Linguistics Anonymous

27 April 2006

ideas needed


Someday I may post something more substantial than this, but today I am trying to finish up a little honors thesis and I really need a title. Since I don't have time to put anything more substantial on here right now, it is basically a synchronic and diachronic analysis of the construction try and +verb, and an analysis of a new pattern which inflects try, producing sentences like "I am trying and write this paper" or "I tried and think up a title myself but had to resort to asking blog-people." Any and all ideas are welcome. If I use yours I will give you a signed copy, bound to be worth thousands of dollars someday. Please help!


Learning LaTeX?

Byron gave me an idea for something to put up here.

I have been meaning to do this for awhile, but now thanks to Justin from Reed College (whom we met at the Harvard LinG Colloquium), I have finally begun to learn LaTeX, a very flexible and powerful markup language for formatting documents and papers. It's widely used in academic circles, as well as being the reason that nearly every paper you ever see at a conference from someone at MIT looks the same.

So here are a few of the resources I used to get myself up and running on LaTeX. One should keep in mind that "up and running" is a gross overstatement: I can still barely do anything. I will, however, update this as I learn more. Also, I own a Mac, so if you don't use a UNIX based system, I'm sorry - you'll have to tailor this to your own machine.

For getting started:
LaTex For Linguists (Just bookmark this - you'll need it)
Beginner's Guide to LaTeX (the best beginner's reference I've found)
Another Beginner's Guide (also very good)

Software for Editing (Mac OSX Only)
TeXShop (easy to use)
TeXMaker (incredibly versatile)
BibDesk (for bibliographic databases)

This should be enough to get folks started. I can't stress how helpful the "LaTeX for Linguists" website was - good stuff on trees, OT Tableaux, IPA Fonts, Semantics notation, etc. Check it out.

Trees in Powerpoint

So, until now, I have had yet to post here. Mostly a product of me being lazy, thesising, visiting grad schools and trying to maintain a semblence of a non-academic life. Anyway, this post isn't much either, but it's something. When I was working on my honors thesis defense powerpoint presentation (that's a really long compound) I decided that I needed to have movement actually taking place on the screen since I have a lot of it and a static picture doesn't necessarily cut it (read: it's too complicated for its own good).

After a little bit of toying with powerpoint, I discovered it's actually fairly simple once you get the hang of it. The biggest piece of advice I can give on the subject is to build your whole tree first on the slide and then and only then start doing the animation process. To give a basic idea of how to do one yourself, I've posted a sample powerpoint on the interweb for your use. Download it here.

26 April 2006

Conference Report

So as Matt mentioned a while ago, he and I presented at the Harvard Undergraduate Linguistics Colloquium this past weekend, along with our colleague Annie. Our presentations all went very well, and we had a fantastic time. The quality of the presentations overall was remarkable, and we met a lot of very cool people.

Both Matt and I promoted this blog in our talks, so it's possible that we have some new readers--I see that Bridget Samuels has already commented on the post I linked above. If anyone else out there found out about us from the conference, please go ahead and leave us some comments; there are quite a few posts now, and many of them (especially Matt's) contain very sharp insights and summaries of important papers. We would definitely appreciate any responses you have to our ideas.

24 April 2006

The Arabic Pronoun Asymmetry: Solved

For reference, see my post: Partial Agreement in Arabic Clauses

In the above post I noted that there was a particular problem with the analysis that I had developed of the Arabic asymmetry given by feature strength in that pronouns seemed to be a marked exception. Concretely, pronouns do NOT raise to a pre-verbal position despite a verbal element which inflects fully for number, unlike other VSO clauses in spoken Arabic.

The major insight is this: a hidden thrust of the previous post was that syntactic movement and agreement of the kind seen in Arabic might be analyzed wholly in terms of feature checking and the mechanism of AGREE, in line with Checking Theory (Chomsky 1995). So in furthering this analysis, it makes sense to ask a rather bold question: to what extent could we forgo major category features and concentrate solely on agreement relations between syntactic constituents?

Luckily enough, this is precisely what Ouhalla (2005) seeks to do. In his analysis, specialized category features have no theoretical motivation (or status). Instead, categorization is driven by context: [PERSON] (on the head Pred) values a syntactic element verbal and [CLASS] (on a head of the same name) values a syntactic element nominal, once an agreement relation is established between the two.

