Hebrew Phrase Structure Rules (PSRs)

[Jase]

As I work through Carnie’s Syntax, I am trying to define some things regarding common structures in Hebrew in comparison with what I’m learning about English structures. Whereas in English, we cannot have two determiners in one phrase (*that the man), in Hebrew it is indeed possible (הָאִישׁ הַהוּא ha-ish ha-hu “the-man the-that”). This means that the normal restrictions on determiners in English are not relevant for Hebrew, though it is true that we cannot have both ha- “the” and the superlative particle הֲכִי hachi “the.most”.

Because of this distinction, I have rendered the demonstrative as “Dem” and its phrase as “DemP.” See the attached image for the proposed tree for the Hebrew NP “this pretty girl” (הַיַּלְדָּה הַיָּפָה הַזֹּאת ha-yalda ha-yafa ha-zot). In the image you can see that I’ve proposed the following PSRs for a basic non-construct Hebrew NP:

DemP → NP Dem'
NP → N' (AdjP)
N' → (D) N
AdjP → (D) Adj'
Adj' → Adj
Dem' → (D) Dem

Because articles (and the superlative particle that I mentioned before) must appear directly before the NP head to which they are attached, they do not modify the entire NP as in English ([D the [N' silly boy in the costume]]). This led me to re-inserting the D into the NP (despite its unseemliness for X-bar beauty) and to placing the demonstrative outside of the entire NP.

Could you look over what I have done in the attached .jpg and let me know if there is anything that I have overlooked this far? Do you know of any place online where I can come across work that people have already done in identifying Hebrew PSRs and specifics about the language?

Thanks!

[Jase]

While the NP has the rule N' → (D) N and the DemP has the ruler Dem' → (D) Dem, I think it wise to remove the D one step out in the AdjP (to give AdjP → (D) Adj'). This is because of the comparative particle (יוֹתֵר yoter) and other specifiers, which may appear between the D and the Adj, as in:

הַיַּלְדָּה הַיּוֹתֵר יָפָה
? ha-yalda ha-yoter yafa
the-girl the-more pretty
“the prettier girl”

I’m not completely sure that this is “grammatical,” but I hear this form used all the time in colloquial Hebrew. I really don’t know if there is a more proper form to express definite comparative adjectives. It’s a question that I will need to ask a native speaker, though I have my inclinations as a fluent user of the language. This is similar to ha-lo nachon, as in ha-tshuva ha-lo nechona (“the wrong answer”). This is like attaching a specifier to the AdjP that falls between the D and the Adj itself.

I’m so excited to be learning this stuff. You have no idea!

[Daniel]

There is no structural reason to have two determiners, so I believe this can be treated as definiteness agreement. In other words, the underlying structure looks roughly like English*, then the additional forms are via agreement (or some kind of copying) rather than phrase structure.
Arguably there is still some more complicated morphological structure, but that would (probably) be within the heads of the phrases, "below" the syntax.

(*It often does, doesn't it? That's what happens when a theory is developed for English then extended for use with other languages. I'm skeptical of this and not necessarily endorsing the approach, but if you follow the basic assumptions in Carnie, you will reach a conclusion along these lines. Whether, starting from Hebrew, you'd make those assumptions is another question.)

I don't know of specific work that deals with this, but note that the same phenomenon is found (almost identically) in Arabic, so that's another place to look for research if you're interested.

I'd recommend this article (Kramer 2010) on Amharic, a related Semitic language:
http://dx.doi.org/10.1111/j.1467-9612.2010.00139.x

In Amharic, there are some complicated (but interesting) distributional facts about the definite article. It always appears in second position, after whatever the first constituent is within the NP. Within (fairly complicated) Minimalism, Kramer argues that this is due to linear displacement: it's actually at the beginning, but phonologically must attach to the end of a constituent. So it just jumps over to the end of the first thing it can. That may seem a bit strange, but the explanation works out well.
Beyond this, additional definite markers sometimes appear in the NP. Unlike in Arabic and Hebrew, however, these are optional in Amharic. So you can tell which is the "real" definite marker, and there is strong evidence for some sort of agreement effect to create the additional copies.

The same analysis would explain Arabic and Hebrew, if we ignore the 1) linear order [the article attaches to the beginning in Arabic and Hebrew] and 2) the optionality of the agreement-- it's optional in Amharic but not in Arabic or Hebrew.


So you may or may not want to get into this level of analysis with other languages yet. Carnie is walking you through the theory slowly and you're not yet ready (from the book, that is) to analyze certain phenomena. Beyond this, Generative syntactic theory does not actually explain linear order in some cases. That may be due to (phonological) displacement or other effects, post syntactic. Regardless, if it is to be explained, it'll be closer to the chapter on movement and so forth.


