not visible in this excerpt

Fig. 1: Spread of foreign born population in New York City by ZIP-Code in the year 2000.

Source: Department of City Planning 2004. The Newest New Yorkers. Immigrant New York in the New Millenium. Briefing Booklet. New York, NY: City of New York, p. 14.

illustration not visible in this excerpt

Fig. 2: New York City’s top immigrant neighborhoods with the largest foreign born groups in the year 2000.Source: Department of City Planning 2004. The Newest New Yorkers. Immigrant New York in the New Millenium. Briefing Booklet. New York, NY: City of New York, p. 24.

illustration not visible in this excerpt

Fig. 3: Selected New York City neighborhoods.

Source: Department of City Planning 2004. The Newest New Yorkers. Immigrant New York in the New Millenium. Briefing Booklet. New York, NY: City of New York, p. 2.

Please read out the following:

bit

cat

bed

shot

book

caught

bat

cut

tip

hot

shed

pick

pack

peck

taught

hood

bucket

fat

dawn

car

marry

cloth

box

good

thank you

felt tip pen

hopscotch

Fig. 4: Word list read out by the informants.

illustration not visible in this excerpt

Fig. 5: Screenshot of the No23 Recorder window.

illustration not visible in this excerpt

Fig. 6: Screenshot of the Audacity window when cutting a sound file.

Appendix B: F1/F2 Collection

Annette, 87f

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Merryl, 61f

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Tim, 48m

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Philipp, 48m

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Jake, 24m

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Patrick, 18m

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Samantha, 17f

Results TRAP -vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT-vowel

illustration not visible in this excerpt

Results CLOTH-vowel

illustration not visible in this excerpt

Results LOT-vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Allie, 17f

Results TRAP-vowel

illustration not visible in this excerpt

Results TRAP -vowel + nasal

illustration not visible in this excerpt

Results DRESS -vowel

illustration not visible in this excerpt

Results KIT -vowel

illustration not visible in this excerpt

Results STRUT -vowel

illustration not visible in this excerpt

Results CLOTH -vowel

illustration not visible in this excerpt

Results LOT -vowel

illustration not visible in this excerpt

Results mean vowels

illustration not visible in this excerpt

Appendix C: Distribution of Each Vowel’s Phonetic Environment

TRAP -vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

TRAP -vowel + nasal

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

DRESS-vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

KIT-vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

STRUT -Vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

CLOTH-Vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

LOT-Vowel

Annette, 87f

illustration not visible in this excerpt

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

Appendix D: Diphthongization of /æ/+N

Merryl, 61f

illustration not visible in this excerpt

Tim, 48m

illustration not visible in this excerpt

Philipp, 48m

illustration not visible in this excerpt

Jake, 24m

illustration not visible in this excerpt

Patrick, 18m

illustration not visible in this excerpt

Samantha, 17f

illustration not visible in this excerpt

Allie, 17f

illustration not visible in this excerpt

[...]

---

^[1] Not only is it an important financial center, but also “the country’s largest retail outlet [...], leading center for theater, music, and the visual arts [...], home of the country’s largest airports, center of international diplomacy and a tourist mecca” (Bronstein 1962: 19). It has more than 80 colleges (cf. Encyclopædia Britannica 2009), including nationally famed institutions like Columbia or New York University (NYU), the latter being America’s largest private university with 40,000 students coming “from every state in the union and from 133 foreign countries” (Online Sources NYU 2009).

^[2] Lankevich, George J. 1998. American Metropolis: A History of New York City. New York: New York University Press, preface: “One-third of all the people and goods that ever entered the United States came through the New York Harbor”.

^[3] The new Census conducted in 2010 was not available yet at the time of writing.

^[4] For a map depicting the spread of foreign-born population in New York City in 2000 see appendix A, fig. 1.

^[5] The terminology referring to race and ethnicity is used by the U.S. Census (2008) for categorization and will be maintained throughout the present thesis.

^[6] For a map depicting New York City’s top immigrant neighborhoods with the largest foreign-born groups in 2000 see appendix A, fig. 2.

^[7] For a map depicting selected New York City neighborhoods see appendix A, fig. 3.

^[8] After the Immigration and Nationality Act in 1965 repealed many previous quotas on immigration that favored European immigrants, New York City experienced an “unprecedented influx of immigrants of color into New York City” (Becker&Coggshall 2009: 757). The latest New Yorkers are, with a few exceptions, mostly non-white and non-European (Department of City planning 2000).

^[9] New Yorkers can get married in 22 different languages, hear the Catholic Mass in 25 languages, follow the mayoral campaign in at least 4 LOTEs and listen to radio programs in 47 languages (cf. Garcia 1997: 5). After a policy by the New York State’s Department of Health, of 83 hospitals in New York City 78 have to provide translation into Spanish, 14 into Italian, 5 into Chinese and 4 into Russian for emergency service within ten minutes. Bilingual Help Telephone Lines, especially in Spanish and Chinese, are, among others, available at the Emergency Telephone Number 911, the Department for the Aging, the Bias Hotline of the Human Rights Commission, the Heat and Hot Water Compliant Line of the Department of Housing Preservation and Development, the Help Centers of the Department of Environmental Protection, and both the Help Centers of the Department of Finance, which provide tax assistance, and the Department of Transportation (cf. Garcia 1997: 37).