This is a fairly hefty proposal, and I cannot do it justice here. However, there is one thing that can be of use to us: Ouhalla points out that it is not a good idea to associate the [PERSON] feature with EPP in his analysis (which one might be tempted to do upon reading his work), since it conflicts with the modern notion of EPP. However, it is still necessary for pronominal elements to check and delete their [PERSON] feature against the Pred head to avoid conflicting feature sets with Pro[PERSON, CLASS]. Regular nominals, on the other hand, do not have a [PERSON] feature, since unless they are a pronoun, they carry default 3rd person (obviously I am skimming over a lot here in the interests of brevity - I direct the reader to Ouhalla's article for a proper treatment of his analysis).

Since we do not want to associate EPP (in other words, the feature associated with raising to [Spec, T]) with [PERSON], but pronominals must check and delete it, we can see that pronominals must enter into an Agree relation with the verbal element (thereby valuing it for number, unlike regular nominals, which do not do so for phi-features) in order to check off [PERSON]. But since [PERSON] does not trigger raising, the pronominal must stay in-situ.

This leaves one open question: how do regular nominals raise, then, if they are not agreeing with the verbal element for [NUMBER] (as was claimed in the initial post). The exact mechanisms of this are still to be worked out, but one might gain insight from Ouhalla:

"...while nouns have [CASE], pronouns have [PERSON], or, to put it differently, [CASE] is to nouns what [PERSON] is to pronouns" (pp. 682).

Thus the conditioning factor on number agreement is the presence of a [PERSON] feature on the nominal that must be checked, and raising is obligatory in the presence of a [CASE] feature on a nominal.


Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT Press.

Ouhalla, Jamal. 2005. "Agreement Features, Agreement and Antiagreement." Natural Language and Linguistic Theory 23: 655-686.

21 April 2006

A Lateral Kick to the Shins: Part 8

I'm done with the substantive posts on this subject, but I thought I should collect them all in order so that new readers (if we ever get any) can read them in order and better understand the sequence. I've also put up a list of references I've used in the course of this project, not all of which are mentioned in these posts. Feel free to use the comments to this post for any questions or complaints.

A Lateral Kick to the Shins: Parts 1, 2, 3, 4, 5, 6 and 7.


Aro, J. (1955). Studien zur Mittelbabylonischen Grammatik (Studia Orientalia, 20). Helsinki: Societas orientalis Fennica.

Böhl, F. (1909). Die Sprache der Amarnabriefe mit besonderer Berücksichtigung der Kanaanismen (Leipziger Semitische Studien, 5/2). Leipzig: J. C. Hinrich.

Faber, A. (1989). On the nature of Proto-Semitic *l. Journal of the American Oriental Society, 109, 33-36.

Gordon, M., Potter, B., Dawson, J., de Reuse, W., & Ladefoged, P. (2001). Phonetic structures of Western Apache. International Journal of American Linguistics, 67, 415-448.

Henderson, E. J. A. (1970). Acoustic features of certain consonants and consonant clusters in Kabardian. Bulletin of the School of Oriental and African Studies, University of London, 33, 92-106.

Jucquois, G. (1966). Phonétique comparée des dialects moyen-babyloniens du nord et de l’ouest (Bibliothèque du Muséon, 53). Louvain: Université de Louvain, Institut Orientaliste.

Ohala, J. J. (1993). Sound change as nature’s speech perception experiment. Speech Communication, 13, 155-161.

Steiner, R. C. (1977). The case for fricative-laterals in Proto-Semitic. New Haven: American Oriental Society.

20 April 2006

A Lateral Kick to the Shins: Part 7

I promised more data, so here it is. The figure to the left shows FFT spectra for the fricatives of Western Apache, a language that has both of the sounds involved in the Akkadian sound change I've been discussing (from Gordon et al., "Phonetic Structures of Western Apache," International Journal of American Linguistics, 2001). As you can see from the labels, the fricatives were pronounced by male speakers before [a], so these examples, like those from Kabardian, are not perfect matches for the environment of the Akkadian change (in medial clusters before coronal obstruents). However, it is well known that consonants tend to be reduced in various ways in clusters, so demonstrating similarities between consonants in highly salient contexts such as syllable-initially should be sufficient to suggest that they could be confused when reduced in clusters.

Note that the two segments (the top two squares in the diagram) are very similar. They both have spectral peaks around 3000 Hz, although the peak for [š] is slightly higher in both frequency and amplitude. The main difference is that [š] has lower intensity in the low-frequency area, although the energy distribution is still remarkably similar, with a steady decline up to around 1000 Hz followed by a fairly sharp rise to the spectral peak. The two are particularly similar when compared to the other two fricatives in the language, whose spectra look very different, with different energy distributions in different frequency ranges. I think this evidence supports my contention that the two sounds were similar enough acoustically to be confused by listeners when heard in medial clusters.