I hope that helps with an overview. Your thoughts seem about right for the tree and so forth, assuming that it made sense to represent it as part of the phrase structure.

I’m not completely sure that this is “grammatical,” but I hear this form used all the time in colloquial Hebrew. I really don’t know if there is a more proper form to express definite comparative adjectives.

Take another look at the early chapters in Carnie
That's exactly what "grammatical" means-- that speakers use it, because it's part of their grammars. Whether something is or is not "proper" is irrelevant (that is, an unrelated question).

[Jase]

I certainly understand the significance of descriptive grammar, but I’m not sure if what I have represented is an utterance of second-language speakers or of native speakers. I know that I hear it a lot, but I have never really paid attention to who is saying it. I want to run the phrases by native speakers to get some feedback.

I know that I still have a lot to learn in the way of movement and all. I’m just looking at the linear structure right now, which is basically where I’m at in the book. As I improve in my approach to Syntax generally, I want to try to make the same generalities in comparison of Hebrew that I’m learning about English. So, when I get to movement and such, I’ll pull together examples of more complex Hebrew structures and look at how things move there. I hope that makes sense.

[Daniel]

I certainly understand the significance of descriptive grammar, but I’m not sure if what I have represented is an utterance of second-language speakers or of native speakers. I know that I hear it a lot, but I have never really paid attention to who is saying it. I want to run the phrases by native speakers to get some feedback.

Oh, ok. I misunderstood. There's a fuzzy area then where something may be produced only by non-native speakers. It's arguably not "language" then. So what you said makes sense.

I know that I still have a lot to learn in the way of movement and all. I’m just looking at the linear structure right now, which is basically where I’m at in the book. As I improve in my approach to Syntax generally, I want to try to make the same generalities in comparison of Hebrew that I’m learning about English. So, when I get to movement and such, I’ll pull together examples of more complex Hebrew structures and look at how things move there. I hope that makes sense.

Yes, makes sense. And don't take anything I said to be discouraging. One thing to be aware of is just that you won't be able to explain every linear order without making the theory much more complex and/or ignoring some effects as non-syntactic, just the result of linearization (in spell-out / PF / phonology).

[Daniel]

Following up on this, I asked someone (a professor who researches Arabic syntax specifically), and there are two approaches to the articles in Arabic:

1. They're definiteness agreement (see above). This is the simpler answer, and the one that makes more sense to me.

2. There really are no "adjectives"-- they're all nouns, marked as definite. The structures that look like Adj-N are actually some sort of reduced relative clause (something along those lines).

Another possibility is that this is a case of multiple exponence or periphrasis, where one element shows up in two places. That's not handled well by tree structures, and one explanation for it might be copying or agreement anyway.

[Jase]

Following up on this, I asked someone (a professor who researches Arabic syntax specifically), and there are two approaches to the articles in Arabic:

1. They're definiteness agreement (see above). This is the simpler answer, and the one that makes more sense to me.

2. There really are no "adjectives"-- they're all nouns, marked as definite. The structures that look like Adj-N are actually some sort of reduced relative clause (something along those lines).

Another possibility is that this is a case of multiple exponence or periphrasis, where one element shows up in two places. That's not handled well by tree structures, and one explanation for it might be copying or agreement anyway.

It might be the case in Arabic that there are not adjectives, but in Hebrew there certainly are. You may be talking about construct phrases (which are only nouns). I specifically mentioned non-construct in this analysis.

In fact, I remember when I did some study into Arabic that they had only one class called al-ismu (اَلْإِسْمُ) that included both nouns and adjectives (as we would identify them). In Hebrew, we have shem etsem (שֵׁם עֶצֶם “noun”) and shem to'ar (שֵׁם תּוֹאַר “adjective”). They are actually different grammatical classes and have different rules.

While Arabic and Hebrew are similar in that they both use generally tri-literal roots, have similar concepts of verbal structures, etc., it is a mistake to analyze what happens in one language all the time by what happens in another. We can do comparative Semitics to look at the similarities between the languages, but there are things that happen in Hebrew that don’t happen in Arabic, and vice-versa.

[Daniel]

It might be the case in Arabic that there are not adjectives, but in Hebrew there certainly are. You may be talking about construct phrases (which are only nouns). I specifically mentioned non-construct in this analysis.

I'm not aware of any differences for Hebrew, and I'm not referring to construct phrases. Arguably we could see adjectival modification as a kind of construct state, but that would fall within what I said above. I'm not defending proposal #2 above, but I'm saying that it has been argued. You'd have to look into that literature specifically to find out how that argument works. The proposal is exactly that: there are no adjectives, because of some complicated workarounds that analyze them as nouns. This isn't talking about what it looks like on the surface, but how it is generated syntactically. Such an argument seems complicated and implausible to me, but it has been proposed.
In the event that it doesn't work out for Hebrew, then it's pretty clear that the agreement analysis is correct.

