Skip to content

Abstracts (TULCON13)

For references and/or footnotes of the following abstracts, click here.


Milena Injac, University of Toronto

The use of the negative particle “ne” in Quebec French news media

The dropping of the pre-verbal negative particle “ne”, as in (1) is a well-known linguistic characteristic of spoken Quebec French (Poplack & St-Amand, 2007).

(1) “Je ne / Ø vois pas.”

“I do not see.” Its use among Quebec francophones is extremely rare in spontaneous speech (e.g. 0.2% in 19xx) (Poplack, 2015; Sankoff & Vincent, 1977); a pattern that has been observed since the 19th century (Poplack & St-Amand, 2007). Since “ne” is stylistically conditioned and is typically only used in formal environments (Poplack, 2015), we wanted to further investigate those rare contexts in which it is retained by examining its use in contemporary media sources. Media sources, such as news reports and debates, represent an ideal source for the analysis of “ne” retention due to their traditionally formal nature. In addition, such sources, which represent somewhat of a middle ground between vernacular speech and written language, have received far less attention than spontaneous speech in the sociolinguistic literature when it comes to investigating the use of “ne” (notable exception: Villeneuve, 2017).

Our corpus was comprised of recent (2019) news reports from “TVA Nouvelles 2018” and two debates, “Fâce à fâce Québec 2018” and “Débat sur les grands enjeux en éducation”, which could be argued to occupy different spots on the stylistic continuum. The former (very formal) discusses professional themes and presents information, and the latter (less formal) is mediated discourse of political and educational themes that provoke emotion, yet attention is still put on speech as it is televised. Through the analyses of 444 tokens, we investigated whether the use of “ne” is stylistically conditioned in these two traditionally formal contexts. We also tested the effect of speaker sex and linguistic factors (inserted element between “ne” or Ø and the verb, and lexical identity of the second negation marker) on the variable use of “ne”.

Our results found that in the most formal context (i.e. the news reports), the usage rate of “ne” (71%) was indeed higher than in the less formal context (i.e. the debates) (35%). In addition, we found that while the presence of an inserted element had no effect on the variable’s retention rate, using negative markers other than “pas” was correlated with “ne” retention, which is consistent with what has been found in spontaneous speech (Poplack, 2015). The results of this study contribute to a better understanding of variation patterns in traditionally formal speech contexts and how such contexts can be positioned along the stylistic dimension between the prescribed norm (standard written) and the vernacular (informal speech).

Justin Leung, University of Toronto

The unified loss of verb particles in Medieval French: a quantitative analysis

Unlike Modern French, Medieval French had a robustly attested and productive system of verb particles that contribute directional (1) and aspectual (2) meanings to the interpretation of an event (examples from Troberg 2019):

(1) Sa lance lessa jus cheoir ‘His lance (he) let fall down’ (Brut, partie Arthurienne, 124)

(2) si le but toute hors. ‘and (she) drank it right up.’ (Beaudouin de Sebourc, 186)

Philological studies claim that the particles disappear from use by the 15th century, and this has supported by a disparate collection of more recent quantitative studies. The importance of determining when and how the particles disappear is related to their possible correlation with the well-known typological shift that involved satellite-framed Latin becoming verb-framed Modern French/Romance (Talmy 2000). Troberg (2019) proposes a formal account of the typological change whereby the disappearance of verb particles results from an underlying syntactic change associated with the shift from weak verb-framed Medieval French to strong verb-framed Modern French. The prediction is thus that the particles should disappear abruptly at the same time. This proposal and its prediction contrasts with claims that the typological change was gradual or that Medieval French directional particles have the same basic grammar as Modern Romance particles (Acedo-Matellán & Mateu 2013).

The present paper tests Troberg’s (2019) account by investigating the time course of change of three verb particles—jus ‘down’, fors/hors ‘out’ and arriere ‘back’—from 1100 to 1599. Measuring the frequency of these three particles over time gives us a statistically informed trajectory of these particles. The advantage of this study is the sheer size and breadth of the corpora (over 200 million words)1 and the treatment of three particles in parallel used in all possible syntactic contexts. One of the challenges of the study is that the data is difficult to analyse using standard methods in historical linguistics and sociolinguistics. Because the data is limited to surviving historical texts, there is a limited amount and type of data to analyse.

