Thursday, July 31, 2025

On Foundational Work in Syntactic Theory

A longtime concern of mine, never voiced until now, is the feeling I get that the field of syntax is not very committed to foundational work. 

Simplifying a lot, the typical linguistics research agenda runs something like the following:

Standard Paradigm for Syntactic Research

1.

Identify some interesting dataset from some language.

2.

Adopt a specific set of syntactic assumptions (the theory).

3.

Show how to account for the dataset within the assumptions.

4.

If needed, tweak the assumptions to account for the dataset.

5.

Show how your account is better than other possible accounts within the theory.

6.

Show how the tweaking does not lead to inaccurate empirical predications.

This is the kind of work that linguists are by far the most comfortable with. Any change in the theory should be directly related to the data being analyzed. If a new theory is proposed, it should be evaluated on the basis of its success in handling data. On this way of looking at things, the link between data and theory is very tight.

I personally have no problem with this kind of work. I have engaged in such analyses many times in my life. But foundational work is of a different character. 

By foundational work, I mean work at the deepest level, analyzing the most basic assumptions of the field. Can the assumptions be understood precisely with no handwaving? What is the logical relationship between the assumptions? Can certain assumptions be eliminated? Are there implicit assumptions in the current theory that need to be brought out? What is the motivation for the assumptions over other assumptions, which may be more or less equivalent in terms of empirical coverage? Why is linguistic theory the way it is (instead of some other logically conceivable way)?

Foundational work is often driven by the desire to understand the underlying order of a formal system, the symmetries and asymmetries, the hidden connections between seemingly unrelated assumptions, and interesting unexpected explanations for why the system is the way it is. The belief is that the human faculty for language is governed by simple and elegant principles, and it is our job as theoreticians to try to uncover them.

The research agenda runs roughly like this:

Foundational Paradigm for Syntactic Researc

1.

Identify a syntactic theory.

2.

Identify a foundational issue within that theory (related to the questions posed above).

3.

Try to gain insight into that foundational issue. 

My characterization of these two paradigms is overly simplistic. Of course, there may be empirical consequences of foundational work. And there are degrees along the scale defined by the two poles above. There is not always a clear-cut distinction.

But the value of the foundational work may just be to gain insight into how the theory works. One of the most interesting papers of this kind was Sam Epstein’s paper on c-command ‘Un Principled Syntax and the Derivation of Syntactic Relations’, which subjected a fundamental notion (c-command) to close conceptual scrutiny, asking questions about why c-command was defined the way it is (instead of some other way), and answering those questions in terms of a derivational model of syntax. My paper on ‘Eliminating Labels’(and related work by Daniel Seely) was written in a similar spirit, proposing that the syntactic notion of label is unnecessary in syntactic theory.

Through many personal experiences, I know that it is very hard to get colleagues and students to engage in thinking about foundational issues. When you start asking foundational questions, people’s eyes gloss over, and they start yawning. In fact, some people even seem to be uncomfortable with foundational work. For example, one reviewer felt obliged to add the following comment to their review of one of my foundational papers:

“I cannot say that this is the kind of paper I care much for. It carefully and clearly discusses an issue that leaves me entirely cold. “

This comment was not about the arguments of the paper. It was about the topic itself. According to the reviewer, the issue left them cold, with the implication that doing such work was not important, and maybe even misguided. I have had similar experiences not just with reviewers, but in many in other areas of academic life (e.g., course enrollment, hiring decisions, selection of colloquium speakers, etc.).

If I am right, it raises the question of why? Why is the field of linguistics so apathetic to foundational work? I think there are a few reasons.

First, foundational work takes a lot of time and effort, and the payoff is uncertain. The question is what counts as progress. You may think for years and years about the ‘copy versus repetition’ distinction or about the definition of c-command or about the nature of workspaces without resolving the issues, even though in the end you have a much deeper understanding. Does that count as progress? Can you write it up and publish it? Does it count as currency in the academic monetary system?

Second, foundational work does not fit into the standard syntactic paradigm, which is the paradigm everybody recognizes as valuable. Meditating on the definition of c-command seems less satisfactory than showing that a particular definition of Agree accounts for more data than a different definition of Agree. Linking to data is a certain way to measure progress. If the work is not tightly linked to data, it may not be clear what the progress is.

Third, the field is beset by what I can call ‘the busy hands syndrome’. If a theory provides many opportunities for analysis of various kinds, including interesting new mechanisms to play with in different ways, it attracts people. It supplies with them things to do. But foundational work does not have this character. Rather, one focuses on some conceptual issue to try to gain insight into it. 

Fourth, and related to the previous points, from the standpoint of a graduate student, the question is what kind of work will they be able to present at conferences and publish before they enter the job market. Committing to foundational work might not do the trick, if the rest of the field feels that it is not important.

