Research

Syntax

My current syntactic research focuses primarily on languages with rich subject- and object-marking systems. I am particularly interested in applying theoretical techniques from generative syntax to under-documented languages. My dissertation work focuses on the argument structure of verbs and nominalized verbs in the Montana Salish/Kalispel/Spokane language, spoken in Northwestern Montana and Eastern Washington. In addition to this I have done some preliminary work, including syntactic elicitation, with several Bantu languages in Northern Malawi.

Dissertation Work: Salish Argument Structure

As an undergraduate at the University of Michigan, Sally Thomason introduced me to Montana Salish, a dialect of the Salish language spoken in Western Montana and Eastern Washington. In my dissertation, I conduct syntactic and morphosyntactic analyses of verbs and nominalized verbs in this dialect and its sister dialects Kalispel and Spoakne, using her years of field notes as well as a published dictionary compiled in the 1870s (Giorda and Mengarini, 1877-79) as my primary sources of data. The dissertation is essentially a case study of the morphological realizations of verbal argument structure phenomena in Montana Salish/Kalispel/Spokane. I test several cross-linguistic models of argument structures that have been proposed in the Distributed Morphology literature by examining how well or poorly they may be applied to these dialects. My work thus provides support for some of these DM models, while the same time highlighting some interesting typological features of Interior Salish languages.

In plainer terms, I am studying the types of verbs that exist in Montana Salish/Kalispel/Spokane, the ways different types of verbs may be modified, and the effects that these modifications have on the rest of the sentence. To give an example of the types of phenomena I investigate, intransitive verbs in these languages may be divided into two categories, illustrated by the pair of examples below:

Čn tiyéš --- 'I crawled'
Čn t̓ɫíɫ --- 'I got dirty (by accident)'

While the subject of sentence 1 is interpreted as the agent, or do-er, of the verb, the subject of sentence 2 is interpreted as the patient, or undergo-er, of the verb. Across the world's languages, verbs that behave like the verb in sentence 1 (taking an agent as the subject in the intransitive) are called 'unaccusative' verbs while verbs that behave like the verb in sentence 2 (taking a patient/theme as the subject in the intransitive) are called 'unergative.' And as in most languages, this essential category distinction has wide-ranging effects on the way these verbs can be used throughout the language.

For example, unaccusative verbs (sentence 2) can be turned into nouns by adding a suffix -min. This suffix acts like the -er suffix in English to create a noun describing a tool used to perform a given action (compare dry/dryer or mix/mixer). Unergative verbs (sentence 2) may be turned into tool nouns as well, but they are nominalized using a different suffix, -tin. Tool nouns formed with -min and -tin generally seem to have the same meaning when used on their own, both describing tools used to perform an action. However, when -min and -tin nouns are possessed (as in my dryer), there is actually a semantic difference between the two sets of nouns. While the possessor of a -min noun is interpreted as the owner or possessor of a tool used to do something, the possessor of a -tin noun is interpreted as the person typically affected by such a tool. In English, a noun like dryer can lead to both these interpretations. My dryer might mean, the tool that I own and use to dry things or, in unusual circumstances, it might refer to a tool (or more commonly a person) commonly employed to dry me. In Salish this ambiguity is not present because -min nouns receive the first interpretation while -tin nouns receive the second.

In my dissertation I investigate phenomena like the -min/-tin distinction and how these phenomena are related to the way verbs fall into different categories.

Bantu Elicitation

In the summer of 2016, I visited Chisenga, a small village in the Chitipa region of Northern Malawi and conducted elicitation sessions in with some of the residents. This is an area of Africa with rich multilingualism. In Chisenga alone at least six different languages, Chinyika, Chindali, Chitumbuka, Chichewa, Chilambya, and English, are in regular use, with immigrants to the region speaking even more. This is an interesting region to study because many of these languages are closely related and yet display different syntactic characteristics which suggest that while most sentences in these language look similar on the surface, speakers of each language may be analyzing that underlying structure in different ways. This type of variation between related languages is known as microvariation. Studying this sort of microvariation has the potential to teach linguists a great deal about the different potential structures which can underly a given speaker's knowledge of language

In the long-term, I am interested in investigating the microvariation within the "object agreement" systems of these languages. Object agreement is a system by which verbs are marked in a way that must match some feature of the verb's object. In English, the verb in a sentence "agrees" with the subject of a sentence (for example, we pronounce the verb read differently depending on the person and number of the subject, you read vs. she reads). In object marking languages, however, the verb agrees not only with the subject but also with the object of the verb. For example, in Chindali, the verb a-a-mú-bek-a, meaning "he/she tricked him/her" contains a prefix, -mú-, that only occurs when the object third-person singular and human.

The object marking patterns of the languages in Chitipa are interesting because speakers of each of these languages seem to place slightly different restrictions on what kinds of linguistic environments make it necessary for you to mark the object on the verb. In some languages, for example, the object marker is only included if the object is particularly important ("under focus") while in others that marker is necessary whenever the object is animate.