In short, no one proposes an analysis with nouns and adjectives where each has a D element in the phrase.

In fact, I remember when I did some study into Arabic that they had only one class called al-ismu (اَلْإِسْمُ) that included both nouns and adjectives (as we would identify them). In Hebrew, we have shem etsem (שֵׁם עֶצֶם “noun”) and shem to'ar (שֵׁם תּוֹאַר “adjective”). They are actually different grammatical classes and have different rules.

Irrelevant. Those are traditional analyses, and certainly in Arabic there is a word for adjectives, even though they are noun-like in a way.
The traditional grammatical analyses for Arabic (and I assume Hebrew) are very interesting and in my opinion should have a bigger impact on syntactic theory (specifically I've done some research regarding 'verbless' clauses in Arabic), but as for this argument, the traditional labels don't apply (for the same reasons that traditional analyses of Latin or Greek don't apply).

While Arabic and Hebrew are similar in that they both use generally tri-literal roots, have similar concepts of verbal structures, etc., it is a mistake to analyze what happens in one language all the time by what happens in another. We can do comparative Semitics to look at the similarities between the languages, but there are things that happen in Hebrew that don’t happen in Arabic, and vice-versa.

It's a mistake to assume there cannot be differences. It's not a mistake to apply and test an analysis for one language on another one that seems to work similarly. The details may vary, but often at least parts of the analysis will carry over.

In short: start by assuming an agreement analysis for definite markers in Hebrew. Test that, and see what needs to be adjusted.



Anyway, a quick search revealed these (among other research):
http://linguistics.huji.ac.il/IATL/27/Abstracts/McNabb.pdf
This is basically the same as the argument in Kramer (2010), and it has several references you could look at for more information.

https://faculty.biu.ac.il/~danong1/papers/Danon1996-MA_thesis.pdf
This is a little more complicated, and I've just skimmed it. It claims definiteness is a "phi feature" (basically a kind of agreement), while some other complicated things are going on with the determiners themselves. Again, plenty of references in there.

[Jase]

In short, no one proposes an analysis with nouns and adjectives where each has a D element in the phrase.

Have you looked at the attributive position of adjectives in ancient Greek? There the adjective can share an article with the NP head or be held by its own article (as in Hebrew). I will provide examples in a bit. I don't have my regular keyboard right now and am typing on a tablet.

[Jase]

I’ve just jumped into the living room where I have a keyboard, and I wanted to post a couple of examples to compare what options there are in Greek.

The same expression in Greek (for example, the good man) can be expressed in three ways:

(1) ὁ ἀγαθὸς ἄνθρωπος. As in English, the adjective (ἀγαθός) follows the D and precedes the N.
(2) ὁ ἄνθρωπος ὁ ἀγαθός. As in Hebrew (האיש הטוב), the adjective follows the noun and both are tagged with a D.
(3) ὁ ἀγαθός. The adjective with a masculine definite article can represent man without it being written, thus it becomes a substantive on its own.

If, as you wrote, “no one proposes an analysis with nouns and adjectives where each has a D element in the phrase,” what would happen to the second case above? Everyone is in agreement that the repetition of the article in Greek NPs does not change the meaning of the phrase, whether we find written ὁ ἀγαθὸς ἄνθρωπος or ὁ ἄνθρωπος ὁ ἀγαθός.

[Jase]

Anyway, a quick search revealed these (among other research):
http://linguistics.huji.ac.il/IATL/27/Abstracts/McNabb.pdf
This is basically the same as the argument in Kramer (2010), and it has several references you could look at for more information.

https://faculty.biu.ac.il/~danong1/papers/Danon1996-MA_thesis.pdf
This is a little more complicated, and I've just skimmed it. It claims definiteness is a "phi feature" (basically a kind of agreement), while some other complicated things are going on with the determiners themselves. Again, plenty of references in there.

I’d already had the last of these PDFs downloaded to my system, but I hadn’t (and haven’t) yet read it. I’ll make sure to devote some time to reading through. Perhaps some of this needs to wait until I’ve worked through the latter parts of Carnie’s book. I’m now just in chapter 7, trying to take my time in this chapter and work through the exercises to make sure that it’s cemented and correct before moving on. Be a bit patient with me. I’m still new to syntax.

[Daniel]

(1) ὁ ἀγαθὸς ἄνθρωπος. As in English, the adjective (ἀγαθός) follows the D and precedes the N.
(2) ὁ ἄνθρωπος ὁ ἀγαθός. As in Hebrew (האיש הטוב), the adjective follows the noun and both are tagged with a D.
(3) ὁ ἀγαθός. The adjective with a masculine definite article can represent man without it being written, thus it becomes a substantive on its own.