The main issue remaining is also the most baffling: why did this change only happen before coronals? After all, /l/ could easily have been devoiced before all voiceless obstruents, and the same perceptual confusion I have posited here could have occurred in those environments. So why was there no sound change there? Why was it limited to just this narrow context? And why did it spread so far within that context without spreading further?

I don't have answers to these questions, important though they are. I only have suggestions, which could explored in further research. The main idea I have, in keeping with the acoustic explanation I have given in these posts for the sound change as a whole, is that there was something about the transitions going in to the coronal segments that made a perceptual confusion more likely. Transitions are very important for perception, so I think this is most probably the direction to go to figure out why the change was so restricted. One way to investigate this, which I didn't do for this project due to time constraints, would be to find native speakers of a language with these segments (for best results, one in which /l/ was phonologically devoiced in comparable contexts) and record tokens with medial clusters like those of Akkadian. Examining those recordings would go a long way toward finding a solution to the issues mentioned above; any particular acoustic features of the clusters could be tested for salience through perception experiments or other techniques of experimental phonetics. My presentation is on Saturday, though, so that will have to wait for another time.

A Lateral Kick to the Shins: Part 6

So what is this mysterious phonetic evidence I've been hinting about? How can you possibly use phonetic evidence to examine a language that has been dead for over 2000 years, anyway?

Well, the short answer is that you can't. We don't know exactly how Akkadian was pronounced and we probably never will. Nonetheless, sound changes like this one and other indirect evidence can give us a pretty good idea of how most phonemes were realized, and we can then look for living languages that have those same sounds and examine recordings of them to try to find out about what Akkadian may have looked like acoustically.

The languages I have used here all have the two sounds involved in this sound change, [š] and [ɬ], although they have both of them as phonemes whereas the latter was only allophonic in Akkadian (according to my interpretation). This is a matter of abstract phonology, though, and shouldn't matter for acoustic phonetics. I should note that I was unable to find any examples of these phones in the most useful environment, in clusters before coronal obstruents, so the following spectral evidence will not quite be identical with how I'm picturing the sounds involved in the Akkadian change. I think the basic points are still valid, though.

We start with Kabardian, a Northwest Caucasian language spoken in the Russian Caucasus. This language has the sounds in question, and they are reflected in these spectrograms (from E. J. A. Henderson, "Acoustic features of certain consonants and consonant clusters in Kabardian," Bulletin of the School of Oriental and African Studies, University of London, 1970).

The top image is of the Kabardian word /šɨ/, and the bottom image is of the word /ɬɨ/. Note that the main difference is the frequency at which the high-intensity noise begins; in the top image this is around 1500 Hz, while in the bottom image it's more like 2500 Hz. Otherwise the two images are pretty similar. Both show high-intensity noise concentrated around 5000 Hz, as well as rather well-defined formant structure. This paper was published in 1970, so the technology used for the analysis was a far cry from the advanced digital methods we have now, but it still shows the similarities between these two sounds very well.

I was originally going to try to put all the phonetic evidence in this one post, but it's pretty late, so I'm going to just finish this post and leave the rest for another. Stay tuned, though, because the other evidence is even more exciting (if you can believe that).

19 April 2006

A Lateral Kick to the Shins: Part 5

John Ohala is a professor at Berkeley (now emeritus) who has done an enormous amount of work on the phonetic mechanisms of sound change. He has concluded that the primary source of sound change is incorrect parsing of speech sounds by listeners, either by dissociating signals that are meant (by speakers) to represent the same sound or by associating signals that are meant to be distinct sounds. In the Akkadian change we are discussing here, it is clear that the origin of the change must be the second of Ohala’s processes: listeners hearing [št] and [ɬt] and assuming they are the same, then adjusting their own speech to pronounce the two sequences identically as [ɬt]. Although the two fricatives are distinct articulatorily, there is considerable evidence that they are similar acoustically. I will present some such evidence (from languages which have both as phonemes) in a later post; right now the important thing to consider is just that they are indeed similar, and this similarity led to a perceptual confusion.