Though I hope to go back and study the syntax of these languages and learn more about the dialects spoken in Northern Malawi in future, the elicitations I have so far been able to conduct in this region have been mostly exploratory. In elicitation sessions with four consultants of varying age, we discussed wh-questions, the interaction between agreement and conjunction, and some features of these languages' derivational morphologies. In addition to this I collected a few personal narratives. If you would like access to any of this data, please contact me. I have the following:

Syntactic Elicitation Session (Recording and Transcription) - Chinyika (3) and Chindali (1)
Narratives (Recording and Rough Transcription) - Chinyika (2) and Chitumbuka (1)

Lol and the Pragmatics of Written Conversation

A second major area of interest for me is the pragmatics of written conversation. I am particularly interested in the ways we adapt the way we use language to accommodate for the fact that when talking online or through text message we are taking part in (near)-instantaneous conversation in the absence of visual or auditory cues. The characteristics of written conversation I examine most carefully in my research include punctuation, emoji, and various letter-based discourse markers which are primarily used on the internet (ex: lol, jk, omg, smh).

My past research has focused in particular on the various ways we use letters and symbols to replace laughter in text. My senior thesis at the University of Michigan as well as several papers I currently have in preparation examine the extent to which written laughter in written conversation does and does not parallel spoken laughter in spoken conversation. These papers all use conversational data taken from Twitter.

The first paper, largely adapted from my senior thesis, is based solely on English data and uses a large corpus to evaluate the extent to which various written laughter forms can be considered a single category. This analysis includes several onomatopoeic forms (ex: haha, hehe, tehe), several abbreviation forms (ex: lol, lmao, rofl), and several emoji forms (ex: 😂, 😆, 🤣). I conclude that letter-based forms of written laughter show enough similarities in terms of their conversational distribution to justify treating them as a single discourse-marking category. This means that though there are slight differences between the social meanings of different sorts of written laughter, they all are placed at similar locations within sentences and conversational turns. Furthermore, English speakers seem to place these spelled-out forms at similar locations within written conversations to the places we use physical laughter in spoken conversation, suggesting that we may be using physical laughter as a model for the appropriate placement of written laughter in conversation. Interestingly, though my research clearly shows that emoji show less flexible distributions within conversational turns from spelled-out forms, they seem to preform similar functions to spelled-out forms when they occur in the same locations within a conversational turn. This may mean that emoji laughter is written laughter with more extensive restrictions on its distribution.

Putting aside the similarities between the distributions of different laugh forms, for the moment, there still some very interesting differences, between the use of different written laughter forms. Based on my observations, just as when choosing any word, a speaker's choice of a written laughter word can perform social functions. Some forms mark the speaker as a member of a group (age, gender, race, class...), others may mark the tweet as being of a certain register (more formal, more casual...), and others mark to frame the purpose of the conversation (flirtation, making friends, making fun of a third party...). The tables below illustrate some of these framing and register differences, indicated by words that tend to occur more often in tweets containing some laugh forms than others. In the three charts below I show that words of different discourse categories (negation, pleasantries, and harsh/profane language, respectively) tend to be associated with different forms of written laughter. All of the words with stars beside them are statistically significant.

The second paper, which I presented at the International Pragmatics Association conference in 2017, looks at written laughter cross-linguistically. This paper examines written laughter forms of the three categories discussed above both in English and in Hebrew, a language from a different language family using a different orthography. In this study I found that not only do written laughter words in English fall into a single category, but written laughter words in Hebrew follow many of the same patterns. Because physical laughter is one of the forms of communication most deeply embedded in human psychology, its conversational distribution is largely the same cross-linguistically. The similarities between the conversational distributions of English and Hebrew forms further suggests that speakers of both languages may be basing their use of written laughter on their use of physical laughter.

The graphics to the left show some of the patterns displayed by written laughter. Each of the bars below represent tweets where the laugh form in question occurs in a different position within the tweet: initially, medially, finally, and "solo" (meaning it is the only word in the tweet). The height of each bar indicates the percentage of tweets with that form of laughter in that location that are targeted at a specific individual (meaning they begin with @username and are designed with a specific recipient in mind). Note that for most forms of written laughter in both languages, tweets in which the laughter word occurs initially or alone are the most likely to be targeted at a specific recipient. This is a result of the indexicality of written laughter. Initial-position laughter must reference something playful in external conversational context, requiring another conversational participant, while medial- or final-position laughter may only reference something within the tweet, which does not require previous context. Note that these patterns hold for spelled-out forms as well as emoji forms!

Future projects which I hope to conduct in this area include an analysis of how bold text, italics, and underlining are used in online communication, the use of punctuation forms such as ellipses (...), asterisks (*) and greater-than/less-than signs (>,<) in conversation. I am also interested in comparing the methods used by individuals who communicate regularly (or have communicated reguarly) in Morse Code to those used on the internet.

If you are interested in conducting research using either my corpus data (which is quite extensive) please contact me for more information about what I can offer you. I'd be happy to share! I also have code which allows for the swift and easy collection of conversational data on Twitter along with indexed personal data about each of the users involved in the conversation. If you are interested in using this code, please contact me!

Google Sites

Report abuse