To end this blog post, we can ask the question of how foundational work should fit into the future of the study of natural language syntax. If everybody is chasing data, competing on the best definition of Agree, or the scope of the impoverishment operation, or the number of projections in the left periphery, are we really making progress understanding the fundamental principles at play in syntactic theory? 

I am definitely not claiming that such empirical work is unimportant. Far from it, it is the bread and butter of syntactic theory. But contrary to popular belief in our field, having an empirically adequate theory does not mean that it is a good theory. If your theory provides easy explanations for X, it does not necessarily add to the value of your theory. This is why, years ago, Chomsky introduced the notion of ‘explanatory adequacy’, in order to go beyond the clever organization of data into rules. 

The purpose of foundational work is to push us out of our comfort zone with the goal of helping us to find deeper explanations for syntactic phenomena.

Wednesday, July 23, 2025

Evedada -- A Traditional Ewe Game

 July 22, 2025 Tuesday

It rained all morning, so we could not really do any serious lexical recordings.  It was a tropical raining, beating down heavily. The background noise on the recordings was just too great. Instead, we decided to play the game evedada (eve-throw-throw), which is a traditional game amongst Ewe boys and men. We got a video of the game. The players had a great time.

Here are my notes. This is the first time I have studied the game, so there are probably a few misunderstandings.

Monday, July 21, 2025

Ghanaian versus Togolese Ewe

Abstract: This paper shows that there are features that distinguish dialects of Ewe spoken in Ghana from dialects of Ewe spoken in Togo. Ewe dialects spoken in Ghana are collectively referred to as Ghanaian Ewe, and those in Togo are collectively referred to as Togolese Ewe. 

Keywords: Ewe, dialect, Ghana, Togo

Ghanaian versus Togolese Ewe


Friday, July 18, 2025

Linguistic Subfields and Genetics

I am not a natural born lexicographer. Far from it. 

I have written one small dictionary for Sasi, a Khoisan language of Botswana. Now, I find myself starting another dictionary, for the Kpele dialect of Ewe. So I have some idea of what it takes to write a dictionary of a certain sort. But I am forcing myself to do it, and it does not come naturally. 

I believe that every linguist is basically born into the field that they adopted. Their innate disposition, including various skills, lead them to the field that they choose: syntax, phonology, semantics, lexicography, sociolinguistics. There would only be only a correlation between genetics and final behavioral outcome, but my hypothesis is that a correlation exists. 

To be more concrete: My hypothesis is that within the set of professional linguists, it will be possible to discover a correlation between genetic traits and career choice (defined broadly into one of several categories: syntax, phonology, semantics, sociolinguistics, lexicography). This is an example of a polygenic index, as described in Dalton Conley’s recent book ‘The Social Genome’. For a summary of the concept, see: https://en.wikipedia.org/wiki/Polygenic_score

My prediction is that if there were a study of identical twins born apart, both of whom became linguists, then they would choose the same field in linguistics (with a greater than random probability). For example, they would both be syntacticians. Since linguistics is a relatively small field, I doubt any actual cases exist, but the prediction is clear.

It is not a question of IQ, because there are absolutely brilliant linguists (genius level linguists) in all subfields. IQ seems to be a different notion. Rather, it is question of propensity to do certain kinds of analysis. If anything, I am a natural born syntactician. I love to manipulate sentences, and witness the effect on acceptability and meaning. I could play with sentences all day long, and never tire of it. It is a source of endless fascination. 

But keeping track of all the intricacies of lexical items and making sure to create uniform lexical entries becomes tiring for me. It does not inspire me in the same was as syntax. For me, lexicography is an obligation as a field linguist, not an adventure. It is something I need to do, because the information is important. For the languages I work with, nobody else is in a position to do the work. So I am obliged.

Sunday, July 13, 2025

Languages I have Studied

This is a list of the languages I have studied in my life. Of the 42 languages below, 29 are African languages, showing the clear focus of my career on African languages. I feel very blessed to have been given a life where I was able to investigate the mysteries of these languages and of the human capacity for language more generally. For a fieldwork oriented linguist, I think this list is pretty typical. 

They can be classified as: 

(a) languages I have studied in the field (Fieldwork), 

(b) languages I have studied in a Field Methods course (Field Methods), 

(c) languages I have studied in a course somewhere (Course), 

(d) languages I learned in the Peace Corps (Peace Corps), 

(e) languages I written a paper about (Paper). 

I also list the location where I studied the language. Of course, the list would be much larger if I added all the languages that I had read about in books or papers.