^[10] Wells, John C. 1982b, 503: “The New York accent extends only to the boroughs of the city itself, together with a sharply restricted area of adjacent New York State and New Jersey.”

^[11] Social markers are variants which “show clear-cut social stratification, but they do not show the level of conscious awareness found for the social stereotype. Various vowel shifts, such as the Northern Cities Vowel shift [...] seem to function as social markers. [...] participants in the community may even recognize [...] [the social] distribution, but the structure does not evoke the kind of overt commentary and strong value judgements that the social stereotype does” (Trudgill 1986: 187). Besides the linguistic role of a social marker and social stereotype, a variant may become a social indicator. The latter signifies “linguistic structures that correlate with social stratification without having an effect on listeners’ judgment of the social status of speakers who use them. Whereas social stereotypes and social markers are sensitive to stylistic variation, social indicators do not show such sensitivity, as shown by the fact that levels of usage remain constant across formal and informal styles” (Trudgill 1986: 187).

^[12] Trudgill names parameters for the definition of an ethnic group, established by the National Council of Social Studies, Task Force on Ethnic Studies 1976. By definition ethnic groups are: “(1) origins that or are external to the state (e.g. Native American, immigrant groups); (2) group membership that is involuntary; (3) ancestral tradition rooted in a shared sense of peoplehood; (4) distinctive value orientations and behavioral patterns; (5) influence of the group on the lives of its members; (6) group membership influenced by how members define themselves and how they are defined by others

^[13] The lexical sets and their phonetic symbols for American English are taken over from Wells (1982a: 121), i.e. the TRAP-vowel /æ/, the DRESS-vowel /ɛ/, the KIT-vowel /ɪ/, the STRUT-vowel /ʌ/, the CLOTH-vowel /ɔ/ and the LOT-vowel /ɑ/, and will be used throughout the present thesis.

^[14] Gordon, M. J. 2002. “Investigating Chain Shifts and Mergers.” In J.K. Chambers, P. Trudgill & N. Schilling-Estes The Handbook of Language Variation and Change. Oxford: Blackwell, 244: “Chain Shifts and mergers can be seen as alternative outcomes of a change situation. Both involve the encroachment of one phoneme into the phonological space of another. If the second phoneme changes so that the distinction between the two is maintained, then the result is a chain shift. If, however, the second phoneme does not change, the distinction is lost, and a merger occurs.”

^[15] The New York State Canal Corporation 1998. The Erie Canal: A Brief History: “At the beginning of the nineteenth century, the Allegheny Mountains were the Western Frontier. The Northwest Territories that would later become Illinois, Indiana, Michigan and Ohio were rich in timber, minerals, and fertile land for farming. It took weeks to reach these precious resources. Travelers were faced with rutted turnpike roads that baked to hardness in the summer sun. In the winter, the roads dissolved in a sea of mud.”

^[16] The location of New York City is indicated by a black line at the right hand side.

^[17] The Quantitative Study of Sound Change in Progress is printed in two volumes, the first one describing the theoretical approach, research methods and results, the second containing all figures of the study as well as vowel charts based on formants F1 and F2 which depict each speaker’s vowel ellipses.

^[18] Labov 2007, 63: “Newark, along with Jersey City, Hoboken and Weehawken, is fully representative of the NYC system.”

^[19] The term tense is used here to cover a complex association of phonetic features, i.e. raising, fronting, lengthening and the development of an inglide, as opposed to lax which denominates a short low front monophthongs (cf. Labov 2007: 63).

^[20] Note that č = tʃ, ǰ = dʒ, š = ʃ, ž = ʒ.

^[21] Labov, William 2007. “Transmission and Diffusion”. Language 83, 378: “There is evidence that the complex short-a tensing system of New York City has diffused outward to four different areas. The resulting systems resemble that of New York City in its superficial outline – the phonetic conditioning of tensing by the following segment – but differ from the original model in the absence of grammatical conditioning, the open syllable constraint and specific lexical exceptions.”

^[22] Some speakers gave their age, others their year of birth. For comparability, all age data has been converted into age, which may have resulted in deviations for informants who were born in the second half of the year. However, this has not created tolerance exceeding one year, which is regarded as nonrelevant, since speakers were made sure to fit into two different age groups, i.e. younger than 25 and older than 40 years, in order to have the gap of at least one generation between them.

^[23] Fasold 1990, 229: “Vowel sounds are characterized in sound spectrography by intermittent concentrations of energy at different frequency levels. These energy concentrations show up on the patterns a sonogram draws as more or less wavy dark horizontal stripes, with white areas in between. The dark stripes represent formants.”

^[24] Praat is a freeware available online at www.praat.org

^[25] http://groups.yahoo.com/group/praat-users

^[26] All Excel tables with the collected data were added to appendix B in the same order as they are examined in chapter 5.

^[27] This visual realization of sounds is described by several literary predecessors, such as Labov (1972), Fasold (1990) and Gordon (2001).

^[28] The single formant values of all the phones and their contexts which were measured and used to calculate the mean vowels are all listed in Appendix B. The order of the informants is the same as here.

^[29] All vowel grids depicting the phonetic environments of all vowels and all speakers can be found in appendix C.