Additionally, there are no clear replacement or competing forms for any of these particles, meaning that neither Labov’s (1972) Principle of Accountability from variationist sociolinguistics nor Kroch’s (1989) Constant Rate Hypothesis from historical linguistics can be applied to the data. In order to overcome these challenges, I analyse the relative frequencies of each particle within each text using a Poisson regression model to determine the trajectory of these particles and to test whether the three particles disappear uniformly in the predicted manner. This method is based on Gadanidis & Denis (2019), a sociolinguistic study that similarly examines count data from different sources over time. The preliminary findings appear to support a unified and thus underlying syntactic change.

Christopher Legerme, University of Toronto

Antimarkedness constraints in the lexicon? : Evidence from Haitian Creole

Haitian Creole (HC) features five different realizations of the postnominal definite determiner (LA): la, lan, nan, a, an. This alternation is not arbitrary and is linked to various phonological processes, namely those relating to vowel hiatus and nasalization. The goal of this paper is to explore HC postnominal determiners through the lens of Optimality Theory and to work toward a system of constraints accounting for forms as they occur given some general patterns of the language. Words like pitit la [pititla] ‘the child’, machin nan [maʃinnã] ‘the car’, lanp la [lãp+la] ‘the lamp’, lanme a [lãmɛa] ‘the sea’, and vwazen an [vwazɛã] ‘the neighbor, clearly illustrate a complex system influenced by phonological context, for example, they show that vowel hiatus is preferred over [la], and that the nasalized vowel [ã] is preferred over [a] when beside other nasal segments. However, the alternation of LA in HC presents a serious challenge to the conventional model of OT stipulating a mapping of only one underlying input to one surface realized output by filtering competing outputs (candidates) through serialized, critically ranked markedness and faithfulness constraints. The removal of the liquid segment /l/ of [la] in forms like [fija] ‘the girl’ creates a context for glide insertion to repair vowel hiatus. In fact, a common question that arises when looking at the distribution of LA has to do with the deletion of /l/ as resulting in the vowel sequence needing to be repaired by glide insertion.

Why not just keep /l/ intact and avoid hiatus in the first place? Since phonological processes triggered by the postnominal determiner are not necessarily automatic to all of HC, the general idea of a deleted /l/ is not enough to motivate the application of a markedness constraint. The classical OT model does not easily accommodate for this morphological dimension of [la] ~ [a] alternation given that markedness constraints look to capture universal phonological phenomena. To deal with this issue we may look toward Chris Golston’s Representation as Pure Markedness model (RPM) and Thomas B. Klein’s extension of said model, Lexical Representation as Pure Markedness (LRPM). The idea of the surface form (SF) as derivative of a prosidified underlying representation (UR) gives us a basis for representing inputs solely in terms of violated markedness constraints so that in essence an utterance can exist underlyingly as the very markedness constraint violations needed in order to evaluate said utterance as a successful candidate. This in mind, we can formulate a way to incorporate allomorphy in OT by allowing for constraint-induced input variability. RPM and LRPM will provide the theoretical motivation behind a constraint we might impose onto a traditional OT model of LA alternation in HC. The term “anti-markedness” expresses a tendency toward a marked phenomenon by respecting the violation of an unmarked feature of language stipulated in the lexicon which is precisely what we find with HC postnominal determiners.

Nicoline Butler, York University

Intonation in Thai-Accented English

Thai-Accented English (TAE) is an intentionally accented register used by Thai- Canadians to facilitate understanding in Thai L2-English speakers of lesser proficiency, and to interact with others of Thai descent. Speakers of TAE exhibit an L1 proficiency in both Thai and English, and can ‘put on’ an L2-like Thai-Accented English. Standard Thai has 5 tones which are realized based on syllable types, in accordance to a series of restrictions postulated by Moren & Zsiga (2006). This study discusses a default tone pattern found in TAE nouns which exhibit similar syllable based restrictions. This study also discusses possible evidence of Thai tonal transfer in TAE.

In 2009, Lim postulated that in some ‘New Englishes’ such as Singaporean English, tonal features have been transferred from loan-words not only at the word level, but also in the intonation contour. Lim attributes this transfer to her model of language ecology, where lexical tone is considered both highly marked and pragmatically dominant. The discoveries made by Lim (2009) and the tonal restrictions of Thai outlined by Moren & Zsiga (2006) raise several questions for intonation in TAE. Firstly, whether the tone features that exist in Thai are considered highly marked and pragmatically dominant enough to be carried into TAE. Secondly, if so, do they manifest only on loanwords, such as in Lim (2009), or are they applied more systematically and extend to non-loanwords?

