The sentences in (1) contain the same words, but differ in word order. Nevertheless, the sentences have very similar, if not identical, meanings.

(1a) I am home today.
(1b) Today, I am home.

Although the sentences differ in surface form, I understand that syntacticians usually say that there is a deeper level of representation ('deep structure') at which both sentences have the same syntactic structure? For instance, would people invoke some type of (optional) movement and hold that ‘today’ was moved in either (1a) or (1b), so that pre-movement, they have the same structure?

What would this mean cross-linguistically? For instance, there’s no German sentence with the same surface structure as (1a) or (1b). Sentence (2c) and (2d) are grammatical in German, but their English counterparts are not.

(2a) *Ich bin daheim heute.

(2b) *Heute ich bin daheim.

(2c) Heute bin ich daheim.

(2d) Ich bin heute daheim.

Although no common surface structure exists, would people still want to say that there is a level of representation at which the German and English sentences have the same structure?

I realise this a rather broad question, but perhaps people could point me in the right direction for further research.

  • 3
    Some linguists would want to talk about deep and surface structures, but not that many. In any event, almost nobody believes that there is some "universal" deep structure that's the same from language to language. Languages are very different; even languages as closely related as English and German. And individual and cultural variations are just as important as language differences; even people who share a native language may parse their own language quite differently.
    – jlawler
    Mar 26, 2018 at 19:36
  • 3
    This is an interesting question. Some theories posit a “deep structure”. (1a) and (1b) have the same structure when it comes to dependencies, what they differ in is topic-focus articulation. Cross-linguistically, predicate-argument structures are universal but this moves the question beyond the syntax-semantics boundary.
    – Atamiri
    Mar 26, 2018 at 21:01
  • 1
    On the eng front, there are vector representations. Mar 27, 2018 at 19:35
  • 1
    @A. M. Bittlingmayer Could you elaborate on that?
    – MarkOxford
    Mar 27, 2018 at 20:30
  • 1
    @MarkOxford Yes, but hard to do concisely. There are vector representations of words (see word2vec). We can also build vector representations of sentences, by averaging the vectors of the word in the sentence or in other ways. Mar 28, 2018 at 12:39

1 Answer 1


Language is constrained by its serial channel, but the memory network it represents has more dimensions. Here is a plausible deep net for all your versions of "I am home today": (sorry I can't draw a nice picture)

det(one) @ Mark

___ TAM(have) @ det(one) @ home

___ TAM(be [at]) @ [the] (one home)

_____________ [at] @ [this] (one day)

The nesting (indents) can't be represented in speech unless we use case markers, strict positioning, or syntax clues. "Today" is temporal, which is a clue that it is adverbial, and because of that clue its position is lightly constrained. We are then permitted to use relative position for 'focus'.

  • Thanks, but could you explain the symbolism you use? Also, how does this account for the fact that ‘today’ and ‘heute’ are constrained differently in the two languages? If we postulate a common syntactic structure ‘beneath’ all four sentences, don’ we have to say: ‘Although these sentences start off with the same ‘deep’ structure, we end up with different surface structures because XYZ’ – and then filling in XYZ is the hard part?
    – MarkOxford
    Mar 28, 2018 at 9:11
  • '@'s represent non-local links, and the indents are to show that an item connects locally to the '@' above it. [] shows elided (or optional) items. Sentences=trees because words are sequential, but the deeper net connects to previous items (even loops within a clause). 'Heute' (hiu tagu = this day) has the same adverbial clues as 'today', but German has a stricter rule that the adverbial must remain near the verb. Elision and free(ish) order without case markers are signs of an advanced language, where the cost of complex internal rules is paid for by the reduction in external bandwidth.
    – amI
    Mar 29, 2018 at 17:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.