Here I would just like to detail the process I am proposing for this sound change, since I'm not sure it was very clear in my original paper. Basically, I'm arguing that there was a phonological rule in Akkadian starting at an early stage (it's impossible to tell exactly how early, but certainly by the Old Babylonian period) which devoiced /l/ before /t/ and /s/. This rule, although it is necessary for explaining the sound change, is separate from it, and may well have operated before other voiceless obstruents as well; there's just no way to tell from the textual evidence since no sound changes occurred in those environments. Later, during the Middle Babylonian period, the perceptual confusion I mentioned above began to develop, with texts showing sporadic cases of for <š> before . As this confusion continued through generations, the pronunciation of /š/ moved more and more in the direction of [ɬ], until finally the two allophones (of different phonemes, remember) became identical in pronunciation and the allophone of /š/ in this environment actually became [ɬ], which was also the allophone of /l/. This is thus a total merger in a highly specific context. Later, by the time of the Neo-Babylonian period, the change had spread analogically to environments before the other coronal obstruents /d/, /ṭ/, /s/, and /z/; the resulting phone was presumably pronounced as a voiced [l] before the voiced segments (its realization before the emphatic is unclear due to our lack of knowledge of its pronunciation).

This is a complicated, multi-step change, but I think it is the only way to adequately explain this unusual shift. In my next post I will present some phonetic evidence to bolster my contentions here.

07 April 2006

A Lateral Kick to the Shins: Part 4

This change is unusual from a phonological perspective, since /š/ and /l/ share few features and it only occurs in a very restricted environment, so it has attracted much attention from scholars over the years. Some early Assyriologists proposed that the answer lay in reassessing the phonetic values of the signs in question; for instance, C. F. Lehmann proposed in 1892 that /l/ was voiceless in Akkadian, while Yehiel Gumpertz suggested in 1942 that it was rather the sound written <š> that was pronounced as a voiceless lateral fricative [ɬ]. These ad hoc proposals, although they do explain the sound change, don't make much sense in the context of the language as a whole, and there is no reason aside from this one change to believe either one.

A more realistic solution is to propose a simple devoicing rule: /l/ --> [ɬ] / _ [(+cor,) -son, -voice] (or something similar). That is, the voiced lateral approximant becomes voiceless before a voiceless (coronal) obstruent. I have placed "coronal" in parentheses because it is not clear how restricted this rule is, and there is probably no way to tell from the evidence available if it operated generally or just before coronals. This is, as you can probably tell, my preferred solution, but I didn't come up with it on my own; it is also proposed in Alice Faber's article "On the Nature of Proto-Semitic *l" in the Journal of the American Oriental Society 109:1 (Jan.-Mar. 1989), p. 33-36. However, her thesis that there were two lateral phonemes in Proto-Semitic, an alveolar /l/ and a velarized /lˌ/, I find dubious at best, and her explanation of how, exactly, this change supports that thesis is muddled. Despite these problems with her proposal as a whole, I have adopted her basic explanation of the sound change, which is to propose the above phonological rule and a subsequent shift of /š/ due to perceptual confusion with the voiceless allophone of /l/ in this environment. I will support this by giving phonetic data from languages that have both postalveolar and voiceless lateral fricatives and showing how they could be confused in medial clusters. For theoretical support I will turn, like Faber, to the work of John Ohala. But he deserves a post of his own.

02 April 2006

Meet the Authors of LA!

Two of the authors of LA have recently been accepted, along with one of our colleagues, to a conference for undergraduate Linguistics Majors at Harvard University. Please come support us as we give our first talks. More information on the conference can be found at:

Jedediah: "A Lateral Kick to the Shins, Explaining an Akkadian Sound Change", Saturday 10:30 AM.

Matt: "Derivational Reflexivity in Spoken Arabic", Saturday, 2:45 PM

Annie (their colleague): "Try and C", Sunday, 11:15 AM.

Hope to see you all there!

FOCUSing on WH Interpretability

Note: this post treats the analysis presented in Boskovic, Z. 1999. "Oh multiple feature-checking: Multiple wh-Fronting and multiple head-movement." in S. Epstein and N. Hornstein (eds.). Working Minimalism. 159-187. MIT Press, Cambridge, Mass.

In his 1999 paper on multiple wh-fronting in Slavic languages, Boskovic deals with the issue of what forces wh-words beyond the first to front after the [+wh] feature on the COMP has been checked by the first wh-word. As a solution to this problem, Boskovic proposes that each of the wh-words beyond the first moves into the left periphery because of [FOCUS] features. This post is a collection of thoughts on this idea.