Two production experiments were conducted with a bilingual Thai-English speaker who regularly uses TAE. First, in order to explore tone assignment in TAE and whether this is influenced by syllable type, a nonce-wordlist of 20 words containing CV, CVS, and CVO syllables were presented in the form of nouns in a framing sentence. Sentences were recorded by the consultant in both Standard English as well as TAE, and then analyzed in Praat for pitch patterns. A second experiment was designed to confirm the patterns evidenced in the nonce-word test, and compared the three syllable types in both nine English loanword nouns regularly used in Thai and nine English nouns which were not. Also presented in framing sentences, the TAE and English recordings of the second experiment were both analyzed and contrasted in Praat.

The results of the initial experiment suggest that TAE has an overall default falling tone (HL) word finally in nouns. The final low tone is truncated in CVO syllables resulting in a rising tone (LH) word finally. The second experiment confirms this overall tonal pattern but contains exceptions in several loanwords. When compared, the pitch patterns of TAE closely resembled those of the Thai English-loanword, and suggest that the default tone found in TAE can be overridden by Thai tonal patterns. These preliminary results point to a potential similarity between TAE and Lim’s (2009) findings for Singaporean English. Upon further study, TAE may be able to provide additional evidence towards Lim’s model of language ecology.

Anissa Baird, University of Toronto

I am not my body: a preliminary study of gender identity, gender expression, and their effects on sex-based and gender-based variability

With regard to sociolinguistic variation, there are many types influenced by a variety of social factors. One such type is gender-based variability, as explored in studies such as Milroy and Milroy (1978), who looked at a selection of phonological variables in Belfast, Ireland and found that women tend to have a wider range of linguistic styles and use stigmatized forms less. Another type of variability is sex-based variability. This has been explored in studies such as Maccoby & Jacklin (1978), who found that women are found to statistically score higher on verbal tasks. What all of this research has in common is a tendency to only work within a binary: male and female, for both types of variability. However, sex and gender are not binaries at all, an idea that is, fortunately, becoming more and more widely accepted in more recent times. Thus, the goal of this study is to be an exploratory first-look at determining how the introduction of this non-binary aspect of sex and gender affects the trends of these variabilities.

The analysis relies on sociolinguistic interviews of 13 AFAB (assigned female at birth) individuals from North America and the UK. These individuals are categorized under one of two categories: cisgender (6 interviewees) and non-cisgender (7 interviewees), the second category being an umbrella term used solely in this study in order to generally group everyone whose gender identity does not match their anatomical sex. The interview itself consists of three parts: a single-paged questionnaire, a personality test, and a two-part verbal test. The questionnaire portion of the interview requires the speaker to answer questions with regard to their anatomical sex, gender identity (whether one considers themselves a man, woman, or otherwise, despite anatomical sex), and whether or not they are (or previously have been) on hormone therapy. Additionally, there is a 2-part section of the questionnaire asking about the interviewee’s swearing habits. The personality test conducted is the AQ test (Baron-Cohen 2001), while the 2- part verbal test is a standard Verbal Fluency (VF) test.

Two hypotheses were conceived. The first hypothesis was that gender expression (the interpretation of one’s actions based on gender roles) correlates with gender-based variability. Therefore, cisgender participants were expected to have more consistent and, overall, lower scores than non-cisgender participants. Admittedly, this hypothesis was speculatory and based on “social norms.” However, as there were no known tests or previous research regarding this subject, this was only meant to be a starting point for research. The second hypothesis was that gender identity correlates with sex-based variability. Therefore, cisgender participants were expected to have a left-hemisphere specialization and do better at verbal tasks, meaning that they would have lower AQ scores and higher VF scores than non-cisgender individuals.

Mixed-effects modeling shows that the results of the first hypothesis are inconclusive, indicating a need to design a more well-suited test in order to obtain significant results. The main finding, however, is that gender identity is considered to be a significant factor influencing AQ scores. This ultimately supports the hypothesis that gender identity correlates with sex-based variation, inferring that it is a stronger predictor of this type of variation than anatomical sex. While gender identity is not found to be significant for VF scores, the expected gap in scores between cisgender and non-cisgender individuals is found. Therefore, it has been concluded that the lack of gender identity as a significant factor for VF scores is more than likely due to the small sample size of participants in this study. The implication of the results is an important one: gender identity, just like anatomical sex, is completely innate and prenatally determined, and, in cases regarding the brain and cognitive function, is an even stronger predictor.

Yanfei Lu, University of Western Ontario

How do Adult L2 Learners Acquire Pronominal Prefixes of Oneida (Iroquoian)

Through this research, I explore how adult native English speakers acquire Oneida as a second language, specifically focusing on the acquisition of the pronominal prefix feature of Oneida. The Oneida people live mainly in three communities at present day: Oneida of the Thames (near London, Ontario) and two Oneida reservations, one in Wisconsin and the other in New York (Michelson & Doxtator, 2002). My research focuses on Oneida of the Thames community in London, Ontario, and the data is collected from their language immersion program, organized by the Twatati Oneida Language committee.