Interesting. This looks like a case where the adjective can function as a noun (as in 3), so in 2 it probably is actually a noun marked as definite (or, possibly just agreement as discussed above). I don't know much about Greek, so I can't comment further. Is 2 an NP (=DP)? That is, could that be, for example, the subject of a sentence? That's not an equative clause, right?

If, as you wrote, “no one proposes an analysis with nouns and adjectives where each has a D element in the phrase,” what would happen to the second case above?

No, I don't think so. Two options:
1. It's an adjective within the NP that is marked via agreement.
2. (What I guessed above) it's actually acting as a noun (as in 3), so it's N-N both marked with an article.
Either way, I don't see this as marking an adjective with an article.

Everyone is in agreement that the repetition of the article in Greek NPs does not change the meaning of the phrase, whether we find written ὁ ἀγαθὸς ἄνθρωπος or ὁ ἄνθρωπος ὁ ἀγαθός.

What "meaning"? The translation test is a bad one because it often misses intricate details and entailments. Even if it is equivalent in meaning, it may not be equivalent in syntax. Consider in English:
The blue book.
The book that is blue.
Those are equivalent in meaning, but have different structures. The same may be true in Greek. (One idea that comes to mind is that the second one is some sort of relative clause.)

I’d already had the last of these PDFs downloaded to my system, but I hadn’t (and haven’t) yet read it. I’ll make sure to devote some time to reading through. Perhaps some of this needs to wait until I’ve worked through the latter parts of Carnie’s book. I’m now just in chapter 7, trying to take my time in this chapter and work through the exercises to make sure that it’s cemented and correct before moving on.

Reasonable. Take your time.
And I'd recommend focusing only on English while trying to understand Carnie. It doesn't always apply well to other languages. Then it's a great exercise to try to extend it. Basically, for now, think about how it would apply and make some notes about problematic languages/constructions to consider later. These are good thoughts!

Be a bit patient with me. I’m still new to syntax.

Of course. You're asking good questions. I'm only trying to provide what is a (relatively) standard analysis to consider-- you're welcome to disagree with it, but start by understanding what that standard analysis is to see whether there's a better one to be proposed.

[Jase]

If I can be forgiven for not taking your advice immediately, the attached image is what I was playing with today, and it illustrates what I’m thinking. With this, I’ll move on to focus only on English in the meantime (until I cover movement and transformation rules).

In Greek, though, the predicate position is used to create verbless clauses using a NP and an AdjP. For example:

καλὸς ὁ καρπός and ὁ καρπὸς καλός are equivalent and both mean “the fruit [is] good.”

In the predicate position, the adjective cannot be bound by an article to the noun phrase. The word order is fluid, with the adjective either preceding or following the noun, which is accompanied by the article. In the case of the attributive position (which I placed above), the adjective must be preceded by an article, whether it is between the article and the noun (as in English) or following the noun with its own article.

I’ll try to stay focused after this on learning the rules as they come out of English usage, understanding the process of determining the PSRs and how they go through movement, etc., and mastering what we have to deal with in my own native tongue before going back and getting into the deeper matters of Hebrew.

Thanks for the advice and direction.

[Daniel]

Your tree suggests that the D head is definite and then causes the adjective and noun to become definite. As a simpler answer, why not consider this to be agreement? Of course agreement itself might be explained in various ways, but I don't think you're proposing something radically new.
One problem is that there are very rarely cases where a single node is doubled. There is some work in copy theory (within Minimalism these days) where there are two copies and (almost always) one is deleted, but not where something moves down a tree or to two new spots from one. In fact, the kind of movement you have there is unexpected.
More than that, the structure is not proper X-bar format because there should be no D node without a DP. Looking at this at the level of morphology (agreement) would work out almost identically, but within the word level.


Don't give up on analyzing languages other than English! That's the last thing I'd recommend. But waiting a bit before applying the theory from Carnie's book is probably a good idea. (Or, again, apply it where it makes sense and make some notes about potential difficulties.)

By the way, this could be helpful:
http://linguistforum.com/morphosyntax/syntactic-trees-in-your-posts!-(instructions)/
It's up to you of course. Attachments are fine, just probably more work for you.

[Jase]

No, I know how to use the trees on the forum, but (1) they are kinda ugly because of the lines that it forms between the nodes and the labels and (2) they portray X' as X\' for whatever reason. I know it’s less work, but I just like how I’m able to make the trees attractive. This way I can also use arrows and other decorative pieces (like setting up the glosses). I can also adjust them since I create them with PowerPoint.