For the most part, Shankara tows the party line of the Vedanta: the world has brahman as its cause. But he also says that the world arises out of what he calls "unmanifest name and form." Structurally, this parallels the Samkhya's idea of prakriti forming the basis of the world. But Shankara wants to have it both ways here. He wants to support the brahmanic orthodoxy, but he also wants to make use of several ideas that have their basis in the unorthodox schools, the Samkhya and the Mahayana schools in particular. After Shankara, the later Advaitins will assert that maya or avidya is the cause of the world, or that brahman in conjuntion with avidya is the cause of the world. Shankara does not speak this way. For him avidya is an "epistemic" principle only; we might say that it is only "metaphysical" to the degree that it is involved in our conceptual construction of the world, that is, to the degree that it is the root of the transcendental illusion.
As we noted, Shankara is able to have it both ways by ranking "realism" and "illusionism." Now there are some interpreters who have argued that Shankara is actually more of a "realist." As I noted before, I think that we are justified in saying so as long as by "realism" we mean something like the realism of the ancients. Those who argue that Shankara is more of a realist in the general sense do so on the following three grounds: 1. on the basis of the interpretation of "ananyatva" that takes it as saying, "the world is real insofar as it exists in brahman"; 2. on the basis that the world does not "dissolve" with realization; 3. and on the basis that the world "becomes" brahman with realization.
Now, the first two of these arguments can be challenged. The first can be shown to be a propaedeutic view. And the second does not necessarily imply anything, other than the fact that the Advaitins were not pralaya-vadins (who thought the world and the mind "dissolve" with realization). The third point, though, is interesting. It refers to a single tract in Shankara's works, the comments on Brhad Upanishad 2.4.12. There the Upanishad refers to dissolution of the world into the "Mahabhuta" or great reality. Shankara comments that when discrimination arises "the world becomes one without a second" and "merges" with the Mahabhuta. He concretizes this idea saying that this means that one's separate existence dissappears and one returns to the "womb" or own-source (yoni). (In a similar way, at the end of chapter 3 of the Gaudapada Karikas, we read that all dharmas are "always already" non-dual and inherently quiescent.) But the language of "merging" used in the above is metaphoric for Shankara, so I don't think this final point stands scrutiny either.
At the same time, it is necessary to point out that there is indeed a pronounced "realist" streak in Shankara's writings (at least in those writings that were written after the commentary on the GK). Time and again he rejects the "idealist" arguments of the Yogacharins, and he continually refers to brahman as a "vastu" or real thing. More importantly, following an unnamed master referred to at the beginning of the Brahma Sutra commentary, he emphasizes that for the practitioner, the world is real until realization occurs.
We have already touched briefly upon early conceptions of personal eschatology and proto-soteriology in Vedic religion, and we can now relate some of those conceptions to Indian conceptions of cosmology. On page 135 of his book, Muktananda mentions Indra-loka. Vedic mythology speaks of the world of Indra, or Indra-loka and it appears also in later mythology; in the Mahabharata, for example, Arjuna travels to Indra-loka. By Indra-loka is meant, more or less, Svarga, the Vedic "heaven". It is often depicted as sitting atop mount Meru, the "axis mundi" of the Vedic cosmology. It serves as the home of the gods -- much like mount Olympus in Greek mythology -- and as an abode for the blest in the afterlife. [...]
We can see in the classification of the three worlds -- the worlds of men, forefathers, and gods -- an instance of the Indian penchant for triads and tripartite divisions. An even older conception also makes use of a tripartite division. The oldest Vedic texts refer to three domains: an underworld populated by Ashuras, a middle world populated by men, and a celestial world populated by the gods or devas. Yet another conception of three worlds describes a celestial realm, an earthly realm, and an intermediate realm. Brhad Up 5.5.3-4 describes these three worlds in relation to the cosmic person. It says, "bhur is its head, bhuvar is its arms, and svar is its feet." These three terms refer to the three lokas of Brahmanic lore. "Bhu" refers to the earthly world or Bhu-loka. "Bhuvar" refers to the intermediate realm of the air, that is, to the "atmosphere" which is known as Bhuvar-loka. Bhuvar loka, which is between the earth and sky, is said to be a world populated by semi-divine (cf. daemonic) munis and siddhas. "Svar" refers to the celestial realm, Svar-loka, which is the same as Svarga, heaven, or Indra-loka. It corresponds to the sky and is populated by gods. At some point in history, various other lokas are added to these three. The next to be added is Mahar-loka, which is said be populated by Brghu and other saints who survive the periodic dissolution of the lower worlds. Taittiriya 1.5.1 says, "Besides these three, the seer Mahacamasya knew a fourth, called Mahar." Eventually, two more worlds are added to these four: Jana-loka, a term that refers not to a world populated by sexy Kiwi babes but to Brahma's sons; and Tapar-loka, in which deified Vairagyins dwell. Atop all of these is Satya-loka, otherwise known as Brahma-loka, the highest of the lokas, and for some, the domain of release. Eventually, the Indian tradition settles with these seven worlds. Muktananda speaks of a "Siddha-loka" but as we can see here, this could refer to any of the lokas above Bhu-loka.