Oneida is a polysynthetic language, which uses complex morphemes to express meanings that would be expressed by individual words and sentences in languages like English (Archibald & O’Grady, 2016). Therefore, the structure of words in Oneida is complex and difficult to acquire for English speakers. In addition, the usage of the morphemes is not only determined by the grammatical categories but also cultural practices, traditions and conventions of the Oneida community. Oneida has a very small number of native speakers. The UNESCO Atlas of the World’s Languages in Danger has marked Oneida as critically endangered (UNESCO, 2017). The small population of native speakers poses many challenges for adult learners. The lack of opportunities for learners to practice Oneida outside of the classroom creates difficulties for learners to enhance their speaking and listening skills. Furthermore, there is a lack of resources available for the learners as well as for creating a curriculum that is appropriate and based on the Oneida language and the Oneida community.

Pronominal prefixes are of the most important components of Oneida utterances. They indicate the participants of the “sentence”, such as the subjects and/or the objects. There are over 100 variations of the Oneida pronominal prefixes which are based on several grammatical and contextual features at the same time. These features include the person, gender, number and clusivity of the performer(s) of the action and/or the recipient(s) of the action. Meanwhile, the phonological environment also triggers variations of the pronominal prefixes. Based on feedback from Oneida speakers and learners, pronominal prefix is one of the most difficult features of the language to acquire.

The participants in my research are adults who are native speakers of English and are currently enrolled in the Oneida immersion program as students of the year 1 class. The data is collected from unit 1 of the unit assessments of the curriculum. During these assessments, learners’ listening and speaking abilities of Oneida are tested in the oral format. Before conducting the analyses, I transcribed the audio data through using Elan software, with the help of Oneida speakers. Furthermore, factors such as learners’ differences of age, amount of previous knowledge, and the class settings are also taken into consideration, since they can influence the learners’ language learning process in various degrees.

This research will contribute to filling the literature gap of acquiring polysynthetic languages as a second language. In addition, polysynthetic languages are found not only in North America but communities worldwide. Since many of these languages also have a small population of speakers, this research can provide valuable information for language revitalization programs, language courses, language learning software or mobile applications to teach Indigenous languages and consequently benefit not only Indigenous communities in Canada but Indigenous communities worldwide.

Xinyi Zhang, McGill University

Bilingualism, Symbol Grounding, and Consciousness: A Process-Dissociation Procedure in French-English Bilinguals

My presentation will introduce the recent study done by my labmates and me with the title “A Process-Dissociation Procedure in French-English Bilinguals.”

Our research investigated the cognitive and unconscious influences of memory during language processing in bilinguals using the source-monitoring error paradigm. A group of English-French bilinguals studied lists of associated words, some in English and some in French, and then performed a recognition test containing presented list items, translated list items and critical word items which were not present in the list. Participants were instructed to reflect on the nature of their recollection, categorizing it as either a “know”, i.e., an unconscious judgment, or a “remember”, i.e., a more conscious judgment. The effect of the translated list items on accuracy was significantly greater than the effect of the critical word items, and this effect was judged to be mostly unconscious. The present study demonstrates the existence of false memory across languages due to language non-selective processes during word decoding and sheds light on the forms of memory underlying veridical and erroneous recognition.

This research deepens our understanding of how language processing works in the bilingual population, in both language encoding and decoding phases. The growth of the bilingual population is astonishing worldwide, and bilingualism has attracted people’s attention in both the academic and the non-academic world. There have been a series of debates about what the status of the two languages that are in bilinguals’ head and how (or if) they interact with each other; our research addressed these questions in terms of “bilingual lexicon space” in details.

Moreover, our research can be seen as an empirical examination of the Symbol Grounding Problem in bilinguals. The Symbol Grounding Problem asks how words get their meanings. Searle’s Chinese Room Argument has shown that a language cannot be acquired by learning the rules of the symbol manipulations. The symbols (words, for languages) have to be grounded with meaning. (While it is true that we could use symbols to define symbols, it has been shown that some sets of words simply have to be grounded first shown by the recursive analysis done on dictionaries.) It has been argued that this grounding of meaning is achieved by the non-symbolic sensorimotor experience that is associated with the symbol. The participants in our experiment were all early, simultaneous bilinguals, and the words we used in our memory tasks were all high frequency, simple words that are highly likely to have been grounded in this language non-specific sensorimotor experience. For them, the symbol grounding process is more complicated than for the unilingual. One referent that formed by the sensorimotor experience would have to be mapped to (at least) two words, one in each language. Or, there are two referents, one in each language? If that is the case, how do we resolve the referent problem for different words that mean the same thing in one language? Our research attempted to answer these questions.