1. Brazilian Portuguese (NYC, Course)

2. Cua (Botswana, Fieldwork)

3. Chuukese (NYC, Field Methods)

4. English (Minnesota) (Native Language)

5. English (AAE) (NYC, Paper)

6. Ewe (Kpelegbe) (Togo, Fieldwork)

7. Ewe (Standard) (Togo, Course, Peace Corps)

8. Ewe (Danyigbe) (Togo, Peace Corps)

9. Ewe (Gengbe) (Togo, Peace Corps, Field Methods)

10. Ewe (Pekigbe) (NYC, Field Methods)

11. Ewe (Wudu) (Togo, Fieldwork)

12. Ewe (Tongugbe) (NYC, Paper)

13. French (Togo, Peace Corps)

14. German (NYC, Paper)

15. Greek (NYC, Paper, Field Methods)

16. Kabiye (NYC, Field Methods, Paper)

17. Khoekhoe (NYC, Field Methods)

18. Kua (Botswana, Fieldwork)

19. Icelandic (Cambridge, Paper)

20. Ife (Togo, Fieldwork)

21. Igbo (Ithaca, Field Methods)

22. Ju|’hoansi (Namibia, Fieldwork)

23. Lubukusu (Ithaca, Field Methods)

24. Masalit (NYC, Field Methods)

25. Medumba (NYC, Field Methods)

26. N|uu (South Africa, Fieldwork)

27. Okri/Cherepong (Ghana, Fieldwork, Field Methods)

28. Sasi (Botswana, Fieldwork)

29. Setswana (Botswana, course, Fieldwork)

30. Sherpa (NYC, Field Methods)

31. Shughni (NYC, Field Methods)

32. Shupamem (NYC, Field Methods)

33. Sinhala (NYC, Field Methods)

34. Songhay (NYC, Field Methods)

35. Spanish (NYC, Course, Paper)

36. Swahili (Ithaca, Course, Paper)

37. Thai (NYC, Field Methods)

38. Twi (Ghana, Course)

39. Tshila (Botswana, Fieldwork)

40. Wolof (Ithaca, Field Methods)

41. Yoruba (Ithaca, Course)

42. =Hoan (Botswana, Fieldwork)

 

Thursday, July 10, 2025

Using Whatsapp for Linguistic Fieldwork

Whatsapp is a smartphone app that allows people in different countries to communicate over the phone for a very reasonable price. 

In addition, Whatsapp also allows the creation of Voicemail messages that can be used in linguistic fieldwork. Here are the basic steps for using Whatsapp for fieldwork. I thank Claire Bowern for some helpful suggestions. 

1.

Buy a smart phone and install Whatsapp.

2.

Have your consultant buy a smart phone and install Whatsapp.

3.

You should be able to call each other and write messages back and forth. 

4.

Install Google Drive as an app on your smart phone (using the Google Play Store app).

5.

You may also have to change the permissions on Google Drive. I went to the Google Drive app (installed on the smart phone), and disabled ‘Transfer uniquement en Wifi’. You may have to play with it a bit.

6.

In a Whatsapp message, write the word or the phrase that you want your consultant to speak. For example, “Say the word for ‘dog’ three times.” Send the message.

7.

Your consultant will create a voicemail message with the response to your question. To do so, they click on the mic to the right of the message window, and start to record. For example, your consultant will speak the word for ‘dog’ three times. Once the voicemail message is complete, they will send it to you. 

8.

When you get the message, press down on and select the voicemail. Then in the top right corner of your Whatsapp app, click on the menu and chose ‘share’.

9. 

Since you have installed Google Drive as an App, one of the options that appears should be Google Drive. Select it.

10.

Go to the folder (e.g., ‘Whatsapp Sound Files’) you have created in Google Drive, and download the voicemail (I clicked ‘Importer’ in my Google Drive app).

11.

You can also send the voicemail to your e-mail account. As in step (8), chose ‘share’. Then select Gmail from the options given. This process is a little more awkward, since you will need to open a new e-mail message for each sound file you produce (instead of just having them in a Google Drive folder right away).

12.

The file type will be .oga, which is not readable by Praat. So you need to convert it to .wav.

13.

Download Audacity to your computer (it is free).

14. 

Open the .oga file in Audacity, then save it as a .wav file.

15. 

The file name should be: XX_dog.wav, where XX are the initials of your consultant (e.g., CC for ‘Chris Collins’). Since you are using the file in Praat, there should be no special characters other than underscore.

16. 

You should now be able to open your file and analyze it in Praat (e.g., transcription)

17. 

Using these sound files, you can start to create a dictionary or a database of recordings for future use. 

18.

I do not yet know the size limitation on sound files, but you should be able to record three repetitions of a word, and any sentence. I am not sure whether you can record a five-minute long oral text yet. I need to experiment with that.

19.

I am not entirely sure whether the sound files are up to the standards of phonetics research, but they sound and look pretty good in Praat. Some verification of the quality of the sound files is needed. That is, a direct comparison is needed of sound files created by a ZoomH4n and by Whatsapp voicemail.

20.

I have yet to carry out a real project using this process, so it may have to be fine-tuned in the future.