First, let us summarize Boskovic's analysis for a sentence such as (1), from Bulgarian:

(1) Koj kogo obica
who whom loves
"Who loves whom?"

This analysis allows Boskovic to capture the overt movement of the wh-words in multiple wh-fronting (MWF) languages, as well as the superiority effects that subsets of these languages exhibit (this data is not repeated here. The reader is referred to the original article). The interesting claim that Boskovic makes in this analysis is that all wh-elements are focused elements for these languages. This means that the [FOCUS] feature in the tree for (1) will be strong on each of the wh-phrases.

What I would like to comment on is the nature of interpretability of the features Boskovic proposes. He does not comment in his paper on which of the features are interpretable and which are not. This will become crucial, however, in order to predict the correct word order. In order for the FOC head to be able to attract multiple wh-phrases (in the case of three or four-wh-word sentences) the [FOCUS] feature on the FOC head must be interpretable (note that it would also be possible to derive a phase-based analysis of feature deletion in order to account for this, but this solution is not pursued here). If the [FOCUS] feature is not interpretable, it would necessary delete under Minimalist accounts of feature checking, and not be able to attract wh-phrases beyond the first.

In a way, this is also an implicit argument for the existence of a FOC head in the left periphery. If there were not this head, either COMP or some other functional projection would be forced to bear an interpretable feature representing, semantically, focus. Since this is probably something we would like to avoid (not to mention issues of having functional heads represent more than one semantic role...), we can posit the FOC head to bear this feature and attract each of the wh-phrases beyond the first in MWF sentences.

Finally, there is one question left to be resolved, and that is how to make this analysis work vis-a-vis Rizzi's account of the FOC head. In his analysis, this head is unique - it is not iterated multiple times in the periphery as TOP is. This raises the obvious question in light of what we have discussed here: if there are more than two wh-phrases in the MWF language sentence, how can we account for them all being moved to the left periphery? The obvious issue would lie in whether we want to allow multiple wh-words to occupy [SPEC, FOC] or to allow FOC to be iterated beyond the first (contra Rizzi). I believe this is a matter that needs further investigating.

01 April 2006

A Lateral Kick to the Shins: Part 3

Sorry for the delay. In the interim, my paper was accepted by the conference, so I really need to get going on this. The remaining posts will describe the change and previous explanations for it, then give a detailed account of my proposal including the phonetic evidence. This first post will give evidence for the change itself.

The change of š to l in Akkadian can be seen between the Old Babylonian and Middle Babylonian periods. It is particularly clear in certain verbal forms (especially the so-called Š-Stem, which forms causatives) and in certain verbs like šaparum "to send" and šakanum "to place." Some examples of these forms from the Amarna letters (from Franz Böhl, Die Sprache der Amarnabriefe mit besonderer berücksichtigung der Kanaanismen): iltapranni "he sent to me" (from ištapranni), altapra "I sent" (from aštapra), ulzizušu "I set it up" (from ušzizu-šu, Š-Stem of izuzzum "to stand" with the clitic pronoun -šu), altakanma "I settled (them)" (from aštakan with the enclitic conjunction -ma).

Some other examples (from Richard Steiner, The Case for Fricative-Laterals in Proto-Semitic): ildu "foundation" (from išdum), ilti "with" (from išti). These indicate that the sound change is general rather than morphophonemic, as might be expected from the verbal forms alone.

I should note that this change is never fully complete in the written language; forms with š continue into very late texts. However, I do maintain that the change was complete in the spoken language by the Middle Babylonian/Middle Assyrian period and was only reflected in writing sporadically because of the extremely conservative character of the cuneiform writing system and scribal traditions. One piece of evidence for this contention is the fact that different types of texts show different patterns of attestation of the change: royal inscriptions, which tend to be more formal and use more archaic language generally, show it much less than letters, which tend to be less formal and closer to the spoken language. Jussi Aro (Studien zur mittelbabylonischen Grammatik) reports that a sample of Middle Babylonian letters showed lt 52 times and št 18 times (8 letters had both forms). This suggests that the change had become widespread in the spoken language at that point (at least in certain dialects; some peripheral areas don't show the change at all).

So that's the basic structure of the change as attested in the texts. Although not all possible environments were shown in the examples above, it occurs before /t/, /d/, /ṭ/, /s/ and /z/. These are all coronal segments (their exact phonetic realization as dental or alveolar is of course impossible to determine), so this definitely looks like a conditioned sound change. But more on that later.