Our research also aims to answer some questions in the field of consciousness. By using the bilinguals as subjects, we hoped to create a (weak) “periscope” from language to consciousness. People tend to think using a language is a conscious state of mind, yet, since language seems to “always” co-occur with “thoughts” (or, (a type of) consciousness), it is hard to investigate them separately. We think there is further application of the approach that we took in this experiment in research in consciousness.

Rosie Owen, University of Toronto

Sweet Songs and Soft Hearts: Metaphor in Cuzco Quechua

Much literature in cognitive semantics has investigated conceptual metaphor, though typologies have only covered Indo-European and Asian languages ( Shen & Gil, 2008) . Under conceptual metaphor theory, metaphor is not merely a linguistic expression but a result of cognitive processes connecting conceptual domains and part of a shared “conceptual apparatus” within a culture (Lakoff & Turner, 1989). Conceptual metaphor has yet to be explored in Quechuan languages, spoken in the South American Andes. With a culture rich in mythology, poetry, and unique forms of arithmetic (Urton & Llanos, 1997), many epistemic fields have been influenced by the so-called “mythopoetical” worldview, including the conceptualization of time (Almeida & Haidar, 2012). Quechua may conceptualize time in the opposite direction of languages like English (i.e. on the continuous path of time, the past is associated with anteriority and the future posteriority) and group space and time into a single unit (Almeida & Haidar, 2012). This study addresses these questions in Cuzco Quechua: in which domains do metaphors manifest? Have epistemic traditions influenced the conceptualization of time and space, into a single unit? Is the reverse time model present?

Study: This study used Shen & Gil’s questionnaire on the Typological Aspects of Figurative Language, as well as Levinson’s Time Questionnaire, to investigate expressions in metaphor.

Predictions: given past anthropological studies, (i) all domains were expected to be rich in metaphor, (ii) time and space were expected to be conceptualized as a single unit as reported, but (iii) the reverse time model was expected to be absent, as per previous controversy.

Methods: The speaker was interviewed using the questionnaires and contributed other culturally relevant or related metaphors. Additional metaphors were found in storyboard elicitations and translated poetry and folktales.

Results: This paper presents a new set of data, including the domains: perception, sensory, body parts, texture, travel, emotions, mental states and activities, and nature terms, as well as SPACE-TIME metaphors. The data shows that space and time are conceptually distinguished and provides evidence of a reverse time model.

Discussion: Expressions of time suggest influence from the historic arithmetic system and contradict previous claims about the direction of time. Perception and sensory metaphors suggest a significance of taste. Cross-linguistic similarities with European languages were seen within phrases for mental activities and the heart as a locus of emotion.

Amelia Fineberg, University of California, Berkley

Semantics and Grammar: Gender in Liturgical Hebrew

Several branches of Jewish practice, such as Reform, Renewal, and Reconstructionist movements, alter the words in traditional liturgy to make prayers written in past times more inclusive and representative of their current beliefs and congregations. Specifically, many congregations choose to use gender-neutral or feminine (rather than exclusively masculine) language, especially ones which express an interest in feminist or LGBTQ issues. Since Hebrew has a robust obligatory grammatical gender system, different strategies are available within the grammar of the language, as well as in various translational choices. In this study, I examine how several Bay-Area synagogues, each with their own prayer compilations, choose to adjust texts. I identify four major strategies:

• Leaving Hebrew text as-is but changing accompanying English translations

• Using the same vocabulary, but changing grammatical gender, such as changing barukh ‘bless (3sg.m)’ to brukha ‘bless (3sg.f)’

• Keeping overall prayer format but changing vocabulary to replace words with strong gendered associations, such as replacing melekh ‘king’ with ruakh ‘spirit’, often triggering grammatical changes (in this case, ruakh is a feminine subject, changing the conjugation of its verb)

• Radically rewriting prayers using “collective” forms Each strategy involves trade-offs: each congregation has competing interests in preserving Jewish tradition and serving current members. Furthermore, especially in the two later strategies, changing vocabulary can change the cognitive frames activated, changing the fundamental meaning of a prayer. I use the tools of cognitive linguistics, including frame theory (Fillmore, 1982), conceptual metaphor (Lakoff and Johnson, 1980), and blending theory (Fauconnier and Turner 2002) to analyze how meanings change when prayer texts are altered, and how those new meanings might serve the congregations which employ them, reinforcing new concepts of what it means to practice as a Jew in the modern world.