The Basic Tripartite Structure of Vedanta: We can now return to what is perhaps the best known of the metaphysical triads in the Indian tradition, the division between the gross, subtle, and causal. The three states of consciousness are actually the original basis of this division. Descriptions of the three states occur in various places in the Brhadaranyaka Upanishad. Initially, a contrast was developed between the waking state and the state of deep dreamless sleep. This pairing functioned as a kind of metaphor for the polarity between life and death, with deep dreamless sleep acting as a kind of analog or foreshadowing of the "sleep" of death. Dreaming initially appears as a kind of mediating or "daemonic" third term, a kind of metaphor for the imaginal realm situated between the two, not quite one or the other. [...]
In the Chandogya, the "fourth" self functions as a kind of transcendental ground of the other selves. Although he vacillates somewhat on this point, Shankara generally agrees with Kant, and says that the transcendental self cannot be an object of experience since it is the transcendental condition of experience as such. In terms of turiya, this idea is referred to in Advaita via the dictum that turiya is not a state among other states but the "truth" of the rest. And yet in the Gaudapada Karikas, and elsewhere, this "fourth" begins to be treated as if it were some kind of "state" of consciousness. Karika 1.15 definitely refers to turiya as a "pada", not in the Mandukya Upanishad's sense of a metaphoric "part" of the self, but in the sense of a "state" or "stage". Also, in an explicit sense, turiya begins to designate liberation (moksha) itself. But if this is so, then we have a problem. To use Wilber's terminology, the other three states function as "structures" of consciousness. And yet moksha cannot be a "structure"; it can only be a "stage". Thus, if turiya is moksha, there is a qualitative difference between the first three padas and the fourth "pada". To make matters more complicated, something else enters the picture. Various traditions begin to speak of certain states of consciousness as "transcendent", that is, as states that go beyond conditioned reality, beyond the "world" of samsaric existence epitomized by the first three states. [...]
The Four-Fold Scheme of the Grammarians and Tantrikas: While the tantric traditions such as those of the Naths and Kashmiri Shaivas, accept the three (or four)-fold scheme of gross, subtle and causal states which is derivative of the Upanishads, there is another scheme derivative of the grammarians that is also operative in tantric works. This scheme is the three, or four, -fold division of the nature of mantra. The grammarians had posited that the "word" (vak), in its most inclusive sense, has four dimensions: para vak, the supreme, which is soundless; pashyanti vak, at which stage the word begins to manifest at an intuitive level; madhyama vak, or the middling stage, at which the word manifests at the mental level as thought; and vaikhari vak, which is physical speech. The tantric tradition assimilates this theory of mantra of the grammarians and integrates it with the theory of the three selves of Vedanta. In fact, the tantric traditions incorporate under this four-fold scheme every conceivable facet of doctrine and ritual of their system in a vast edifice of correspondences. The synthesis of the Vedantic and grammarian schemes within the tantric manifold figures most prominently in the works of Abhibnavagupta, the great Kashmiri Shaiva yogin, polymath and philosopher. This four-fold structure occurs elsewhere in tantric literature and can be found, for instance, in chapter three of Jnaneshvar's Amrta-anubhava, The Experience of Immortal Bliss, a work clearly written under the influence of the Naths.
In the tantric version of the theory of mantra, these four stages of the manifestation of the word are associated with the cakras in an interesting way that may seem counterintuitive. The "supreme" vak is associated with the root cakra, muladhara. The "causal" stage is associated with the navel cakra, manipura. The "subtle" stage is associated with heart cakra, anahata, and the "physical" stage is associated with the throat cakra.
In the tantric formulation the term "madhyama" is clearly associated with the mind, either manas or buddhi. This is the domain of the subtle. What, though, does "pashyanti" refer to? Here is my take: The verbal root "pash" means "to see". This particular ocular verb is sometimes associated with the "spiritual" seeing of the ancient rishis. The term "rishi" means, literally, "seer". It was thought that the ancient rishis "saw" the Veda (which means "knowledge") and then composed their hymns and mantras accordingly. This "seeing" is contrasted in the later tradition, which emphasizes the more derivative "hearing" (sravana). The ancient rishis, however, did not hear the Veda, they "saw" it; that is, they intuited its meaning directly. By "intuition" here we do not mean some fuzzy faculty of "feeling", but rather unmediated cognition, such as the apprehension at this moment that we awake. This, I think, is what is meant by the term "pashyanti": it refers to the intuitive mode of knowing in which subject and object, though distinguished, are in direct relation with each other. In Plotinian terms, it corresponds to immediate relation between noemata (Plato's "forms") and noesis (Plato's episteme) within the sphere of nous.
The Upaniṣads are a particularly delicate case; the problem, stated in simplified form, has been whether the Upaniṣads were pre- or post-Buddhist. Their subject-matter and method of presentation have much in common with Buddhistic writings; the Pāli style seems, indeed, to be a diluted imitation of the Upaniṣadic style. The secular approach of the Upaniṣads is characteristic also of Buddhism and Jainism, those religions of princes. If we work on the presupposition that in India progress is from the simple to the complex, from brevity to elaboration, the Upaniṣads must be regarded as earlier. This is my own view. Louis Renou, Religion of Ancient India (London: Athlone Press, 1953) p. 7.