Ryann Soltero, Gordon College

Why-wh-movement in Korean

Introduction: Wh-movement is a feature found in many, but not all, languages. The feature determines whether or not the wh-pronoun moves from its base position to a structurally higher position (van Gelderen, 2017). Shin (2005) asserts, however, that the Korean language deviates from this pattern. Unlike other wh-movement languages, such as English, Korean does not demonstrate movement with nominal wh-words. The only wh-word seeming to require movement is ‘why.’ The sentences Shin uses as tests to argue for this conclusion, however, are unclear. When shown to a native speaker, the sentences used were unnatural and seemed forced. In order to better understand this phenomenon, the presentation will use more natural data and provide a cross-linguistic perspective on why-wh-movement.

The Puzzle: Although nominal wh-phrases in Korean are generally in-situ, why-wh-movements can move. In this case, ‘why’ (or, ‘wae’) is fronted. The following data shows a comparison between a sentence with standard wh-movement in (1), and a why-wh-sentence in (2):

너의               아이들은            누구니?

Nuh-e              nu gu-ni?

Your-POSS               who-question

‘Who are your kids?’

Above, ‘who’ can remain in the wh-in-situ position without much question.

(2)  왜    그는               떠나는거야?

 Wae   geu-neun     dduh na-neun.geo ya?

 Why  he-is             leaving-is.question

‘Why is he leaving?’

In (2), while Korean is a free order language, standard SOV word order requires that ‘why’ is fronted, thus undergoing wh-movement.

Proposal: While Shin’s research primarily explains the semantic reasoning behind this phenomenon, this paper looks to explain the syntactic differences that trigger this occurrence of why-wh-movement while also clarifying the data. Not only does ‘why’ ‘s bypassing of island effects contribute to this phenomenon, but so does its function as a question. In other words, if a speaker was to take the nominal wh-word out of a wh-sentence and rearrange it, the gap left behind from the nominal wh-word would inhibit the sentence from having a full meaning. In the case of why-wh-sentences, however, despite the absence of the why-wh-word, the sentence still has meaning. The disparity between the two seems to be an argument versus adjunct difference. Through various island tests in Korean, Mandarin, and English, this paper will answer questions regarding the syntactic differences between why-wh-words and nominal wh-words that provoke this movement.

Ernest Leung, University of Toronto

Derogatory Terms in Hong Kong English’s First Dictionary and their Sociocultural Impacts

Hong Kong English (HKE) is gaining international recognition as an independent variety of English. One attempt to codify these lexicons is Patrick J. Cummings and Hans-Georg Wolf’s Dictionary of Hong Kong English: Words from the Fragrant Harbor. This study assessed to what extent this dictionary helps stabilize HKE endonormatively, as well as to evaluate its social and cultural impacts on Hong Kong English and its users. The study identified entries in the dictionary that contain derogatory meaning in Chinese culture, and then compared and contrasted which of these terms the dictionary recognizes as derogatory. The analysis used Edgar Schneider’s dynamic model to understand endonormativity (37). In assessing how well the dictionary recognizes Chinese culture impacts HKE lexicons, which is a criterion in establishing a variety of English endonormatively, the analysis found that while the dictionary recognizes HKE derogatory terms that refer to speakers of European background (“gweilo” and “gweipo”), it fails to capture terms that are distinctively derogatory in the local Hong Kong context, for example, “gaylo.” This result shows that the dictionary lacks an insider awareness of the words’ cultural connotation, revealing its European cultural in the dictionary that only stabilizes HKE exonormatively; it fails to take the influence of Chinese culture into consideration while attempting to codify and standardize HKE. By unsettling the seemingly settled HKE in this dictionary, this study questions whether a dictionary always implies an endonormative stabilization for a variety of English, as Schneider’s dynamic model proposes (2007). The findings will inform the theoretical approaches for future dictionaries as they codify other emerging varieties of English.

Jingshu Yao, University of Toronto

Mom, Talk to Me in My Mother Tongue: SES and Heritage Language Maintenance of East and South Asian Canadian Community

According to Statistics Canada (2018), 20.6% of Canadians have a primary language other than English and French. However, in the linguistic environment dominated by the two official languages, it’s hard for heritage language speakers to maintain their languages and cultures. This research analyzed the connection between parental support for heritage language maintenance (HLM) and the immigrant parents’ Socio-economic States (SES). Previous studies suggest that SES of a family is an external factor that influences the literacy of its future generation and parents with higher SES are more willing to have their children learn heritage languages (Reese, 2006, Mori & Calder, 2017). The data was collected from 91 immigrant parents among East and South Asian communities in Toronto, Ontario through an online survey, all of who have raised children in Canada. The survey gathers background information about the participants, their usage of heritage languages in their daily life, whether they have ever registered their children to HLM program, and what motivates then to pass down their heritage language to the next generation. They parents’ ESE are evaluated based on occupation and education. After comparing their answers based on SES, length of living in Canada, and ethnicity, the results correspond with the hypothesis. Even though High-SES and Low-SES parents are both supportive for HLM, but the different income, environment, and culture value results in differences in their motivation. More LSES parents believe that HLM is a method for the children to communicate with family and community members, while HSES parents put more emphasis on culture value, identity, and career opportunity. Those LSES children are more likely to acquire HL from daily life but more HSES children are registered for HLM programs at school.

Poster Presentations

Ryan MacDonald,University of Toronto

Onset and Coda Repair in Phonetic Loanwords from English into Mandarin Chinese

In this paper I analyze how complex onsets and simple and complex codas in phonetic loanwords from English into Mandarin Chinese are repaired in order to fit the restricted syllable structure of Mandarin. In Mandarin syllables, the only complex onsets allowed are consonant-glide sequences, complex codas are not allowed, and the only segments allowed in coda position are the nasals /n/ and /ŋ/ (Trísková, 2011, p.103-107). English syllables, in contrast, allow complex onsets of varying sequences, complex codas, and many consonants are allowed in the coda position for simple codas. So, in borrowing words from English which contain syllables that violate the syllable structure of Mandarin, phonological processes occur to repair disallowed syllables. I look at what processes occur for each of the disallowed structures, and analyze what factors cause each process to occur, and why there is variation between similar instances of syllable violations.

In order to analyze these processes, I searched through online Chinese newspapers published for speakers living in Canada, the US, and the UK and compiled words which were borrowings of English place names or transliterations of peoples’ names from English news sources. With place names and names of people I could ensure that the words were borrowed from English and not through other languages. I grouped the data into simple codas, complex onsets and complex codas and further grouped the data within each of those categories for each of the separate types of segments – obstruents, liquids, nasals, etc. I analyzed the data using Optimality Theory. I found the following: Mandarin employs both epenthesis and deletion in reparation of onsets and codas containing obstruents; liquids behave differently than obstruents – they are changed or deleted; nasals are unaltered in some cases and changed into a different nasal in others; when not deleted, liquids are repaired differently when they are in onset position versus when they are in coda position – they change to /l/ in complex onsets and change into /ɚ/ in codas; and the majority of complex onsets and almost all of the complex codas are repaired with a combination of processes used for simple codas.

I conclude that Mandarin has both epenthesis and deletion in order to create syllables no larger than C(G)VN. Epenthesis makes closed syllables that do no have a simple /n/ or /ŋ/ nasal coda into separate open syllables by changing the coda(s) into an onset of a new syllable. It is also used to make complex onsets into separate simple onsets of multiple syllables. Deletion, on the other hand, is used to make closed syllables open without making new syllables. It is also used to make complex onsets into simple onsets. Voiceless coronal stops in coda position are more likely to be deleted than other obstruents which are more likely to have epenthesis. Liquids in coda position or in complex onsets are more likely to be changed into another liquid or remain unchanged in the case of /l/ in complex onsets rather than have epenthesis or deletion. For nasals, neither epenthesis nor deletion occurs they remain unchanged or changed into another nasal in the case of /m/. So, there seems to be some preference for epenthesis or changing segments over deletion, but ultimately the most important thing is to adhere to the strict syllable structure of Mandarin which is why deletion also occurs.

Joseph Jaros, University of Pittsburgh

Learning Vocabulary through Variable Input: How Interactions with Multiple Speakers Affects Lexical Consolidation

Natural speech is variable. two speakers can say the same utterance, but have very different outputs, with variation in the phonetic features of their speech. Despite this variation, we still have a remarkable ability to comprehend meaning behind the speech. The lexicon must be able to identify words despite the variability in production. How lexical representations encode phonetic variability is largely unknown. The representation may only encode minimal information, bypassing the variability in the speech signal. Alternatively, multiple variants of one lexical item may be stored, facilitating identification of novel phonetic forms. The goal of this research is to understand how phonetic information is stored in the lexicon. By tracking how speakers learn new words, we can understand the structure of a single lexical unit. If phonetic variability is encoded within the lexicon, then I hypothesize that learning new words with variability will create a more robust lexical representation compared to words learned without phonetic variation.

Participants learned 72 novel words and associated meanings (ex. Cathedruke – To type quickly) over 2 days. Half of the words were presented consistently by one speaker (no speaker variability), while the other half were presented by 6 different speakers (high variability). Participants were then tested on their explicit recall of the novel words on the second day of learning and a week after learning. Participants also competed a lexical decision task on the second day of learning. This task measures interactions between novel and existing words and serves as a proxy for the strength of the newly formed lexical items.

The explicit recall task showed that there was no added benefit of speaker variability in remembering the newly learned words. The lexical decision task showed a marginally significant effect for variability (F=3.097, p=0.089, ηp2=0.094), indicating that lexical representations differed based on the amount of variation that the learner was exposed to. When the memory tests were factored into the lexical decision task, variability in the novel word primed lexical retrieval of existing words by 50ms, indicating a more robust lexical entry compared to words learned with no variability.

Learning new words with variation changed the way the lexical unit is created. Results support the hypothesis that phonetic variation is encoded within the lexicon, as opposed to being filtered out or not specified in our mental representations. My future work will focus on analyzing the pronunciations of the participants in the study. If phonetic detail and variation is encoded directly into the lexicon, then participants should show generalized pronunciations for words learned with variability, but mimic the speaker for words learned without variability.

Ewen Lee, University of Toronto

Fuzhou Tone Sandhi: Complications and Implications of an OT Account

Fuzhou is an Southern Min language that exhibits a complex pattern of right-dominant tone sandhi where the final syllable of a word group retains its citation tone, and the tone of the syllable preceding it surfaces with a sandhi tone. Some examples are provided in (1)-(3) to illustrate this process. These examples also include instances of initial consonant assimilation and vowel alternations conditioned by tone; these will not be accounted for in this analysis.

(1) /nieŋ53/ + /xuoi213/ → [nieŋ21 ŋuoi213] ‘age’

(2) /paiŋ242/ + /tsiu31/ → [pɛiŋ53 tsiu31] ‘to host a banquet’

(3) /thœy21/ + /xuɔŋ44/ → [thœy44 uɔŋ44] ‘annulment’

Multiple synchronic analyses have been attempted by researchers in the past (e.g., Chan 1985, Yip 2002, Donohue 2013), but these analyses have been largely rule-based. I propose an Optimality Theory account of tone sandhi in Fuzhou. Researchers have noted that the natural classes of the sandhi tones are somewhat opaque, and are also subject to inter-speaker variation. Most importantly, researchers report different tone sandhi patterns, further complicating the data. Donohue (2013) gives a phonetic analysis of the tone system based on the speech of 4 speakers, and derives a rule-based analysis to create a basic analytical framework for the tone sandhi process. I will use this set of rules as a foundation to formulate an OT analysis of this system.

To illustrate Donohue’s analysis, (3) will be used. A pitch value of [21] is assigned the label L in the [-upper] register, and [44] is assigned H in the [+upper] register. In (4), the tone of the first syllable deletes when it is followed by [+upper] in the next syllable. In (5), the H tone on the second syllable is associated with the now toneless first syllable. Finally, a register rule applies in (6) where all tones become [+upper]. The resulting surface form is [44 44].

Surface form: [HH] or [44] This analysis is not perfect, and fails to generate some of the sandhi forms. By modeling this system under a system of constraints, I plan to resolve at least some of these inconsistencies. Proposed constraints and the motivations for them are listed below. (7) *RISE – Rising tones are more marked in the sandhi forms compared to falling tones. (8) *CONTOUR – Surface sandhi forms generally do not cross register boundaries, so marking a violation for any time a register boundary is crossed ensures that this pattern is maintained. (9) OCP – Most of the tones dissimilate in the sandhi forms; OCP helps to explain this. (10) MAX-(TONE)LONGER-SYLL – This is taken from Hsieh (2005) in their description of Taiwanese tone sandhi to relate tone sandhi to phrase-final lengthening. The rightmost element must be the longest syllable, and if this constraint is ranked high, it will produce right-prominent tone sandhi. Crucially, it must be ranked above MAX-(TONE) to prevent the final tone from deleting.

By applying an Optimality Theory framework, this research aims to account for the complex tone sandhi that has not been accounted for in a satisfactory way. Similar endeavours have been successfully applied to related Southern Min languages.

For references and/or footnotes of the abstracts, click here.

Notes: The abstracts above are provided to TULCON13 by the presenters. We have tried our best to maintain the original formats and we are not responsible for the contents. If you would like to modify your abstracts and/or references section on our website please email us at